Datasets:

mimiklee

/

masterthesis-longt5-sum-8k

like 0

INTRODUCTION The Cameroon Anglophone Civil Society Consortium (CACSC) emerged on the background of the Anglophone Problem. This can be traced from the colonial era. Following the defeat and ousting of the Germans out of their Kamerun Protectorate in 1916 (after the First World War), Britain were victorious partitioned German Kamerun Protectorate into two unequal halves. The French had four-fifth (4/5) and the British took a fifth (1/5) of the territory. The British for administrative imperatives divided her own part into British Southern Cameroons and British Northern Cameroons and administered it as an appendage to Nigeria. In February 1916 two diplomats Oliphant Lancelot of Britain and George Picot of France met in London and provisionally partitioned the territory in what came to be referred to as Oliphant-Picot Line. The line drawn on the map of Kamerun the former German Protectorate ran from Lake Chad in the North to the Mungo River in the South created a boundary between British and French spheres. On 10 th July 1919, the Milner-Simon Agreement settled the 1916 partition. In 1922, the League of Nations (LON) recognized the partition and the international boundaries that separated the two territories. 1 In the course of ruling the territories as Mandate B territories of LON, the French introduced the Policy of Assimilation in French Cameroon and the British on their part introduced the British system of Indirect Rule. Two distinct cultures and attitudes then developed. The people of British Southern Cameroons adopted the Anglo-Saxon culture and French Cameroon was introduced to the Centralized Republican System of administration tied to "Napoleonic Code". This greatly alienated the two territories in the areas of language, social attitudes and lifestyle, system of administration, education and judiciary system. 2 When French Cameroon achieved independence on 1 January 1960 as La Republique du Cameroun, British Southern Cameroons political atmosphere was still very tensed and characterized by political ideologies and shades. Due to the differences in the political path to follow between the leaders and main political parties at the time such as J.N. Foncha leading Kamerun Peoples Democratic Party (KNDP) and E.M.L.Endeley of Cameroon Peoples National Convention (CPNC) and One Kamerun (OK) under Nde Ntumazah, the United Nations organized a plebiscite in British Southern Cameroons. The territory had two options of vote either to gain independence in association with the Independent Federal Repuplic of Nigeria or voting to reunify with the Independent La Republique du Cameroun. Following the results of the UN organized plebiscite of 11 February 1961 in British Southern Cameroons, those for reunification won. According to Ebune, there was an overwhelming vote to gain independence by reunifying with their "brothers" of La Republique du Cameroun with 233,571 for against 97,741 votes for union with Nigeria. 3 Foncha's KNDP won 4 and by this victory, British Southern Cameroons reunified under an already independent La Republique du Cameroun. 5 Ngoh posits that the Foumban Constitutional Conference of July 1961 sealed the union of these two states with federation as the form of government which as best was to preserve the cultural values of Publisher: The USA Journals each state as well as their institutions. 6 On October 1, 1961, the Federal Republic came into existence with British Southern Cameroons regarded and called State of West Cameroon and La Republique du Cameroun called State of East Cameroon. However, postindependence developments and adjustments left the British Southern Cameroons with a series of politicoeconomic and social grievances as the constitutional agreements that were concluded at Foumban were gradually being scrapped off by the Francophone majority regimes and there was need for the Anglophone indigenous people to come together and act in civic space to address the situation. By 1961 in October, West Cameroonians had already been provoked with the appointment of J.C Ngoh as a Federal inspector answerable to the Yaoundé government of Amadou Ahidjo and many agitated. 7 Benard Folon in 1964 in his article titled "will we make or mar" enlightened the Anglophone community on the excessive use o of power by the Yaounde regime and called for liberty, democracy and equality and the rule of law. 8 A.N. Jua the Prime Minister of West Cameroon openly showed his opposition to the imposed Federal Inspector to West Cameroon and was out to preserve the West Cameroon identity. Albert Mukong another Anglophone Cameroonian through writing of petitions and other anti-government literature expressed his dissatisfaction to what the Anglophones went through. He organized meetings in order to draw the attention of the Anglophones on what was going on. In 1984 Mukong founded the Cameroon Anglophone Movement (CAM) which called for separation. According to Ngoh in 1972 Anglophone Cameroonians were challenged in their marginalization when the Federal structure was dismantled, 9 and prominent West Cameroonians protested and criticized the move publicly in the likes of Albert Mukong, Gorji Dinka and Bernard Fonlon. 10 Ngoh in another argument pointed out that in 1984, through the modification of the constitution's articles 1, 5, 7, 8, 26 and 34, the name of the State was changed from the United Republic of Cameroon to the Republic of Cameroon. 11 He went further to say that Law No. 84-001 of 4 February 1985 abolished the name "United Republic of Cameroon" to replace it with "Republic of Cameroon". 12 According to the Voice Newspaper quoting some Anglophone hardliners "It was a country that already existed since January 1960". 13 According to Mutanga, "The move by Biya in 1984 was a clear signal to Anglophone Cameroonians that they have been dissolved in an already existing nation and they ought not to exist". 14 To some respondents many English speaking Cameroonians stomached so much and in 1990s, the wind of change that blew across the continent opened more opportunities for the Government of Cameroon to right the wrongs that had vexed the Anglophones over years but it was still dashed into the sea with outright killings, arrest and imprisonment of Anglophones during and after the creation of the Social Democratic Front Party in 1990 15 a party with a pro-federalist position and a national following. 16 The regime crushed outspoken opposition 17 in Anglophone Cameroon and in 1992 when elections were rigged (presidential) according to some respondents, Anglophones showed their togetherness in opposing the government. Achille posits that in 1993 the government failed to take proposals of the Southern Cameroons' delegates in the Tripartite Conference on the return of a two State Federation. He went further to say that The All Anglophone Conference (AAC) of 1994 led to the formation of the Southern Cameroons National Council (SCNC) with a mandate to force the regime to move to a two State Federation but the Anglophone efforts were watered down and to him, the "principle of equality was undone by law". 18 This discontent sipped through the vent of the demands of the Common Law Lawyers of Anglophone extraction and the Anglophone teachers strike in 2015 and 2016 to degenerate into a point where it could not be handled through simple dialogue. --- CONTEXTUAL ISSUES Within the Cameroon Federation as early as 1961, the Anglophones started witnessing aspects of marginalization, assimilation as well as exploitation. They complained bitterly against disparagements which were in the form of appointments of Francophones into key positions in West Cameroon, imposition of the French language and administrative style on the Anglophones, closure of economic institutions and the violation of the Foumban Constitutional agreements of 1961 which put the two Federated States on equal stand. These resentments and other aching issues led to what became known as the Anglophone problem in Cameroon. Ngoh postulates that 1961 is the logical historical date for the start of the Anglophone problem. 19 Anglophone problem through the lens of this paper is the struggle by the ethnic Anglophones (former British Southern Cameroons) to uphold and preserve their cultural identity especially the Anglo-Saxon style of administration, educational system, the Common Law Judiciary System and the English language by the deliberate attempts at eroding or assimilating them by the Francophone regimes. The English speaking Cameroonian grievances were varied and owe their roots to the Federal Constitution which gave so much power to President Ahijo who ruled by decrees and purposefully violated the Constitutional agreements of 1961. Same administrative procedures were carried out by his successor President Paul Biya from 6 th November 1982 when he became President of the United Republic of Cameroon and the Anglophones continued to fight back through concerted efforts. For instance in 1983, the English speaking students in the University of Yaounde I went on strike due to the use of the French language in all aspects of the university and an attempt to eradicate the English educational system and the government responded by creating the Buea University Centre (for students in the 1985/1986 academic years to decentralize the University of Yaounde) 20 Cameroon General Certificate Examination Board ( CGCEB). 21 The fear for an eventual elimination of the unique Anglo-Saxon judicial and educational system by the Francophone system caused the lawyers and teachers strike of the Anglophone faction in 2016. Their desire to restitute the past in an old tradition of marginalization only met with a Government bed rock of intolerance, denial, 22 excessive use of intrigues and brutality, humiliation and above all, resistance to embrace truth. However, the need to effect change and reverse the socio-economic and political situation of the Anglophones through the lawyers and teachers only became a new venue and opening for all Anglophones to converge and break new grounds for the former West Cameroon. The strike call from lawyers who gave the Government a period from the date of deposit of their worries was also supported by Cameroon Education Forum (CEF) who decried among other things the violation, neglect and partial implementation of a twenty years old ordinance on education. 23 Together with Cameroon Teachers' Trade Union, Teachers Association of Cameroon, Teachers of Higher Education (the University of Bamenda was not a signatory to the strike decision but was involved only when the Prime Minister asked that, the representative of the teachers attend a meeting called by him) a strike was called. Later, other Associations in the Anglophone area 24 joined the strike action as called and to some personalities, the government was well informed of it but did not take action. The lawyers and teachers civil strike was a recurrent of protests that had for decades been staged by Anglophone Cameroonians in demand for a return to the Federal System, an Independent Anglophone State, equal opportunities and attention to be given to the social and economic needs of the Anglophone minority who were undervalued according to many Anglophones. --- CONCEPTUAL ORIENTATION An Anglophone in the Cameroon context according to Ngoh refer to a person who after the partition of German colony of Kamerun in 1916 found him/herself in the British Cameroons and they and their off springs grew up under the British in British Southern Cameroons and embraced the Anglo-Saxon culture. 25 Nfi on his part holds that, an Anglophone is one whose ancestry is Southern Cameroons and ethnic bases are in the former territory of British Southern Cameroons whether they speak the English Language or not. 26 To this paper, an Anglophone refers to people of the defined territory of British Southern Cameroons with a defined culture, history and ancestry. The concept of strike has been variedly scaled by researchers to suite their variant ideologies. This term had hardly been defined without violent. Borrowing from the Hegelian 27 school of thought, when reality 27 Hegelianism was a philosophy of G.W.F. Hegel which can be summed by the dictum that "the rational alone is real" which means that all reality is capable of being expressed in rational phylum. He aimed at reducing reality to a more synthetic unity within the embodied system of absolute idealism. and rationality are absent, a people become pruned to chaos and eventually revolutionary ideas are planted which leads to an uprising. Some respondents have it that prior to the strike action, the Anglophones had been resilient to the assimilation machinery put in place by the regime and the accumulated vexation overtime caused an outburst of a strike (protest) in 2016. It was a rise up (mass) against the constituted authority of Cameroon. --- The American The civil society belongs to that expanding and variegated composition of all institutions, voluntary organizations and corporate bodies within a state. Civil Society according to European Union is the involvement of "all forms of social action carried out by individuals or groups who are neither connected to, nor managed by the state". 28 Van Dyck on his part opines that Civil Society is "an organized social and cultural relationship that exist within a State build on indigenous values, tradition and principles to foster collaboration and the achievement of specific goals among citizens and other stakeholders". 29 He went further to say that the actors usually operate in civil space. To Evers, Civil Society is the coercive association which cuts across the borders of the state, family or community and which influences and strengthens policies through its operation and leadership. 30 In the context of this paper, Civil Society refers to the coming together of labour unions, social entrepreneurs, academic and research workers, judicial workers and indigenous people of Anglophone extraction in an attempt to position development dialogue as a means to provide opportunities, collective action and mobilize the people to articulate demands, voice and address the concerns and common plights of the society at local, national and international levels. Thus, Anglophones came together as a body in 2016 (though they have been meeting as a group) through an outburst of an agitation to change the status quo. The Cameroon Anglophone Civil Society Consortium (CACSC) was the mass mobilization of the people of Anglophone extraction of Cameroon composed of formal and informal, non-institutionalized and nonpolitical groups who seek to restore lost liberties and privileges 31 as a result of their differences witnessed in the union which was formed in February 1961. The Consortium on the other hand was an association of Anglophones formed during the suppression of the lawyers' strike of 6 th October 2016 to establish a common front against the government. It was a society of Anglophones business community, (Traders and Buyam Sellams), Teachers Associations, (Higher, Secondary and Basic) motorbike riders, drivers syndicates and other interested persons formed for the purpose of the fight against marginalization and other injustices. It was also widely asserted that the development of the CACSC and their stand point was only understood as an embodiment of grievances in the economic, social and political domain. The CACSC for a period of time dominated political discourse in Cameroon and became a liaison between the people of English speaking Cameroon and the State of Cameroon. Evidences show that the creation of the Cameroon Anglophone Civil Society Consortium (CACSC) on 6 th December 2016 was the outcome of an --- ORIGIN OF THE CONSORTIUM The Consortium draws its origin from the different stages, modus operandi and diverse weapons and methods used by the Anglophones to resist the extermination and assimilation policies of the Central Government towards the Anglophone cultural identity. According to Neba, the Teachers Association of Cameroon (TAC) led by Azong Wara in 1993 launched a serious campaign to have an independent examination board for Southern Cameroons' examination and this move was supported by the Confederation of Anglophone Parents-Teachers Association (CAPTAC) and a cross section of Anglophones (a sign of unity and community spirit expressed by the Anglophones). 32 He went further to say the people demonstrated publicly their wants and desire to have their Anglo-Saxon culture preserved just like the Consortium demands led by Agbor Balla and co in 2016. Teachers in the Anglophone regions in 1993 had also expressed their discontentment with the educational system in Anglophone Cameroon through petitions to international bodies. 33 Through groups like Cameroon Anglophone Movement (CAM), Free West Cameroon Movement34 and others, Anglophone solidarity had been expressed and the government had tried to play ignorance and created division among the people. Moreover, in 1995, lawyers had articulated their worries through a strike and individuals like Barrister F. Alowbwede, E. Elad and others were arrested because they demanded for a return to Federal and Anglo-Saxon Court systems. The Voice newspaper went further to expose that in the heat of the crisis, brothers on the other side made statements and calls for the "complete annihilation of Southern Cameroonians". 35 When the lawyers strike began and lawyers were arrested and beaten, in solidarity with them other Anglophones joined and this goaded the brutality which was used by government to dismantle such togetherness. In the same line, television journalists of Vision 4 in Yaoundé in the persons of Jacque Ze and Ernest Obama likened the Anglophones to "roaches and cockroaches who deserved to be wiped out". 36 Others called Anglophones "terrorists" and called the government to impose a state of emergency in the two regions. Banda Kani a politician expressed his worries as to why government had not used military techniques and weapons that could completely delete every single soul of Anglophone origin from surface of the earth. It was alleged that Bernard Okalla Bilai called protesters "dogs" and Tchiroma called Anglophones "secessionists". As exposed in the Voice Newspaper, the scar was also made deep on the Anglophones when names were tagged on them such as 'Anglo fools", "Biafrans", "les ennemis dan la maison" (in English enemies within) only went a long way to spark flames of an already existing vexation. 37 These utterances and hate speech only hardened the Anglophones who were undeniable marginalized, suppressed and humiliated but togetherness remained their strength in a spirit to disassociate with Francophone Cameroon. In fact some Anglophone leaders such as Chief Ayamba, F. Alobwede, James Sabum and others had assembled Anglophones to a first ever held session of the Constitutional Assembly of the Federal Republic of Southern Cameroons in Bamenda on May 1-2, 2000. 38 This signaled the coming together of Anglophones in another dimension to use such togetherness and meetings as weapons to propagate and disintegrate themselves from their French "brothers". --- The American Class antagonism as propounded by Karl Max 39 was at the heart of the formation of the Consortium. The system of inheritance (political appointments), bad governance, 40 economic exploitation with Southern Cameroon territory contributing 60% of the Gross Domestic Product(GDP) but were least developed all projected the subjugation of the Anglophones and caused them to come together; form a common force to clearly show their social economic, cultural and political dissatisfaction and interest to free themselves. Neba argues that the Anglophone discontent was a result of systematic neglect and alleged bad faith of the president who had the "master card" to solve the presented problems. 41 Based on the above, the "Anglophone minority" who also worked within the ambit of Tasang's statement; "The fight at hand is not a teacher's fight. If it were, it would talk about living and working conditions. This is an Anglophone struggle... is time for us all to rise up as one man." 42 lawyers strike but pregnant with other grievances which saw all endemic, brave and charismatic Anglophones coming together and was within this backdrop that defenders of the course representing the Anglophones such as Agbor Balla, 43 Tasang Wilfred, Fontem Neba and others put in their all to pilot the CACSC activities. In 2016, the move took another twist and Anglophones from all quarters came together to form what was called the Cameroon Anglophone Civil Society Consortium (CACSC). A group formed to defend the course of Anglophones striking lawyers and teachers and to discuss the way forward with the government of Cameroon. Many have argued that, government delay; threats and empty promises caused CACSC to emerge and eventually demanded separation. --- THE CONSORTIUM AND GOVERNMENT OF CAMEROON The lawyers and teachers strike in 2016 was supported by the entire Anglophone community and they got seriously involved after the ill-treatment of lawyers in Buea on the 6 th of October 2016 when some were beaten and robes seized. It was in solidarity with the call of lawyers and teachers that made the situation a "community problem" 44 The government could not pretend to be ignorant of the problem. They fully had it in front of them but they lacked the political will to honestly solve the problem for fear of losing grip over the territory and a people already impoverished, disgraced, cheated, robbed of their economic resources and systematically suppressed by the regime. 49 It was within this premise, of denial of facts, mistrust, tension and wearisomeness that the Cameroon Anglophones Civil Society Consortium (CACSC) was formed on 6 December 2016 as an umbrella organization which involved all the various teachers' trade unions and lawyers of common law. 50 Agbor 46 Professor Paul Ghogomu was the Director at the Prime Minister's office. Leader of inter-ministerial Ad-hoc Committee to examine and bring out solutions to the issues raised by the striking Anglophone Teachers' Trade Union. 47 Ngoh, Cameroon 1884-Present, 392. 48 Vally Tum, 53years Barrister of law based in Bamenda interviewed on 21 December, 2019. Balla Nkongho represented the lawyers and was president 51, Tasang Wilfred backed up the teachers and was program coordinator and the Secretary of the Consortium was Fontem A. Neba. A press release No. 3 of 13 December 2016 confirmed these persons as leaders of the Consortium and had Eyambe Elias as deputy secretary general and advisers were Harmony Bobga, Abangma James, Abia D and George Ngwane. 52 The "Consortium" now managed the strike actions of the lawyers and teachers and pressed hard on the Government as classes for the second term 2016/2017 academic year was to begin. 53 The Cameroon Anglophone Civil Society Consortium in one of their demands asked for the return to federalism. The Government argued that, it was unconstitutional to change the form of state and settled on the use of repressive and political tricks to handle the problems tabled by the Consortium. The Government solved some of the grievances; for example provided the English version of OHADA Uniform Acts, created a polytechnic, and released some Anglophones arrested in connection to the strike; but the contagious nature of the strike saw pro-Anglophones activists and movements who hijacked the stage in demand for secession and not even federalism as was demanded by the Cameroon Anglophone Civil Society Consortium (CACSC). on the other hand brought about armed conflict in the Anglophone regions. Provoked by delay and the wind of change that blew across the country, the strike action took another twist as the already angry lawyers and teachers were poised to bring the government on its knees. As aforementioned, it will be good to present some of the grievances of the Common Law Lawyers and those of the teachers that finally bought them together in a Consortium (6 December, 2016). There were eleven points from the teachers which moved up to twenty five and the government had almost agreed to handle twenty one of them before the Consortium was banned. the Anglophone regions declared an indefinite strike until their tabled grievances were looked into and the government did not take serious the strike call. Some Anglophone politicians like Paul Atanga Nji fuelled the flames of the strike when statements such as "there is no Anglophone problem were made." 58 The government through the Prime Minister Philemon Yang convened a meeting in Bamenda from the 25-26 November 2016 to discuss with the teachers and it was in the meeting that he rejected Atanga Nji's statement that, "there was no Anglophone problem". 59 To him, Atanga Nji's position was not the Government's stance. 60 Some political heavy weights even openly articulated that there was confusion within the government ranks because of the counter statements from these political gurus. On this count, the Government intensified its efforts through meetings with opinion leaders and visits to traditional rulers such as Fon Agwafor III of Mankon which did not curb the situation. The Governor of the North West Region Adolphe Lele L'Afrique together with top Government officials had talks with church authorities, teacher's trade union leaders and political leaders but the talks failed as evidenced with "ghost towns" and closure of schools and the Government sent more officials to help solve the problem. --- a-Common law lawyers The Consortium asked parents to continue to keep their children at home. The Government of Cameroon made efforts to ensure school resumption on 7 January 2017 and a team of ministers stormed Bamenda; the The Government felt the least threatened with the above statement and in an effort to allow schools resumption for the term in the North West and South West Regions, the Ad-hoc committee of Paul Ghogomu held another meeting on the 12-13 January 2017 to convince the leaders for schools to start while solutions would be made to the grievances but the meeting as well failed as demands were increased with federalism as an option which later also echoed by one of Anglophones reunification activist of the 1960s, Victor Mukete. 63 In another vein, the government sent politicians of Anglophone origin to discuss with the people and their arrogance instead brought mayhem between them and nothing was achieved. Prime Minister Yang and Jean Kwete were sent to Bamenda to calm the striking teachers and lawyers in December The 65 Ibid 66 Harmony Bobga was the President of North West Lawyers' Association, a member of the steering committee of the Cameroon Common Law Bar Association and equally one of the advisers of the Consortium. and teachers but instead had glued themselves to their comfort in Yaoundé without going to the scene of event to handle issues. The reason why the people saw Federation as the only solution and a means to disconnect themselves from a system that refuses change. 67 Another personality sent by the Government was Baba Danpollo (Ndawara) to talk with the striking lawyers but the talks failed because the Government position was that Federation was not going to be an option. Government feared that Federation was "equal to secession." The CACSA viewed federation as the only way out but the Yaoundé regime saw it as a "taboo" a word that should not be used or mentioned. 68 The Government of Cameroon made it clear to the Consortium that the idea of a Federation was irrelevant and not debatable. That the "State of Cameroon was one and indivisible." 69 The Consortium on the other hand blamed the Government and said "the Ad-hoc committee created has not achieved anything" and the "Government does not want to move the process forward" and argued that the Consortium was ready to dialogue and solve the problem but Government seems to be engaged in other priorities. 70 As observed by many Anglophones, the State of Cameroon used intimidation as a weapon instead of dialogue but the CACSC stood on a two State Federation which was popular 71 and the State through the governments' spokesman Isaa Tchiroma used the media to fan flames of hatred by insisting "that the form of the State cannot be changed" thus prepared the stage for Government use of force in these parts of the country (North West and South West regions). 72 The Government of Cameroon used force and militarized the two English speaking regions and the CACSC called on the Government to withdraw the thousands of troops especially in Bamenda, Buea and Kumba. According to a press release from the CACSC at some moments, during discussions, the Minister of Higher Education ordered that the military be locked in the room of the ongoing discussions just to intimidate the members of the Consortium. 73 It was alleged that the Government through the chairperson of the Adhoc Commission instead struggled to manipulate the process when he had a meeting with the teachers and this exacerbated tension with the call for intensive "ghost towns." In December 2016, Paul Ghogomu said "dialogue could only be possible if the teacher's first call off their strike action". To the CACSC it was provocation. According to an executive member of CATTU, "the money I saw on the table during one of the meetings was what I have never seen in my entire life as a teacher". 74 To her, this diluted the extent to which the talks could be genuine because the money was to intimidate Consortium members. According to some respondents, the Government resorted on the use of force, intimidation and the non respect of engagements that were subtle and philosophical. Some went on to say that the discussions on Friday January 13, 2017 between the Consortium members and the Government were ruined (that night) due to social media gossips. Ngoh in his book Cameroon 1884-Present says that the population of Bamenda was misled through rumours on the social media that the CACSC members were held hostage and forced to sign the awaited resolutions. 75 According to Tata, --- We were in the conference room and almost concluded to call off the strike though Tasang did not see himself calling it off when some personalities dashed into hall from rumours on social media that we were held hostage to sign the document. The rumour was a false alarm masterminded by some individuals to set in confusion which they did. 76 Tata went further to say that when the crowd outside the Governor's office heard this armed with bibles, started singing and praying. 77 Independence" 78 mentioned singing and prayers in the days of Angola civil war and this was the same scene outside the hall of discussion. Confusion erupted, police ordered for the block of roads to Upstation because of the large crowd and the population misinterpreted it for an attack. According to the Post, calls and text messages went out for the block of major entrances into Bamenda town and City Chemist Round About. 79 The Voice News paper reported that at mid night, John Fru Ndi of the Social Democratic Front (SDF) Party and Ben Muna of Alliance for Democracy bashed into the conference room of the Governor's office 80 only to realize it was false information. But the streets were already full and the reaction "seemingly caused police to open fire" 81 and tension mounted beyond control either by the Consortium or the Government. The Post equally wrote that acts of violence that night caused arrest and shooting and four youths were seriously wounded around Ntarikon area. 82 According to Stephen T, "...I got a forwarded message on my phone that my brother and friends of the Consortium were under pressure from the Government, I immediately started running towards Up-Station Bamenda just to meet another large crowd at Finance Junction. The crowd was also moving to Up-Station with chants like 'we no gree oh, we no go gree...'and by the time I arrived the Governor's entrance there was no space to pass due to the crowd but my brother later told me it was not easy with 78 Brinkman Inge, "Dreams and Agency during Angola's war of independence" in African Dynamics, Struggle Beyond Structure Social and Historical Trajectories of Agency in Africa (ed) Mirjam de Bruijn, Rijk Van Dijk, Jan-Bart Gewald(Leiden: Brill, 2007), 73. --- 79 The Post, No. 01789, Monday, January 16, 2017, 3. 80 The Voice, Issue 180, Wednesday, October 9, 2019, 2. them in the meeting but they refused to call off the strike...." 83 This scenario resulted to a ruined deliberation and the opportunity of handling the problem was eminently frustrated and produced other results. The legality of the Consortium to function was questioned and despite the confusion; they operated until it was banned on Tuesday January 17, 2017 together with the Southern Cameroon National Conference (SCNC). 84 To many respondents, the leaders of both sides did not go into the negotiations with an open mind and given the fact that these leaders had the interest of their people to protect, did not consider the outcome. Some argued that if the Consortium and the Government tactfully looked at the angle from which they approached their problems, the argument of force and the force of argument could not have had a place in their discourse. The lack of political will to solve the problems and accept suggestions spoiled the discussions and thus a glimmer to the armed conflict. --- THE CONSORTIUM AND ITS RAMIFICATION During the period of deliberations, in an effort to show proof of some seriousness and an expression of a will to do more if the Consortium respected their own end of the bargain, the Government made strenuous efforts and carried out some of the demands of the Consortium; but to many Anglophones who thought the Consortium was the only forum that could handle the Anglophone grievances soon saw their hopes dashed when Tuesday January 17, 2017 turned into a The Government addressed some of the grievances of the teachers and lawyers such as the release of some Anglophone leaders arrested, creation of a Polytechnic for the English subsystem of education, admission of students into the University of Buea and Bamenda as set by applicable regulations, provided the English version of the OHADA Uniform Acts and the Cameroonian Penal Code, established a Common Law Bench at the Supreme Court of Cameroon; appointed English speaking Magistrates to the post of President of the Judicial Bench at the Supreme Court; integrated and promoted Magistrates throughout the country which took into account their mastery of the two official languages. 85 The government announced in December 2016 the recruitment of one thousand bilingual teachers, 86 promised two billion francs as grants to private schools among other things. In another frantic effort to ensure the pacification of the Anglophone community, the Government created the National Commission of Bilingualism and Multiculturalism 87 under Peter Mafani Musonge as coordinator by Decree No. 2017/013 of 23 January 2017 read over the 5pm bilingual broadcast over Cameroon Radio Television (CRTV). 88 The commission was to report and propose recommendation on the protection and promotion of bilingualism and multiculturalism to the president of the Republic..." 89 Nico H. said; "even a hundred of such commissions will not do a thing except there was peace, justice and reconciliation." 90 Though many complained that it had no use in the address of the problems raised by lawyers and teachers, it however was a step by government to calm down some other English speaking Camer

After the reunification of Cameroon in 1961, some West Cameroonian political class began to cast their doubts on the centralized Federal System put in place. This system according to some Anglophones gave room for marginalization and assimilation of West Cameroon especially with the closure of its development agency. This somehow nursed grievances and in 2016, the bon a fide English Speaking Cameroonian lawyers and teachers rose up in a "peaceful civil disobedient protest" to express their disillusion, over the erosion of the Common Law Judicial System and the Anglo-Saxon Education System in Cameroon. This paper from this angle looks at the origin of the Consortium, Government position and the impact. The paper argues that, Government's responses and policies towards the corporate grievances and the ban on the Consortium laid the foundation of a radicalized armed conflict in the Anglophone region and within this dispensation; other discourse towards a lasting solution became valuable. The major instrument used for data collection was interviews, secondary literature, author's experience and a multidisciplinary approach to present the facts. This paper concludes that the failure to break even between the Government of Cameroon and the Cameroon Anglophone Civil Society Consortium (CACSC) introduced armed conflict with costly consequences.

on; appointed English speaking Magistrates to the post of President of the Judicial Bench at the Supreme Court; integrated and promoted Magistrates throughout the country which took into account their mastery of the two official languages. 85 The government announced in December 2016 the recruitment of one thousand bilingual teachers, 86 promised two billion francs as grants to private schools among other things. In another frantic effort to ensure the pacification of the Anglophone community, the Government created the National Commission of Bilingualism and Multiculturalism 87 under Peter Mafani Musonge as coordinator by Decree No. 2017/013 of 23 January 2017 read over the 5pm bilingual broadcast over Cameroon Radio Television (CRTV). 88 The commission was to report and propose recommendation on the protection and promotion of bilingualism and multiculturalism to the president of the Republic..." 89 Nico H. said; "even a hundred of such commissions will not do a thing except there was peace, justice and reconciliation." 90 Though many complained that it had no use in the address of the problems raised by lawyers and teachers, it however was a step by government to calm down some other English speaking Cameroonians. In spite of the efforts made to address Consortium grievances which were embedded in the teachers and lawyers concerns, many Anglophones still were vexed with the fact that, the Government did not resolve the problems as tabled to them (State). The Consortium condemned all acts of violence and equally cancelled all 11 February and 20 th May celebrations in former West Cameroon while they called on a sit in strike and the Government to clear the streets of uniform men. 91 Violent reactions erupted in Bali where the Divisional Officer Divine Kamera was asked to leave his office and return home. 92 The State Counsel of Bali was locked up in his office, together with his staff. On Tuesday January 17 2017 at 5 o'clock a communiqué over national news banned the Consortium and SCNC. 93 The arrest of Agbor Balla and Fontem Neba a few minutes after the ban increased violence in Buea around Malingo Junction and barricades raised. 94 Government use of water cannon, tear gas and gun shots only irritated the angry mob who charged into the streets to block the transfer of Consortium leaders to Yaoundé. Ngo Bakang described the scene in the following words; 100 Lawyers Right Watch Canada was an NGO with special consultative links with the Economic and Social Council of the United Nations and condemned the government of the President of the National Assembly. 99 Although it was generally said Parliamentarians have immunity, it could easily be concluded by this situation that those in Cameroon might not have because a parliamentarian could not be begging for protection even to the extent of hiding for fear of arrest. I The arrest of Consortium members sparked condemnation from various quarters. Some Human Rights groups such as the Lawyer's Right Watch Canada a non-governmental organization with special consultative status with the Economic and Social Council of the UNO, who objected the arrest of Felix Agbor and co and argued that it was a wrong use of "Criminal Law Powers." 100 The Social Democratic Front (SDF) party also condemned government in the management of the crisis and the detention of Anglophone Cameroonians. 101 In an interview, Tchiroma said; "people would face consequences of their actions" this buttressed the fact that; there would be more arrests and detention. And after the ban of the Consortium, government launched a man hunt for other Consortium members and Wilfred Tasang following a tip-off ran into hiding 102 and disappeared from public scene for fear of being picked up to join the duo 103 Diaspora community became more violent and disturbed delegations from Cameroon sent to talk with them a few months after the Consortium was banned. In Maryland USA, the Ministers' delegation was stopped from having a meeting. There was violence in Belgium, and in South Africa, the delegation was vandalized. According to Lucas T, "sending such a team out there was a scheme to waste State money when they knew well that they will not be received" 105 In UK and Canada it was unsuccessful. A State of emergency was imposed by the governors of the two regions from 29 September to 3 October 2017 with curfew, restricted movements, and gatherings of more than four people. 105 Lucas Ntang,Tasi 75years, retired member of Cameroon Parliament interviewed at Ntamulung 14/01/2020. 106 The Voice, No. 155, 3. 107 The Post, No 1146, Tuesday April 25, 2017,3. Internet connection was shut down 106 for a period of three months in the Anglophone regions of Cameroon which according to some Government officials was a threat to "peace". To some, it was a means for character destruction, destabilization of public opinion and it misled many Cameroonians. The internet through the social media was the rallying point for those who wanted change especially at the helm of the Government as pointed out by some Cameroonians. As someone remarked, "internet was gone for three months nothing changed we switched to other connections it was reconnected on April 20, 2017 we still did not have a change." But it should be noted that the mayor of Buea Ekema Patrick thanked President Biya for internet restoration. 107 There was victimization and intimidation of people on the social media especially those who supported the Consortium stance for Federation. And Fontem Neba quickly pointed out that, "because of this, those who stood for Federation will no longer talk of it even on the social media." 108 This made some activists to pipe down for fear of being tracked especially as the police and military searched phones on daily basis. We have understood that the Government would not take the consortium serious and would not solve the problem fast so in support of the course, we shut down our shops during the days of ghost towns as fixed by our leaders though the days were many and it weighed down on us economically. Though we suffered, it was a good sacrifice. 113 The above statement shows proof that the business community was in solidarity with the Consortium though it had economic repercussions on them. Another said; The ghost towns were effective because each time there was to be a ghost town, we were informed through the social media and we bought food and other items and stocked at home. And if I was to travel to buy, I rescheduled my trip and it did not affect my business life in any way though there were many of such ghost towns called such as the Mancho Bibixy. 114 This also confirmed the above statement that social media was a means through which the Consortium circulated their information. In fact, ghost towns continued ceaselessly in the entire Anglophone territory. As a result of the failed discussions between the Government and the CACSC, some Anglophones settled that their children would not attend school and political elites of the Anglophone extraction made fruitless campaigns on back to school as many schools closed their doors. On Mondays in observance of ghost town calls by CACSC no school dared to open its doors. Statements from some high profile politicians such as; "we owe our children a duty and our utmost responsibility is to educate them" did not make a change to the school situation. Senators, Parliamentarians and members of Government from South West known as "Living Force" also called on teachers to stop the strike to enable parents send children to school. 115 Total insecurity seized the people of the North West and South West regions when the Consortium and the SCNC activities were banned by Minister Rene Emmanuel Sadi of Territorial Administration and Decentralization with the motive that, "their activities were liable to jeopardize the security of the State." This was in relation to Tchiroma's declaration in a press conference shortly before the banning order that "the Head of State has affirmed without any ambiguity that the unitary form of the State is intangible and Cameroon is one and indivisible and shall so remain." 118 What a paradox, the leaders whom the Government negotiated with to put an end to the strike action were labeled "terrorists" and terror took over the country as troops flooded the towns and every nook and cranny was invaded by the troops with armored cars "manned by soldiers armed to the teeth" a kind of situation Mukete described in his work My Odyssey: The Story of Cameroon Reunification. 119 Water tankers paraded the streets as the Government tacitly declared a state of emergency in the two Anglophone regions. 120 The situation reached fever height and some Anglophones seek means for self-defense as horror swept across the North West and South West areas. According to Munjo A., 116 Many people who had hoped that the Consortium will bring a solution to the crisis were stormed with the ban and the rapidity with which troops took position in Buea, Bamenda town and their actions took the struggle to a more radicalized stage which became more frightful even to the troops. 121 Homes were invaded, property looted by soldiers, torture and indiscriminate arrests without warrants increased in the towns of Bamenda, Kumba, Mamfe, Kumba, Santa, Kumbo and others. Awuro Esthela was arrested and taken to Yaounde, Abonge Eric was picked up at "Big Brother bar" at Commercial Avenue Bamenda and ferried to Yaounde where he spent nine months in detention in Secretariat D'etat a la Defense (SED). In March 2017, Pen Terence was bundled up at mile four Nkwen Bamenda by security forces and taken to Yaounde. It was estimated that by 1 st October 2017 over 500 persons were arrested. 122 The poor suffered a great deal and blackmail due to enmity stepped in and no one was safe. Taxis, motorbikes, the major streets and quarters became insecure. --- CONCLUSION This paper exposes the view that Government inability to quickly handle the problem of the teachers and lawyers resulted to the formation of the Cameroon Anglophone Civil Society Consortium on 6 December, 2016 which became the canopy under which all Common Law Lawyers and various Teacher's Trade Unions operated. The hanky panky game between the Government and the Consortium where mistrust, dishonesty and threats became instruments for settlement for the two groups only nursed a breeding environment for the Anglophone diaspora and those termed by the government as secessionists to take advantage and finance the already crippling situation. Their entry worsened the already bad situation and their demands through the Consortium angered the State and it took wrong decisions. The Ad-hoc committee refused to entertain talks on the Federal System and the release of all Anglophone detainees linked to the crisis. This paper from this point, affirms that the poor management of the crisis from the start to include dialogue and failure to listen to the voices of the people and international community; and the out law of the Consortium laid the foundation of armed struggle in Anglophone Cameroon. The paper found out that insecurity griped the people and some schools in the two Anglophone Regions closed their doors. Though some grievances were looked into by government, it did not satisfy the Consortium. This paper further argues that crisis management had lapses on both parties; the Government and the Consortium. The crisis was a hard pill for the Government to swallow and through the various narratives, the Consortium grievances were unresolved and the Anglophone population looked up on the Consortium. Peace became a "scarce commodity" dialogue could have been a remedy in the face of the deep mistrust but genuine dialogue was absent. This paper thus maintains that the Government of Cameroon and CACSC were the actors that could have solved the burning issues at the time but allowed conflict to take central stage. It thus calls the attention of the Cameroon Government and stakeholders to revisit their strategies in handling and addressing problems of the people. --- NEWS PAPERS The Voice --- Eden The Post --- Breaking News Bilingual The Guardian Post The Median The Horizon Life Time --- Cameroon Tribune The Independent Observer The

After the reunification of Cameroon in 1961, some West Cameroonian political class began to cast their doubts on the centralized Federal System put in place. This system according to some Anglophones gave room for marginalization and assimilation of West Cameroon especially with the closure of its development agency. This somehow nursed grievances and in 2016, the bon a fide English Speaking Cameroonian lawyers and teachers rose up in a "peaceful civil disobedient protest" to express their disillusion, over the erosion of the Common Law Judicial System and the Anglo-Saxon Education System in Cameroon. This paper from this angle looks at the origin of the Consortium, Government position and the impact. The paper argues that, Government's responses and policies towards the corporate grievances and the ban on the Consortium laid the foundation of a radicalized armed conflict in the Anglophone region and within this dispensation; other discourse towards a lasting solution became valuable. The major instrument used for data collection was interviews, secondary literature, author's experience and a multidisciplinary approach to present the facts. This paper concludes that the failure to break even between the Government of Cameroon and the Cameroon Anglophone Civil Society Consortium (CACSC) introduced armed conflict with costly consequences.

INTRODUCTION As the prevalence of social media in the world around us increases and the number of users on these online platforms grows, so too grows the rate at which scholarly content is being proliferated and discussed in these venues. More and more, academics are finding it rewarding to look to these platforms for the insight they provide into research problems. One reason scholars have turned to social media is to measure the influence their work is having in those spaces; this has become known as alternative metrics, or altmetrics [1]; another reason is for Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]. ACM/IEEE, June 2019, Urbana-Champaign, Illinois USA <unk> 2019 Association for Computing Machinery. the knowledge online platforms provide about human behavior-an area of research known as social-media analytics [3,6,9]. Studies in social-media analytics tend to focus either on text, using approaches such as Natural Language Processing (NLP), sentiment analysis, or opinion mining to arrive at and support research conclusions [12], or on the proliferation of content through online communities [5]. These approaches have proved effective for understanding or predicting many aspects of human behavior; but they leave a number of other expressive signals unexamined. Click-based reactions, on the other hand, are a relatively underutilized resource in social-media research. Examples of quick-draw, ready-made expressive features are becoming increasingly prevalent across many platforms, and as such have attracted some amount of attention from researchers in the past few years [2,4]. In this paper, we present a new dataset of click-based reactions to scholarly articles on Facebook and use it to gain insight into how users are interacting with scholarly articles on that platform. In addition to information about the articles themselves, our dataset records the count of each click-based feature we could access through Facebook's Graph API. We use our newly developed dataset to train and test two machine learning algorithms, and our analysis of the results shines some light on surprising relationships between features. --- BUILDING THE DATASET Before going any further, it will be useful to define a few terms and features: • Click-based reactions -non-textual user interactions with shared content-sometimes referred to simply as reactions; includes Facebook Likes and Reactions, Re-shares, and Page visibility (definitions for these last two are below). • Reactions -the five click-based reactions: Love, Amazed, Laughing, Sad, and Angry; will be distinguished from the common term "reaction" by capitalization. • Page visibility -the number of followers a Facebook page has. • Re-shares -the number of times users have re-shared a public post of an article into another location. The roots of our dataset lie in the online resource Altmetric 1, which tracks the impact scholarly articles have across a variety of social media platforms. We used Altmetric as a "jumping-off point", querying their API 2 for information on articles we were interested in and for the public pages onto which they had been shared. It gave us access to the titles, publication dates, subjects, and the URLs of Facebook shares for nearly 1.5 million scholarly articles. We targeted content shared on Facebook rather than other socialmedia platforms for several reasons. First, Facebook offers its users a variety of click-based interactions with which they can personalize their response to content; other platforms we considered targeting such as Twitter have more limited palettes of click-based reactions available to users. Second, Facebook's enormous population of active users increases the likelihood that content shared there will receive more attention: it has about 2.27 billion active monthly users, almost seven times Twitter's active population of 330 million. Third, the impact of scholarly articles on Twitter users has been the subject of many recent studies [5,8,11], whereas the response to this type of content on Facebook remains largely unexamined. With our list of Facebook URLs for article shares, we queried Facebook's Graph API 3 for the reaction counts on each post. Our dataset records their responses, and was collected during the period of December 1-13, 2018. Constraints in the number of queries allowed by Facebook's API (200/hour) determined the rate at which we could work. The resulting dataset is publicly available on OSF 4 as a comma-separated-value file (CSV). We limited our collection efforts only toward scholarly articles published in 2017. Choosing this year accomplished three goals: (i) Reactions were released by Facebook in February 2016 [7], so any articles we looked at had to be published after that time to have meaningful data on this feature; (ii) any time a new feature is unrolled, it takes some amount of time for users to learn how to use it; Prah [10] finds that use of Reactions increased from 2.4% of all interactions in April 2016 to 5.8% by June 2016, and up to 12.8% of all interactions by June 2018; by the time of our data collection in December 2018 a large enough subset of users were comfortable expressing themselves with the feature to warrant more scholarly attention; and (iii) by the time we began our data collection, a sufficient interval of time had passed for articles to be widely shared and reacted to (between 11 and 23 months). Of all the articles tracked by Altmetric, we found 296,052 were published in 2017 and had been shared on Facebook at least once. We eliminated entries that were missing data on the pages to which the articles had been shared; this reduced our set to 135,635 articles. We further limited the scope by selecting only articles with Scopus 5 subjects in the scientific domain. We chose to focus only on articles in the Health Sciences, Physical Sciences, Social Sciences, and Environmental Science. Figure 1 shows that these four categories, article counts fall within one standard deviation of the mean number of articles, as do the total number of Facebook shares (Health sciences is the only exception, exceeding one standard deviation greater than the mean of article counts). It also displays the full list of subjects in all the 2017 articles and gives a sense of their distribution. The mean and two standard deviations are indicated there with blue lines for both axes, and the four subjects we target are indicated 2 http://api.altmetric.com/. 3 https://developers.facebook.com/docs/graph-api/. 4 https://osf.io/4kh7r/. 5 https://www.scopus.com/. with arrows in the plot. Limiting the scope of subjects reduced the number of articles needed to process to just over 31,000. When we removed articles with missing features such as abstract and title, we were left with 11,474 articles: these are the articles recorded in our dataset. In our data collection process, we took the utmost care to respect Altmetric's and Facebook's specifications for how and why their data can be accessed and used and to protect the personal information of social-media users. Our interests are only in the ways that people are interacting in the aggregate with scholarly content on social media platforms-not in specific ways users' beliefs or opinions may influence their behavior. We recognize that identifying information could in some instances be inferred a posteriori from some of the data we collect; however, our method of data collection does not target anything that could be used to consistently identify individual users and avoids collecting identifying information about individuals. --- DATA EXPLORATION The click-based features of our dataset are displayed along the axes in Figure 2; also displayed are the Pearson r correlation coefficients for all feature pairs. Highly correlated pairs are: Like and Love (r = 0.82), Sad and Angry (r = 0.81), Like and Amazed (r = 0.77), Love and Re-shares (r = 0.71). We can infer that high positive correlation is a sign that users employ features in similar contexts, and that the emotional expressions represented by those features overlap. For example, a Like seems to have a meaning comparable to a Love or (to a lesser extent) an Amazed, or (to an even lesser extent) a Laughing reaction. These relationships may not surprise us because they are all positive emotional states; but other feature pairs that have related expressive values in usage, such as Angry and Sad reactions, are not so intuitive. Likes and Re-shares are correlated with the most other features; this might be explained by the fact that these two are the oldest reactions-but we also notice they are correlated with other emotionally positive reactions such as Love or Amazed and not with the negative emotions Sad or Angry. It follows that by Liking or Re-sharing a post, a user expresses a positive emotional reaction to its content. Looking at this from another angle, we infer that content that is more likely to inspire a negative reaction from users is less likely to be Re-shared or Liked. High correlation between features can lead to increased variance in model results. To counter this, modelers often eliminate one of a pair of correlated features. Rather than removing features and losing data in our sparse dataset, we combined Love/Wow and Sad/Angry Reactions into two new composite features for our models. Low correlation signifies that features have relatively distinct use values. Among the lowest r coefficients are Love/Angry (r = 0.145) and Laughing/Sad (r = 0.224); this makes intuitive sense, as these reactions nominally encompass opposite emotions. Laughing/Page visibility (r = 0.229) is another low-correlation pair, suggesting that articles that inspire humor are more likely to be posted to public pages with relatively low follower counts. It is likely that this relationship may be a result of our choice to limit the articles we include to those in the scientific domain, where humor is an under-utilized affect. Our dataset also contains outliers in each feature category; to correct for these, we re-scaled the features to a range from 0 to 1, then took the cube root to these new values. Our root normalization function is demonstrated in Equation 1; it helped to smooth the distribution of values, raising the lower values by more than it raised the higher values. The result after combination/normalization is displayed in Figure 3. rt_norm(F i ) = 3 F i -F min F max -F min(1) Even after transformation, our dataset is still sparse-zeros are unchanged by the transformation; yet features with greater variance, such as Visibility or Likes, have less spread between the IQR and outliers. The median value of all Reactions is zero, and non-zero values in those features are all in the fourth quartile. Likes have the largest interquartile range (IQR), though the median is still close to zero. Page visibility and Likes have the highest median values of all features. --- SUPERVISED-LEARNING MODELS To explore the relationships in our dataset further, we isolated two feature subsets and trained two supervised learning classification algorithms with them. We used Decision Tree and Random Forest algorithms because of the insight they provide into the relationships between features, and our feature sets are detailed in Table 1. We were interested in gaining insight into the extent to which users' interactions could be related to articles' subject matter; and so we selected article subjects as the class labels for our models. This gave us four targets for our multiclass classification models to predict. With the first set (A) we were interested in finding the extent to which click-based reactions that are immediately available to Table 1 displays the accuracy and Area Under the Curve (AUC) of our models, and Figure 4 shows the results of our models using several different metrics. For reference, scores are shown against the baseline, which represents random guesses at which of the four class labels an article belongs to. Feature set B produced significantly better results than A with both algorithms. Average accuracy of models with feature set B is 160% greater than the baseline, while feature set A is only 58% greater. Figure 5 shows the relative importance of each feature in our models. In feature set A, Likes have the greatest weight, accounting for 51% of the result on average between the two algorithms; the weight of Re-shares is the second highest importance, accounting for an average of 27% of the result. In feature set B, Visibility is the most important feature accounting for an average of 94% of the result; the remainder of the weight is spread relatively evenly among the remaining features. --- DISCUSSION AND CONCLUSION Our new dataset of click-based reactions to scholarly content on Facebook offers a wealth of possibilities for researchers interested in social media analytics. We have demonstrated how it can be used in the exploration of user interactions with scholarly content on Facebook, and how click-based reactions are an effective data source for investigating indicators of user emotional attitudes. Results from the models trained and tested on our dataset suggest that the number of followers a page has (Visibility) may be predictive of article subject matter; this indicates that there may be patterns in the content shared on Facebook pages and the number of followers these pages have. It may prove useful for researchers to explore the ways in which Facebook page popularity is stratified by the type of content displayed on their pages. We have also suggested some interpretation of Facebook clickbased reactions that are not immediately apparent, notably that Re-shares convey an emotionally positive feelings toward content, and that Sad and Angry Reactions express similar affects. These relationships are not at all obvious, and give us insight into how these features are being used in practice.

Research on social-media platforms has tended to rely on textual analysis to perform research tasks. While text-based approaches have significantly increased our understanding of online behavior and social dynamics, they overlook features on these platforms that have grown in prominence in the past few years: click-based responses to content. In this paper, we present a new dataset of Facebook Reactions to scholarly content. We give an overview of its structure, analyze some of the statistical trends in the data, and use it to train and test two supervised learning algorithms. Our preliminary tests suggest the presence of stratification in the number of users following pages, divisions that seem to fall in line with distinctions in the subject matter of those pages.

Introduction Poor perinatal health is a global public health problem and reflects issues of inequality and injustice. Perinatal deaths include both stillbirth and neonatal death [1,2]. Worldwide, over 14,500 perinatal deaths occur each day [3,4] almost 99% of them in developing countries. Among developing countries, Sub-Saharan Africa and South Asian countries record over three-quarters of the world's stillbirths [5] and neonatal deaths [6]. Furthermore these regions have experienced slow progress in reducing perinatal mortality rates in the past two decades [7]. With only a 2.7% annual rate of reduction in neonatal mortality rate (NMR) between 2000-2012, it is estimated that it will take 110 years for an African and about 80 years for a South Asian woman to experience a similar chance of saving their newborn baby as is currently experienced by the average woman in developed countries [8]. Moreover, at current rates of reduction of the stillbirth rate, about 160 years and 100 years will pass before the average pregnant woman from Africa and South Asia respectively experience the same stillbirth rates as a woman in a developed country [9]. These differences are principally attributed to the weak health systems and poor coverage of maternal and neonatal care provision in Africa and South Asia [10]. National statistics from Nepal show that, despite significant progress over time, the rates of neonatal and perinatal deaths are still high at 33 per 1,000 live births and 37 per 1,000 live births respectively [11]. Within Nepal, significant disparities exist in the distribution of perinatal mortality rates. The mountain region has a neonatal mortality rate of 46 per 1,000 live births, with 85% of all neonatal deaths occurring within the first week after birth [11], while the perinatal mortality rate is likely to be higher due to the underestimation of stillbirths. These figures are equivalent to the highest mortality rates in Sub-Saharan countries. Nepal's national policies on maternal and child health acknowledge equity, rights-based approaches, family/community based care, woman-friendly, 24 hour birthing services, social inclusion in maternal and newborn care, and counting every perinatal death [12][13][14][15][16]. However, strategies developed to tackle poor perinatal health are predominantly medically-oriented and emphasise the treatment of newborn infections and promotion of health facility births. Studies undertaken in Nepal are also focussed on medical causes of perinatal deaths [17][18][19][20] and on epidemiological investigations of socio-demographic and health service utilisation variables and their distribution across the country [21,22]. There is little evidence examining the social determinants of poor perinatal health, particularly in the country's most disadvantaged areas. The literature suggests that biological events such as pregnancy and childbirth are also socio-culturally constructed and therefore we need to understand tradition, society and culture and to examine the socio-cultural contexts of pregnancy and childbirth [23][24][25][26]. Traditions, social values and culture shape pregnancy and childbirth experiences and have a strong impact on women's choice and control over both maternal and child health [23]. Values and cultures are of course different in different societies [25]. Perinatal deaths are linked to social, cultural and religious beliefs and values. For example, Hmong women living in Australia believe that disharmony in personal health as well as in the supernatural world causes miscarriage, stillbirth and neonatal death [27,28]. Likewise, studies from Ethiopia, Tanzania, Uganda and Ghana found similar perceptions among women and their families about neonatal and infant death. In Ethiopia, supernatural forces are believed to cause perinatal death and dead babies are often buried in the house or in the backyard without any notification of birth or death [29]. This Ethiopian study identified that it is not permitted to mourn a perinatal death as it is considered to be against God's will. Similar findings are reported in Uganda [30] and Ghana [31]. A range of studies discuss religious affiliations related to, or associated, with both positive and negative health outcomes [32][33][34][35][36]. The impact of religion on health is described mainly through two pathways: by affecting individual beliefs and norms related to health practices, and by increasing people's social capital/connectedness and empowering people to effectively utilise relationships and resources for their health and wellbeing. Positive health outcomes are described mainly in supporting people to live through and cope with the tragedies of bereavement, HIV/AIDs, non-communicable diseases such as cancer, positive mental health and reduced addiction/substance abuse and crime. Negative outcomes are about not seeking, or delayed seeking of, medical care due to rigid religious beliefs. Lay beliefs and lay knowledge have proven crucial to understanding and addressing the social determinants of health [37][38][39][40][41]. Lay beliefs offer an explanation for what people do and why they do it (including health behaviour), which could contribute to more effectively planning or tailoring health policy or interventions. Although the data highlight that perinatal mortality is high in Nepal's mountainous region [11], no study to date has examined women's and families' views about poor perinatal survival within the broader sociocultural contexts. Studies conducted in Nepal mostly cover the plains or semi-urban hilly regions, and are often limited to describing patterns of mortality or service utilisation [42][43][44][45][46][47][48], or medical causes of perinatal deaths [49][50][51][52][53]. An in-depth socio-cultural understanding of what lies beneath the continuing high perinatal mortality rates is lacking both at the national level and more specifically for the mountainous region which exhibits the highest rate. This paper aims to explore the beliefs and experiences of local communities about perinatal sickness and death in these mountainous villages. --- Methodology The study was conducted in two rural mountain villages of Nepal which rank the lowest on the Human Development Index (0.304), have one of the worst reported child survival rates in the country, and have no access to transportation. The two mountain villages provided a diversity of context: one village is predominantly Khas ethnic, called Aryans and who follow Hindu religious beliefs, with access to two village health facilities (a community birthing unit and a health facility), district hospital, and the other village comprising mostly Lama people of Tibetan descent who follow Buddhism, and have access to a village health facility with a birthing unit. This selection enabled the principal researcher to conduct fieldwork for data collection within the constraints of practicality in such remote areas. This study does not intend to compare the villages as such, rather the different villages were chosen to present findings as a case about how women and families living in remote mountainous settings explain and deal with perinatal deaths. Another reason for choosing the two villages was to reach the required number of participants in this study-the women and families who had recent perinatal deaths. Qualitative interviews were conducted between February and June 2015 with 42 women and their families who had experienced a neonatal death or stillbirth in the previous four years, nine Skilled Birth Attendants (SBA), two Female Community Health Volunteers (FCHV), two support staff, one Auxiliary Health Worker, and five other stakeholders (S1 File). Women interviewed were in the age range 16 to 35 years. Women continued to be interviewed until a saturation point was reached where no new information was obtained by interviewing additional women. This saturation point was reached after interviewing these 42 women. Separate interview guides were developed for the women and families, the health service providers, and the other key stakeholders (S2 File). The stakeholders comprised local journalists and staff of non-governmental agencies working in the field of maternal and child health, and child marriages. The views and experience of health service providers of their day to day experiences about provision of health services to improve mother and baby's survival, and that of the local stakeholders, supplemented the data from women's interviews and helped to understand a comprehensive picture of ongoing perinatal deaths in the study communities. Interviews covered a range of questions around mothers' experience of perinatal death and sickness. Participants were identified purposively using local volunteers, FCHVs and through contact with health facility staff. Interviews were conducted in the local language by the first author and audio recorded with participants' consent. The first author was assisted by the local health volunteers and a health service provider during interviews with participants from Lama Communities. Although Lama women and their families spoke Nepali, it was felt that helping them to speak in their own local language would encourage them to more naturally explore the phenomenon-contexts of perinatal deaths. The first author is a Nepali national with seven years' experience as a health worker in Nepal's hilly and mountain regions. The study was approved by the Social and Behavioural Research Ethics Committee of Flinders University, the ethical board of Nepal Health Research Council, and the District Health Office of the study district in Nepal. Written informed consent was sought from the participants before the start of the interview. --- Data analysis Interview files were simultaneously translated and transcribed into English by the first author. Six random transcripts were checked by five bilingual (Nepali and English) experts to ensure consistency in transcription and translation into English. NVivo version 10:00 software was used to organise the data and facilitate the development of coding frames. The texts were analysed with a comprehensive coding process, using an inductive thematic analysis technique as suggested by Braun & Clarke (2006). Thematic analysis involves identifying, analysing and reporting various themes from the data, where themes are the central organising concepts about the data. In this study, these are the aspects within the data which have revealed the socio-cultural contexts influencing perinatal deaths in the remote villages under study. --- Findings The analysis of data provided three key themes emerging from the qualitative interviews in relation to religio-cultural contexts that describe predominantly the views of acceptance and fatalism about perinatal death and sickness. These themes are: (1) 'Everyone has gone through it': perinatal death as a natural occurrence; (2) Dewata (God) as a factor in health and sickness: a cause and means to overcome sickness in mother and baby; and (3) Karma (Past deeds), Bhagya (Fate) or Lekhanta (Destiny): ways of rationalising perinatal deaths. --- 'Everyone has gone through it': Perinatal death as a natural occurrence The collective experience of perinatal death among the study participants, their families, neighbours and the whole community has contributed to the notion of perinatal death as a natural and acceptable occurrence. This notion is expressed as follows by a 20-year old mother with experience of losing her child: My grandmother had 10 births (Sutkas-childbirths). All of her babies died...., after 12 years she delivered my father and his three sisters. My father says,'you are young, you can bear babies, you haven't lost anything'. My father-in-law is also the only surviving son in his family. (Interview, mothers) Another participant, while describing her own and her sister's story, labelled perinatal death as a common phenomenon that everyone faces: My sister lost two babies, a boy, and a girl. We both lost our babies. Now she has two living children, and I have four. Everyone has gone through it. We can do nothing about it... (Interview, mothers) Perinatal deaths, particularly for the first or second pregnancies, have been seen to occur in the participants' generation, their parents' generation, their grandparents' generation and so on, and have been accepted as intergenerational life events. This has led to a perceived lack of control over mother and child health. A local auxiliary nurse also confirmed that the village women take the loss of a baby rather naturally, as a 'generational continuum': Village women believe that losing a baby is seen as a generational perpetuation in their families. They say,'my sister also lost one, my mother-in-law had the same experience and my mother had also lost babies. (Interview, service providers) The commonness of the experience of perinatal loss is reinforced further when women and families find local Female Community Health Volunteers (FCHVs) and health service providers also experiencing similar events: That female health volunteer (she points towards a nearby female health volunteer's house) lost a son after birth. And, the two women over there also lost their babies. It is like this here, it occurs with everyone. (Interview, mothers) 1.1 Timing and perinatal deaths: Low level of concern about early life. The study revealed that there is an association between the time of death and the level of acceptance: the sooner the baby dies, the more acceptable it is for parents and community. Stillbirth and death immediately after birth or during the mother's confinement in Gotha (cowshed) is readily acceptable: I didn't feel worried about the stillbirth (Hudaimareko). I felt it was okay. It rather made me easier to resume my day to day work sooner. (Interview, mothers) Gotha is the place of birth, the ground floor of their house where a woman gives birth and resides during the postnatal period until about three weeks after birth. This study explored that both mother and baby are considered impure (Chhuhi) after birth. To keep the main part of the house pure, and not to pollute their Dewata (God), the birthing mother and her baby remain confined in Gotha. By adhering to this tradition it is believed that they are not displeasing their God, and hence protecting themselves from any harm that might result from God's wrath. It is not a matter of significant concern when a baby dies before the last month of pregnancy (Hunemahina). In the local dialect, the term foetal death is used interchangeably with miscarriage which shows the lack of special importance given to pregnancy loss. My wife had a foetal death (Aadanjhadne), but that was not a death, she lost it at six months of pregnancy. (Interview, mothers) The concern towards a baby's death increases when a woman enters the main home after Gotha around three weeks after birth, and family members visit the baby. In these mountain areas, perinatal deaths have no ritual significance. From the viewpoint of health volunteers, the repeated occurrence of perinatal deaths and lack of social or religious concern placed on them render them not worth counting or reporting. Foetal death was also described by participants as Pakhala-literally translated as a diarrhoeal disease-which metaphorically means a nonsignificant, common and natural occurrence. This perception about foetal death is shown in an interview with a Female Community Health Volunteer (FCHV): There are many women losing their babies in pregnancy (Pakhalajane) and neonatal deaths. You can see these in every single house. Are we going to record all such deaths? Two years ago, even the local doctor's (Auxiliary Health Worker) sister-in-law lost her baby. I know two women there who lost their babies in pregnancy (Pakhala). A woman in that house [she points to the house] had twin babies, but both died as newborns. (Interview, health volunteers) Poor recording/reporting of such deaths became obvious when the first author on fieldwork reviewed the health facility records and had day to day conversations with the FCHVs during the recruitment of participants to this study. Moreover, it was not easy to get the data at first. Initially, FCHVs could not remember any such death in their neighbourhood. Over time, during repeated contacts and conversations, they started recalling women who had experienced perinatal losses. During the five months of fieldwork in the two study villages, eventually 42 women voluntarily reported 49 perinatal deaths (16 stillbirths, and 33 neonatal deaths) occurring in the last four years, with a majority of them occurring in the last two years. These 42 women are estimated to represent approximately 3% of the women of childbearing age in the study villages. By comparison, the local health facilities in the two study villages reported only five neonatal deaths in their verbal reports and only three in the local records. Not a single stillbirth was reported in the village health facilities. Based on these crude data, it is estimated that the study villages have a current neonatal mortality rate of 44 per 1,000 livebirths-similar to the official estimation of 46 per 1,000 as reported in the 2011 Nepal Demographic and Health Survey. This field visit also revealed an extended perinatal mortality rate of 63 per 1,000 births if stillbirths and all neonatal deaths until 28 days after birth are included in the calculation [54]. These estimates are based on the reported number of perinatal deaths for the previous four years. It should be noted that these two rates-neonatal mortality rate and the extended perinatal mortality rare-are not strictly comparable because the neonatal mortality rate is based on live births but the extended perinatal mortality rate is based on all births (live births plus still births). With a household survey based on a representative sample, the rates would be likely to be different because women in this study were participants of qualitative interviews who willingly reported their perinatal losses. --- The will of Dewata (God) in health and sickness: A cause of, and a means to overcome sickness in mothers and babies The word 'Dewata' is a collective name of Hindu Gods and Goddesses. The study revealed that believing in God's will as the cause for health and sickness has a very strong impact on views around perinatal death. One can find the Dewata represented in various symbolic forms everywhere: for example, as statues, temples and ribbons in the farmland, on the banks of streams, on street corners, in the forest, inside the houses and in the middle of the villages. 2.1 Dewata (God) and childbirth complications: Seeking faith healers during pregnancy, and childbirth. God's will is believed to be a key cause of problems in pregnancy and childbirth. One of the participants, a 22 year old mother, commented that her breech presentation during delivery and the subsequent neonatal death were due to not worshipping their God: My baby did not die due to breech presentation (Ulto). It is because God (Dewata) was angry with us [she cries]. My family should have called the faith healer (Lama), and prayed to God [Gyana]. They didn't do anything to please God at home [she cries], therefore my baby died. Local faith healer had told that it was not going to be a good fortune if we didn't worship God. (Interview, mothers) It is believed that faith healers can make the necessary prayers to please God. Faith healing is an old tradition practised for generations and has a strong foothold in these villages. Faith healers outnumber health volunteers and health service providers. There are different cadres of faith healers locally known as Dhami-Jhakri, Dangri in the Khasan community; and Lama, Chumba in the Lama community. Women and their families believe faith healers are chosen by God, they can understand God's language and can alleviate any kind of suffering including women's sickness. The faith healers are usually called to be present during childbirth and they are relied upon to relieve a woman from pain and suffering, to speed-up the birthing process and to save the lives of mothers and babies. They are also relied upon to help with other family health issues and pregnancy complications such as fainting during pregnancy; prolonged and severe labour pain; breech position of the baby; and when a woman feels weak during the birthing process. --- God's will and impact in health care seeking Seeking assistance from professional health providers is often not the first choice of treatment. Professional health providers are sought only when the faith healers fail to provide assistance and if the faith healer gives the women and their families permission to contact a health provider. A local health worker described a moment when he had to attend a woman in labour at home together with a faith healer: The faith healer (Lama) was ringing a bell around and reciting prayers (Mantras). I had to deliver her by rupturing her membrane. They invite us only when they are permitted to do so by the faith healers. (Interview, service providers) Believing any sickness is a result of God's will means that formal health care for any health problems during pregnancy and childbirth is rarely sought. A local Auxiliary Nurse confirmed this: She [referring to a pregnant woman] was seven months pregnant. I asked her family to describe to me what had happened. They said that she died due to God's curse (Dewatalagne). (Interview, service providers) 2.2 Dewata (God) after birth-Seeking and receiving care for babies. Women and families also attribute their babies' sickness to God's displeasure. In their daily conversations, the range of sicknesses are together called 'God's wrath (Dewatalagne)'. It is believed that sicknesses happen when God is not pleased. To alleviate the sicknesses, one is expected to please God by prayers, wearing amulets, sacrifice (animal) and offerings made through faith healers. --- Local sickness labels The different sickness types frequently described by the participants are listed in Table 1. --- Type 1 sickness: God attributed as a major cause and a means to alleviate/cure sickness The type 1 sickness labels in Table 1 were attributed mainly to God's will. Participants believed that to overcome these sicknesses they need to please their God. Faith healers were utilised to make offerings to God and to make their God happy. For example, a new mother's lack of breastmilk was believed by one family to be caused by witchcraft/evil eye, for which they called a faith healer: For three days, there was no milk secreted from her breasts. We contacted faith healers (Dhami, Dangri) to avoid witchcraft k; four faith healers (two Dhamis and two Dangris) came to our home....We could not save him [the baby] longer; he died on 27 th day. (Interview, family members) Likewise, for another participant, her newborn with complaints of vomiting was believed to be afflicted by a ghost (Bhut) and the forest God (Bandewata). She lost her newborn baby last year: We contacted local faith healers (Dhami and Lama). They said the baby was under the influence of a ghost (Bhut). They also told me to pray to the forest God (Bandewata). But, the baby died early. I couldn't even manage to pray to the forest God. (Interview, mothers) In addition, participants believe that pregnant or postnatal mothers become unwell if they attract the wrath of God while on the way to, or working in the forest for grazing cattle and collecting fodder, firewood and grass. They believe that the forest God could affect a baby in the womb. The God of a family clan (Kul Dewata) and astrological hindrance (Graha), which is the influence of heavenly bodies, is also believed to inflict sickness on mothers and babies. After her three perinatal losses (1 stillbirth and 2 neonatal deaths), one interviewee mother described how her family worked to make their God happy to save babies: We worshipped to overcome astrological hindrance (Graha), contacted faith healers (Dhami) from around the villages to know why I continued losing my babies. We Table 1. Sickness types reported by women and families in the villages. --- Sickness types Beliefs Type 1 Sickness: God as a cause and alleviator/cure of sickness worshipped the God of my parents' home (Maiti Dewata) as well as the God of my in-laws' home (Poili Dewata). We tried our best to worship and pray to God. I don't know why I continue to lose my babies. (Interview, mothers) Women believe that the unhappy God of their family clan (Kul Dewata) could bring disgrace to the land and the site of their house. One mother also consulted the faith healer when her newborn baby was bleeding from the umbilicus. However, her baby died on the 11 th day after birth: Both the faith healers (Dhami and Lama) were right about me; there is something wrong with this house [the site of the house]. I lost my two children here. They said, 'God (Dewata) of this house is unfavourable to you'. Like they said, both my children died at this house. (Interview, mothers) They also believe that an angry God can affect people through the hungry dead spirits (Muiya) of the deceased family members or close relatives. A mother-in-law described the deaths of her two grandchildren (newborn babies) from this cause: It was not a sign of luck. We contacted local faith healers (Dhami, Dangri) to offer the spirit (Muiya). Yet the baby did not survive. Nothing worked. (Interview, family members) The health seeking behaviour of families is strongly influenced by their beliefs about God's will in disease and death. For an unhappy God, the medicine from a health facility is perceived to have no effect, hence participants are reluctant to contact health workers and believe that seeking care from a health facility could even be harmful. A young husband aged 18 years stated: Here in our tradition, if it [disease] is due to God, medicine doesn't work at all. If it was due to God, and they took medicine, it would further harm. (Interview, family members) Type 2 sicknesses: God as a cause, and sickness alleviated through God and local traditional therapy Women and families refer to God as a main cause of sickness, yet sometimes they seek combined care including herbs from local herbalists, worship, prayers and Mantra recitations from the faith healers (Table 1). They seek local herbalists, called Baiji to overcome sickness, particularly when the sickness types are perceived to be Taplagne (effect of heat), Banlagne (rainbow attack) or Mojhlagne (weakness of the womb). In these sickness types, God is attributed as a main cause, and they prefer to seek a faith healer and herbalist together. Even if it is the herbal medicine from Baiji (herbalist), they still focus on pleasing God by sacrificing animals and making offerings. A baby is perceived to suffer from Taplagne when found hot (feverish), and having diarrhoea or vomiting: Taplagne makes their body hot and febrile, causes diarrhoea (Chherne), vomiting (Ukhalne), and Pneumonia (Sardi). A faith healer, who practices both as a faith healer and a herbalist (Baiji) treats baby with local herbs, throws holy grains (rice) and water over the sick baby's body, and prays and worships God. Pastes made from herbs are applied on the baby's head and body. (Interview, mothers) A sickness is attributed as Banlagne (rainbow attack) when they do not have any other explanation for babies' deaths, such as death soon after birth. One of the participants described: My neighbour also lost her baby boy due to Banlagne (rainbow attack). Experienced herbalists can treat it. We need a broom, Khukuri (a knife) and a bird feather (Garud) to worship God to alleviate Banlagne. (Interview, mothers) When a woman is believed to be suffering from Mojhlagne (weakness of the womb), she is considered more vulnerable to continuous baby losses in pregnancy or soon after birth. The participants believe that this occurs due to God's curse, and believe that it can be treated by transferring it to specific plants or fruit trees with the help of an experienced faith healer and herbalist. 2.3 Perinatal losses: Repeated pregnancies and aversions to family planning-God's will?. Women suffering perinatal losses and stillbirths go through many pregnancies. In the remote mountainous region of Nepal, where the women are often poor, malnourished and less educated, such repeated pregnancies at short intervals are liable to end in stillbirths, or perinatal deaths if the pregnancy goes to term [55][56][57][58]. The use of contraceptives that can prevent frequent pregnancies and consequently may assist in preventing perinatal deaths is believed to be against God's will. This is particularly related to vasectomy, the male method of sterilisation. Participants believe that if they have a vasectomy, their God will be unhappy and bring bad luck to their family, such as disease or death of family members; this includes pregnancy loss and newborn death, as well as damage to livestock and property: One mother, aged 32 lost seven children (including stillbirths, neonatal, infant and toddler deaths) out of her 10 pregnancies. Her husband shared his fear of vasectomy, believing that his family God (the Kul Dewata) does not favour it....because of our family God (Kul Dewata), the operation [vasectomy] won't suit me. My father had the operation, then this didn't work, he died when he was just 42, very young. He was very young. (Interview, family members) It is to be noted that ongoing perinatal deaths in these areas are also associated with gendered expectations about a girl and a daughter-in-law. Young girls are viewed to be secure and settled in the in-laws' family by giving birth to a baby who lives, preferably a baby boy. The context of high mortality in the study areas exerts a pressure to repeatedly conceive. Hoping to have a baby who lives, preferably a baby boy, young women have repeated pregnancies with shorter birth intervals. The weak social position of a daughter-in-law (pregnant woman) renders them powerless to decide on how many children to bear and when. They are often controlled by their mothers-in-law and husbands about pregnancy and childbirth matters. --- Karma (past deeds), Bhagya (fate) or Lekhanta (destiny): Ways of rationalising perinatal deaths According to the Oxford dictionary, the Hindu or Buddhist notion of Karma is defined as "the sum of a person's actions in this and previous states of existence, viewed as deciding their fate in future existences". The Indian scholar, Krishan [59] described Karma in the Hindu religious sense as an 'action potential' manifesting into a result or consequence which influences an individual during her/his next life. Likewise, the nearest English translation of Bhagya refers to fate. Similarly, the English translation of Lekhanta means 'already written' or 'pre-destined'. In these study villages, Karma or fate as reasons for perinatal death is attached personally to a woman's Karma or fate. It is believed that one's good Karma in the past yields good results. If one had bad Karma in the past, the results would be bad, which s/he has to experience during the present lifetime. One 35 year-old's story, who is currently pregnant with her 10 th child, shows how she related Karma and fate to her repeated pregnancies and the deaths of her babies: I lost these babies [her three children]. Had they survived, why should I have had too many births? I am unlucky, this is my fate (Bhagya); this is my Karma. (Interview, mothers) By Bhagya (fate), the participants in this study referred to the current state that a woman has been facing, and which is perceived as a result of one's Karma, therefore the word Bhagya is often used together with Karma. A baby's death is frequently linked to a woman's fate, which in turn is thought to map out from her Karma. The use of the term 'fate' indicates a stronger belief about the lack of control over babies' deaths. A 31 year-old mother did not see any possibility of preventing the death of a baby against her fate: How can we stop this [a death of a baby]? We can't prevent a man dying and a river flowing. If the baby is not in your fate (Chado), s/he will certainly die. Look, these other children, they are here. That baby was not in my fate (Chado), and passed away on the day after birth. (Interview, mothers) A woman's Karma or fate is perceived to be strongly related to her experience of perinatal deaths, particularly when she experiences continuous losses, usually the loss of two or more babies. Karma and fate is often perceived personally, as a woman's personal fate. Believing that this was her personal fate, a 20 year old mother, who lost her 3 babies, even told her husband to marry another woman, and said she would not want him to be sad on account of her personal fate: The perception of Bhagya (fate) is even more complex. A few women perceived fate not only as a reason behind the deaths of their babies, but also as a reason for other, future consequences that they will have to bear. Different from Karma and fate, the participants rationalised destiny solely to a baby's own predestined future. A few women had no idea of the causes of their loss, and simply accepted it, nodding their head on their husbands' comment and staying busy caring for their other babies. One man who lost 2 newborns, is an educated person, works in public office, and still he perceives that it was his babies' Lekhanta (destiny) to die: Our two babies died after birth. What to do [there is no way]. No one can control these deaths. The ones who are to die will die anyway. This was Lekhanta (the destiny) of these babies. The rest of our babies survived. Now, they are growing up. We have a grown up eight year-old daughter, a young boy and a baby girl. (Interview, family members) The belief about destiny implies a complete lack of control over their babies' deaths. Participants likened the babies to a cucumber (Kakadi), which could be picked from the kitchen garden any time before it is ripe: I lost my own children, and also lost my three grandchildren. What disease did those little ones have? There is no other reason, just a destiny (Lekhanta). I don't know, it was neither heat (Taplagne), nor any other problems with these babies. The newborns are like cucumber (Kakadi), they could be picked up anytime as per God's plan. One God gave it and another God took it away. (Interview, family members) During an informal chat in the village, a local faith healer commented that Lekhanta (destiny) is the reason for babies' deaths before Chhaith, a local Hindu ritual celebrated usually on the sixth day after birth: When a woman is hungry, it affects her baby; we call it Hanpiyera (work exhaustion and hunger). The baby dies in the womb. If the baby dies before Chhaith (the sixth day celebration), it is due to Lekhanta (destiny). This is not due to anything related to mother, father and family. (Interview, faith healers) The deaths of babies after birth during a woman's confinement in the birth place, Gotha, are also attributed to destiny. The women and families were more convinced about destiny as a reason for

This paper examines the beliefs and experiences of women and their families in remote mountain villages of Nepal about perinatal sickness and death and considers the implications of these beliefs for future healthcare provision.Two mountain villages were chosen for this qualitative study to provide diversity of context within a highly disadvantaged region. Individual in-depth interviews were conducted with 42 women of childbearing age and their family members, 15 health service providers, and 5 stakeholders. The data were analysed using a thematic analysis technique with a comprehensive coding process.Three key themes emerged from the study: (1) 'Everyone has gone through it': perinatal death as a natural occurrence; (2) Dewata (God) as a factor in health and sickness: a cause and means to overcome sickness in mother and baby; and (3) Karma (Past deeds), Bhagya (Fate) or Lekhanta (Destiny): ways of rationalising perinatal deaths.Religio-cultural interpretations underlie a fatalistic view among villagers in Nepal's mountain communities about any possibility of preventing perinatal deaths. This perpetuates a silence around the issue, and results in severe under-reporting of ongoing high perinatal death rates and almost no reporting of stillbirths. The study identified a strong belief in religio-cultural determinants of perinatal death, which demonstrates that medical interventions alone are not sufficient to prevent these deaths and that broader social

ta). I don't know, it was neither heat (Taplagne), nor any other problems with these babies. The newborns are like cucumber (Kakadi), they could be picked up anytime as per God's plan. One God gave it and another God took it away. (Interview, family members) During an informal chat in the village, a local faith healer commented that Lekhanta (destiny) is the reason for babies' deaths before Chhaith, a local Hindu ritual celebrated usually on the sixth day after birth: When a woman is hungry, it affects her baby; we call it Hanpiyera (work exhaustion and hunger). The baby dies in the womb. If the baby dies before Chhaith (the sixth day celebration), it is due to Lekhanta (destiny). This is not due to anything related to mother, father and family. (Interview, faith healers) The deaths of babies after birth during a woman's confinement in the birth place, Gotha, are also attributed to destiny. The women and families were more convinced about destiny as a reason for a baby's death when the baby died after seeking help from faith healers. --- Discussion This study has examined the religio-cultural contexts surrounding perinatal mortality in two remote mountain villages of Nepal. These villages were selected because they ranked lowest in terms of development and child survival and provide rich evidence of local ways of understanding and responding to the very high levels of perinatal mortality, even though they might not be representative of all the villages in the region. The collective experience of perinatal death in the community has shaped the villagers' construction of perinatal deaths as inevitable experiences which modern healthcare systems cannot address. Some people believe that modern healthcare may even make the situation worse. Individual and collective experiences are viewed as valid sources of knowledge in the social constructionist view [60]. In this study, the past experience of women and the experiences of older family members construct the knowledge of younger women. The persistent occurrence of perinatal deaths is considered as a 'generational continuum', bound to occur with everyone. --- Acceptance of ongoing perinatal deaths This study has identified acceptance as the norm when a baby dies before term (Hunemahina), as a stillbirth (Hudaimareko) or during a mother's and baby's confinement after birth until the third postnatal week. In this study, the women's and families' description of perinatal deaths as Pakhalajane, or Aadanjhadne indicates premature deaths before term are not viewed as lives lost, but perceived simply as a diarrhoeal condition as if women were emptying their bowel. Hence, these are not considered worthy to report to authorities, nor to be mourned. Such perceptions are similar to the views of Hmong women living in Australia who consider stillbirths and neonatal deaths as non-significant events [61]. A recent study from the rural Amhara and Oromiya regions of Ethiopia [62], a study from Tanzania [63] and a study from Uganda [30] have described similar perceptions of low social significance given to stillbirth and neonatal death, as they are considered as deaths of non-humans, deaths of spirits, or events not worthy of sharing with others. A sense of stigma about perinatal deaths has been reported by some studies in African [29,63], Asian [64,65] and migrant Australian communities [27,28]. These studies have discussed stigma as one of the key factors making perinatal deaths invisible in communities [9,63,66]. However, in the present study, such deaths did not appear to carry any stigma; they were simply considered to be of low importance, and therefore not worth reporting. The common experience of perinatal deaths among every family, who also view these occurrences as generational continua, has led them to believe that these occurrences are so common that no one pays any attention to them. Another reason for not attaching importance to such deaths is the lack of any ritual significance of a stillborn baby or neonatal death. A perinatal death is considered a mere biological loss not requiring any death rituals. Such a construction does not consider perinatal death as the loss of a social individual, which means no attention is paid to improving perinatal survival in the study villages. Loss of these early lives is more a question of a nonvalue attributed to their personhood in these villages, reminiscent of Aries' argument in relation to the social construction of childhood that parents will not respond too emotionally to infants who might die early, and hence consider them as "neutral" (or non-person) for some time after birth [67]. This loss is also not considered a psychological burden to women and families to attract a stigma for having experienced such losses. Stigma-related burden is almost non-existent in these communities also because of beliefs about God and Karma in illness causation and death. This indicates a level of acceptance of perinatal deaths to such an extent that parents and families see no reason to take any measures to prevent such deaths. --- Dewata (God) as a factor in health and sickness People believe that God exerts a powerful influence on illness causation and response to recovery. In these villages, God is omnipotent and affects every aspect of day to day life, and is not just a symbolic statue kept inside the house [23]. Such a worldview of God shares similarity with the traditional Akan religious worldview about health and sickness as described in a study from Ghana [68]. The indigenous Akan people in Ghana, and people in Tehuledere region of Ethiopia [69], believe in a host of human (witch, sorcerer), non-human (ghost, ancestor, evil spirit) and supernatural beings (God/Allah and deity) having capability to affect their lives positively and negatively. In the study villages in Nepal, common illnesses of mother and baby are also believed to be due to the displeasure of God, described under the generic name Dewata, and are believed to operate in different forms through human and non-human agents. The study also showed the villagers having their own illness language broadly categorised into two types: (i) illness solely attributed to God, where the cures are shaman healings, prayers, animal sacrifice, offerings, worships, mantra recitations, sprinkling holy waters, exorcism and amulets; and (ii) illness in which God is attributed as cause, yet the care combines the former approaches with the use of local herbs. In both illness types, local faith healers are the key care providers because they are believed to possess Godly powers and have the ability to invoke the mercy of God. Other studies from Nepal also describe how villagers seek the help of traditional healers for general health problems [70,71]. Beliefs about perinatal sickness and death due to supernatural forces are also revealed in other studies in Africa and among Hmong women [31,61,62]. What is added from the present study is that not only baby's illnesses but also mother's common illnesses, birth complications and contraceptive norms are considered to be in God's control. In the villages studied in this research, God is believed to be both a cause and a cure for mothers' and babies' problems along the pre-pregnancy to postnatal continuum. Birth complications which require skilled attendance or immediate referral are believed to be caused by God's wrath, so that the family will invite faith healers to perform exorcism, prayers and offerings to God until the last minute. Importantly for healthcare provision, the study identified the belief that for a sickness due to God's wrath, going to health facilities might negatively impact on the cure. Health workers are therefore consulted only when none of these work, and usually only after permission is obtained from the faith healers. --- Rationalising perinatal deaths as Karma (past deeds), Bhagya (fate) or Lekhanta (destiny) Fatalism has been described by studies in different South Asian countries as a means to accept various events, resulting in an inertia where attitudes and behaviours perpetuate the occurrence of such events. Outsiders might view these as "risky" behaviours which must be changed to prevent the occurrence of events such as perinatal deaths. An Indian study [72] showed that belief in Karma, sin and God's punishment were described as reasons for Leprosy by about two-thirds (65%) of the study's interviewees. Fatalistic beliefs about infant deaths are identified in other studies, such as in the upper Lombok region of Indonesia where infants died due to simple treatable conditions [65]. A multi-country study from South Asia and Africa records fatalism surrounding treatment of neonatal infections [73], healthcare for small and sick newborns [74] and stillbirths [66]. However, none of the studies referred to here have thoroughly explored the specific socio-cultural contexts of the events investigated in as the way which the present study has. In these Nepalese study villages, the religio-cultural base of women's Karma (past deeds) and fate were interchangeably used as reasons for a baby's death. These beliefs intensified the acceptance and fatalism about perinatal death so that women and families remained passive about seeking healthcare for perinatal sicknesses, contributing to the continuing occurrence of these preventable deaths. One of the most common phrases during interviews and informal chats about why a baby died, was 'What can we do? This was my Karma' (Ke garne, mero karma yestai), reflecting deeply rooted fatalistic beliefs. Such fatalistic beliefs came out even stronger when women rationalised their baby losses as Lekhanta, which they often meant as the baby's personal destiny, thus feeling helpless in doing anything to prevent the baby's death. --- Implications for future health provision Raising critical consciousness about Karma: As a fatalistic belief to empowerment. Max Weber, in his book 'The Religion of India: Sociology of Hinduism and Buddhism' described the fundamental values of Hindu and Buddhist religions, Karma and reincarnation as doctrines of fatalism [75]. Weber states that these values do not talk about this world but about a supernatural world, and that they point to past or future lives. Although Weber's interpretation matches the reality of the villages studied in this research, the concept of Karma may be viewed as a universal law of justice, a law of cause and effect. From the present study, it is argued that the perception of Karma and persistent occurrence of perinatal deaths in the study villages is the result of their false Karma consciousness. Karma is not meant to endure inequity or injustice as the women face in the study area of Nepal. In the Bhagavad-Gita, the fundamental religious textbook of Hinduism [76], Karma is described as a great art of performing action in all realms of thinking, speaking and acting. It is not so much about the past Karma of previous lives, it is mainly about present Karma which an individual is considered to have a control over, hence bringing the power into the individual's authority and will. Therefore, the Karma doctrine is described as bringing awareness about one's actions and bringing back cause and effect into his/her control rather than relying on fate or passively waiting for past karma to map out one's fate. It is meant to empower one from weakness, pessimism and escapism, and to help one remain firm as a Kshatriya, a warrior who is mindful of the realms of his thoughts, speech and actions. The values that Weber described as fatalistic refer to teaching an art of living, and a way to liberate one's life. On this basis, in the study villages an active collaboration could be sought with the local religious leaders and faith healers to correct the misconception (or a general popular conception) of these fundamental values and thereby potentially offering a slight reorientation of belief to local women and families. This argument might also at once reinforce their religio-cultural values yet also avoid the false perceptions and associated fatalistic views which have become popular conceptions in these areas rather than the scriptural ones. In the present study, the acceptance and fatalism related to religio-cultural contexts is an emergent theme analysed through the participants' narratives regarding the experiences and beliefs about their stillbirths and neonatal deaths. The deeply interconnected religious, spiritual and cultural values have been termed together as religio-cultural contexts. One may argue that the religio-cultural beliefs of Karma, fate and God have healed the wounds of women and families and any potential psychological burden from experiencing perinatal losses. However, the continuous human losses cannot be justified on the grounds of human rights, or the right to life of every child [77], nor when almost all of these deaths are most likely preventable. Addressing the false perceptions of Karma and Dewata (God) with due respect to local culture may be a key to the way forward. Using a socio-cultural lens in reaching care: Bridging professional dialogues with lay discourses. Current perinatal survival policy and practices in Nepal describe a predominantly bio-medical discourse to improve survival, with less regard to the evidence regarding local socio-cultural contexts in the mountainous regions [12][13][14][15]. The policies are influenced by overwhelming national and international evidence on epidemiology and bio-medical risk factors of perinatal deaths: stillbirths [9,78], and neonatal deaths [8,79]. The available interventions are largely bio-medically oriented, aimed to prevent deaths from sickness, complications of prematurity, infection and asphyxia, based on a Western medical viewpoint of the causes. By comparison, the lay constructions and beliefs surrounding perinatal sickness and death see loss of early life as generational perpetuation and common experiences; they attribute perinatal sickness (and cure) to God; and rationalise perinatal deaths as Karma, fate and destiny, resulting in the belief that either no-one can intervene or only a faith healer can intervene. Both views lead to formal healthcare either not being sought at all, being sought "too late", or being deemed as actually more harmful. The findings strongly indicate a need for bridging professional discourses with lay discourses. The healthcare systems and policies need to acknowledge and negotiate this in their actions about improving poor perinatal survival. Partnership between health service providers and faith healers. The present study argues that it is not just cultural perceptions of safety with God which Kaphle [23] discussed, but it is a false awareness of God in their religio-cultural context, most likely begun as a moral order to instil discipline which became reinforced by faith healing practices. The belief in God's influence in health and sickness has been a predisposing factor preventing families from seeing the medical severity of any sickness or childbirth complications of a mother or baby. Eventually, this influences families in who to choose as their healers [80]. They therefore choose traditional healers as the medium to invoke God, thus delaying, or indeed often preventing, them from seeking skilled and timely care from "formal healers" (in the healthcare system). Furthermore, the perception of religious beliefs about God and Karma are popular concepts rather than scholarly concepts. On this basis, the present study urges health professionals to work in mutual collaboration with faith healers and religious leaders so that a mother and baby could access timely healthcare for any sickness and could be saved from simple avoidable causes, and at the same time also continue to observe their religio-cultural duties. Such interventions have been introduced in Indonesia, where trained midwives work in collaboration with traditional birth attendants to attend women's delivery and both provide for the emotional and cultural needs of local women, refer pregnant mothers, and provide postdelivery services in the communities [81]. There has been a growing realisation for the need to integrate spirituality in medicine/ healthcare [82,83]. This has been considered especially important in addressing the religiosity of patients in societies with diverse faiths. McCormick and Min [83] suggest that a spiritual history should be recorded about every patient as a part of his/her general medical history so that any religious/cultural/spiritual beliefs can be understood, and utilised as a resource for the healthcare, support and wellbeing of the patient. Although the study villages are not open societies with diverse faiths, the engendered context of the perception of God as an aetiology of sickness, and the religious value of their Karma being seen as a reason for perinatal deaths, indicates a lack of consideration of religio-cultural factors in the formal health system in the villages. It is imperative for primary healthcare workers to understand and address such beliefs in these communities in a culturally safe way, not merely focusing on instructing about "danger signs" from a medical perspective during pregnancy, birth and postnatal period to a handful of women attending health facilities. Rather, it would be imperative to revise the entire curriculum for the training of doctors, nurses, midwives and health workers to include training to address religio-cultural issues in health in specific communities. Intensifying community engagement: Revisiting the contents of behaviour change communication. Ending preventable stillbirths and neonatal deaths is an international goal [84]. A range of studies [85][86][87] have suggested a long list of family/community and health facility based interventions to prevent these deaths. Although these interventions are typically characterised as family and community based, they are often prescriptive and do not understand the context and awareness level of women and families in the communities. Women's and families' construction of perinatal deaths as a natural event of low social significance means that they pay less attention to ongoing perinatal deaths. On the other hand, this indicates that they most likely have a low level of awareness and behaviour change for adopting healthy perinatal care (seeking formal health care), even though the national policy strategies [12,13] are aimed at such behaviour change. The key bio-medical focus of the policy discourse about causes of stillbirths and neonatal deaths (infection, asphyxia and intrapartum complications) is intended to limit the content of behaviour change to conveying the knowledge of danger signs (during pregnancy, delivery and postnatal) to mothers. However, this does not link to local understandings of what constitutes danger. Local religious/faith healers could be approached to ask about the possibility of supporting women and families to explore and discuss their constructions related to personhood status of a stillborn and a newborn baby (loss of early lives as natural events of low social significance), illness causation (God) and their rationalisation (Karma). Assisting communities in this way could be a key focus and content of family and community based behaviour change interventions and birth preparedness packages. Otherwise, the prescriptive list of interventions or their packages (newborn/child intervention packages) alone are likely to be much less effective, and this could be one of the reasons for its low impact in Nepal's newborn intervention package [21,88]. Addressing fatalism as a systemic issue. M<unk>lqvist [89] described the invisibility of neonatal deaths in a northern province of Vietnam due to a dysfunctional reporting system. A study from India also described that reporting the actual number of deaths is avoided by health providers as deaths could be judged as being due to their poor performance [90]. The present study suggests that the invisibility of perinatal death in this study's remote mountain villages is reinforced by the fatalistic attitude towards perinatal deaths, not only in the community but also in the local health system. The sum total of deaths reported by participating women from the study villages provided a larger number of deaths than the sum total of the perinatal deaths reported by the District Health Office for the last four years across the 24 villages in the district [54]. This underreporting seems likely due to the low social significance accorded to perinatal deaths. Hence, the service providers and female health volunteers did not find it worth inquiring and reporting perinatal deaths. This systemic fatalism (including among community health workers) has contributed to the invisibility which is certain to reduce the efforts to implement perinatal survival programmes by the local health system. Such rural areas as in Nepal are therefore likely to be overlooked. Methodological implications. The study has also a methodological advantage. A large body of evidence in perinatal survival research is based on structured surveys and verbal autopsy [91] to describe a pattern of mortality and service utilisation across socio-demographic determinants and medical causes of deaths. By comparison, this study has examined the influence and interactions in socio-cultural contexts that lead to perinatal deaths. Recently an interest is growing about the need for social autopsy, which analyses death narratives and focuses on identifying social aspects impacting perinatal deaths [92]. This research has found that indepth qualitative interviews with women and their families who have experienced perinatal deaths are possible and are an appropriate way to uncover the influences of the local socio-cultural context that contribute to poor perinatal survival. In addition, women and families in the villages indicated that being interviewed made them feel respected and valued their experiences. The limitation of this study is that the interview participants were selected purposively; the views of others may be different, such that the study lacks a wider generalisability in other regions. However, the study has provided a thorough examination of the religio-cultural context of perinatal death in the study region and has identified policy implications in implementing interventions to address the ongoing high stillbirth and neonatal death rates in rural villages. --- Conclusion Perinatal deaths are regarded as common occurrences, and religio-culturally constructed as the deaths of foetuses and neonates which have no personhood status or value. Mothers' and babies' sickness and recovery is attributed to Dewata (God)'s will, while Karma (past deeds), Bhagya (fate) or Lekhanta (destiny) are perceived as reasons behind perinatal deaths, which are religio-cultural beliefs deeply rooted in the mountain communities. These fatalistic beliefs are perceived as culturally safe, yet pose a high risk to the survival of babies when the bio-medical evidence suggests that 99 percent of perinatal deaths are preventable. The persistence of fatalism also raises the question of whether perinatal survival interventions have effectively reached the rural communities and villages. The findings of this study strongly indicate the need to invite discussions on the construction of personhood and social significance of foetus and newborn babies in the current behaviour change discourse. The findings further call for starting partnerships of medical health practitioners at primary health care level with faith-based healers and religious figures so that ways can be found to respect the lay worldviews of Dewata and Karma in health and sickness without these continuing to perpetuate poor perinatal survival in the mountain villages. --- All relevant data are within the paper and its Supporting Information files. --- Writing -review & editing: Mohan Paudel, Sara Javanparast, Gouranga Dasvarma, Lareen Newman.

This paper examines the beliefs and experiences of women and their families in remote mountain villages of Nepal about perinatal sickness and death and considers the implications of these beliefs for future healthcare provision.Two mountain villages were chosen for this qualitative study to provide diversity of context within a highly disadvantaged region. Individual in-depth interviews were conducted with 42 women of childbearing age and their family members, 15 health service providers, and 5 stakeholders. The data were analysed using a thematic analysis technique with a comprehensive coding process.Three key themes emerged from the study: (1) 'Everyone has gone through it': perinatal death as a natural occurrence; (2) Dewata (God) as a factor in health and sickness: a cause and means to overcome sickness in mother and baby; and (3) Karma (Past deeds), Bhagya (Fate) or Lekhanta (Destiny): ways of rationalising perinatal deaths.Religio-cultural interpretations underlie a fatalistic view among villagers in Nepal's mountain communities about any possibility of preventing perinatal deaths. This perpetuates a silence around the issue, and results in severe under-reporting of ongoing high perinatal death rates and almost no reporting of stillbirths. The study identified a strong belief in religio-cultural determinants of perinatal death, which demonstrates that medical interventions alone are not sufficient to prevent these deaths and that broader social

replicate all of aforementioned stylised facts simultaneously. It also produces social networks that exhibit salient features of real-world networks; namely, they cannot be 1 Introduction Conventional modern macroeconomics has long recognised the crucial relevance of expectations and belief-formation for aggregate dynamics (Gal<unk> 2015). In particular, beliefs about economic inequality and perceptions of social hierarchy can inform individuals in such diverse fields as consumption decisions (Duesenberry 1949;Veblen 1899Veblen [2001]]; Frank et al. 2014), redistributive preferences and voting behaviour (Gimpelson & Treisman 2018;Kim et al. 2018;Choi 2019) or subjective well-being and ethical convictions (Kuhn 2019;Clark & Senik 2010). Even in the most sophisticated behavioural models, belief-formation is, however, typically either assumed to be atomistic (Gabaix 2020) or does not systematically account for the impact of individual embeddedness within heterogeneous social contexts on those beliefs, even if social interaction is explicitly modelled (Flieth & Foster 2002;Lux 2009). We propose a parsimonious network-based model for the interaction of macro-level inequality, micro-level beliefs and the mediating effects of heterogeneous social contexts. In contrast to the assumption of deductive reasoning in orthodox models, we build on the empirically well-established notion that economic agents reason inductively and generalise from finite samples. Recent theoretical and empirical work has demonstrated the potency of this approach in explaining phenomena in such diverse fields as human probability assessment (Sanborn & Chater 2016;Chater et al. 2020) or regional inequality (Collier & Tuckett 2020). The model is both consistent with several stylised facts about inequality perceptions and the micro-level evidence on the composition of social networks. The relevance of individual beliefs is perhaps best exemplified by spelling out its political economy implications. Across income groups and countries, the public perception of economic inequality and many other macroeconomic variables is empirically wrong, often spectacularly so. Errors in those beliefs might be due to conceptually different problems: uninformed beliefs or misinformed beliefs (Kuklinski et al. 2000). Uninformed voters are ignorant about the actual state of affairs, while misinformed voters' beliefs are consistently deviating from it in one direction. The distinction is a crucial one. Uninformed voters' beliefs would cluster around the actual state of affairs and, with no systematic deviations, be correct in expectations. For uninformed voters, we only need one informed voter to tip elections under majority rule into the correct direction; a majority of ignorant individuals might nevertheless vote for the correct policy, which is now known as the'miracle of aggregation' (Page & Shapiro 1993). However, his miraculous aggregation breaks down when we consider misinformed rather than uninformed voters with beliefs that are no longer randomly distributed but consistently tend in a (false) direction (Caplan 2011). The type of error in perceptions is thus intimately linked to the efficacy of democratic systems. For inequality perceptions, beliefs appear to be indeed the result of misinformation in this technical sense and they are consistently biased across income groups and welfare regimes. In contrast to much of the behavioural literature, we refrain from ad-hoc assumptions about possible biases, e.g., assuming that individuals tend to perceive themselves in the middle of social hierarchies (cf., e.g. Knell & Stix 2020). Instead, we assume unbiased information processing capabilities for all the economic agents. Information is, however, asymmetric and agents form estimates about aggregate variables according to their local information. We show that a parsimonious process can generate sufficiently skewed information sets to replicate the aforementioned stylised facts and generate perceived inequality levels that are quantitatively in line with recent empirical evidence for a large sample of 32 OECD countries (Choi 2019). In essence, we assume that agents (correctly) observe inequality within their local social network and (correctly) form estimates about the total population from them but still generate biased perceptions due to their network contacts not being representative for the overall population. Employing a new variant of a random geometric graph network, the assumption of income homophily alone can generate substantial misperception in line with the empirical evidence. The derived network topology also corresponds to empirically observed social networks across the world and features a small-world structure. Given the ubiquity of these topological features, our homophilic process appears to be a plausible candidate to explain the equally ubiquituous inequality misperceptions. Our contribution is thus threefold: Firstly, we compile a list of four stylised empirical facts about income inequality and its biased perception from the nascent literature on the matter. Secondly, we develop a model that simultaneously replicates these stylised facts building on homophilic linkage and unbiased individual estimates based on local signals. Our model is quantitatively consistent with empirical estimates regarding both the input income distribution and the output perceptions, in contrast to the somewhat stylised models in the extant literature that also fail to replicate the dynamic behaviour of perceptions in response to changes in actual inequality. Thirdly, the network-formation algorithm presents a novel way of generating random geometric graph types of networks which is more intuitive for many application scenarios and allows specifying a minimum degree. The remainder of this paper is organised as follows: Section 2 extracts four stylised empirical facts about inequality perception; furthermore, it reviews the evidence on empirical network topologies and individual belief formation within networks. Section 3 introduces the basic model of homophilic graph formation, reviews the main mechanisms generating heterogeneity in information sets, abd validates the model. Section 4 presents our analytic and simulative results, shows that they are consistent with the outlined stylised facts regarding network topologies as well as inequality perceptions and derives some important implications regarding heterogeneous segregation patterns across the income distribution. Section 5 concludes and discusses several promising avenues for further research, especially regarding consumption dynamics and voting behaviour. --- Related Literature Our model joins three different strands of literature. Empirical findings on inequality perceptions that a single theory or model has not yet explained constitute its main explanandum. As explanans, we develop a network model featuring the current state of research into both the social network structure of empirical networks, mainly their homophily and small-world character, and individual perceptions in networks. The family of random geometric graphs constitutes the third strand of literature as a promising methodological choice in Section 3. --- Stylised Facts on Inequality Perceptions and Middle Class Bias The empirical literature has identified four particular stylised facts for any theory of perceived inequality to be evaluated against: (i) Irrespective of their objective status, all individuals perceive themself to be in the middle of the social hierarchy (Kelley & Evans 1995;Evans & Kelley 2004); (ii) as an immediate corollary of (i), poor individuals overestimate their social position, rich individuals tend to underestimate it (Knell & Stix 2020); (iii) poor individuals tend to perceive inequality to be higher and are closer to objective inequality on average (Osberg & Smeeding 2006;Newman et al. 2018) and (iv) the evolution of objective inequality is detached from the evolution of subjective inequality, that is, increases in objective inequality do not necessarily increase perceived inequality (Kenworthy & Mccall 2008;Bartels 2018;Gimpelson & Treisman 2018;Hvidberg et al. 2020). The ubiquity of misperceptions across states and welfare regimes calls for a common mechanism independent of differences in actual inequality or institutional framework. By way of illustration, we show the empirical frequency of self-perceptions to enable comparisons to the model output in Section 4 for stylised fact (i). The data shows the empirical frequency distributions of self-reported income deciles for Germany and the US from the 1987, 2007 and 2014 wave of the ISSP Group (2016).1 Typically, Germany is considered to be the epitome of a coordinated market economy, while the US represents a liberal market economy (Hall & Soskice 2001). By contrast, the qualitatively similar densities for both countries show that the mechanism behind misperceptions should be independent of the specific welfare regime. Especially the'middle class bias' in the self-perceptions is rather striking. For a representative survey like the ISSP, each decile should, per construction, include exactly 10% of observations and the frequency densiy should therefore exhibit a uniform density at 0.1. Instead, the frequencies display a marked peak at the middle categories, i.e., most Germans and most Americans tend to think they are middle class, even though they are objectively not. This finding holds for all considered countries in the ISSP and all considered years, apart from the three periods and two countries we selected for illustrative purposes (Choi 2019). 2 Notice, however, that the qualitative middle class bias manifests itself in quantitatively rather different frequencies throughout time and between countries. We focus on the phenomenon that is common to all considered densities, i.e., the qualitative middle class bias and leave the direct quantitative calibration of our model for further research. We find this middle class bias to be the major driver of our results, implying the other stylised facts (ii) -(iv) directly but emerging endogenously in our network model. The relevant features of this network are discussed in the subsection below. --- Empirical Social Networks Empirical networks exhibit ubiquitous and salient features that can serve as stylised facts to guide the validation of proposed theoretical graph formation processes. Probably the most prominent one is the small-world property, indicating that paths between nodes in realworld social networks are unexpectedly short. At the same time, those networks also feature high degrees of clustering. Small-worldiness has obvious implications for any contagion process, be it rumours, diseases or information, where contagion across the whole network happens much faster than our intuition would suggest (Watts 1999;Moore & Newman 2000;Kleinberg 2001). The empirical research has identified small-world features across many different social groups, including friendship networks in schools (Weeden & Cornwell 2020), corporate board networks (Kogut & Walker 2001;Borgatti & Foster 2003;Davis et al. 2003;Conyon & Muldoon 2006;Galaskiewicz 2007) and scientific and artistic collaboration (Watts & Strogatz 1998;Newman 2001;Uzzi & Spiro 2005). Given this ubiquity, it appears safe to say that a graph-generating process for social networks needs to simultaneously produce low average path lengths and large degrees of clustering to be consistent with this stylised fact. We situate our model in the random graph literature, where graph formation happens according to a stochastic process and is not the result of deliberate optimisation. Random graphs have been very successful in replicating structural stylised facts about network topologies, with the Watts-Strogatz model famously able to replicate those small-world properties (Watts & Strogatz 1998). Since the graph-generating process is, however, stochastic in nature, it does not feature explicit behavioural microfoundations. Therefore, we extend the purely stochastic notion with a behavioural ingredient, notably, that link-formation is homophilic. First introduced by Lazarsfeld et al. (1954), another salient feature of empirical social networks is homophily, the tendency of similar individuals to connect with each other. This tendency is not only an empirical curiosum but has relevant theoretical implications, e.g., for information transmission, where homophilic segregation can severely slow the speed of learning (Golub & Jackson 2012) or diminish the attention members of minority groups receive (Karimi et al. 2018). The similarity can come in many dimensions such as gender, ethnicity or socio-economic status (McPherson et al. 2001). We focus on the latter in the narrow sense of economic homophily, which is empirically well-established. One strand of literature focuses on friendship networks in schools and colleges and provides evidence significant homophily according to income or social class (Cohen 1979;Huckfeldt 1983;Mayer & Puller 2008;Boucher 2015;Malacarne 2017). Even one of the earliest contributions in the field, however, shows that homophily in socio-economic status is not fixed in time and varies with cultural norms and the importance of class distinctions (Cohen 1979). In light of this result, it appears unsurprising that we find considerable variation in implied degrees of country-level homophily in our model, perhaps reflecting cultural norms not in the structure but the degree of the graph formation process. Even for a constant degree of (economic) homophily, increasing inequality also increases segregation in the population, as then the relative distances in incomes increase and agents becoming more selective in choosing links. For geographical segregation, this effect is empirically documented (Reardon & Bischoff 2011;Chen et al. 2012;Tóth et al. 2021), testifying indirectly to the relevance of homophily as a graph-generating feature. Homophily in income also exists in social media friendship networks (Lewis et al. 2012, for a large sample of Facebook friends), where spatial segregation should not confound findings and becomes apparent in the choice of romantic partners, typically under the label of 'homogamy' (Kalmijn 1991;Kalmijn & Flap 2001). Finally, a very recent contribution by Cepi<unk> & Tonkovi<unk> (2020) for a representative sample of Croatian adults finds evidence for homophilic tie formation according to social class and income, with however considerable variability in cross-class ties, hinting at possible confounding factors we aim to capture with a parsimonious stochastic process. Importantly, Cepi<unk> & Tonkovi<unk> (2020) show that there is also strong variation in cross-class ties, though, which we show might be crucial for unbiased individual inference. A major predictor is there shown to be cross-class sociability is political participation which would, in our model, amount to the prediction that politically active individuals are also better informed on average which is indeed perfectly in line with the evidence (Campbell 2013, for a recent survey). --- Belief Formation in Networks The literature on belief formation itself appears to be much more scarce than the literature on the effects of perceptions and misperceptions. While very different in detail, the two currently dominant theories of public opinion formation suggest that the beliefs an individual holds are in a broad sense averages over the idiosyncratic messages they receive (Zaller 1992; Lodge et al. 1995). 3 This literature has focused on the specific 'averaging' individuals use to process their information sets. Regarding perception formation about societal inequality, there exist to the best of our knowledge only two analytical models so far, namely Knell & Stix (2020) and Iacono & Ranaldi (2021). Both derive biased perceptions under very restrictive assumptions, though, and need to impose some kind of 'biased averaging'. Knell & Stix (2020) assume that agents form subjective income densities over the whole support of possible income levels but let these densities be self-centered such that the mode of the distribution corresponds to their respective own incomes. While they partially succeed in replicating the four stylised facts on inequality perceptions at least qualitatively, their assumption essentially imposes the middle-class bias of stylised fact (i) ex ante and not as an emergent outcome from first principles. Iacono & Ranaldi (2021) also impose potentially biased perceptions and assume that agents only observe their own incomes as well as the minimum and maximum incomes. They continue to derive several important results on voting behaviour and show that information treatments on inequality might alter preferences for redistribution, therefore qualitatively replicating this stylised fact from the empirical literature. While the notion of local information sets appears appealing, using only the minimum, maximum and own income strikes us as unnecessarily artificial and implausible. Most importantly, being static, both models fail to make sense of the -arguably most relevant -stylised fact (iv), i.e., inequality perceptions being very persistent and not responding to changes in actual inequality. Our model is featuring this persistence by exploiting the fact that rising inequality also features rising segregation, as is also empirically established for geographical segregation (cf. the evidence discussed in subsection 2.2). We develop on the notion that information is local but assume unbiased processing with skewed information sets as implied by the well-documented homophilic social network formation on which we expand in Section 3. The psychological literature on'social comparison theory' (Festinger 1954) supports the notion that individual self-perceptions are much more responsive to local knowledge about small groups than to information about aggregates, e.g., knowing the population average (Buckingham & Alicke 2002;Zell & Alicke 2009;Alicke et al. 2010). Thus, belief formation about inequality appears to be indeed primarily based on local knowledge. This does not imply, however, that informational treatments in the form of reported averages do not change beliefs at all. Providing information about the actual degree of inequality seems to exhibit a significant effect on redistributive preferences for Argentina, Sweden and the US (Cruces et al. 2013;McCall et al. 2017;Karadja et al. 2017), with however small and insignificant effects for Germany (Engelhardt & Wagener 2018). Finally, two recent studies for the whole of Europe and Denmark separately demonstrate that individuals indeed tend to know the income levels of their immediate friends and family rather well, with non-negligible effects on inequality and fairness perceptions as well as perceived social positions (Clark & Senik 2010;Hvidberg et al. 2020). --- Model Apart from this indirect evidence from informational treatments, there also exist several studies that measure the impact of local exposure to inequality on perceptions and redistributive preferences directly, therefore offering also direct evidence for our proposed mechanism. 4Thal (2017) demonstrates using a large-scale survey (Soul of the Community (SOTC)) that affluent Americans' perception of social conditions is largely based on extrapolation from their own neighbourhood, as the affluent within homogeneous and isolated neighbourhoods perceive social disparities to be significantly less severe. Dawtry et al. (2015) find robust evidence for the US and New Zealand that individuals base their estimate of average societal income (and other quantiles) largely on their immediate subjective experience or'social sample'. This leads to differences in fairness perceptions and redistributive preferences, whenever the composition of social circles varies by income, as homophily strongly indicates. Kraus et al. (2017) finally demonstrate with respect to racial economic inequality that the homogeneity of the immediate social network appears to mask racial inequities, therefore also testifying to the relevance of immediate lived experience for perception formation. In a series of articles close in spirit to our approach, Chiang (2011Chiang (, 2015a,b),b) exploits this notion and shows experimentally and computationally that individuals base their beliefs about inequality on local perceptions within referent networks and that income homophily has a potentially strong effect on those perceptions. While his approach is exploratory and does not account for the outlined stylised facts on inequality perceptions and empirical social networks, we provide a tractable model, readily calibrated with regards to those phenomena that is introduced below. --- Model This section provides a content-oriented presentation; a technical description following the ODD protocol is avaliable upon request. The model consists of three distinct phases run in sequential order: 1. Agent initialisation and income allocation 2. Network formation 3. Gini perception and network evaluation Each phase runs only once and phases one and two build the structure which phase three then analyses. This sequence implies that during network generation, agents adapt to others' income level. However, there is no reaction to others' linking behaviour or perception and, Model thus, the model does not feature interaction in a narrow sense. Moreover, in the model, an agent's social contacts depend on their income. We choose this direction of causality for technical reasons and because it seems empirically likely (cf. Section 2). Nevertheless, our process scheduling would also be consistent with the opposite direction of causality or positive feedback effects between income and social contacts. The model is designed that way because it focuses entirely on income perceptions given defined income distributions and network structures. Hence, both an agent's income and their social contacts remain constant for the evaluated time frame or, put differently, that the simulation outcome is a snapshot of a certain point in time. --- Agent Initialisation and Income Allocation There are 1, 000 agents in the model; each agent draws their income from an exponential distribution with a mean of <unk> = 1. Such a distribution normalises the empirical observed (pre-tax or market) income distributions in various industrialised countries for the vast majority of individuals (Dr<unk>gulescu & Yakovenko 2001;Silva & Yakovenko 2004;Tao et al. 2019). Thus, one can understand the model population as constituting a representative sample of empirical populations of these countries. The upper tail of 1 to 5 % of the income distributions empirically follows a Pareto law (Silva & Yakovenko 2004). We deliberately choose to exclude this small minority from our model, since their population size would induce another degree of freedom in our model and we want to demonstrate that segregation is indeed endogenous and not driven by differences in actual income regime. We use an identical, pre-validated exponential distribution for all Monte Carlo runs and also all levels of homophily to ensure comparability between simulation runs. Agents store their true income decile for evaluation purposes, too. --- Network Formation Each agent draws five other agents to link to. Like for real-world networks, links are therefore created by agents, not imposed on them. The number of five link choices is also empirically validated, as humans tend to only know the income of close friends or family (Clark & Senik 2010;Hvidberg et al. 2020), with typically only five individuals at this closest layer of emotional connection (Zhou et al. 2005;Hamilton et al. 2007;MacCarron et al. 2016). 5The relative weight in the draws are a function of the homophily strength and the respective Model income levels. Thereby, agent j's weight in agent i's draw is denoted by w ij and determined as follows: w ij = 1 exp[<unk> |I j -I i |](1) I denotes the income of an agent, and <unk> <unk> R + denotes the homophily strength in income selection, externally set, and identical for all agents. <unk> = 0 represents a random graph, and for an increasing positive value of <unk>, an agent becomes ever more likely to pick linkneighbours with incomes being closer to their own. The exponential character of the link function ensures that those others with are large income difference become unlikely picks even at low homophily strengths. The choice of an exponential weighting function might seem arbitrary but upon closer inspection, we find that translated into the probability of i choosing j, this weighting is equivalent to the discrete choice approach developed and popularised by Manski & McFadden (1981). The homophily parameter <unk> <unk> (0, <unk>) is then simply the intensity of choice parameter. To translate weights into probabilities, we normalise by all weights for all agents, i.e., p ij = exp[-<unk> • |I j -I i |] k<unk>M <unk>i exp[-<unk> • |I k -I i |,(2) with M <unk> i as the set of all agents except i with size N -1.6 This formulation in (2) has a rather intuitive interpretation, with <unk> = 0 implying equiprobable picks with p ij = 1/(N -1), <unk>j <unk> M <unk> i, and thus indeed a random graph, while <unk> <unk> <unk> implies that p approaches unity for j with minimum income distance and 0 for all other j. Manski & McFadden (1981) demonstrate that the discrete choice rule above emerges naturally from random utility theory, i.e., agents maximise utility and utility can be decomposed into an observable and unobservable component. In our case, the observable component the agents minimise would be the income differences, with the unobservable part being all the attributes from which our agent in question would benefit due to their social connection. This appears to be rather intuitive, since of course income differences might be a rather salient characteristic and thus observable, while the utility from social connections might in some cases plausibly exceed Model the one derived from merely a good fit or small social distance. 7 In this sense, the weighting function in eq. ( 1) is plausibly microfounded in a utility-maximising framework and can now be considered the workhorse choice rule in behavioural macroeconomics (Franke & Westerhoff 2017). Franke & Westerhoff (2017) also survey evidence from a several lab experiments in different macroeconomic contexts that discrete choice is indeed consistent with the data, while Anufriev & Hommes (2012) and Anufriev et al. (2018) provide laboratory evidence for the discrete choice approach for financial markets. However, there might of course be other potential choice mechanisms that could provide avenues for further research on network generation that can be readily included within our proposed flexible RGG framework. Figures 3 and4 illustrate the linkage probabilities implied by the weighted draw based on the exponentially distributed income levels. As can be seen, the decay within the left tail is always more rapid than for the right tail, indicating differences in the'selectivity' above or below a relative position. We understand'selectivity' according to rank as the effect a decrease in income rank distance of one agent to another has on the linkage probability between them. Consequently, the local maxima of individual linkage probability densities exhibit a bi-modal shape with peaks at the highest and lowest rank but are also heavily skewed to the left, i.e., agents with the high incomes are most selective in their link picks. General selectivity increases with <unk>. Notice also that largest income ranks are extremely selective in all scenarios, in some cases in some cases exceeding linkage probabilities of incomes close to the median by more than two orders of magnitude in linkage probabilities. An anonymous reviewer pointed out that individuals may form links based on relative rather than absolute income differences: For lower incomes, a given absolute gap in units of currency may mean two entirely different lived experiences, while people with high incomes may hardly notice the same absolute gap. To represent this in the linkage function in eq. ( 1), one must simply replace I i by ln(I i ) and I j by ln(I j ). Appendix C analyses this transformation of scale in detail. The altered argument in the choice function is equivalent to assuming that agents aim to minimise the percentage difference in incomes and is, therefore, a natural extension to capture potentially scale-dependent tie-formation along the lines discussed above (Törnqvist et al. 1985). Our findings from Section 4 regarding self-perceptions and aggregate inequality perceptions prove qualitatively robust; quantitatively, the major findings occur at even lower 7 Notice, however, that the derivation of the above choice rule crucially depends on the axiom of Independence of Irrelevant Alternatives (IIA) (Luce 1977), i.e., the probability of choosing between j and k being independent of the probability of choosing l. IIA might be a good first-order approximation for homophilic choice but in friendship networks, knowing one agent j might indeed increase the likelihood of knowing another agent l that is friends with j. It might thus prove interesting to extend and generalise the above choice rule to examine the effects on the network topology in further research. Selected Probability Densities for R between 0 and 899 Selected Probability Densities for R between 0 and 899 homophily levels. However, the logarithmised incomes fail to replicate the greater underestimation of inequality for richer individuals with higher income ranks. The segregation tendency is approximately symmetric for moderate to high homophily strengths, i.e., all agents are approximately equally likely to include agents below and above them in income Model rank. Since there is hardly any differential behaviour according to income rank, all agents tend to perceive roughly equal levels of inequality according to the (local) Gini, in contrast to the stylised fact (iii). Thus, absolute income differences pose the more strongly validated presentation in the present model framework. Nevertheless, the model invites empirical research into whether income homophily is based on absolute or relative differences -or some combination of both.8 R = 0 R = 99 R = 199 R = 299 R = 399 R = 499 R = 599 R = 699 R = 799 R = 899 R = R = R = R = R = R = R = R = R = R = R = 0 R = 99 R = 199 R = 299 R = 399 R = 499 R = 599 R = 699 R = 799 R = 899 R = R = R = R = R = R = R = R = R = R = The resulting network for our baseline specification is a member of the family of Random Geometric Graphs (Dall & Christensen 2002), which Talaga & Nowak (2020) showed to reproduce core features of many social networks efficiently. Specifically, we combine the notions of homophily (Boguná et al. 2004) with pre-setting node degrees (Newman et al. 2001;Newman 2009).9 However, concerning our application, we are able to simplify both approaches by pre-determination of only the global minimum degree, like in Preferential-Attachment networks, and consequently defining relative weights rather than absolute probabilities. Links are undirected and have identical weights for evaluation purposes. Agents pick their neighbours in random sequential order. If an agent i picks agent j who had themself picked i before that, the already existing link between the two agents remains untouched, but i does not pick another neighbour instead of j. Consequently, each agent has at least 5 link-neighbours (i.e. clsoe social contacts with mutual knowledge of income) but may have more. --- Gini Perception and Network Evaluation Agents know about their own income and also their social contacts' incomes. However, they do not possess knowledge about any other agent or structural features of the whole income distribution. Thus, agents judge income inequality in the population as well as their own income position solely based on themself and their link-neighbours. Besides the agents' perceptions, there is a global assessment of various network parameters in order to validate the model. Subjective inequality perceptions mirror standard Gini calculation on the level of individual personal networks: Each agent finds the mean of all income differences between themself and each link neighbour and between any two of their link-neighbours and divides Model this by twice the mean overall income of themselves and all link-neighbours. Then, the overall perceived Gini is simply the arithmetic mean of individual perceptions. To estimate their income decile, an agent compares the number of link-neighbours having a higher income than the agent themself to the link-neighbours having a lower income than the agent themself. --- Validation The simulation results of our model are in line with our theoretical expectations and we can explain their emergence in terms of the mechanisms sketched in Section 4. Moreover, we carried out sensitivity analyses that revealed no unintended consequences of changes in any relevant model feature like homophily level, number of links or actual income distribution. Thus, we consider the model design and implementation to be internally validated (Gilbert & Troitzsch 2005, p. 22) as a tool for explaining inequality perceptions in the model population. Transferring these explanations from the model to the real world requires external validation of our model. However, there are different accounts of what constitutes an explanation in the first place. The current discussion of the concept of explanations in the philosophy of the social sciences highlights two types of explanations: How-actually and how-possibly explanations, also known as candidate explanations (Epstein 1999). While how-actually explanations aim for identifying the actual mechanism driving the dynamics in a specific case, how-possibly-explanations provide mechanisms that could possibly bring about the explanandum in question (Reutlinger et al. 2018); they enquire for mechanisms that potentially cause the observed phenomenon. In case of epistemically possible how-possibly explanations, these mechanisms are in line with the knowledge about the real world (Grüne-Yanoff & Verreault-Julien 2021). Our model yields an epistemically possible how-possibly explanation of inequality perception because it "produces quantitative agreement with empirical macrostructures, as established through on-board statistical estimation routines" and also "quantitative agreement with empirical microstructures, as determined from cross-sectional and longitudinal analysis of the agent population" (Barde & Van Der Hoog 2017): Simulation outputs of a societal structure close to a small-world one with self-segregation of highest-income agents and severe underestimation of the income Gini across income levels mirror the corresponding empirical findings. Following the suggestion by Fagiolo et al. (2019), we use empirical micro-data to calibrate the model. Namely, it relies on an exponential income distribution that characterises industrialised countries. Furthermore, the extent of agents' closest layer of interaction ('Dunbar's number') that means mutual knowledge of income, their linking behaviour, and individual perception formation follows rules that are theoretically established in rational choice theory but also empirically grounded in the referenced lab experiments and surveys. The exponential weighting function from the discrete choice framework is also analytically convenient and lets us represent the probability densities of ties in closed form. This allows us to e.g. demonstrate conclusively that the combination of discrete choice in graph formation and an exponential income distribution leads to the endogenous emergence of echo chambers for top-income earners whose isolation increases in the intensity of choice <unk>. 10 This empirical input calibration and output validation jointly guarantee resemblance (Mäki 2009) between our model and the real world. We develop a specific parallel reality (Sugden 2009) that features generating mechanisms for empirical findings in our reality, and hence our results present a candidate explanation for the stylised empirical facts. There may be different, more adequate, parallel realities featuring either these or even better mechanisms, despite to the best of our knowledge there being no existing models that fulfil these characteristics. Overall, the following section presents an epistemically possible how-possibly explanation of inequality underestimation that "constitutes epistemic progress on the way towards HAEs [how-actually explanations, A/N]" (Grüne-Yanoff & Verreault-Julien 2021) of the phenomenon. The model simultaneously features technical verification and external validation based on input and output measures. Gräbner (2018) considers this combination desirable albeit rarely possible for model development. Since our model features a range of proposed micro-mechanism

Across income groups and countries, individual citizens perceive economic inequality spectacularly wrong. These misperceptions have far-reaching consequences, as it might be perceived inequality, not actual inequality informing redistributive preferences. The prevalence of this phenomenon is independent of social class and welfare regime, which suggests the existence of a common mechanism behind public perceptions. The literature has identified several stylised facts on how individual perceptions respond to actual inequality and how these biases vary systematically along the income distribution. We propose a network-based explanation of perceived inequality building on recent advances in random geometric graph theory. The generating mechanism can * Daniel Mayerhoffer's contribution was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) -430621735. Furthermore, financial support for the paper by the University of Bamberg through the Fres(c)h grant no. 06999902 is gratefully acknowledged. We would also like to express our gratitude to two anonymous reviewers and Martin Everett for his excellent editorial work, Arndt Leininger and his students, Miriam Bömer, Bettina Gregg, Johannes Marx, Moritz Schulz, Eleonora Priori and Jan Weber as well as the participants of the Networks 2021 conference, the 9th PhD conference for the renewal of constitutional economics, the 9th ECINEQ Meeting, the 33rd annual EAEPE conference, the annual PhD conference by the Hans Böckler Foundation 2021, the 2nd Scientific Workshop by the Network for Pluralist Economics, the 8th Workshop on Complexity, Innovation and Knowledge (WICK), the Colloquium for the Advancement of Knowledge in Economics (CAKE) of the University of Utah and 28th DVPW congress, especially our designated discussants Macartan Humphreys, Claudius Gräbner and Pietro Terna which were of great help at crucial junctions of this investigation. Finally, we also thank Carsten Källner for his able research assistance. All remaining errors are, of course, ours. statistically distinguished from small-world networks, testifying to the robustness of our approach. Our results, therefore, suggest that homophilic segregation is a promising candidate to explain inequality perceptions with strong implications for theories of consumption and voting behaviour.

s for top-income earners whose isolation increases in the intensity of choice <unk>. 10 This empirical input calibration and output validation jointly guarantee resemblance (Mäki 2009) between our model and the real world. We develop a specific parallel reality (Sugden 2009) that features generating mechanisms for empirical findings in our reality, and hence our results present a candidate explanation for the stylised empirical facts. There may be different, more adequate, parallel realities featuring either these or even better mechanisms, despite to the best of our knowledge there being no existing models that fulfil these characteristics. Overall, the following section presents an epistemically possible how-possibly explanation of inequality underestimation that "constitutes epistemic progress on the way towards HAEs [how-actually explanations, A/N]" (Grüne-Yanoff & Verreault-Julien 2021) of the phenomenon. The model simultaneously features technical verification and external validation based on input and output measures. Gräbner (2018) considers this combination desirable albeit rarely possible for model development. Since our model features a range of proposed micro-mechanisms (e.g., on endogenously evolving segregation, cf. subsection 4.5), we also hope to inform empirical research to further examine their external validity. 10 This combination of analytical convenience that leads to internal validity and empirical plausibility that affirms external validtity is also one of the reasons why we deliberately choose not to use an Exponential Random Graph (ERGM) or Stochastic Actor Oriented Modelling (SAOM) framework (Snijders 2011, for a recent survey) but situate our model in the RGG framework: Firstly, the application of these types of models would require merging relational data with the socioeconomic status of the respective agents which is rarely achieved in practice, as De Paula (2017) notes. In our case, the problem of data availability is compounded by the fact that we require the graph data not only to report all social ties but also to identify the closest layer of emotional connection. Only there we can reasonably expect agents to exactly observe incomes as is required by our model mechanism. We are currently not aware of any dataset fulfilling these constraints but welcome any empirical attempt in this direction, as the external validity of our proposed model mechanism can ultimately only be established empirically. Secondly and more importantly, the estimated coefficient estimates and tie-level probability densities from ERGMs and SAOMs are purely phenomenological and need to be simulated by Monte Carlo techniques, while we are able to express them analytically and thus precisely determine the effect of our model parameters. We thank an anonymous reviewer for pointing us to ERGMs. --- Results The homophilic graph model will be evaluated against the five stylised facts outlined earlier. As we have shown in Section 3, we only require the homophily strength parameter <unk> <unk> R + 0, the number of links each node chooses C and the income distribution as inputs for initialisation. Since link formation is stochastic, we run the graph formation routine 100 times and report model averages, if not otherwise indicated. Most of the results are obtained with initialisation by the same set of incomes generated from an exponential distribution with location parameter <unk> = 1 and 1, 000 observations for C = 5 choices of link-neighbours each agent undertakes to make results comparable for variation in <unk>. The overall Gini coefficient for these 1, 000 randomly generated income levels is with G <unk> 0.50701 within 1.5 % deviation from the theoretical Gini of G = 0.5, indicating that the observed effects of <unk> are not artefacts of initialisation. Results are also robust for different numbers of links chosen per node, as long as C N. We also evaluated the null model for <unk> = 0, where we did not find any significant deviations in the mean inequality perceptions and the actual overall inequality of G = 0.5, testifying to the robustness of our approach.11 --- Small-Worldiness We use state-of-the-art methods to test for the existence of small-world features against an appropriate network null model, here an Erd<unk>s-Rényi (ER) graph with the corresponding number of nodes and mean degree first introduced by Erd<unk>s & Rényi (1960). ER graphs appear to be the correct null model for two reasons: Firstly, they are a particular case of our model with <unk> = 0, i.e. without homophily. Hence, the procedure allows isolating the impact of homophily and examining whether the model indeed tends to yield'smaller worlds' for homophilic formation in the precise sense outlined below. Secondly, we can establish an exact one-to-one correspondence between a graph generated by our model and the ER model, as ER graphs only require the number of nodes and a linkage probability for initialisation that is fully determined by the mean degree of the correspondent network. Other prominent generating models such as Watts-Strogatz graphs have additional degrees of freedom like the'rewiring probability' without clear correspondence to our model. We construct three summary metrics to test our model against, as introduced by Humphries & Gurney (2008). Firstly, <unk> measures the deviation in average path lenghts L, that is, <unk> i := L i L ER i,(3) where L i is the average path length of network i with L ER i as the average path length of a correspondent ER graph with equivalent number of nodes and mean degree. 'Smallworldiness' requires <unk> <unk> 1, as our network should not deviate too much from the random benchmark that indeed features short paths. E[L ER i ] = (log[N ]-<unk>)/(log[k] ))+1/2 with <unk> as Euler's constant, N as the number of nodes and k as the average degree can be analytically derived which we use in our calculation (Fronczak et al. 2004). Secondly, we also require a high clustering coefficient which an ER graph cannot generate. The deviation in the clustering coefficients <unk> is defined as <unk> i := C i C ER i,(4) with C i as the clustering coefficient of graph i and C ER i as the clustering coefficient of the corresponding ER graph. Here, again, analytical results are available which we utilise, mainly that E[C ER i ] = k/N with again k as the average degree and N the number of nodes (Watts 1999). Since ER graphs typically do not exhibit clustering, we require here that <unk> i > 1 for a small-world to be present. Finally, we use a summary measure <unk> introduced by Humphries & Gurney (2008). We define <unk> as <unk> i := C i C ER i / L i L ER i = <unk> i <unk> i.(5) Humphries & Gurney (2008) show that <unk> i features desirable statistical properties when confronted with the conventional Watts-Strogatz model for graph formation and shows a unique maximum between the extreme cases of a random network and an ordered lattice. This is in line with our intuition that small-worldiness results from the interaction of order (in the form of high clustering near the lattice) and randomness (in the form of the random graph featuring low average path lengths), as shown by Watts & Strogatz (1998). We require <unk> > 1 for small-worlds. Note that <unk> > 1 is an immediate corollary of the two requirements <unk> > 1 and <unk> <unk> 1, but <unk> > 1 does not imply the two individual requirements. We call the first sufficient condition'strong small-worldiness' and <unk> > 1 with a violation of either --- Results <unk> > 1 or <unk> <unk> 1 'weak small-worldiness', where we now only require normalised clustering to increase faster than average path lengths. Note: The Figures report violin plots for the relevant statistics for'small-worldiness'. The average path length is significantly higher than the ER benchmark for all depicted <unk>, indicating violation of the'strong small-worldiness condition'. Normalised clustering coefficients are for <unk> > 1 significantly higher than the ER benchmark and increase at a much faster rate than average path lengths, indicating that indeed the 'weak small-worldiness' condition is fulfilled. The ER benchmark is nested in our model for <unk> = 0, as is also readily visible from the fact that both <unk> <unk> 1 and <unk> <unk> 1 for <unk> = 0. We indeed find that homophily induces path lengths to grow significantly above this ER benchmark. Normalised clustering coefficients, however, increase much more rapidly with homophily than average path lengths, demonstrating that our model can achieve relatively high clustering without simultaneously increasing path lengths in the same way. The proposed process thus violates the strong condition but fulfills the weak condition for small-worlds and is therfore broadly in accordance with the topological patterns found in real-world social networks. We note further the symmetry to the canonical Watts-Strogatz approach (Watts & Strogatz 1998). While we build on a random network with short average path lengths and interpolate to the desired high clustering through homophily, Watts and Strogatz start from an ordered state with high clustering and approach the random graph benchmark by rewiring to generate shorter average path lengths. Arguably, however, our approach starts from a plausible and empirically well-established behavioural principle in contrast to the purely stochastic process in the Watts-Strogatz world without such behavioural foundations. Besides providing empirical validation, this finding might also point to relatively rapid contagion throughout the homophilic network, be it in the form of rumours or 'expenditure cascades'. --- Results --- Perceived Social Hierarchy and Middle Class Bias For unbiased hierarchy perceptions, the reported frequency of perceived social position would coincide with the actual positions. Unbiased perceptions thus entail reported perceived positions of equal frequency, as they coincide with the actual population shares. As we show both analytically in Appendix B and by simulation, perceived social positions for homophilic graph formation are far from the equiprobable benchmark. We find a tendency of the vast majority of individuals to place themselves in the middle of the perceived hierarchy, in line with the empirical evidence. We prove that the tendency exists for all <unk> <unk> (0, <unk>). Its strength is a function of <unk>, though, as we show exemplarily in Figures 8 to 11. The figures plot the empirical densities of income ranks which the respective the individuals perceive to hold. For <unk> = 1, the tendency is relatively weak, while for <unk> = 4, <unk> = 8 and <unk> = 14, the densities display a distribution that notably peaks at the centre. In fact, the displayed densities indeed seem to feature all the salient features of the densities of empirical perceived social positions, as shown in Choi (2019) and also in Figures 1 and2 Note: The Figures report the perceived social positions for <unk> <unk> <unk>1; 4; 8; 14<unk> with 10 bins each. All Figures exhibit significant deviation from the benchmark with equal frequencies. The tendency for individuals to place themselves in the middle of the income hierarchy is, however, only apparent for the middle and right panels, indicating that a homophily strength <unk> of 1 might be too low to replicate the empirically observed tendency. For <unk> = 4, 8 and 14, the densities approximate the empirical densities rather well, though. Notice that this a necessary outcome of homophilic graph formation under very mild and general conditions and based on a well-established utility maximisation framework, in contrast to models that take this tendency as an assumption. The latter strand of literature has typically taken a bounded rationality stance on the issue and argued that it is failures in information processing which explain the persistent errors in perceptions of social positioning. Our model replicates stylised fact (i) purely by virtue of the network formation process. In contrast to the literature on bounded rationality, we can hence show that stylised fact (i) is consistent with purely rational actors that form correct beliefs based on their available infor-Results mation, as long as homophilic graph formation constrains their information sets. Our model thus entails very different policy implications to improve self-perceptions. Since information processing is assumed to be correct in our model, information treatments, i.e., increasing the information received from nodes with heterophilic incomes, have mitigating effects on perceptions. Influencing information processing itself, as implied by the established models, is arguably a much harder task for policy. Figure 12: Errors i show the difference between perceived position q i and actual position a i for all individuals i, and <unk> = 4. The superposed line corresponds to i = 0.5 -a i or the belief for all individuals to be in a median position of the income distribution. Except for the boundary regions close to the minimum and maximum income, the theoretical fit approximates the trend in the data reasonably well. This indicates that the trend to the median is indeed present for the vast majority of the population. An immediate corollary of the population perceiving themselves to earn the median income is the tendency for rather poor individuals to overestimate their position and the rich to underestimate it, as all perceive themselves to be in the middle. Hence, the fit for the median perception tracks the trend in the simulations reasonably well for the vast majority of observations (cf. Figure 12). As we discuss in more detail in Appendix B, there is no tendency to the median for the left and right tail of the distributions which the simulation results reflect, too. Indeed, approaching the minimum or maximum improves the accuracy of individual estimates. The intuition for this is quite simple: The poorest and the richest individual will always correctly perceive their social position, independent of <unk> <unk> R + 0. The rationale for this is that the actual minimum (maximum) of the whole will Results always be the minimum (maximum) of any potential non-empty subset of the population. Apart from such boundary effects, however, we indeed replicate stylised fact (ii) insofar as the poorer half of the population seems to overestimate their social position, while the richer half underestimates it. This finding is in line with the empirical evidence and suggests that total whole population tends to underestimate the degree of inequality, as we will show in the upcoming subsection. --- Perceived Individual Inequality + <unk> = 8 + <unk> = 4 + <unk> = 1 Figure 13: Plot of inequality perceptions against the income rank. Almost all individuals underestimate true inequality with a Gini of 0.5. Degrees of underestimation vary, though, where bias increases approximately monotonically in income rank. The intuition for this is that homophilic graph formation lets unweighted inequality (absolute income differences) increase only linearly in income rank, but the reference standard (mean income) increases exponentially. We define perceived inequality as the Gini coefficient calculated over the perception set of a given individual i. In Figure 13, we plot those perceived Ginis against the income ranks of our individuals with a higher rank indicating a higher income. In line with stylised fact (iii), we find that inequality perceptions decrease approximately monotonically in income rank, while almost all individuals underestimate the actual degree of inequality significantly. As a result of our homophilic graph formation process, perceptions are most accurate for the poorest which either over-or underestimate actual inequality of G = 0.5 slightly. The Gini coefficient is conventionally defined as the ratio of (unweighted) mean differences in Results the incomes within the perception set of an individual to twice the mean income within this group. Homophilic graph formation now lets those unweighted mean differences increase linearly at most, while the mean incomes increase exponentially due to the exponential distribution by which incomes are initialised. As a result, the ratio falls almost monotonically. This results is not only plausible due to its accordance with stylised fact (iii) but might also correspond with the empirical evidence on perception formation. One of the most prominent hypotheses on perception formation from stimuli is the Weber-Fechner law (Fechner 1862) which indicates that perceived differences in stimuli need to be proportional to the baseline of a given stimulus to be recognisable. The phenomenon is well-established not only for sensory stimuli (Formankiewicz & Mollon 2009;Pienkowski & Hagerman 2009) but also finds use in marketing research on price responses (Sirvanci 1993;Snell et al. 1995). In this framework, one can also understand a decreasing perceived Gini as the change in stimuli (the unweighted differences in incomes of the perception set) do not increase in the same way as the baseline of stimuli (the mean incomes of this perception set) and is thus also consistent with the psychological microevidence. The dashed vertical lines correspond to the empirical sample minimum and maximum, while the bold line corresponds to the sample mean. We find that varying the homophily <unk> parameter can fully quantitatively account for the variation in empirical perceptions. --- Results --- Perceived Global Inequality For further validation, we also examine whether our graph generating process can quantitatively replicate empirical perception patters. We use the mean, minimum and maximum for inequality perceptions on a national level calculated yearly for a large sample of 32 OECD countries in a 30 year time-span by Choi (2019).12 Over all countries, they find a minimum perceived Gini of G min = 0.1276, a mean perceived Gini of G mean = 0.1708 and a maximum perceived Gini of G max = 0.2534. In analogy to their empirical results, we average over the Gini perceptions of all individuals. As we show in Figure 14, we find that our process can fully account for their empirical findings and the variation between inequality perceptions by only varying the homophily parameter <unk>. We also note that the sample average of national inequality perceptions implies a homophily degree <unk> <unk> [8; 9]. Yet, our findings imply considerable cross-country variation in homophily that ranges between <unk> <unk> 4 to <unk> <unk> 14. Apart from extremely low homophily level, increases in actual inequality induce a much lower increase in perceived inequality. For moderate and large degrees of homophily, the schedule reaches a plateau rather quickly, i.e., perceived inequality responds extremely slowly to changes in actual inequality. --- Results --- Perception Dynamics + <unk> = 8 + <unk> = 4 + <unk> = 1 To analyse perception dynamics, that is, the reaction of inequality perceptions to changes in actual inequality, we need to initialise the model with another distribution, as the exponential has a fixed Gini of about 0.5, irrespective of its precise parametrisation. We use the log-normal as another benchmark and vary the dispersion parameter <unk> to simulate changes in the Gini coefficient which is another distribution typically used to describe the skewed nature of empirical income distributions (Knell & Stix 2020). As we see, apart from implausibly low degrees of inequality, changes in actual inequality cause far less than a one-to-one change in perceived inequality. Especially for higher degrees of homophily, we find that the schedule quickly reaches a plateau, where inequality perceptions are now extremely persistent with respect to increases in actual inequality. Our model thus is consistent with stylised fact (iv) as our last test of validity. --- Results The persistence in perceptions occurs because homophily becomes more binding and segregation stronger when actual inequality increases. This mechanism leads ceteris paribus to a decrease in perceived inequality which offsets a direct impact of objective inequality on subjective perception. Compare, for illustration, the two regimes close to a completely egalitarian income distribution near G = 0 and relatively high degrees of inequality near G = 0.5. The egalitarian state is close to a random network, as homophilic segregation presupposes income differences. Small changes in actual inequality are thus not strongly reflected in segregation and almost fully impact perceived inequality, leading to a one-to-one correspondence of perceived and actual inequality in this neighbourhood. For large degrees of actual inequality and large homophily, changes in actual inequality immediately impact segregation, leading to a plateau and very persistent perceptions. Notably, this mechanism is not only consistent with the empirical evidence in terms of its emergent outcome; several recent studies by Reardon & Bischoff (2011); Chen et al. (2012) and Tóth et al. (2021) examine the mechanism directly and show that economic inequality tends to increase (spatial) segregation. A fruitful avenue for further research could be the time-scale on which this channel works, with more laggard segregation responses obviously decreasing the space for inequality-enhancing policies. --- Segregation Patterns We measure segregation as the proportion of links an individual i has in their own decile as <unk> i as one particular way to measure'selectivity' without access to behavioural linkage parameters. This constitutes a straightforward but standard way to measure segregation and is easily transformed into normalised measures of segregation like the E-I index <unk> i which is defined as the difference between the share of between-group links and the share of within-group links (Bojanowski & Corten 2014). 13 Figure 16 plots the simulated segregation statistics as well as a theoretical fit for <unk> = 4. For analytical convenience, the superposed red line plots the probabilities that an individual chooses another agent to link to within their own decile as a first pick, so the total choice set consists of 999 other individuals, and does not account for the possibility that other agents already link to the agent in question, in contrast to our algorithm. Appendix A details the derivation. The goodness of fit demonstrates that these incoming-links do not exhibit a significant effect on segregation patterns and tend to average out in the aggregate, showing that our analytical approximation is indeed reasonable. Segregation exhibits two distinct patterns along the rank distribution. Firstly, we find that segregation exhibits a skewed U-shape and increases, especially for the richest decile, Figure 16: The figure plots our segregation measure, measured as the proportion of links of a node to nodes in the same decile, along the rank distribution. The theoretical fit is obtained for the assumption that nodes choose their neighbours themselves without other nodes choosing links incoming to them. The goodness of fit shows that this is indeed a reasonable assumption. Segregation exhibits distinct and non-trivial patterns both regarding global as well as local maxima. which is almost completely disconnected from the other groups. In this sense, our graph formation process endogenously creates echo chambers for the richest whose information sets do not cover the poorer population at all. This results from the fact that the richest part of the population is extremely selective in choosing their link-neighbours, as we have shown in Section 3. Secondly, we also find a rather strong variation together with local maxima within deciles. This finding might be, however, spurious and a partial artefact of boundary effects at decile boundaries. As Appendix B shows analytically, individuals will choose those sets of link-neighbours with the highest probability that are distributed symmetrically around them in rank. Thus, individuals exactly at the decile boundary will most likely select a set with half of their neighbours across the boundary. Individuals closer to the centre of a decile, on the other hand, will by the same token choose with highest probability link-neighbours within their own decile. Arbitrarily pre-defined group boundaries can thus create within-group variability in commonly used indices like the E-I index that nevertheless exhibits desirable statistical features at an aggregate level (Bojanowski & Corten 2014). These findings extend well beyond income deciles, as variables like age group, place of living, gender, education or ethnicity are likely strongly correlated with income. Studies using E-I type indices to detect homophily in other variables might hence create spurious results if income homophily Results is also present. The relevance of such boundary effects has increasingly also been recognised in applied work (Hvidberg et al. 2020). Whenever dimensions have a cardinal scale like income, it might therefore prove more fruitful to use a rolling-window type of estimation, where within-groups are defined in relation to the individual in question, such as a fixed number of income ranks or a fixed income rank interval around theirs. For non-biased individual samples, one would expect agents to both estimate global inequality correctly in aggregate and also locate their true income quantile. However, homophily triggers link selectivity and hence biased samples which in turn causes inequality perceptions based on an agent's income level and rank (cf. Figure 17) --- Summary of Generating Mechanisms Selectivity in link formation depends on global income inequality. Furthermore, agents whose income is further from the global median income are more selective in their linkneighbours and so are agents with the higher income ranks, as an artefact of the cut-off in the exponential selection function at the low end of the income distribution (cf. the skewed Discussion U-shape in Figures 3 and4). Such link-neighbour selection generates personal networks for each agent in which this agent tends to have the median income and where income rank differences are relatively small. Moreover, the extent of relative income differences in one's personal network now depends on the characteristics of the agent in question. Firstly, the impact of link-neighbours with great rank differences to the perceiving agent is larger if this perceiving agent and consequently the majority of link-neighbours have a low income themselves since local inequality calculations weight income differences by the local mean. Secondly, agents close to the global income distribution median ceteris paribus perceive lower inequality levels, for the greatest rank distances tend to be smaller in these cases. Overall, in a homophilic linking regime, the complex interaction of actual income inequality and of individual absolute income and distribution rank cause biases in income level self-rating and inequality assessment that aggregate to biased underestimation of inequality. Due to the interplay of factors that feed into individual selectivity in choice of link-neighbours, the relation between objective income structure and individual perceptions is non-monotonic and not trivial but requires case-based assessment. --- Discussion Our parsimonious model provides an epistemically possible how-possibly explanation of the stylised empirical facts regarding inequality perceptions that we identified in the literature. Individuals who evaluate their immediate social environment without bias can misperceive their own rank in the overall income distribution as well as global inequality. Homophilic formation of the immediate environment suffices to fully explain the discrepancy between actual and perceived inequality since a rising level of actual inequality causes higher selectivity in link-formation. Moreover, the further away someone's income rank is from the global median and the higher their income, the more selective they are in their choice of link-neighbours. Thus, public misperceptions are not necessarily driven by limitations in information processing, e.g. a behavioural tendency to place oneself near the median of social hierarchy, but by limited information sets the individuals exhibit for inductive reasoning. The seemingly subtle distinction between constraints on information processing and information sets carries important policy implications: When it is the limits of available information and not limits in cognitive ability driving misperceptions, informational treatments may be successful, as also the empirical literature suggests (Cruces et al. 2013;McCall et al. 2017;Karadja et al. 2017). Such treatments can either consist of delivering information about income inequality itself or facilitating the formation of more diverse contacts in order to overcome the segregation by income that our model finds. In other words, this means breaking up echo chambers that are caused by humans drawing confidence in their beliefs only from repeated observations while ignoring a potential lack of diversity in sources (Foster et al. 2012;Schwarz et al. 2016). Educating individual citizens about their information deficit and providing ways of overcoming it is important from a democracy theory perspective. For example, (Rawls 2005, p. 224) requires "presently accepted general beliefs" as basis for arguments in the public forum. However, while one can asses the income inequality objectively without any room for disagreement if using all globally available information, citizens who work only with their individual information will agree on a belief about the Gini that underestimates its actual value. Hence, the lack of individual information access inhibits deliberation about the level inequality and its changes, e.g., in response to past policy measures. For the first time, we are able to infer the composition of these reference groups from readily available observational data on perceived inequality to inform both empirical investigations as well as more comprehensive model-building in other regards. Directly investigating perception networks might provide a possible remedy for the problem that identification of interaction effects within survey data in the form of time series is hindered by sample sizes that are typically one or two orders of magnitude too low to distinguish noise from true interaction (Alfarano & Milakovi<unk> 2012). Our main empirical prediction is that the homogeneity of social groups, the fraction of links to agents within the own income decile, roughly follows a U-shaped pattern with a massive decrease in diversity for the richest and (a quantitatively much more modest one) for the poorest agents. This finding might point to an endogenously emerging 'elite discourse' with almost no transmission of information to the poorer 90% of the population. Our cross-country analysis shows that there exists considerable variability in implied homophily levels. There are several possible candidates to explain this variability, such as cultural norms, diversity in media and political representation or spatial segregation. Regarding spatial segregation patterns, Thorstein Veblen made the farsighted observation as early as 1899 that urbanisation should increase diversity in social contacts, since cities are the place "where the human contact of the individual is widest and the mobility of the population is greatest" (Veblen 1899(Veblen [2001]], p. 66). Thus, perceived inequality should ceteris paribus be higher in urban areas resulting from the higher average income diversity per perception network, a testable hypothesis and thus a possible avenue for further research in spatial economics. This is also what the rather scarce existing evidence for Vietnam and Central and Eastern Europe suggests (Mahajan et al. 2014;Binelli & Loveless 2016). In this way, spatial and perception network segregation might therefore overlap and interact (Newman et al. 2018;Tóth et al. 2021), and policy affecting the allocation of land could thus also exhibit unintentional effects on perceptions. We leave detailed analyses on these determinants for further research. In terms of theory, our empirically validated random geometric graphs might provide an ideal microfoundation for theories of consumption as dependent on the relative income position and for which shocks that affect local income compositions lead to 'expenditure cascades' (Duesenberry 1949;Veblen 1899Veblen [2001]]; Frank et al. 2014). This new mechanism might shed light on the disputed link between economic inequality and growth. We will address these questions in further research. Finally, our model presents a way of generating random geometric graphs defining both the distribution of the feature that determines linking probability between any pair of nodes and a minimum degree value for each node. Put differently, we apply a Barabási-Albert (1999) Preferential-Attachment type procedure, which is intuitive for social scientists, to features other than degree and get network graphs that can be analysed using readily available methods from random geometric graph theory. To the best of our knowledge, there is no such way yet. While it is reassuring that a scale-transformation to a log-scale delivers the same salient'middle-class bias' as our baseline specification, log-transformations in general strike us as a very parsimonious way to capture scale-dependence in choice. This is not only relevant for inequality perceptions but for essentially all variables and features where perception of stimuli is plausibly dependent on scale. In general, it is possible to apply the generating procedure to features other than income that exhibit different distributional patterns. This approach will hopefully inspire future studies of expectation formation, e.g., regarding inflation or business sentiment. In these fields, identification of the relevant perception networks might be a crucial step to bring macroeconomic theory currently mostly building on atomistic rational expectations and empirical studies, that find little support for those types of expectations, closer together (Pesaran & Weale 2006). Hence, we also provide a toolkit for analysing the impact of homophily regarding any specified feature on network generation (and potential interactions on the resulting network) given a particular distribution of this feature that is well-grounded in the behavioural economics literature. In sum, we anticipate our theoretical framework to generate numerous avenues for further studies both theoretically, regarding graph-and expectation formation, as well as empirically, pertaining to the determinants of homophily and possible policy measures, to information transmission and to the effects of inequality on aggregate consumption. Analogously, we get for d <unk> 0 w ij = exp[-<unk> log(1 -((R + d)/N ) -log(1 -(R/N )) ] (13) = exp[log N -R -d N -<unk> ) -log N -R N -<unk> ] (14) = N -R -d N -R -<unk> (15) = N -R N -R -d <unk> (16) To translate w ij into probabilites, we need to normalise by all weights. Note that this is still a (close) approximation of the probabilities of link-formation of a given node i. Nodes draw their C link-neighbours from the set of all neighbours. This implies that draws are not independent, as we assume here. Since C N, however, the effect is marginal. The approximation for the probability below, however, seems to perform quite well which we verify in our subsection on segregation. The probability p that i chooses j as a link-partner can therefore be approximated as p ij (N, R, d) <unk> <unk> <unk> <unk> N -R N -R-d <unk> ( -1 d=-R+1 N -R N -R-d <unk> + N -R-1 d=1 N -R- d N -R <unk> for d <unk> 0, N -R-d N -R <unk> ( -1 d=-R+1 N -R N -R-d <unk> + N -R-1 d=1 N -R- d N -R <unk> for d > 0. Notice that the function behaves as expected and is monotonically decreasing in |d| <unk> N +. The strength of selection also increases monotonically in the homophily parameter <unk>. For <unk> = 0, we recover the equiprobable case without any decay. The precise functional form of the decay for <unk> <unk> R + is, however, far from trivial and changes along the rank distribution. The right tail of the correspondent density is always a power transformation of a linear function, whereas the left tail for any given R is a power transformation of function with hyperbolic decay. In this sense, all nodes

Across income groups and countries, individual citizens perceive economic inequality spectacularly wrong. These misperceptions have far-reaching consequences, as it might be perceived inequality, not actual inequality informing redistributive preferences. The prevalence of this phenomenon is independent of social class and welfare regime, which suggests the existence of a common mechanism behind public perceptions. The literature has identified several stylised facts on how individual perceptions respond to actual inequality and how these biases vary systematically along the income distribution. We propose a network-based explanation of perceived inequality building on recent advances in random geometric graph theory. The generating mechanism can * Daniel Mayerhoffer's contribution was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) -430621735. Furthermore, financial support for the paper by the University of Bamberg through the Fres(c)h grant no. 06999902 is gratefully acknowledged. We would also like to express our gratitude to two anonymous reviewers and Martin Everett for his excellent editorial work, Arndt Leininger and his students, Miriam Bömer, Bettina Gregg, Johannes Marx, Moritz Schulz, Eleonora Priori and Jan Weber as well as the participants of the Networks 2021 conference, the 9th PhD conference for the renewal of constitutional economics, the 9th ECINEQ Meeting, the 33rd annual EAEPE conference, the annual PhD conference by the Hans Böckler Foundation 2021, the 2nd Scientific Workshop by the Network for Pluralist Economics, the 8th Workshop on Complexity, Innovation and Knowledge (WICK), the Colloquium for the Advancement of Knowledge in Economics (CAKE) of the University of Utah and 28th DVPW congress, especially our designated discussants Macartan Humphreys, Claudius Gräbner and Pietro Terna which were of great help at crucial junctions of this investigation. Finally, we also thank Carsten Källner for his able research assistance. All remaining errors are, of course, ours. statistically distinguished from small-world networks, testifying to the robustness of our approach. Our results, therefore, suggest that homophilic segregation is a promising candidate to explain inequality perceptions with strong implications for theories of consumption and voting behaviour.

N -R N -R-d <unk> ( -1 d=-R+1 N -R N -R-d <unk> + N -R-1 d=1 N -R- d N -R <unk> for d <unk> 0, N -R-d N -R <unk> ( -1 d=-R+1 N -R N -R-d <unk> + N -R-1 d=1 N -R- d N -R <unk> for d > 0. Notice that the function behaves as expected and is monotonically decreasing in |d| <unk> N +. The strength of selection also increases monotonically in the homophily parameter <unk>. For <unk> = 0, we recover the equiprobable case without any decay. The precise functional form of the decay for <unk> <unk> R + is, however, far from trivial and changes along the rank distribution. The right tail of the correspondent density is always a power transformation of a linear function, whereas the left tail for any given R is a power transformation of function with hyperbolic decay. In this sense, all nodes are more'selective' regarding individuals that are poorer than regarding the richer part of the population. To see this, compare the decay for the minimum and the maximum of the distribution for <unk> = 1 as a special case. For R = 0, p ij <unk> 1-(|d|/N ) with linear decay in |d|, as there exists only a right tail, while for R = N -1, p ij <unk> 1/(1 + |d|) which decays extremely fast in |d| by a power function, as there exists only a left tail here. In this sense, the richest individual is far more'selective' in choosing their (poorer) link-neighbours than the poorest individual choosing their (richer) ones. The theoretical expected segregation index we compare against our simulation results can be straightforwardly computed from those probabilities. Let <unk> i be the set of nodes that are in the same group as node i such as an income quantile. The probability to connect with a link-neighbour pi can then again by approximated as arg min <unk> ij I k <unk> ij |I k -I i |.(19) It remains to be shown that this minimisation leads to the choice of a set <unk> ij for which I i is the median value. The median requires the same number of values above or below I i in <unk> ij. With M links for node K i of income rank R and M even, this requires M/2 values above and below I i. For I i as the median being minimising for the absolute distances, this requires i) that there exists no node with rank R + M/2 + 1 such that their income distance to K i is less than the income distance from node K i to the node ranked R -M/2. If i) is violated, the node with rank R + M/2 + 1 is part of the distance-minimising set and thus, I i is not the median of <unk> ij. The symmetrical condition ii) requires that there is no node with rank R -M/2 -1 such that its distance to K i is less than the distance of K i to the node with rank R + M/2. In terms of a quantile function, we require <unk>( R + M/2 + 1 N ) -<unk>( R N ) > <unk>( R N ) -<unk>( R -M/2 N )(20) and <unk>( R N ) -<unk>( R -M/2 -1 N ) > <unk>( R + M/2 N ) -<unk>( R N ).(21) Rearranging yields <unk>( R + M/2 + 1 N ) + <unk>( R -M/2 N ) > 2<unk>( R N ) > <unk>( R + M/2 N ) + <unk>( R -M/2 -1 N ). (22 ) Expressing the left-hand side of inequalities for a generic distribution in ( 22) for an continuous exponential such that R/N <unk> p and substituting the quantile function, we require -log[1 -(R + M/2 + 1)/N ] <unk> + -log[1 -(R -M/2)/N ] <unk> > 2 -log[1 -(R/N )] <unk>(23) The condition R/N <unk> p presupposes N to be sufficiently large for the discrete realisations of the sample to approximate the quantiles of the continuous exponential distribution. We find this condition fulfilled for several numerical experiments. It is easy to see that the left-hand side condition in ( 22) is fulfilled for a quantile function whose first derivative is monotonically increasing which is the case for d<unk> <unk> (p)/dp = 1/((1 -p)<unk>) for p <unk> [0, 1) and <unk> > 0. We can also show this by manipulation of ( 23 N - R -M/2 N + (R + M/2 + 1)(R -M/2) N 2 -1 + 2 R N - R 2 N 2 <unk> 0 (26) - 1 N + R -M/2 -(M/2) 2 N 2 <unk> 0 (27) R -M/2 -(M/2) 2 -N N 2 <unk> 0.(28) Since R <unk> N per definition, condition ( 28) is trivially fulfilled. Notice that this implies for an exponential initial distribution, I i cannot be below the median in the most likely set. The right hand-side of inequalities ( 22) is a bit more demanding. Stating the condition in terms of the quantile function for an exponential, we get 2 -log[1 -(R/N )] <unk> > -log[1 -(R -M/2 -1)/N ] <unk> + -log[1 -(R + M/2)/N ] <unk>.(29) Simplifying yields (1 -(R/N )) 2 <unk> (1 -(R -M/2 -1)/N ) • (1 -(R + M/2 -1)/N ) (30) (1 -(R/N )) 2 <unk> (N -R + M/2 + 1)(N -R -M/2) N 2 (31) (N -R) 2 <unk> N 2 -RN -N M/2 -RN + R 2 + RM/2 + N M/2 -RM/2 -M 2 /4 + N -R -M/2 (32) 0 <unk> N -R -M/2 -M 2 /4 (33) R N <unk> 1 - 1 2 M N - M 2 4N. (34 ) For our discrete sample, F <unk> (I i ) <unk> R/N which reveals that condition (34) is only a slightly more demanding condition than boundary condition (18) that guarantees the possibility of I i being a median in the first place and only differs by M 2 /4N. Since we typically assume M N, this term vanishes. Indeed, for a realistic baseline scenario with N = 1, 000 and M = 5, the condition is fulfilled for the poorest 99% of the population and thus for the vast majority. Together with the lower boundary condition (18), the tendency to place themselves in the middle should exist for about 98% of the population and thus the vast majority. Minimising absolute deviations for an exponential income distribution and M N thus entails choosing sets that let I i be the median of <unk> ij <unk> I i for almost all I i. While the strength of this mechanism will of course be dependent on <unk>p ij /<unk> I k <unk> ij |I k -I i |, the median is the most likely outcome for any homophilic network as the perceived quantile for the vast majority of nodes. Note: The Figures report the perceived social positions for <unk> <unk> <unk>1; 4; 8; 14<unk> with 10 bins each and for a choice function with log incomes in its argument. All Figures exhibit significant deviation from the benchmark with equal frequencies. While the tendency is apparent in all panels, its strength expectedly increases and perceptions are more narrowly clustered around the middle categories the higher <unk> is. homogeneous inequality perceptions for all considered <unk> <unk> 4. For low <unk>, the behaviour of local Ginis is thus broadly consistent with one aspect of stylised fact (iii), namely that perceived inequality tends to decrease in income rank, as is also shown in Figure 22 for <unk> = 1. However, poor agents then drastically overestimate inequality in violation of the second aspect of stylised fact (iii) that (almost) all agents somewhat drastically underestimate inequality. Heuristically, the behaviour of the local Gini coefficients is a direct consequence of the behaviour of an exponential income distribution on a log-scale. Incomes increase locally linear in rank around the mean income (of <unk> = 1) but increase much faster superlinearly near the upper and lower tails of the distribution, as Figure 23 shows. + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + This shape of the distribution of logarithmised incomes implies that the agents with income at the lower tail will have a relatively strong probability of observing incomes around the mean, where income ranks do not change inclusion probabilities very much. Since these values are, from the perspective of low-income agents, extreme values, this higher probability of observing incomes within this locally linear region translates into higher expected inequality perceptions, as is evident from the boundary effects in Figure 22. A symmetric argument applies to the upper tail of the distribution, although with a more attenuated effect, since the mean income is closer to the maximum value in income rank due to the skewness of the exponential. Hence, the values around the mean are not as extreme from the perspective of the agents with maximal incomes, leading to a less pronounced increase in upper boundary inequality perceptions. Agents with log incomes located in the locally linear region exhibit the strongest tendency to perceive incomes close to them in rank and exhibit the most pronounced dislike of extreme perceptions at the upper and lower tail. In this sense, the bias against extreme values increases with decreasing distance to the mean income which indicates that inequality perceptions decrease until the mean income is reached and increase afterwards. Increasing <unk> then disproportionately affects extreme distances due to the exponential nature of the weights, mitigating any differences in perceptions caused --- Appendix A Linkage Probabilities for Homophilic Networks Heuristic Derivation. Linkage Probabilities. Consider an arbitrary node K i indexed i <unk> 0,..., N -1 as their rank R increasing in income that is part of a graph G with N nodes characterised by adjacency matrix A. Let I i denote their income, where f <unk> (•) defines the PDF of a exponential probability distribution defined over the real half-line (0, <unk>) with parameter <unk> > 0 and F <unk> (•) the corresponding CDF. The quantile function for any population share p and with parameter <unk> > 0 for an exponential distribution is given as We assume without loss of generality that <unk> = 1 for normalisation. Calculated <unk> values therefore need to be scaled up by the inverse of the mean income, <unk> -1 for empirical application. The quantile of a node with income I i can be approximated by their rank R, such that p <unk> R/N as a discrete approximation of the continuous probability density which holds for large N. We want to derive the probability that a node i with rank R <unk> N + 0 connects to a node j with a distance of d to node i. Expressing the weights as defined in Section 3 now in the form of quantiles, we get Assume first that d > 0, that is, j is richer than i. Simplifying the weights yields for d > 0: Appendix B Perceived Quantiles in Perception Networks Proof Sketch. Pure Homophily implies a Tendency to the Median in Perceived Quantiles. Consider an arbitrary node K i indexed i <unk> 0,..., N -1 in a graph G characterised by adjacency matrix A. Let I i denote their income, where f <unk> (•) defines the PDF of a exponential probability distribution defined over the real half-line (0, <unk>) with parameter <unk> > 0 and F <unk> (•) the corresponding CDF. Let M be the number of links of node K i with M even. This leaves us with N -1 M = S possible permutations of link-neighbours. Assume further for Let now <unk> ij be an arbitrary realisation of a permitted set of incomes of nodes to which K i linked, indexed by j out of the set of permitted sets <unk> i with <unk> i = <unk> i1,..., <unk> iS <unk> and size S. Assume further that all incomes in <unk> ij are distinct. If link formation is independent of I i as the sole characteristic differentiating K i from all other nodes, all sets <unk> ij of the same size M are equally likely with probability 1/S by extension, since K i connects to any other node with equal probability. This would be the case for both standard preferential attachment models as well as ER random graphs. In our model, the probability p ik that A ik = 1 depends negatively on the absolute distance |I i -I k |, such that <unk>p ik /<unk>|I i -I k | <unk> 0. By linearity, the probability p ij of node i to have <unk> ij as their chosen set of incomes to which she is linked decreases in the sum of absolute differences, that is, It follows, that p ij as a local probability of a set of a given length being chosen by homophilic preferential attachment is maximised for a minimisation of I k <unk> ij |I k -I i |. Since the benchmark without homophily is equal probability of 1/S for all sets of a given size M, this condition also maximises the global probability that this set is chosen for a given size M. Formally, the minimisation problem chooses a set or sets <unk> ij such that Appendix C Linkage Based on Logarithmised Income --- Differences In this appendix, we discuss the case where agents seek to minimise relative rather than absolute income differences in tie formation, i.e., weights for tie-formation are inversely proportional to the absolute distance in log income. Agent j's weight in agent i's draw wij is thus determined by The scale-transformation by logarithmising tends to offset the property of the exponential income distribution to exhibit much higher (absolute) income differences in its upper tail than in the lower parts of the distribution, since the natural log has negative second derivative. For moderate to high levels of <unk> agents will have a strong tendency to choose an equal number of agents above or below them in income that is roughly homogeneous across the income distribution (apart from agents located at the boundary). For self-perceptions, this implies a'middle-class bias' per stylised facts (i) and, by extension, (ii), as is also readily verified by simulation with the results below in Figures 18 to 21. Logarithmising income in the weight function as in eq. ( 35) bears enormous consequences for inequality rather than self-perceptions, though. This is what Figure 22 below shows. Apart from agents at the respective upper and lower boundaries, the approximately homogeneous segregation tendency across the income distribution manifests itself in roughly + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + --- + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Figure 22: Plot of inequality perceptions against the income rank. Almost all individuals underestimate true inequality with a Gini of 0.5 in all cases of <unk> > 0. For moderate to high homophily strenghts <unk> between 4 and 14, there is little covariation of income rank and perceptions (except for the inflated perceptions at the upper and lower boundary). by them. For <unk> <unk> 4, any differential behaviour is then barely visible in local inequality perceptions within Figure 22, in contrast to both stylised fact (iii) and our results for choice with absolute income differences.

Across income groups and countries, individual citizens perceive economic inequality spectacularly wrong. These misperceptions have far-reaching consequences, as it might be perceived inequality, not actual inequality informing redistributive preferences. The prevalence of this phenomenon is independent of social class and welfare regime, which suggests the existence of a common mechanism behind public perceptions. The literature has identified several stylised facts on how individual perceptions respond to actual inequality and how these biases vary systematically along the income distribution. We propose a network-based explanation of perceived inequality building on recent advances in random geometric graph theory. The generating mechanism can * Daniel Mayerhoffer's contribution was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) -430621735. Furthermore, financial support for the paper by the University of Bamberg through the Fres(c)h grant no. 06999902 is gratefully acknowledged. We would also like to express our gratitude to two anonymous reviewers and Martin Everett for his excellent editorial work, Arndt Leininger and his students, Miriam Bömer, Bettina Gregg, Johannes Marx, Moritz Schulz, Eleonora Priori and Jan Weber as well as the participants of the Networks 2021 conference, the 9th PhD conference for the renewal of constitutional economics, the 9th ECINEQ Meeting, the 33rd annual EAEPE conference, the annual PhD conference by the Hans Böckler Foundation 2021, the 2nd Scientific Workshop by the Network for Pluralist Economics, the 8th Workshop on Complexity, Innovation and Knowledge (WICK), the Colloquium for the Advancement of Knowledge in Economics (CAKE) of the University of Utah and 28th DVPW congress, especially our designated discussants Macartan Humphreys, Claudius Gräbner and Pietro Terna which were of great help at crucial junctions of this investigation. Finally, we also thank Carsten Källner for his able research assistance. All remaining errors are, of course, ours. statistically distinguished from small-world networks, testifying to the robustness of our approach. Our results, therefore, suggest that homophilic segregation is a promising candidate to explain inequality perceptions with strong implications for theories of consumption and voting behaviour.

Introduction In 1985, the World Health Organization declared in Fortaleza, Brazil, that "There is no justification for any region to have a caesarean section rate higher than 10%-15%" [1]. The statement was based on the good maternal and perinatal care outcomes of the Nordic countries. In middle/low-income countries, the optimal caesarean section (CS) rate is influenced by preferences regarding delivery, available medical services, family income, and the health care professionals' qualifications, as well as parents' education level. Nevertheless, CS rates show a significantly increasing trend worldwide [2]. The most likely reasons include: reluctance to undergo a vaginal delivery (particularly amongst pregnant women with previous CS experiences), little improvement in facilities for vaginal delivery, and increased antenatal indications for CS consequent on improved equipment and technology (such as 3D ultrasound, fetal monitoring, fetal mirror, etc.) [3]. There is a growing trend in China for CS, from 2.0% (14/701) in 1978-1985 to 36.6% (813/2224) during 2006-2010 [4]. A 2011 survey conducted in multiple regions of mainland China found that the CS rate was 54.5%; 24.6% of these were performed in the absence of any medical indication [5]. Reports from 2014 show CS rates in China of 54.90% [6], and 58.5% [7]. In the United States, CS has become the most common surgical procedure; the CS rate increased from 20.6% in 1997 to 31.5% in 2009 [8]. In recent years, the CS rate in Europe was 19%-33% [9], in South America rates were 30%-50% [10], and there is a rising tendency in Africa and other regions [11]. Molina et al., further investigated the CS rate for all WHO Member States; 45 countries had a CS rate <unk>7.2%, 48 countries 7.2%-19.1%, 48 countries 19.1%-27.3%, and 53 countries >27.3% [3]. The increasing trend was illustrated by a significant average annual increase in primary (4.3%) and repeat (4.8%) CS rates from 1998 to 2008 in Australia [12]. Socio-economic factors have contributed to the increase in "unnecessary" CSs in China [7]. A large retrospective survey in China identified factors which led to CS delivery; these were not only personal (education, vocation, age of childbearing, residence) but also medical (for example hypertension, above average fetal weight, fetal malposition) [4]. The increased preference for CS is likely to be influenced by multiple additional factors, such as the financial status, underlying diseases, level of education, family/social environment, reproductive knowledge, media reports, feedback from social circles and medical staff, as well as the previous delivery experience [13]. Despite the greater number of complications and risks of CS as compared to vaginal delivery, women still tend to choose CS. Litorp's investigation concluded that these women focused on a perceived benefit for their child, and overlooked the risk to themselves of going through a CS [14]. In respect of the women's preference, health workers should also monitor the influence of different health care models [9]. In certain high-risk pregnancies, CS delivery is an effective measure to reduce maternal and prenatal mortality, however, high CS rates do not relate to any decline in maternal and neonatal mortality rates. Moreover, CS conveys risks of complications that are greater than those of vaginal delivery [15], (such as infant and maternal mortality, severe maternal complications (post-partum bleeding, organ damage, infection, pain, etc.). CS also triggers other socio-economic challenges, such as the excessive costs for infant intensive care, as well as for other health care resources. In Europe, it has been estimated that if appropriate decisions regarding mode of delivery were applied to those delivering after a previous CS, not only would 160,000 unnecessary caesarean sections be avoided annually, but about €150 million of additional expenses would be saved [16]. The increase in CS delivery over time has been maintained. As for the primipara, the increase may be related to differences in clinical decision making or maternal request. However, other reasons, such as anxiety, various pregnancy complications, a painful previous childbirth experience, as well as the neonatal morbidities caused by the vaginal delivery, may influence a choice to opt for CS [17]. On 29 October 2015, the Chinese government announced their new policy: an amendment to its 1978 single-child family policy would fully allow couples to have two children. Some expect that this may trigger the next baby boom in the mainland China. This study explores the socioeconomic status of the different populations on their preference for the caesarean sections. In particular, we compared the delivery preferences of women in their first and second pregnancies and analyze socioeconomic factors that impact on delivery preference. --- Participants and Methods --- Ethical Approval All subjects gave their informed consent for inclusion before they participated in the study. The study protocol was approved by the Ethics Committee of Chongqing Medical University (record number 2015008). --- Research Method A cross-sectional analysis of expectant mothers from June to August in 2015 from 16 hospitals in five regions of Mainland China was undertaken. The sample consisted of pregnant women in five regions, namely, Chongqing, Chengdu, Zunyi, Liaocheng, and Tianjin, China. Chongqing, Chengdu, and Zunyi are in south China, Liaocheng and Tianjin are in north China. Face-to-face interviews using a survey questionnaire were conducted by the investigators who were specifically trained, medical students. The participants were categorized into two groups based on delivery times: Expectant Mothers in their First Pregnancy and Expectant Mothers in their Second Pregnancy. --- Population and Sample Participants were those pregnant women who want to get examined in obstetrics clinic. Data were stratified sampling selected pregnant women (both expectant mother of first and second pregnancy). In 2455 target interviewers, 55 participants declined to answer any questions, and the preliminary response rate was thus 97.76% (2400/2455). Among 2400 respondents, the final analysis sample included the 2345 persons who answered all questions. --- Face Validation and Content Validity The questionnaires were established by a panel of nine experts (three nutrition experts, two target population experts, two epidemiology experts and two health education experts). --- Sampling Framework This study was conducted in obstetrics clinic in selected hospitals. The following guidelines were implemented during the Hospitals survey. To reduce admission rate bias, hospital level was divided into Level 3A hospital, Level 2A hospitals, and Level 2B hospitals and below. According to the hospital level, we randomly sixteen hospitals in five regions of Mainland China were included in this study. --- Survey Administration --- Participants-Pregnant Women Feeding Service Social Media-QQ Group To better understand the needs of pregnant women and give better service for pregnant women, we established a social media communication group (QQ group). This QQ group conveniently provided research personnel to answer questions raised by pregnant women and facilitate communication between pregnant women and researchers. Our research group's members answered pregnant women's questions in a timely manner. --- Network Communication among Team Members Network communication: we used a social media mobile app QQ (Tencent Group, Shenzhen, China) to strengthen communication among team members. We promptly collated and shared problems encountered during the investigation through the group, in order to exchange knowledge and experience. The network was used to share relevant data, and to provide online training for investigators. --- Investigators Team members were from different grades, including undergraduate and graduate students, academics from Chongqing Medical University, Chengdu University of Traditional Chinese Medicine, Jackson State University and the University of Adelaide, and hospital obstetricians. All investigators underwent standardized training and were familiar with the objectives and methodology. --- Process of Development of Questionnaire We designed and modified the questionnaire as follows. First, after the first draft of the questionnaire, students met to modify the questionnaire. Second, teachers from School of Public Health and Management, Chongqing Medical University modified the questionnaire. Third, Invited obstetricians revised the questionnaire. Fourth, foreign experts modified the questionnaire. --- Pilot Study A total of 24 individuals participated in a pilot test in June 2015. The questionnaire was subsequently modified, according to results of the pilot. Also, it became apparent that investigators required further training; this was provided. --- Work Manual of Investigators The manual was divided into two parts. The first part contained the overall plan; it included the background, purpose, technical route, research method, data processing, etc. It was written in the relatively simple language, to ensure that it was understood by all of the students. The second part contained details regarding organization and implementation, including the preparation of our materials, application of various funds, progress in other work, and our modification consequent on the pilot test. In this section, some of the important aspects of the investigation were stressed, to avoid mistakes during the investigation, and also to add relevant literature concerning the interview skills. We delivered the work manual to each member of the research group. --- Questionnaire The questionnaire was customized for the target population, with modifications based on the pilot study. The final draft of the questionnaire was agreed after several discussions with experts after review of the pilot investigation. We modified the questionnaire, especially the presentation of questions and improved the answer options of the questions. The revised questionnaire had an acceptable level of face and content validity and readability. Demographic data included age, residence (Urban/Rural), per capita income of the family (<unk>4500/<unk>4500 and <unk>9000/><unk>9000), occupation (Rural migrant workers/Urban and rural unemployed, unemployed/Industrial workers of Non-agricultural registered permanent residence/Individual business/Business services staff/Civil servants/Senior manager and Middle-level manager in large and medium enterprise/Private entrepreneur/Professionals/Clerks/Students/Others), advanced age for pregnancy(aged 34 years old and above) (YES/NO), chronic diseases (YES/NO), hospital level (Level 3A hospital/Level 2A hospitals/Level 2B hospitals and below), nationality (Han nationality/Minority), only child (Yes/No), husband is the only child (Yes/No), marital status (Unmarried/Marriage/Remarried/Divorced/ Widowed). Pregnancy was divided into three trimesters. Education level was categorized as <unk>primary school, junior middle school (basic education), <unk>a senior high school (including vocational/technical secondary school and junior college), (secondary education) and <unk>senior college and university (higher education). CS delivery intention among pregnant women were divided into 3 categories "Normal delivery intention", "Unsure prior to delivery", and "Caesarean delivery intention". --- Data Analysis The data was carefully reviewed prior to entry into the database using EPI Data 3.1 software (The EpiData Association, Odense, Denmark). Data analysis was performed using statistical software (SAS version 9.1; SAS Institute, Cary, NC, USA) after careful data sorting and cleaning. The characteristics of the participants were summarized using either means and standard deviations or frequencies and percentages, and were presented using descriptive analysis (means, standard deviations, and percentages). Chi-square tests were employed for comparisons when appropriate. Ordinal logistic regression analysis [18] was conducted to examine the risk factors of CS delivery intention among expectant mothers. --- Results --- Demographic Characteristics of Expectant Mothers In total 1755 and 590 expectant mothers in their first and second pregnancies were enrolled. 354 (15.1%) intended to deliver by CS and 585 (25.0%) participants were unsure prior to delivery. Furthermore, 156 (8.9%) expectant mothers in their first pregnancy and 198 (33.6%) expectant mothers in their second pregnancy intended to deliver by CS. 346 (15.4%) Han nationality and 8 (8.6%) minority nationality intended to deliver by CS. 42 (14.3%) in the first trimester of pregnancy and 97 (13.8%) in the second trimester of pregnancy, and 215 (15.1%) in the third trimester of pregnancy intended to deliver by CS. 263 (12.5%) participants who were advanced in age for pregnancy, and 91 (37.5%) participants who were not, intended to deliver by CS (see Table 1). Notes: Education level was categorized as <unk>primary school, junior middle school (basic education), <unk>a senior high school (including vocational/technical secondary school and junior college), (secondary education) and <unk>senior college and university (higher education). --- Ordered Multivariate Logistic Regression for Caesarean Delivery Intention To further investigate the factors that affect the characteristics of women with particular delivery preferences, we chose the following parameters: hospital level, nationality, only child, husband is the only child, marital status, education level, residence, per capita income of the family, career, trimester of pregnancy, advanced maternal age in pregnancy, parity. CS delivery intention (Normal delivery/Not sure/Caesarean delivery) was a dependent variable; we then conducted ordinal logistic regression analysis. In the ordinal logistic regression analysis model, partial regression coefficient (ß) = Estimate. OR = e ß. Compared with Han nationality, minority nationalities were less likely to have CS delivery intention (95% CI ( <unk>0.9588, <unk>0.0398), p = 0.0374). Compared with Expectant Mothers in their First Pregnancy, Expectant Mothers in their Second Pregnancy were more likely to have CS delivery intention (95% CI (0.6467, 1.0584), p <unk> 0.0001). Compared with women in early pregnancy, women in late pregnancy were less likely to have CS delivery intention (95% CI ( <unk>0.5215, <unk>0.0209), p = 0.0327). Women with advanced maternal age in pregnancy were more likely to have CS delivery intention (95% CI (0.4593, 1.0308), p <unk> 0.0001) (see Table 2). --- Discussion In view of the increasing CS rate in mainland China and the rest of the world [2], especially among pregnant women without any medical indications for CS, we investigated the child delivery preference among Chinese pregnant women during the transition of China's one-child policy. Our study shows that the preference of the natural childbirth accounted for 59.8%, CS 15.09% and "not sure" 24.94%. Consistent with these findings, Liu et al. [19] conducted a similar study among Shanghai pregnant women over last six years and reported that more people preferred vaginal birth (61.2%) than CS (24.7%). However, our study indicated that the proportion of intended CS delivery among second-time mothers is much higher than that of the first timers. In the ordinal logistic regression analysis model, this study found that compared with expectant mothers in their first pregnancy, expectant mothers in their second pregnancy were more likely to have CS delivery intention. Pregnant women of advanced age were more likely to have CS delivery intention; they may have had more experiences and psychological pressures to undergo a vaginal delivery. Of course, women childbearing for the second time is generally older than women childbearing for the first time, with higher pregnancy risk as well. Advanced maternal age is associated with higher risks of miscarriage, premature birth, stillbirth and higher rates of gestational diabetes mellitus, gestational hypertension and preeclampsia-eclampsia [20,21]. Overweight and obese older pregnant women have a significantly high risk, particularly for stillbirth and preterm delivery [22]. Therefore, it is critically important to provide the appropriate prenatal care (disease detection and prevention) to women who are pregnant for the second time. Effort should also be put to improve the quality of the medical professionals and their continuing education, strengthening the obstetric management, and monitoring the CS criteria. In order to reduce the unnecessary adverse outcomes, effort should be increased to strengthen the monitoring and management of older pregnant women so that reasonable dietary guidance and weight monitoring [23] can be provided, as well as improving health care awareness of physical health and social and psychological wellbeing. The Chinese government implemented the one-child policy nearly 40 years ago, and the CS rate gradually increased. The two-child policy will increase birth rates; this study found that expectant mothers in their second pregnancy were more likely to choose CS delivery. Consequently, the CS delivery rate may further increase. This study provides more important implications for the control of CS deliveries after China's new universal two-child policy. An interesting phenomenon in the survey is that the preference for CS among ethnic Han women is higher than that of ethnic minorities. Possible reasons include special phenomena, such as a desire for children to be born on special days named "Auspicious day", even "Auspicious hours"and thus, choose CS delivery. Women's awareness of childbirth is also subject to cultural, social norms and expectations, as well as their local medical conditions and the medical, advice they receive [24,25]. Other studies also demonstrated the influence of religions and communities on the attitude to CS [14]. For example, Janevic et al. [26] surveyed the delivery preference of women from different races and birthplaces in New York; after adjusting for multiple risk factors, women's CS preference differed by race and birthplace. All ethnic groups except East Asian women were at an increased risk of CS delivery; the highest rates were among Hispanic Caribbean women and African American women. Janevic et al. suggested that some potential factors should be further explored, including hospital environment, provider's bias, and patient preference. Efforts to reduce CS rates should address these disparities. Henderson et al. investigated the quality of obstetrics services for minority groups in England and Wales. They found that compared with whites, minorities have poorer obstetric services, and they argued that these services should be improved [27]. Compared with women in early pregnancy, women at the late stage of pregnancy were less likely to have CS intention. This may be consequent on prenatal pathology (such as malposition, pregnancy-induced hypertension, gestational diabetes, fetal growth restriction, an oversized infant, twins, umbilical cord around the neck and other fetal anomalies) and psychological changes. As they get closer to their due date, pregnant women may be more likely to pay attention to the decision of delivery mode. Women under anxiety and psychological tension are particularly vulnerable to external influences; for example, inpatients could be affected by women who are in labour, as well as by information from the medical staff. A 2010 report is worthy of mentioning; this re-defined the concept of term infants from 39 to 40 weeks (+6 days) [28]. The report pointed out that infants born less than 39 weeks may not perform optimally in the long run in reading and maths. It was believed that week 37-39 of the pregnancy is a critical development period of fetal brains [28]. Therefore, in the hospital, unless there is a medical exception, it is important for medical staff to nurture the correct delivery attitude, and organize periodic training, and to foster patient's confidence and a positive attitude. Although CS could effectively reduce the high-risk pregnancy complications and neonatal mortality, unnecessary CS should still be avoided for low-risk deliveries. Of course, patient's needs and the precondition of mother-baby safety must be taken into account in decision making. To decrease the C-section rate, we should first decrease the rate of CS through maternal request. Appropriate policies and guidelines should be developed to accomplish this goal [6]. CS delivery rates positively correlate with infant mortality rates among high-income industrialized countries. One cause of this phenomenon is iatrogenic preterm delivery [29]. Global health care professionals should put actions to strengthen the obstetric care and the accurate assessment of CS criteria [9], including providing of options of painless childbirth and education and psychological interventions, increasing of quality of natural delivery services, proper culture and prohibiting of doctors from professional opinions and profit [30]. This study has certain limitations. First, cross-sectional survey data reduced the ability to make direct causal inferences, to explore whether unmeasured factors may better explain the observed relationships we observed, and to determine the direction of causality. Second, the face-to-face survey administration design may convey information bias. Respondents may not have answered the questions truthfully. That said, all questions in the survey were reviewed by a panel of researchers and participants in the pilot study, and thus, the questionnaire was less likely to include items that could be perceived as sensitive by the study participants. During the face-to-face interview, investigators asked questions one by one to assure that respondents would answer seriously. Third, the authors have not actually asked the about the reasons for participants' choice of delivery, this study mainly focuses on the characteristics of the participants who chose caesarean delivery. Fourth, our study was not exactly nationally representative. The sample consisted of pregnant women in five regions, namely, Chongqing, Chengdu, Zunyi, Liaocheng, and Tianjin, China. Chongqing, Chengdu, and Zunyi are in south China, Liaocheng and Tianjin are in north China. --- Conclusions Estimated 8.9% of expectant mothers in their first pregnancies and 33.6% of mothers in their second pregnancies intended to deliver by CS. This study provides more important implications for the control of CS deliveries after China's new universal two-child policy. Any intervention programs to reduce the rate of CS should focus on the Han minority, pregnant women of advanced age, women in early pregnancy and, expectant mothers in their second pregnancy. --- Author Contributions: Lianlian Wang participated in the design of the study, provided interpretation of study results and drafted the manuscript. Xianglong Xu participated in the design of the study, performed the data analysis, provided interpretation of study results and drafted the manuscript. Yong Zhao participated in the design of the study and helped draft the manuscript. Philip Baker, Chao Tong, Lei Zhang, and Hongbo Qi, contributed to the interpretation of study results and helped draft the manuscript. All authors read and approved the final manuscript. --- Conflicts of Interest: The authors declare no conflict of interest.

Objective: This study explores the basic demographic characteristics of expectant mothers in the context of their intentions regarding mode of delivery, in particular, the preference for caesarean delivery, and analyzes the social and psychological factors that influence delivery preference. Method: A cross-sectional survey of pregnant women was conducted during June to August in 2015. This study adopted a stratified sampling method, and 16 representative hospitals in five provinces of China were included. Results: 1755 and 590 of expectant mothers in their first and second pregnancies, respectively, were enrolled in this study. 354 (15.10%) intended to deliver by caesarean section and 585 (24.95%) participants were uncertain prior to delivery. 156 (8.89%) of expectant mothers in their first pregnancy and 198 (33.56%) expectant mothers in their second pregnancy intended to deliver by caesarean section. Ordinal logistic regression analysis found that nationality, parity, trimester of pregnancy, and advanced maternal age were factors associated with intention to deliver by caesarean (ordered logistic regression/three-level caesarean delivery intention criterion; odds ratios p < 0.05). Conclusions: 8.89% of first pregnancy expectant mothers and 33.56% of second pregnancy expectant mothers intended to deliver by caesarean section. Any intervention program to reduce the rate of Caesarean delivery should focus on the Han population, older pregnant women, and expectant mothers in their second pregnancy, at an early gestation.

The resilience of family seems to be getting more challenges. The continuous development of technological advances and an era of competition that is getting tougher make it increasingly difficult for family members to interact well and warmly. Thus, to build "Home Sweet Home" will be more difficult. This is triggered by the widespread associations that violate norms and have no shame to show off their household problems. Both husbands and wives used their household problems as topics for conversation that become "pride". Therefore, they feel that it is normal and consider it as a common issue. Furthermore, there is an affair which can trigger quarrels and disharmony in the household. Finally, it ends in divorce. 1 Every couple want the integrity in building a household. However, the reality shows that the divorce rate is high. The existence of social pressure in society that divorce is neither a prohibition nor disgrace in society, divorce has become a common issue. 2 Divorce is the breaking up of a legal marriage before a court judge based on conditions stipulated by the law. As published in Article 113 of the Compilation of Islamic Law, that one of the reasons for breaking up of a marriage is divorce. In accordance with the law, divorce cannot just happen, but there must be reasons justified by the law to divorce. That is very basic, especially for courts which have the authority to decide whether a divorce is appropriate or not feasible to be implemented, including all the consequences that occur as a result of divorce. Therefore, it is necessary to understand the soul of the divorce regulation and the causes and consequences that may arise after the husband and wife have terminated their marriage. 3 Another definition of divorce is separation between husband and wife as a result of their failure to carry out their respective role bonds. In this case, divorce is seen as the result of an unstable marriage in which the husband and wife then they live separately and are legally recognized by the applicable law. In other words, divorce is a break in the family because both partners decide to leave each other so that they stop doing their obligations as husband and wife. However, the occurrence of divorce for children is a "sign of death" of the integrity of the family, it feels like half of the "self" of the child has been lost, life will not be the same again after their parents' divorce and they have to accept the sadness and feelings of deep loss. 4 The divorce rate in Indonesia is getting more worrisome from time to time. In 2013, the BKKBN asserted that the divorce rate in Indonesia was already the highest in the Asia Pacific region. In fact, in the following years the number of divorces continued to be high. Data from the Ministry of Religion submitted by the Head of the Kepenghuluan Sub-Directorate which was published in the national newspaper are as follows: Responding to data, a sample of last two years was taken, namely in 2012 and 2013, then the average divorce rate was taken every year as many as 350,000 cases, it indicated that in a day there were 950 cases or 40 cases of divorce every hour. Then, according to the Deputy Minister of Religion, dealing with the number of divorces, 70% of divorce lawsuits were filed by the wives, it implied that from the 40 divorce cases, 28 cases were proposed by the wife's side. The female-initiated divorce is called divorce by litigation, while male-initiated divorce is called divorce by thalaq or male repudiation. Then, cases of divorce by litigation occurred in Bantul Regency, Yogyakarta were higher than divorce by thalaq. The interesting thing from these data is that the majority of divorce cases were in the urban buffer districts. Meanwhile, based on the district area, in 2013 there were 55 cases of divorce by litigation occurred in Banguntapan district, 70 cases in Kasihan district, and 39 cases in Sewon district. Then in 2014 cases of wife-initiated divorce in Banguntapan district were recorded 79 cases, in Kasihan district 53 cases, in Sewon district 58 cases. In 2015, in Banguntapan district there were 21 cases of divorce by litigation, 49 cases in Kasihan district, and 19 cases in Sewon district. Meanwhile, in other districts the number was below the 3 urban buffer districts. There are several reasons of divorce. According to the Islamic Law Compilation in the article 116 (Zulkifli, 2018), the reasons of divorce allowed are: a. One of the parties commits adultery, drunkenness, gambling, drinking, and others which is difficult to be cured. b. One of the parties leaves the other party for 2 years without the permission of the other party and without valid reasons or other matters beyond the capabilities. c. Either party gets 5-year imprisonment or a harsher sentence after the marriage. d. One party conducts cruelty or serious maltreatment that endangers the other party. e. One of the parties gets a physical disability or disease that is difficult to cure which causes the person cannot carry out the obligations as husband or wife. f. There is a dispute or quarrel between husband and wife that cannot possibly be reconciled or reconciled again. g. Husband prohibits taklik divorce. h. Religious conversion or apostasy which causes disharmony in the household. Perception towards something can influence one to make a decision including decision to file a divorce. Therefore, perception of societies in Bantul regency towards divorce can also influence them in making decision in divorce. Perception (from Latin perception, percipio) is the act of compiling, recognizing and interpreting sensory information to provide an overview and understanding about the environment. Perception covers all signals in the nervous system that result from physical of chemical stimulation of the sense organs. For example, vision, in which the light is hitting the retina in the eye, smells use odor or aroma molecules, and hearing which involves sound waves. Perception is not a passive acceptance of cues, but is formed by leaning, memory, hope and attention. Perception depends on the complex functioning of the nervous system, but it appears to be non-existent because it occurs outside of consciousness. 7 The terms of perception is often referred to as a view, description, or assumption. Because perception relates to the response about one thing or object. Perception also has several definitions, including: a. According to Bimo Walgito: perception is a process that is preceded by sensing, which is a process that takes the form of receiving a stimulus by a person through the senses or also called the sensory process. b. According to Slameto: perception is a process that involves the entry of messages or information into the human brain. c. According to Purwo Darminto: perception is a direct response from an absorption or the process of someone who knows several things through sensing. d. In the dictionary of psychology, perception is defined as a process of observing someone's environment by using their own senses so that they become aware of everything in their environment. 8 Perception has a subjective character, due to it depends on the abilities and circumstances of each individual so that it will be interpreted differently by one individual to another. Thus perception is a process of individual treatment, namely giving responses, meaning, description or interpreting what is seen, heard, felt by the senses in the form of attitudes, opinions, behavior, or referred to as individual behavior. Selective perception is interpreting selectively what someone sees based on the individual's interest, background, experiences and attitudes. Factors that affect perception are divided into internal factors and external factors. Internal factors contained in the individual include several things, namely: a. Physiological: the information that come in through the senses, then the information obtained will influence and complement efforts to give meaning to surrounding environment. The capacity of the sense to perceive each person in different so that interpretations of the environment can also be different. b. Attention: namely individuals need a certain amount of energy spent to show or focus on the physical form and mental facilities that exist in an object. Energy of each person is different and this will affect the perception on the object. c. Interest: the perception of an object varies depending on how much energy or perceptual vigilance is moved to perceive it. Perceptual vigilance is a person's tendency to pay attention to certain types of stimuli or can be mentioned as the interest. d. Undirectional needs: this factor can be known from how strong an individual is to look for objects or messages that can provide answers in accordance with itself. e. Experience and memory: experience can be said that it depends on memory, it means that to what extent person can remember past events in order to know a stimulus in a broad sense. f. Mood: an emotional state affects the behavior of person, this mood shows how feelings at a time can affect a person in receiving, reacting and remembering.9 External factors that influence perception are characteristics of the environment and the objects involved in it. These elements can change the perspective of the individual around the world and influence how someone feels or receives it. Meanwhile, external factors that affect perception, including: a. Size: the bigger of the size, it is easier to understand, so that it is more effective to form perceptions. b. Color: contrasting colors, i.e. objects that have sharper light are easier to understand (to be perceived) than objects that have less sharp light (slightly). c. Stimulus uniqueness and contrast: external stimuli whose appearance against the background and surroundings completely outside the expectations of other individuals will attract a lot of attention. d. Motion or movement: individuals will pay a lot of attention to moving objects within sight rather than at stationary object. Based on the explanation above, the writer carried out the research to find out the factors contributing to the high number in cases of divorce by litigation and exploring the perceptions of societies in Bantul towards divorce. This research was expected to give recommendation for the problem of the high number of divorces particularly those by litigation. --- B. RESEARCH METHODS This was a qualitative research conducted in Bantul Regency Yogyakarta. Data collected was data from 2013 to 2015 in Banguntapan District, Kasihan District, and Sewon District. Data was collected through documentations from Bantul Religious Court and Bantul Ministry of Religion, and interviews with women filling for divorce, religious leaders, and public figures. --- C. DISCUSSION --- Factors Contributing to Divorce by Litigation and Perception of Societies in Bantul towards Divorce Bantul Regency is located in the south of Yogyakarta Special Region Province, which borders to the north with Yogyakarta City and Sleman Regency, south of the Indonesian Ocean, east to Gunung Kidul Regency and west to Kulon Progo Regency. The total population in Bantul Regency is 919,440 people or 299,722 families, with a population density of 2,012,193 people per kilometers. The population is spread across 17 districts, 75 villages and 933 hamlets, with an area of 506.86 km. From the 17 districts in Bantul area, Banguntapan District has the highest population density, namely, 4,218 people per kilometers, it is followed by Sewon District that has of 3,835 people per kilometers, then Kasihan District. The level of population density in the three districts is due to the large number of immigrants who buy land for residential houses or occupy housing that have emerged in the three districts because the location of these three districts are closest to the city center. Thus, Banguntapan District, Sewon District, and Kasihan District are called Urban Buffer District. The three districts that support the city should have social life dynamics that are different from the other 14 districts in Bantul Regency, in various fields that bring positive and negative consequences. Social dynamics that occur in society are in the form of changes in social values, prevailing norms in society, patterns of individual and organizational behavior, the composition of social institutions, strata or classes in the power and authority. It can be asserted that there have been social changes which include changes in social organization, social status, social institutions and structures. The findings of the study in three districts in Bantul regency indicated that the factors causing the high number in divorce by litigation were economic factors, level of education, consumptive lifestyles, and changes in societies' perceptions towards divorce. Head of the Office of the Ministry of Religious Affairs in Bantul Regency argued that the high number in cases of wife-initiated divorce was dominated by economic factors. Economic problems often result in disharmony between married couples, then the wife proposes divorce. Besides, the factor of the maturity of the husband and wife is still lacking. Including the lifestyle in the era of freedom and it eases for married couples to choose divorce when their household life is no longer getting along. 10 The spread of divorce cases from various factors will directly damage the younger generation, such as children of divorced families. Because they live in an unpleasant environment, full of contention, lack of attention and protection. As in Kartini's perspective, there are several causes of juvenile delinquency, one of which is the theory of psychogenesis.11 This theory emphasizes the causes of delinquent behavior or delinquency from a psychological aspect. Some of the factors that come from the theory of psychogenesis are broken home or divorces parents who started with a cold war in the family. 12 In the past, the public perception of divorce was that divorce was taboo and shameful. Especially if the divorce occurs at the wish of the wife. A woman who asks for a divorce from her husband is viewed negatively by society. So that divorce is considered something taboo and it is embarrassing in the society. As a majority of people who embrace Islam, the underlying view of divorce is through an understanding of the teaching of Islam itself. Divorce is an act that is hated by God. However, if there is no other alternatives, while divorce is a way out in a marriage that can no longer be maintained. 13 Factors that affect divorce are there is no existence of mitsaqan ghalidzan or the absence of strong marital ties due to the influence of third party intervene and the lack of responsibility of the husband. 14 Divorce is also shortcut in solving problems that exist in family. According to Ponijah, her husband does not ever think about the needs of the family. If Ponijah asks for money for family needs, for her child's school fees, her husband just promises and never fulfills it. 15 Apart from being triggered by family problems due to the economy and the lack of responsibility of the husband. Divorce is also caused by an excessive lifestyle. This is due to the influence of newcomers from the city, the majority of which have better economic stability than the original population. So that it affects the thoughts, behavior, attitudes and even appearance of the local person who wants to imitate the newcomers. 16 This lifestyle also affects the socialization of teenagers. In accordance with the current development, there are many cases of pregnancy before marriage. This seems to be a natural issue due to society's perceptions have begun to change and the decline in ethical values. 17 The lack of the role of parents as role models for their children and the ease of means of transportation and access to social media encourages the society to do many activities, both positive and negative. 18 Based on several interviews with various sources of information and data that have been collected, it can be concluded that there is an increase in economic independence and knowledge and social status of women, in this case the wife as a factor in the increase in cases of female-initiated divorce. The educational opportunities for women that are increasingly open are national assets that need to be optimized for its potential. However, on the other hand, the advancement of women's thinking, it gives opportunities for women's independence in various sectors, such as the economic, social, cultural and legal political sectors. By becoming more independent, women are more confident and better prepared to face problems including the decisions to sue for divorce. Responding to this conditions, it is not surprising that the number of female-initiated divorces is always higher than male-initiated divorce. Although the disharmony of family begins with the husband's irresponsible behavior, and as a source of commotion in the household, the husband does not take the act of divorce. While the wife and children feel depressed and suffer due to the husband's behavior, with all her courage, the wife proposes the decision to initiate divorce for maintaining the peace of the household and children. In addition, societies of Bantul regency believed that divorce was a common issue and considered as the best solution in solving family problems. Based on the research findings, that perceptions of society toward divorce were highly influenced by: 1. The parenting styles of parents and or other adults for the children and or their students have a very big influence on the development of mental attitudes later when the children or students have been adults. 2. The atmosphere in the house and the family that is not conducive, the habit of the family is looking for comfort, pleasure, peace outside the home. So that the function of home and family does not need to maintain the comfort and calmness. 3. Divorce is shortcut to break the chain of problems that are entwined in a household. With a divorce: as if" the problem will be resolved. Therefore, when the cracks in the household start to appear and have been going on for a long time and there is no solution that benefits the wife (especially divorce is the last way to be undertaken). 4. The status of a widow or widower is no longer taboo and shameful, but society can accept openly and respect the status decision. 5. The crisis of role model, good advice is easier to be said by anyone, anywhere and anytime. However, the good advice does not give the maximum effect, because people pay more attention and believe in actions seen. Therefore, as role model figures, children look for figures outside the home according to their wishes. Thus, a lot of teenage children who admire artists, public figures who are not necessarily good and in accordance with their personalities and family conditions. 6. The crisis of spiritual, religion is limited to knowledge and it is not an act or personality, so religious value are not a shield of life but merely ceremonial. 7. Social media, various types of social media from Twitter, Facebook, blog, Instagram, path will make it easier for someone to communicate and get information. However, with his convenience, both adults and adolescent, both educated and mediocre, they are less selective and less wise in responding to the advantages of the facilities offered by this information technology. 8. The Era of Women's Independence, Educational opportunities and achievements that have been achieved by woman have encouraged woman to have higher and stronger position in various aspects. By having a high and strong position, women are very independent. The independence of women makes them to have the courage for expressing opinions and making the decisions, as well as when they are treated unfairly. --- D. CONCLUDING REMARKS The problems of unwise use of social media, early marriage, sexual violence, irresponsible parents, unspoken marriages, consumptive attitudes, and excessive pleasure must be prevented and prohibited. Thus, that the result of this study is expected to contribute thoughts and input to parents in order to provide correct knowledge and understanding of married life to children so that they can form a family that is sakinah mawaddah wa rahmah and having the proper perception towards divorce.

This study took place in Bantul district since there were many divorce cases in this area, particularly more cases of divorce by litigation compared to those by thalaq every year. Moreover, it had the highest cases in the Special Region of Yogyakarta. According to data from the office of Ministry of Religious Affairs in Bantul, in 2013 the number of divorce by thalaq was recorded 194 cases, while the number of divorce by litigation was 456 cases. In 2014 the number of divorce by thalaq was 186 cases while the number of divorce by litigation was 553 cases. In 2015, there were 55 cases of divorce by thalaq and 198 cases of divorce by litigation. This study aimed at finding out the factors contributing to the high number in cases of divorce by litigation and exploring the perceptions of societies in Bantul towards divorce. This study used qualitative methods, so then data was collected from documents from Bantul Religious Court and Bantul Ministry of Religion, and interviews with women filling for divorce, religious leaders, and public figures. The findings of the study indicated that the factors causing the high number in divorce by litigation were economic factors, level of education, consumptive lifestyles, and changes in societies' perceptions towards divorce. In addition, societies of Bantul regency believed that divorce was a common issue and considered as the best solution in solving family problems.

Introduction With frequent and frequent natural disasters, China is one of the countries most severely affected by natural disasters. According to the data released by the Ministry of Emergency Management, in 2022, a total of 112 million people were affected by various natural disasters in China, 554 people were killed and missing due to disasters, 2.428 million people were urgently relocated, 120,71.6 thousand hectares of crops were affected, and direct economic losses amounted to 238.65 billion yuan [1]. How to deal with the impact of natural disasters and minimize the risk and loss is still an issue that governments at all levels need to pay attention to. However, due to the uncertainty, suddenness, potential severity and other characteristics of natural disasters, as well as other objective conditions, it is far from enough to rely on the power of the government to carry out disaster relief, and it also needs the joint participation of multiple subjects such as social organizations, the public and enterprises. Hebei Province is a province where natural disasters occur frequently. There are many kinds of natural disasters with high frequency and great harm. Flood, geological disaster, typhoon, drought, hail, low temperature freezing and snow disaster and other types of disasters have different degrees of occurrence. In the past ten years, the cumulative number of people affected by natural disasters in Hebei Province was 110 million, and the direct economic loss was nearly 162.48 billion yuan, posing a certain threat to the life and property safety of people in Hebei Province. So far, although the government of Hebei Province has made some progress in disaster relief, the disaster relief work is increasingly inseparable from the joint participation of the society. Based on this, this paper attempts to analyze the role and realization mechanism of multiple subjects in disaster relief by combining the current situation of multiple subjects participating in natural disaster relief in Hebei Province, explore the difficulties faced by multiple subjects participating in disaster relief, and put forward relevant optimization suggestions, so as to build a more scientific and effective multiple disaster relief mechanism. --- Literature Review About disaster relief, domestic and foreign scholars have done a lot of research on it. In the context of frequent natural and man-made disasters, disaster relief has been paid more and more attention by the government and scholars. In China, disaster relief mainly relies on the government, and social participation is less, but in recent years, multiple subjects such as social groups, enterprises, the public and the media have actively participated in disaster relief and played an important role. In this context, scholars have discussed the multi-subject participation in disaster relief from different angles, mainly in the following aspects. (1) Multiple subjects participate in the scientific research of disaster relief At present, the theoretical research on multi-subject participation in disaster relief in the academic circle mainly focuses on three aspects. The first is governance theory, represented by Rosenau [2] and Gerry Stoker [3]. Stocker believes that governance is an activity in which entities with a common goal participate, either government or civil society organizations. The concept of governance proposed by the United Nations Governance Committee in Our Global Partnership is widely used in academic circles, which emphasizes the diversification of the subjects managing public affairs, including not only the government, but also nongovernmental social organizations, enterprises and individuals [4]. In short, no matter which concept is adopted to explain the theory, it provides theoretical support for multi-subject participation in disaster relief. The second is the multi-subject theory. Multi-subject participation in social public affairs is a new governance mechanism and an inevitable requirement for building a harmonious society in the new era. This theory proposes that in the governance of social public affairs, in addition to the state or government organization, other social forces should also actively participate, such as market organizations, social organizations and the public. In terms of governance methods, it is not only to obey the orders of government organizations, but also to adopt consultation and other ways to break the pattern that the government is the only power center [5]. Allowing multiple subjects to participate in the framework of disaster relief is conducive to making up for the shortcomings caused by "government failure" and better safeguarding the interests of the broad masses of people. Finally, there is the theory of emergency management, which was first used in the military and national security fields. Emergency management refers to the process in which the government, in response to emergencies, analyzes the causes, processes and impacts of emergencies in order to minimize the harm, and effectively collects information from all parties to optimize decisionmaking [6]. In recent years, natural emergencies have become more and more frequent, and the pressure on the government to deal with such incidents is also increasing. However, in the practice process, due to various practical conditions, emergency management has been expanding from the government to multiple subjects such as the private sector and the third sector. Therefore, through the participation of multiple subjects, it is an effective way to form an effective emergency management mechanism. (2) Research on the necessity of multi-subject participation in disaster relief In the field of disaster relief, social organizations are not a simple supplement to the government, but an indispensable force in the process of cooperative emergency response, which can provide important assistance to government organizations in terms of resources, technology and other aspects (Kapuca, 2011) [7], and from the perspective of medical rescue based on examples. The role and importance of ngos and the private sector in disaster relief work were analyzed (Kimiko Yamashita, 2019) [8]. At the same time, some scholars believe that disaster relief is not only a matter for the government, but also requires the participation of ngos, enterprises and individuals as governance subjects (Nasution Bahrul Ilmi et al., 2022) [9]. It is necessary to build a cooperative model of linkage between the government and local governments. When ngos, enterprises and ordinary people are absorbed into the disaster relief system, such a whole-society coordinated disaster relief model is effective (Rosenzweig et al., 2021) [10]. In China, there is no lack of research on the necessity of multi-subject participation in disaster relief. For example, in the book On Disaster Relief, the author did a lot of research on the necessity of social organizations participating in disaster relief, and explained the necessity of social organizations participating in disaster relief from the two aspects of participation motivation and the relationship between social assistance and the government (Li Xiaojin, Li Qiang, 1996) [11]. Some scholars used figurative metaphors to illustrate the necessity of multisubject participation in disaster relief, comparing disaster relief to "a wall against the wind" and ngos to "cement for the wind wall" (Jia Xijin, 2005) [12]. In addition, multi-subject participation in disaster relief also plays an important role in providing service, supervising the government, timely detecting crises and providing countermeasures and suggestions (Zheng Anyun, Dai Yaling, Miao Danmin, 2008) [13]. It can be seen that disaster relief not only needs the power of the government, but also the participation and support of social forces (Zhang Yong, 2016) [14]. (3) Research on the advantages and difficulties of multi-subject participation in disaster relief In terms of advantages, foreign scholar Salamon (1981) believes that the advantage of social organizations is flexibility, which can provide different services according to different needs [15]. Yan Wenhu (2010), a Chinese scholar, analyzed the role of government organizations in disaster relief and believed that government organizations have advantages such as fast response speed, high rescue quality and wide social resources [16]. Zhang Yong and Liu Jun (2011) pointed out that ngos play an indispensable role in disaster relief due to their public welfare, flexibility and professionalism, and play a positive role in disaster relief combined with the overall governance of the government [14]. In terms of difficulties, foreign scholar Salamon (1981) put forward the voluntary failure theory, believing that social organizations have limitations such as insufficient charity resources and unprofessional charity work [15]. John Twigg and Diane Stenner (2011) also believe that social organizations have difficulties such as unreasonable organizational structure [17]. Chinese scholar He Xiao (2009) pointed out in his research that social organizations still have problems such as insufficient legitimacy of volunteers, poor organizational management, weak organizational influence and surplus volunteers when participating in disaster relief [18]. Based on the performance of social organizations in the Wenchuan earthquake, Meng Tian (2014) affirmed their advantages such as fast response and meticulous work in his research, and pointed out that social organizations still have problems such as legitimacy dilemma, participation dilemma, cooperation dilemma, resource dilemma and capability dilemma [19]. --- The Effectiveness of Multi-subject Participation in Disaster Relief (1) Government level Regarding disaster prevention and relief, the Hebei Provincial government has always put the interests of the broad masses of the people first. In recent years, in combination with the specific deployment and relevant requirements of the CPC Central Committee and The State Council on the reform of the disaster prevention, reduction and relief system under the new situation, Hebei Provincial Government has achieved certain results in disaster prevention and relief. In terms of policies and regulations, the Hebei Provincial government follows The State Council's "National General Emergency Plan for Public Emergencies". The Regulations on Natural Disaster Relief, the National Comprehensive Disaster Prevention and Mitigation Plan and other policies and regulations, It has successively promulgated the "Hebei Provincial People's Government General Emergency Plan for Public Emergencies", "Hebei Provincial Natural Disaster Relief Emergency Plan", "Hebei Provincial Comprehensive Disaster Prevention and Reduction Plan", "Hebei Provincial Natural Disaster Relief Measures", "Hebei Provincial Major Meteorological Disaster Emergency Plan", "Hebei Provincial Flood Control and Drought Relief Emergency Plan", "Hebei Provincial Flood Control and drought Prevention and Typhoon Emergency plan", and "Hebei Provincial geological emergencies Policies and normative documents such as the Disaster Emergency Plan not only clarify the responsibilities and division of labor of local governments at all levels in disaster relief and prevention, but also provide policy and regulatory guarantees for multi-subject participation in disaster relief, and promote the specific implementation of disaster prevention and reduction and relief work. It can be seen that the disaster relief of Hebei Provincial government has achieved certain results in the construction of policies and regulations. In terms of disaster relief expenditure, Hebei Province is one of the provinces with the most serious natural disasters due to its complex geological structure and frequent natural disasters such as flooding, drought, hail and cryogenic freezing. Therefore, the Hebei Provincial government has always attached importance to the investment in disaster relief. According to the existing data of the National Bureau of Statistics, the disaster relief expenditure of the Hebei provincial government in 2019 was 4.689 billion yuan, 5.293 billion yuan in 2020 and 5.892 billion yuan in 2021, which was generally higher than that of other provinces in the country, which undoubtedly provided a solid material guarantee for disaster relief work and ensured the effective and smooth progress of disaster relief work. In addition, the Hebei provincial government has also implemented agricultural insurance against natural disasters in some cities, and established a unified rural housing insurance system to further reduce the losses of vulnerable groups in natural disasters. In short, for disaster relief, the Hebei provincial government has achieved certain results in both superstructure and material basis. (2) The social dimension As mentioned above, due to the characteristics of public welfare, flexibility and professionalism, social organizations play an indispensable role in disaster relief, which can be combined with the coordination and integrity of the government to play a positive role in disaster relief. Of course, the participation of social organizations in disaster relief is also based on relevant laws and regulations. For example, Article 3 of the Charity Law stipulates that non-governmental organizations may voluntarily carry out disaster relief work by donating and providing services, and Article 8 of the Law of the People's Republic of China on Earthquake Prevention and Mitigation stipulates that the government should actively encourage and guide nongovernmental organizations and individual citizens to carry out earthquake mass detection and mass prevention activities. It is necessary to rely on the strength of multiple subjects such as social groups, volunteers, communities and enterprises to form an emergency management mechanism with overall command, quick response, complete functions, orderly coordination and efficient operation. At present, Hebei Province has a considerable number of professional organizations that can carry out disaster relief tasks. For example, the Hebei Red Cross Society, the Hebei Charity Federation, the Hebei Charity Fund Joint Foundation, the Hebei Love Rescue Team and the volunteer service federation of various cities. Among them, the most representative is the Hebei Red Cross Society, which has so far established 57 rescue teams in different categories such as medical, search and rescue, and public health, ranking third in the country. In addition, in recent years, some enterprises and volunteers in Hebei Province have also actively participated in disaster relief work, and achieved good results. Taking the Zhuozhou flood relief situation in Baoding as an example, various social groups, enterprises and individual citizens uphold the spirit of "one party in trouble, eight parties support" and actively respond to the government's call to participate in flood relief. Among them, many social organizations such as Hebei Photovoltaic New Energy Chamber of Commerce, Baoding Female Photographers Association, Baoding Electric Power Equipment Industry Association, Langfang Wenzhou Chamber of Commerce and Cangzhou Boai Life Volunteer Service Association actively carried out donation activities to timely transport necessary living and medical supplies for the disaster area. In addition, a total of 120 volunteer service teams and 18,000 volunteers from Handan, Zhangjiakou, Langfang, Cangzhou, Shijiazhuang, Xingtai and other places participated in Zhuozhou flood relief work. According to preliminary statistics, the Zhuozhou flood relief work, a total of 800 social organizations participated in the province, all units and individuals donated a total of about 105 million yuan, carried out more than 760 rescue tasks, rescued more than 20,000 trapped people. It can be seen that the participation of various social groups, enterprises and individuals not only reduces the pressure of the government in disaster relief, but also improves the efficiency of disaster relief work, which is exactly the effect of multi-subject participation. --- The Problems of Multi-subject Participation in Disaster Relief From the practice of multi-subject participation in natural disaster relief work in Hebei Province, although each subject has made great efforts in disaster relief, and the multiparticipation has also played an advantage in disaster relief, there are still some problems in practice. (1) The legal status of multiple participants is unclear Although the rules and regulations of the Hebei Provincial government on disaster relief are relatively perfect, mainly including Hebei Provincial Natural Disaster Relief Emergency Plan, Hebei Provincial Natural Disaster Relief Measures, Hebei Provincial People's Government General Emergency Plan for Public Emergencies, etc., there are few detailed regulations on multiple participants in disaster relief. In particular, there is a lack of institutionalized provisions on the coordination mechanism among the participants and the division of powers and responsibilities. The government is in the leading position in disaster relief, but the legal status of other participants is vague, which will discourage its enthusiasm in disaster relief. On the other hand, with the continuous expansion of the volunteer service team, some rescue teams organized by enterprises, individuals or social groups to rescue in the disaster area, due to the absence of relevant volunteer laws, there will be chaos in the disaster relief process, greatly reducing their rescue capabilities and collective centripetal force. (2) The information communication mechanism of multiple participants is not perfect Due to the high frequency of natural disasters in Hebei Province, which have brought different degrees of harm to society and citizens, it is necessary for multiple parties to participate in coping with the impact of natural disasters. However, since the division of labor and responsibility of multiple entities in disaster relief is not clear, it is difficult to achieve the effect of coordination and cooperation in disaster relief. Although there are many channels for the release of disaster warning information in Hebei Province, it is mainly released to the public through TV, Internet, mobile phone SMS, electronic display and other channels, which can enable social citizens and rescue organizations to prepare for disaster prevention and relief in advance. However, after the disaster occurred, there was a lack of information exchange and feedback platform, insufficient disclosure of disaster information, and information communication and sharing between multiple participants and between the government and multiple participants could not be realized, resulting in unclear task distribution among various participants, ineffective coordination of rescue forces, and inaccurate implementation of the "last mile" rescue. (3) The lack of professional knowledge and skills of multiple participants in disaster relief At present, the people's Liberation Army and the armed police forces under the leadership of governments at all levels are mainly responsible for natural disaster relief tasks. In recent years, non-governmental rescue teams have gradually joined the disaster relief team. However, due to the lack of special disaster relief training and professional knowledge learning, some rescue teams blindly carry out disaster relief tasks without understanding the situation of the disaster area, which not only fails to cooperate with the government's disaster relief operations, but also causes chaos at the disaster relief site. In addition, the public's own professional knowledge and skills of disaster relief are insufficient, and they are often in a passive state when facing natural disasters, resulting in poor self-rescue and mutual rescue ability of the public. At the same time, since citizens learn about the disaster information of natural disasters mainly through the early warning issued by the government before the disaster, they fail to pay timely attention to the spread of the disaster or the external relief situation, which will greatly affect the rationalization of the victims and hinder the orderly development of disaster relief work [20]. --- Strengthen the Multi-subject Participation in Disaster Relief Countermeasures (1) Improve the rules and regulations for multiple entities to participate in disaster relief On the one hand, it is necessary to constantly improve and improve the rules and regulations for multi-subject participation in disaster relief. Hebei Province can fully learn from the rules and regulations for multi-subject participation in social relief in other provinces and combine the relevant practical experience of the province, so as to clarify and strengthen the scope of power of social organizations, enterprises, volunteers and the public under the legal framework. Further clarify the scope of responsibilities of multiple entities. According to the characteristics of different participants, the rules and regulations with high feasibility and strong operability are formulated to provide a good institutional environment for multiple participants to participate in disaster relief and ensure the legitimacy and enthusiasm of multiple participants. On the other hand, the government should pay full attention to the legal status of social organizations, enterprises, volunteers and the public to participate in disaster relief, give sufficient financial support, and further improve the corresponding reward and incentive policies. At the same time, we should improve the two-way supervision mechanism, not only introspection but also mutual supervision. The government should accept the supervision of itself and multiple subjects, and multiple subjects should also accept the supervision of the government and themselves. Maximize the integration of human, material and financial resources of the whole society, achieve efficient cooperation of multiple subjects, and carry out rescue work in an orderly manner. (2) Improve the information coordination mechanism for multi-subject participation in disaster relief First of all, we need to smooth the channels of information communication. As the leading force of disaster relief, when natural disasters occur, the government will announce the early warning information of natural disasters to the society through official channels, so that other participants can timely understand the disaster level and the situation of the disaster area and formulate disaster relief plans. After the occurrence of natural disasters, the government can announce the disaster relief process, rescue and relief consultation, and material needs of victims to the public through the official website, broadcast, social platforms (Tiktok, wechat public account, Weibo) and other channels, so as to increase the participation of the public and multiple subjects. Secondly, build a multi-subject information sharing platform based on digital technology. Information sharing platform is a bridge connecting the government and multiple subjects, and it is particularly important to build an information sharing platform led by the government and participated by multiple subjects. At present, the Hebei Provincial government has its own egovernment platform, which can realize the exchange and cooperation between various government departments. On this basis, it can be combined with blockchain, big data, Internet of Things, artificial intelligence and other technologies [21] to build a multi-subject information sharing platform to realize the two-way flow of information and data interoperability in disaster relief, and ensure the timeliness, accuracy and authenticity of disaster information. (3) Improving the professional level of multi-subject participation in disaster relief Considering the characteristics of current disaster relief and the existing problems of multisubject participation, first, the government should fully encourage multi-subject participation in disaster relief, provide appropriate financial or expert support, regularly conduct disaster relief lectures, encourage the establishment of professional disaster relief teams, and encourage them to carry out timely and appropriate simulation exercises in peacetime. Strengthen the capacity building of relevant rescue workers, and improve the response ability and professionalism of disaster relief teams. Second, regular disaster relief exercises and training are organized for the people in cities and counties prone to natural disasters in Hebei Province, so as to improve their mutual and self-rescue ability and psychological quality in natural disasters, so that the people can truly apply what they have learned and cope with natural disasters calmly [22].

The prevention, reduction and relief of natural disasters is a worldwide problem. In recent years, with the vigorous promotion of the social governance system of "government-led, social coordination and public participation", Hebei Province has achieved phased results in the field of natural disaster relief by formulating relevant regulations and policy documents, actively guiding multiple social entities to participate in disaster relief in an all-round way, establishing disaster early warning mechanisms and giving play to the role of the market. But at the same time, there are some problems, such as unclear legal status, lack of interactive platforms and lack of professional skills of disaster relief participants. In view of the above problems, this paper puts forward some countermeasures to improve the rules and regulations for the participation of disaster relief subjects, improve the information communication mechanism among disaster relief subjects and enhance the professional level of disaster relief subjects.

Introduction A large number of studies have demonstrated that poor nutrition is disproportionally experienced by those with socioeconomic disadvantage (e.g. low income, low educated). For instance, individuals of lower socioeconomic position (SEP) reportedly consume diets higher in fat, lower in micronutrient density and have lower intakes of fruit and vegetables than those of higher SEP [1][2][3][4]. In children, an inadequate diet is associated with higher body mass index (BMI) and disease, both in childhood and adulthood [5][6][7]. A key prerequisite to promoting good nutrition among low SEP children is to better understand the mechanisms underlying healthy eating behaviours among this group. Social ecological models suggest that aspects of the home or neighbourhood environment, as well as personal factors, are likely to predict health behaviours such as healthy eating [8,9]. It is well recognized that the family environment plays a key role in the development of eating behaviours and food consumption among children [10][11][12][13]. Parents are primary socialization agents who transmit messages to children about their eating [14]. Over the past decade, the association between parental feeding style and practices and children's eating behaviour has received increased empirical focus. For instance, adolescents who describe their parents as authoritative (strict yet involved and supportive) have been found to have healthier dietary behaviours than those who describe their parents as authoritarian (also strict yet uninvolved) [15,16]. Parenting practices that encompass both direct (e.g. verbal encouragement to eat specific foods) and indirect (e.g. modelling consumption of specific foods) communications have been found to influence children's eating [17]. For instance, prohibiting palatable foods and exerting excessive control over access to unhealthy foods has been found to increase taste preference and desirability for that food, leading to weight gain and poor eating habits [18]. In addition, implementing 'food rules', such as offering dessert as a reward for consumption of vegetables, have been found to increase the child's preference for the reward food [19][20][21]. Furthermore, a recent review of qualitative data on parental perceptions regarding healthy behaviours for preventing overweight in young children revealed that several parenting strategies such as lack of time, lack of motivation and decreased perception of responsibility for child weight management acted as barriers to behaviours for promoting healthy eating and preventing overweight [22]. In addition to parental feeding styles and practices, parents' own eating styles have also been show to influence children's eating behaviours. Parents' food preferences, the foods they consume and make available to their children and their restrained eating (e.g. dieting) influence children's eating behaviours, potentially as a result of social modelling and by the choice of foods made available to their children [14]. In addition to the family environment, many factors within the broader environment have also been found to encourage or inhibit healthy eating among children. For instance, healthy eating behaviours among children have been associated with good accessibility to quality healthy food options in their local neighbourhood and involvement in the growing process of fruits and vegetables (e.g. from their home vegetable garden) [23]. The media has also been consistently reported as exerting a negative influence on children's eating and weight status. For example, children exposed to food advertising of unhealthy foods have been found to show increased preference for promoted foods and higher consumption and purchasing requests to parents for advertised foods [24,25]. However, although the majority of food advertisements promote unhealthy foods [25], there is some evidence to suggest a relationship between advertisements promoting nutritious foods and positive attitudes and beliefs about these foods [26,27]. Friends and peers have also been found to influence children's food preferences and eating behaviour. For instance, a recent qualitative study highlighted parental reports that peer consumption of certain foods (i.e. from their school lunchboxes) encouraged preferences for obtaining and consuming these foods [23]. While there is good evidence for an association between various family and environmental influences on children's eating behaviours, the majority of studies have focussed on parent report measures only and on child unhealthy BMI or risk of overweight/obesity as the main outcome. Rather than focussing on obesity-inducing behaviours, it may be fruitful to examine'resilient' children; socioeconomically disadvantaged families with children who manage to eat well, despite increased risk of poor nutrition. To our knowledge, no study has simultaneously assessed parent and child perceptions of barriers and supports to healthy eating exclusively among a sample of low SEP children of normal body weight and with a diet including aspects of good nutrition. Enhancing our understanding of supports for healthy eating and methods employed to overcome unhealthy eating among disadvantaged children who eat adequate intakes of fruit and vegetables may inform nutrition interventions and obesity prevention among this group. The current study was designed to gain a better understanding of the barriers and supports assisting children to eat well. Specifically, we aimed to gain indepth insights into the family and environmental factors underlying resilience to poor nutrition from children and their mothers living in disadvantaged neighbourhoods. --- What helps children eat well? A qualitative exploration --- Methods --- Participants The participants were 38 mother-child pairs. A summary of the sociodemographic characteristics for the sample is provided in Table I. Briefly, the mean age of mothers and children was 38 and 9 years, respectively, and the majority of the sample (79%) resided in rural regions of Victoria, Australia. Participants were drawn from a larger study of 4349 women who participated in the Resilience for Eating and Physical Activity Despite Inequality (READI) study, a longitudinal cohort study examining resilience to obesity among socially and economically disadvantaged women and children residing in rural and urban areas of Victoria, Australia [28][29][30]. Disadvantage was defined using area-level disadvantage, which has been shown to have associations with poorer health outcomes, independent of individual markers of disadvantage [31]. Women in the READI study were randomly selected from neighbourhoods ranked in the lowest Victorian tertile of relative disadvantage, an index that considers area-level income, education and employment [32]. For the READI study, all women completed a baseline survey that assessed individual, social and environmental factors potentially associated with physical activity, diet and weight. From the READI baseline cohort, women with young children (n = 1680) were invited to complete a survey about the health and lifestyle of their child (n = 685, response rate = 59%). For the current study, participants were selected using purposive criterion sampling. From the cohort of women who had completed only the baseline survey and a survey for their child (n = 685), 67 children aged 8-12 years were identified as resilient. For the current study, resilience was defined as: healthy weight range (using mother reported height and weight to calculate BMI for age percentiles where children rated between the 5th and <unk>85th percentiles were considered within the healthy weight range [33]), regular consumers of fruit (>1 serves per day) and vegetables (>2 serves per day) and physically active (ranked top 50% for participation in moderate-to-vigorous physical activity measured using objective accelerometry physical activity measures). The criteria for healthy eating were based on the distribution of fruit and vegetable consumption for this group. Increasing the criteria further (i.e. to include more than 2 serves per day), the size of the eligible sample would have been reduced too much to recruit sufficient participants for this study. Fruit and vegetable consumption was used as a marker for a generally healthier diet in selecting participants and was assessed from survey data from an item that asked 'How many serves of fruit (in a separate item-vegetables) do you usually consume each day?'. The child age range was selected, as we were interested in prepubescent children who were considered old enough to participate in an interview session. An analysis of qualitative data relating to physical activity and sedentary behaviours and their determinants in mother-child pairs is to be the subject of a separate paper. From these 67 eligible children, 12 parents did not respond to the study invitation, 8 women refused participation and 6 women had relocated, and following recruitment, there was a dropout rate of 3 (7%) due to work commitments. The remaining 38 children (response rate 57%), along with their mother, participated in the current study. --- Procedure In July 2009, mothers were mailed a letter explaining the study and inviting them (and their child) to participate. After receiving the letter, each potential participant was contacted by phone (no more than twice) by the field manager in order to arrange a time and date for the interview. Interviews were conducted from July to October 2009. Two trained research staff conducted each interview at participant's homes. At each visit, two separate interviews were conducted, one for the mother and one for the child. The child interview was conducted without the mother immediately present (i.e. the mother was in the family home but not sitting in on the interview). The child was not present for the mother interview. With the participant's permission, an electronic dictaphone was used to record each interview and hand written notes were also taken L. K. Williams et al. by a staff member. Interviews followed a specific interview schedule and lasted approximately 20-45 min. Participants were advised that they had been selected because the child consumed relatively more fruit and vegetables compared with other children in the READI cohort and that we were interested in finding out how and why this was the case. At the end of the interview, mothers' were presented with a $25 gift voucher and children a $10 voucher in recognition of and gratitude for their time. Ethics approval for this study was granted by the Deakin University Human Research Ethics Committee. --- Materials The social ecological model guided the development of a semi-structured interview schedule that included questions that assessed mother and child perceptions on supports and barriers to fruit and vegetable consumption, patterns of healthy eating more generally and physical activity. The existing evidence, our own previous research and a series of pilot interviews were used to develop the interview schedule items. Only the healthy eating questions were examined for this paper. Different interview schedules were used for mothers and children. Sample questions from the interview schedule include (mother) 'What do you think helps your child eat a healthy diet?', 'Is there anything that you think makes it difficult for your child to eat a healthy diet?', 'Do you actively do anything that you think helps your child to eat a healthy diet?', (child) 'What do your parents do to help you eat healthy The children's interview also involved the researcher presenting the child a slide show on the computer that displayed various pictures, such as family meals, breakfast, fruit choices and afterschool snacks. Pilot interviews with children indicated that the inclusion of visual aids, such as a computer slide show, helped to make the interviews feel less threatening and also provided children with a prompt that helped them to think and comment on their eating habits and food preferences. The mothers' interview included similar types of questions regarding what they believed was helping (and making it difficult) for their child to eat a healthy diet and what they did as parents to influence their child's eating behaviours. --- Coding and analysis Interviews were transcribed verbatim. The first author then read all the transcripts to develop a detailed hierarchical numerical coding scheme that was used to code all transcripts. Open, axial and selective coding, utilizing NUD*IST (QSR International, 2002) version six qualitative software program, was used to code, subcategorize and unify coding of transcript text [34]. Transcripts were coded to identify mother-child pairs [e.g. 1 (mother) and 1.1 (child of mother 1)]. Template analysis [35] and inductive thematic analysis [36] were used to develop and interpret the themes [37]. For instance, some themes emerged from a list of codes (template analysis) identified in the textual data of the interview schedule, while our knowledge of the literature shaped a further set of themes that emerged from transcript data, not directly related to the interview schedule items (thematic analysis). Researcher triangulation [38] was employed to increase the validity of the data and its interpretation. This was achieved by having the second author read the transcripts and then select a random sample of 10 mother-child transcripts that were cross-coded to check for inter-coder agreement. No major discrepancies in coding or interpretation were observed. --- Results The results are presented around two main themes that emerged from the data: (i) active strategies from parents to promote healthy eating behaviours and (ii) external barriers and supports to healthy eating. Active strategies from parents to promote healthy eating encompassed four sub-themes: parental control and moderation, support for healthy eating, eating rules and parental role modelling. External barriers and supports to healthy eating included three sub-themes: access, advertising and friends/family. Each of the above main themes is illustrated with excerpts from participants. Excerpts from children include their code (calculated as their mothers code plus 0.1), gender and age in parentheses. Excerpts from mothers include their code and the gender and age of their child in parentheses. The code is used in replacement of pseudonyms (pseudonyms are used in excerpts where names have been provided by participants) and to identify mother-child pairs (e.g. participant 10 is the mother of participant 10.1). Theme 1: active strategies from parents to promote healthy eating behaviours Parental control and moderation: limiting and controlling access to unhealthy food The most salient theme that emerged from the interviews from both mothers and children was the perception that children consumed a healthy diet because parents had control over the amount and type of food the children consumed. In some instances, this was directly related to the child's age, with mothers articulating an anticipated decrease in control as their children became older. Parental control over food was enforced by limiting access to unhealthy food, limiting food choice, emphasizing moderation and restricting unhealthy food options to special occasions (e.g. weekends, celebrations). For many mothers, the responsibility of consuming a healthy diet was reportedly in their control and not L. K. Williams et al. the responsibility of the child. This involved much more than simply providing and encouraging consumption of healthy food, it was also about refusing requests for less health food choices, emphasizing moderation and maintaining an 'eat this or nothing' approach to food and healthy eating. I do give them treats occasionally,... I try and get them to eat really healthy. I do buy the odd treat for them. Like, there's a box of Nutrigrain up on the shelf. but the kids know that during the week they're to eat Weetbix or something healthier and on the weekend that's a treat for them. So, it's still. they still get their treats. But, in moderation. (11: mother of son aged 7) The impact of such parental control of food was clearly articulated by many children. For the majority of children, food preferences (whether unhealthy or healthy) were directly dictated by taste. Although a small number of children reported eating certain foods to reduce the risk of weight gain or to sustain energy, most children reported taste as the deciding factor for preferred foods rather than health benefits. However, while taste strongly dictated preferences, preferences did not always dictate consumption. Food consumption was reportedly largely governed by parental control. Hence, while a large proportion of children articulated a strong preference for unhealthy foods, they reported that access to these foods was restricted or limited by their parents. Chocolate balls (cereal), they're not allowed to have that as breakfast because I consider them lollies because there's so much sugar in them... we buy a jar of Nutella and it's for school holidays... (24: mother of son aged 8) My mum sometimes buys chocolate balls which are these round things that are chocolate. It's actually a type of cereal that we're not allowed to have as cereal, we're just allowed to have them as snacks, treats. A treat, it's very rare. (24.1: male aged 8) Well I would like ice-cream [for breakfast] but I know I wouldn't be able to have that. (28.1: female aged 10) Support for healthy eating: encouragement to consume fruit and vegetables, healthy eating education and provision of fruit and vegetables Some mothers reported that they verbally encouraged their children to eat fruit and vegetables, a theme that was reiterated by a small number of children. Maybe if we go down the street and she wants afternoon tea I try and coach her into choosing something that's healthier than something else. Mothers reported that encouragement to eat well was also communicated by educational messages about the benefits of healthy eating. However, although many children reported that they received encouragement to eat well, none of the children specifically reported receiving messages of an educational nature. Although children did not report receiving messages of an educational nature, many children appeared to have a good knowledge of foods, in addition to fruit and vegetables that were and were not healthy. And yeah we had that discussion, the content of sugar in cereals and I showed them on the side of the packet the rice bubbles and they were amazed at that at the time. (24: mother of son aged 8) Support for healthy eating from mothers was not always communicated through verbal encouragement and healthy eating education. Both mothers and children reported the positive impact parental provision of adequate fruit, vegetables and healthy food options had on healthy eating. Mothers reported that their child maintained a healthy diet What helps children eat well? A qualitative exploration because they provided healthy food: in lunchboxes, for dinner and by having a well-stocked supply of fruit and vegetables for snacking. When children were asked what their parents did to help them eat healthy food, almost all respondents reported that their parents provided them with fruit and added vegetables to their meals. Look there's a big fruit bowl I keep on the bench and the kids know they can eat that any time. There are certain foods they can graze on whenever but then there are other foods that it's, you know, off limits. With fruit I mean yeah always two or three pieces in their lunch box. (37: mother of daughter aged 10) They usually make healthy food like vegetable soup and when they make food they put in peas and carrots and lettuce. (7.1: female aged 8) --- Eating rules Many mothers reported that they mandated eating certain meals, namely breakfast. Some mothers also mandated three meals a day. While almost all the children reported consuming breakfast, lunch and dinner, they did not articulate an awareness of this eating rigidity or report any concerns with eating the standard three meals a day. Sometimes Jessica will be ''I'm running late for school, I haven't had breakfast yet'', or ''I've still got to have breakfast''. 'Yes you do have to eat breakfast before you leave, you don't skip breakfast''. You can skip other things if you're running late but yeah you don't skip your breakfast or your tea. (19: mother of daughter aged 10) In addition, many mothers reported implementing specific rules to promote healthy eating. These included prohibiting dessert until vegetables were consumed, consuming leftover fruit from lunchboxes before further food is made available, prohibiting 'junk food' before lunch and mandating a minimum fruit and vegetable daily consumption quota (e.g. must eat two pieces of fruit a day). Many children also articulated an awareness of these eating rules. The rule is you've got to eat three different coloured vegetables... You can choose which three but you've got to have three. --- Parental role modelling Compared with direct influences on healthy eating, indirect influences, namely role modelling (both positive and negative) reported by mothers emerged as a less salient theme. Only one child reported an awareness of her parents' eating habits. [I] lead by example. I think we eat a fairly healthy diet. (23: mother of son aged 9) Mum is very healthy. Dad's not completely healthy... Dad's nowhere near as healthy as Mum. Mum's always having the healthy food. And Dad's just happy with a sandwich or if he's down the street he might get a pie. (20.1: female aged 10) --- Theme 2: external barriers and supports to healthy eating Access The majority of mothers reported good access to fruit and vegetables within their local community. Some mothers reported the poor quality and range of fruit and vegetables available at supermarkets, yet overcame this barrier by growing their own fruit and vegetables or travelling further to access better quality produce. Some of the mothers resided in rural and provincial areas where fruit and vegetables are the towns primary industry and hence access to fresh seasonal produce (e.g. from orchardists and 'pick your own' produce outlets) acted as an environmental support to healthy eating. Residing in these rural areas, for some, also resulted in less access to mainstream take-away options and fast food, which mothers reported reduced their child's consumption of these unhealthy options. --- L. K. Williams et al. We basically have to buy our fruit and vegetables from a supermarket whereas if we went to much of the bigger towns that are nearby you've got access to the mini-markets and more of a range. I think that we're very limited. So we're going to extend the veggie patch this year. Yeah, we've got lettuce, at the moment we've got silver beet in there, garlic, we've got the herbs out the front. and peas, yeah. We grow it to eat it. (20: mother of daughter aged 10, rural resident) The school have the Stephanie Alexander program so they cook and they've got the garden and they grow everything at school as well. And living out here you've got the market gardens, and the fruit stalls and everything out there. So the special treat when they go shopping is they used to get given an apple. (24: mother of son aged 9, rural resident) Yeah seasonal fruit is fantastic and so this area for seasonal fruit is really good. It's free and it's fresh and it's off the tree so you can't get better than that. A good thing about this region is there's beautiful peaches, apricots, like all our friends are mostly orchardists so we just duck around and pick some peaches or pick some apricots. Yeah so seasonal is a big deal. We try and avoid having apples or things that are out of season. (29: mother of daughter aged 13, provincial resident) For those children who had access to a school canteen, it was reported that canteens acted as a support for healthy eating with almost all school canteens banning or limiting unhealthy food items. There's no lollies or chocolates or chips or anything like that [at the canteen] anymore, it's all just. healthy options now. [1: mother of son aged 11] [The] school canteen, which is just starting this, just the end of last semester. It's, it's pretty healthy... I'm sort of part in there, helping out. And we're all doing healthy food. --- Advertising The majority of mothers and children did not report that advertising negatively influenced their food purchasing or consumption behaviours. In fact, many mothers and children reported the positive influence that some advertisements had on healthy eating. Some children also reported that while some of the unhealthy advertised foods appeared appealing, they did not pursue attaining advertised products due to knowledge of them being unavailable to them (due to geographical or parental restrictions). Actually we both enjoy watching. there's a Safeway ad that tells you about what fruit and vegies to eat now and we both watch it. If one of us misses us the other will come and say, ''Look this is the one that's in season. Let's get this and try this''. Or if it's a new fruit we haven't seen we always try to get it. --- Friends/Peers Children and mothers reported both the negative and positive influence of their siblings, friends and peers on healthy and unhealthy eating predominantly through verbal comments, peer modelling and from the presence of desirable healthy and unhealthy food items in friends' lunchboxes. If you're over at a friend's house they'll go and get chips or something... Then I'll normally have chips with them. (25.1: male aged 12) [My child] is in prep and he'll see something in someone's lunch box and he'll come home and What helps children eat well? A qualitative exploration say ''I want that tomorrow''.he definitely does that. [1: mother of son aged 11] I made some dairy gluten free vegetable slice and it was lovely cold. We sat there and had it together. She came home from school, ''I'm never having that again''. ''Why not?'' ''The kids said it was disgusting and they was teasing me''. I said, ''That's okay, you just tell them that they're missing out on good stuff''. [2: mother of daughter aged 8] --- Discussion The purpose of the current study was to explore the family and environmental factors underlying resilience to unhealthy eating. Individual interview discussions with mothers from disadvantaged neighbourhoods and their children revealed the presence of parental strategies and external barriers and supports to promoting healthy eating behaviours. This is one of the first studies to include both mother and child reports and focus exclusively on low SEP families of children who eat well. Our study underscores the importance of focussing specifically on'resilient' children (i.e. those eating relatively well) to further elucidate potentially effective parent-child attitudes and behaviours in preventing unhealthy eating. Compared with previous studies that have predominantly focussed on the barriers to healthy eating and a healthy weight status, our results highlighted the active role mothers from disadvantaged neighbourhoods played in promoting healthy eating. For instance, almost all the mothers in the current study believed that parents were the main vehicle for influencing healthy eating and as a result, they exercised significant control over their child's food by implementing 'food rules', providing access to fruit and vegetables and restricting unhealthy food items. Although there is some evidence to suggest that excessive control over access to certain foods and implementing food rules has a negative effect on eating and weight [18,19,21], the children did not report their mothers to be too strict or controlling. Perhaps, this was because mothers also offered education and explanations about unhealthy food items and promoted the importance of being healthy. It is also possible that the children in the current study, particularly those who were younger, were also accustomed to this parenting style (or unaware of anything different) and shared similar attitudes about food and eating. These parenting attributes (i.e. strict, restrictive, yet encouraging and supportive) are consistent with an authoritative parenting style, which has been demonstrated as a positive predictor of healthy eating and weight status [16]. The emotional climate created by authoritative parents is one of the high strictness and involvement, warmth, emotional support, appropriate granting of autonomy and clear, bidirectional communication [39]. As parenting is a learned process, teaching parents to implement practices inherent in an authoritative parenting style (e.g. by accepting responsibility for their child's diet, prohibiting unhealthy foods and encouraging and modelling healthy food choices) is a viable avenue for nutrition promotion interventions among low SEP families. In addition, children in our study had a good knowledge of healthy and prohibited foods, despite not articulating an active awareness of specific food-related education. The finding that transference of healthy eating education from mother to child was likely internalized by children highlights potential limitations of relying on children's selfreport in evaluation of health promotion interventions. The results from our study also highlighted some environmental influences as both barriers and supports to healthy eating among families residing in disadvantaged neighbourhoods. Previous research has indicated the negative impact on eating of advertising and poorer access and availability of healthy food options [40,41]. Some mothers from our study reported instances of poor availability and quality of healthy produce yet many had developed strategies to overcome these barriers, namely, through responding to and creating more sustainable access to fruit and vegetables and other healthy food options. For instance, many families had their L. K. Williams et al. own fruit and vegetable garden, a practice consistently associated with increased fruit and vegetable consumption [23,42]. Although a number of families in the current study benefited from residing in rural or provincial areas where fruit and vegetables were the town's primary industry and accessibility to larger garden space was more available, it is possible that provision of skills and resources for home-grown produce is a potential avenue for increasing fruit and vegetable consumption among low SEP families. Previous research has consistently highlighted the negative impact advertising has on children's eating, yet our results indicated that although some children reported an awareness of the negative influence of food advertising, most children did not feel negatively influenced by televised food advertisements. There are three plausible explanations for this finding. Firstly, many children reported low access to outlets selling unhealthy food, a likely outcome given the large proportion of rural residencies (e.g. limited fast food outlets, small supermarkets). Therefore, children may not be affected by advertising of foods unavailable to them. Secondly and unsurprisingly, given the high control over unhealthy food practices exhibited by mothers when children reported desirability for unhealthy advertised foods, they perceived these foods as prohibited by their parents. Thirdly, parents who restrict access to unhealthy food may also be more likely to restrict television viewing time, making exposure to these advertisements less prominent. It is possible that these factors offer protection against the negative influence of unhealthy food advertising. Another external support for healthy eating articulated in the interviews was school canteens. At present Victorian Government only stipulates canteen 'guidelines', which are not consistently implemented at each government school [43]. Our data suggest little evidence of canteens serving unhealthy food or acting as a negative influence on children's healthy eating. Although it has been previously reported that consumption of foods from Australian canteens contributes a small amount of daily food intake for children [44], the suggestion that schools are limiting their options to healthy choices is positive. Strengths of this qualitative study are the large sample, inclusion of both mother and child reports and a focus on healthy eating among resilient children from low SEP families. Some limitations of the current study warrant consideration. Reports of external supports and barriers particularly, rely on participants' perceptions of their own local environment, which may differ from objective supports/ barriers. They may also differ across samples, population groups or geographic areas. This limitation was most notable given the large proportion of participants residing in rural and provincial areas. It is interesting to observe that the majority of children identified as resilient to overweight, obesity and unhealthy eating from the larger study resided in rural areas. Although there is no clear evidence that a difference in BMI exists between Australian rural and urban children [30,45,46], the results from the current study suggest certain factors that may be protective of unhealthy eating and potentially overweight and obesity among rural children from disadvantaged neighbourhoods. Given that the study focused exclusively on women and mothers, it does not offer any insight into the role that fathers and other extended family members have on their child's eating. For instance, almost 95% of the sample was married, and although only one child mentioned her father, it is possible that fathers are potential sources of influence on children's healthy eating. A further limitation present in most research that involves interviews is that participants may answer in a socially desirable manner. This limitation is perhaps more salient in the current study where participation in the larger survey had occurred and exposed adult participants to the broad aims and objectives of the research conducted. In the current study, effects of social desirability were minimized by conducting interviews independently for parents and children. In addition, children were not privy to the main aims and objectives of the project. Finally, we only assessed barriers and supports to healthy eating among children who consumed a relatively healthy diet rather than to draw comparisons of those who What helps children eat well? A qualitative exploration consume less adequate diets. There has been much research that has focussed on predictors of unhealthy behaviours and by using a qualitative study design and restricting our sample in this way, it enabled us to generate hypotheses from those who manage to remain resilient about potential avenues for nutrition promotion among disadvantaged children. The results of the current study offer insight into potential avenues for nutrition promotion among disadvantaged children residing in socioeconomically disadvantaged areas. Further research should focus on comparisons of the barriers and supports to healthy eating identified from the current sample among disadvantaged families with children who donot manage to eat well. Although our findings need to be replicated in a larger sample using survey methods, the results tentatively suggest that active parental strategies such as exercising control over access to unhealthy food, provision of education and encouragement for consumption of healthy food, provision of healthy food options and positive role modelling may aid the prevention of unhealthy eating among disadvantaged children. Lessons learnt from families equipped with strategies to overcome environmental barriers to healthy eating, such as skills in sourcing local and home-grown produce, may also facilitate healthy eating among disadvantaged children. --- Conflict of interest statement None declared.

It is well known that persons of low socioeconomic position consume generally a less healthy diet. Key determinants of unhealthy eating among disadvantaged individuals include aspects of the family and external environment. Much less is known about family and environmental determinants of healthy eating among social disadvantaged children. The aim of this study was to gain insight into the family and environmental factors underlying resilience to poor nutrition among children and their mothers living in disadvantaged neighbourhoods. Semi-structured interviews were conducted with 38 mother-child pairs (N 5 76) from disadvantaged neighbourhoods. Children were selected if they were a healthy weight, consumed adequate intakes of fruit and vegetables and were physically active. Two main themes emerged from the interviews: active strategies from parents to promote healthy eating and external barriers and supports to healthy eating. Mothers believed that exercising control over access to unhealthy food, providing education and encouragement for consumption of healthy food and enabling healthy food options aided their child to eat well. Children did not perceive food advertisements to be major influences on their eating preferences or behaviour. The results of the current study offer insight into potential avenues for nutrition promotion among disadvantaged children.

Introduction Political, social, technological, and economic changes have brought new paradigms and challenges to the organizational world, and it is crucial that organizations adapt in order to remain competitive in the market. CC have emerged in this context and may be characterized as an industry with a high rate of growth, at both international and national levels [1]. CC have become the main form of contact and interaction with customers [2,3], whereby communication is made through a number of channels (i.e., emails, telephone calls), reducing costs for organizations and improving customer service [4]. CC therefore offer various services, ranging from problem solving (and dealing with complaints) to providing additional information [5] and fostering customer loyalty [6], as there is closer proximity between the organization and its customers [7] in so far as there are no geographical barriers and round the clock service is offered [8]. According to the Portuguese Association of Contact Centers (APCC), there are over 80,000 employees in Portugal. However, although CC contribute to a personalized and higher quality customer service, they are associated with high levels of employee turnover, absenteeism, stress and burnout [9]. These consequences are the result of monotonous and repetitive duties, comparable to a modern form of Taylorism [10], where employees have little autonomy over their work and tasks [11] which are neither complex nor challenging [12]. This simplification of work results from a structural division where, in most cases, an employee only performs front-office (answering calls) or back-office (administrative tasks) duties. The lack of rewards is compounded by low pay and a high workload which can have a negative impact on employees' stress and well-being [13]. It should also be noted that the scripts used by these employees, with detailed instructions that structure and organize their intervention [14], enable a high level of control by the organization. Nevertheless, some research has shown that Human Resources practices can be developed [15][16][17] which, combined with the establishment of positive relationships with the supervisor (leaders) and co-workers (peers) [18] can mitigate these negative effects on employees' stress and well-being. In this study, the aim is to ascertain the extent to which the different back-office and front-office duties differ both in terms of the psychosocial work environment, and the levels of stress and well-being experienced by the employees. Considering the nature of the work of front-office and back-office employees, job characteristics, social support, HR practices, work-life conflict, workplace attitudes and well-being and general well-being were compared in both duties. Additionally, this study seeks to ascertain whether these characteristics are identical across all the companies, or whether there may be differences in the tasks performed or in their management. To this end, the employees of a total sample of 15 CC companies were studied, thus making it possible to estimate the proportion of the psychosocial work environment conditions, stress and well-being attributable to the characteristics of the organization. Hence, a further aim is to clarify whether the characteristics inherent to the work of CC are inevitable or whether they depend on the duties performed (i.e., back-office and front-office) or organizational context characteristics. Since there is no theoretical development according to which it might be possible to establish potentially significant differences, this study is of an exploratory nature, seeking solely to ascertain whether there is consistency between the job, context and well-being characteristics of back and front-office employees in fifteen distinct companies. Consequently, this study contributes to the construction of a healthier working environment, in an area characterized by constant growth. --- Theoretical Framework 2.1. Job Characteristics According to the JD-C (Job Demands-Control) model, the demands and control (i.e., autonomy) job characteristics can explain the stress and well-being of employees, and situations of stress occur when the job is characterized by high demands and low control [19][20][21]. Previous studies have shown that CC are associated with high demands related to the way employees are constantly monitored and evaluated [10]. The evaluation of employees is mostly based on quantitative criteria, which consider several factors such as the number of calls, their duration and also the number of calls on hold [12]. Therefore, this excessive workload, resulting from HR systems' pressure to meet the pre-established goals of the organization and the constant monitoring [22] make CC work highly demanding. As for autonomy, CC employees are generally considered to have low control and to be dependent on the planning and organization of the tasks they perform [8,11,23]: In addition to the fact that employees do not control when or with whom they speak, they are obliged to follow scripts that organize and structure their intervention [14]. Thus, and with recourse to the job characteristics model of Karasek [20,21], CC are predicted to be environments characterized by high demands and low control (i.e., autonomy) [2,3,10]. --- Social Support Social support has been recognized as a way of mitigating stress and reducing the negative effects of high demands and low control situations [14,[24][25][26] and the JD-C(S) (Job Demands-Control-Support) model identifies social support as a third dimension that can influence stress at work [20,21]. Social support occurs as a result of the fact that social relationships and interactions at work act as resources to combat job demands [27], as employees receive the information they require and develop different coping strategies that can be used in their daily lives [18]. Since employees with access to more resources are better able to effectively respond to any demands that may arise [2,28], the CC that promote social support may be predicted to promote employees' resources to combat the demands of the task and therefore promote their experience of well-being and reduce their levels of stress. --- HR Practices In the same vein, the HR practices adopted by the organization can also contribute to the well-being of employees [16]. Employees' perception of job characteristics (i.e., demands, control) are influenced by the Human Resource (HR) practices in place [1] which may lead to increased or decreased levels of stress [22]. Several Human Resource Management (HRM) models do not focus on employee performance alone [29,30], such as the HIM (High Involvement Management) Model, which highlights the importance of empowering employees through power, information, knowledge and rewards, while equally ensuring their performance and well-being [31]. The challenge of HRM in relation to CC lies in establishing a balance between HR control practices, geared towards standardizing the work, and HR practices compatible with the HIM model that seek to reduce employees' stress [32,33]. Therefore, it is important to study a broad range of HR practices and analyze their impact on employees' stress and well-being [34,35]. To this end, the recruitment and selection process, the welcoming and integration process, training opportunities, rewards and the performance evaluation process were deemed HR measures that can directly influence CC employees' levels of stress. --- Recruitment and Selection One way to contribute to the well-being of employees is by recruiting people with a suitable profile for the job and the organization [36]. Thus, it is important to invest in the recruitment and selection process, since it can influence the quality of the work, the interpersonal relationships of employees and also the services provided by the organization [37]. In the context of CC, and since these employees perform unchallenging, monotonous and repetitive tasks, the recruitment focus should not be on seeking highly skilled employees, but rather on identifying individuals with soft skills, thus giving priority to behavioral/social skills over technical skills [38]. On the other hand, Chapman and Webster [39] also highlight the fundamental role of recruiters in this context, since they can positively influence employees' perception of the job characteristics. In addition to the responsibility of recruiting people with a suitable profile for the job and the organization (facilitating their integration), recruitment also contributes to employees' adjustment of expectations and behaviors by clarifying their role in the organization [40]. --- Welcoming and Integration Process The welcoming and integration process is important, as it reduces the levels of stress of the new employees, thus providing them with a positive experience at the beginning of their new job [41] and contributing to their well-being [42]. This process fosters the construction and development of the employee-organization relationship, facilitating the sharing of information, internal communication, and team cohesion [43], while also accelerating the new employee's adaptation, since it stimulates the acquisition of knowledge regarding the culture, values and goals of the organization [44]. --- Training Training promotes personal development [45] which contributes to the development of employees' personal resources (i.e., self-control) [46]. Thus, through a set of duly planned learning experiences, individuals acquire new knowledge and technical skills that can facilitate the execution of their tasks, which in turn reduces job demands [47]. In fact, training can be a strategy used by HRM to alleviate stress, as the sharing of knowledge and strategies are tools which better prepare employees to respond to job demands [48]. --- Compensation and Rewards Several CC have adopted a variable salary component for all the employees who meet the pre-established objectives and goals. Batt [49] has identified this incentive compensation as an HR practice that is compatible with the HIM model, which equally values employee performance and well-being [31]. This HR practice aligns the interests of the organization with the interests of all its employees [50], contributing to job satisfaction and the well-being of employees [51] who feel that their effort is being rewarded [52]. --- Performance Assessment Performance assessment consists of the continuous monitoring process of employees' behavior and performance, which enables an assessment of how efficiently they perform their duties [53]. CC use high levels of monitoring with quantitative and qualitative criteria by which employees are assessed [54]: These criteria include not only customer satisfaction, but also the number of calls made, their duration, and the number of calls on hold [3,4,12]. This practice ensures standardization of the job but can have a negative impact on employees' well-being [9]. In fact, according to Deery, Iverson and Walsh [14], high monitoring levels are for several reasons associated with increased levels of stress among employees. Firstly, stress may result from high demands which may lead to a role conflict [55], as employees are expected to establish a positive relationship with customers. However, on the other hand, they are also obliged to meet quantitative criteria (i.e., quantity and speed of calls), leading to an intensified workload [8]. Moreover, there may be additional pressure to meet the pre-established goals, since in most cases CC employees work according to an incentive compensation system [56]. Secondly, the high degree of monitoring reduces employees' autonomy, as they are obliged to follow scripts that structure and organize their interaction [57]. Finally, besides a heavy workload and limited autonomy, the constant monitoring to which they are subject also implies high emotional regulation on the part of these employees, as their performance is also assessed through customer satisfaction [15]. Employees therefore use the few resources they have to combat the additional stress they experience as a result of being observed, instead of focusing on providing a quality service [58]. However, Grant and Higgins [59] maintain that performance assessment can have a positive impact on the well-being of employees. According to these authors, monitoring can be a means of identifying training needs, thus promoting the development of employees' new skills and knowledge, which, as previously mentioned, is associated with the reduction of stress levels [47]. In this regard, the effects of monitoring and its impact on employees' well-being depend on how the performance assessment data is used. --- Work-Life Conflict According to the role conflict theory [60], an individual's resources are finite and decrease according to the roles they play. Thus, and based on the resource scarcity (i.e., time, energy) hypothesis, role conflict arises when the demands of each domain are incompatible, and the individual is obliged to choose where to apply these resources [61]. As resources are finite, when individuals participate in one domain (i.e., work), this implies an investment of their resources (i.e., time and energy), and consequently their participation in other domains is compromised [62][63][64]. The job characteristics of CC are associated with high levels of stress, impairing employees' participation in other fields. --- Attitudes and Well-Being at Work 2.5.1. Organizational Commitment Organizational commitment is a psychological state which defines the employee's level of identification with the organization and its objectives [65] As an attitude, organizational commitment reflects the bond which links employees to the organization for which they work [66]. According to Meyer and Allen [67], this bond may be represented in different ways (i.e., affective, normative and continuity commitment), which condition the behavior of employees. Affective commitment is negatively associated with stress [68], as employees develop a positive emotional relationship with the organization and regard its goals as being compatible with their own [69], leading to a reduction in the ambiguity of their role [70] which enhances their well-being [71]. It has been acknowledged that CC are associated with high levels of stress and therefore affective commitment may be used as a resource [72] to combat stress [73]. The studies of Schmidt [74] conclude that, due to the job characteristics of CC, employees with an emotional connection to the organization display lower burnout levels than their co-workers. --- Work Engagement Work Engagement is a stable and persistent psychological state which reflects the wellbeing and motivation of employees at work [75]. According to Schaufeli and Bakker [76], it is through work engagement that employees' energy levels are expressed, reflected in their effort and persistence in the face of difficulties (vigor), their enthusiasm, pride and inspiration (dedication) and also the intrinsic pleasure and concentration associated with the performance of their duties (absorption). Thus, it may be concluded that a workforce with high engagement may constitute a competitive advantage [77], since this variable is positively associated with job satisfaction [78], general well-being [77] and is, consequently, negatively related to stress [76]. However, engagement depends on the resources (social, physical and organizational characteristics) obtained by individuals and used in the work context [79]. Therefore, the organization should provide the resources required by all its employees in order to promote their intrinsic satisfaction and enhance their well-being. --- Burnout According to Maslach and Leiter [80], burnout is a means of identifying stress in the workplace, reflected in the employee who has not been able to adapt to the duties/organization. This may be operationalized as a prolonged response to emotional and interpersonal stressors at work and may be analyzed through two core dimensions: Exhaustion and cynicism [75]. Exhaustion refers to feelings of extreme fatigue, emotional overload, and a lack of energy and emotional resources to perform one's work. Cynicism consists of adopting negative, cold, and distant attitudes towards work [80]. Maslach, and Leiter [80] have identified a number of burnout risk factors such as excessive workload, lack of control, and low pay. Thus, low control, a high workload and low pay contribute to the onset of burnout in CCs [81]. Moreover, burnout is negatively associated with employee satisfaction and well-being, and positively related to stress [81]. Thus, one of the challenges faced by organizations is that of adopting measures that contribute to the reduction of burnout. --- General Well-Being According to Johnson, Cooper and Cartwright [82], there is a correlation between job satisfaction and the physical and psychological well-being of employees. Thus, it is important to analyze dimensions such as job characteristics, social support and the HR practices adopted by the organization in the context of CC, as these variables can explain and predict the satisfaction and general well-being of employees [11]. Considering job design and the job characteristics model [23], it may be said that work in CC is monotonous and demanding and employees have a low level of autonomy. Therefore, low job control, high job demands, and the limited diversity of tasks have a negative impact on employees' satisfaction and are also associated with high levels of stress [12]. Although there is little flexibility in terms of monitoring and job design in CC, several studies have pointed to a solution being found in the HR practices adopted by the organization [8] and the promotion of social support, as both these features can mitigate the effects of stressors [22]. Such is also the case with organizations that implement measures to foster a work-life balance, as they increase employee satisfaction and, consequently, contribute to their general well-being [83]. --- Method --- Procedure and Sample The data collection for this study was carried out as part of a research project conducted within the scope of a partnership with the Portuguese Association of Contact Centers (APCC), with the purpose of identifying and diagnosing psychosocial risks at work in the context of CC. To such end, associated companies were contacted by APCC management to participate in the study. The employees of the CC companies who agreed to participate were notified by HR of the objectives of the study and were invited to take part in the study. Through the SurveyMonkey platform, a link was generated which directed participants to an online survey. Finally, the employees were informed that their participation was voluntary, confidential and anonymous. A convenience sample was obtained, corresponding to a total of 2232 employees from 15 different CC companies, with a response rate of over 70% (ranging from 71% to 81% among the companies, corresponding to 32-432 respondents per company). However, due to a lack of responses to some of the assessed scales, only 1440 participants were considered for the study. The characteristics of the sample are presented in Table 1, in which the characteristics of the whole sample of front-office and back-office employees are presented. --- Measures Job Characteristics. Job demands and control (i.e., autonomy) were measured by means of the Job Content Questionnaire (JCQ) [84], as the Portuguese version had already been used in previous studies [85]. Therefore, using a Likert scale of 1 (I totally disagree) to 5 (I totally agree), the participants responded to a questionnaire composed of 7 items that analyzed job demands (e.g., I have too much work to do) and 4 items referring to the level of autonomy they had at work (e.g., I have control over what happens in my work). Thus, high scores in these two scales indicate high demands and high autonomy, respectively. The two scales have a good rate of internal consistency, as Cronbach's <unk> was always above 0.7 [86] (0.88 and 0.88 for demands and 0.84 and 0.85 for autonomy, for front-office and back-office employees, respectively). The JCQ was also used to measure social support [86] through 5 items that analyzed supervisor support (e.g., My supervisor is concerned about the well-being of his/her employees) and 6 items regarding peer support (e.g., The people I work with help in the accomplishment of tasks). The participants assessed the extent to which they agreed with each statement using a Likert scale of 1 (I totally disagree) to 7 (I totally agree). High scores correspond to a high level of supervisor support and peer support. Internal consistency rates were 0.88 and 0.89 for supervisor support and 0.86 and 0.87 for peer support, for front-office and back-office employees, respectively. Human resources practices. Human resources practices were analyzed with recourse to an adaptation of the scale used by Chambel, Castanheira, and Sobral [87], based on the scales of Lepak and Snell [88], Slattery, Selvarajan, and Anderson [89], Takeuchi, Lepak and Wang [90] and Chambel and Castanheira (2012) [1]. The questionnaire consisted of a total of 22 items which analyzed the various human resources practices adopted by the organization to which the participants responded using a Likert scale of 1 (I totally disagree) to 7 (I totally agree). Recruitment was measured by 4 items (e.g., When I was recruited by this company my specific knowledge was analyzed), presenting a Cronbach's <unk> of 0.80 and 0.83 for front-office and back-office employees, respectively. The welcoming and integration process consisted of 4 items (e.g., When I started working in this company I had initial support from my supervisor) and Cronbach's <unk> was 0.83 for front-office and 0.81 for back-office employees. Training was analyzed by means of 5 items (e.g., With the training/experience I have received I can easily change roles within this company), with a Cronbach's <unk> of 0.91 for front-office and 0.90 for back-office employees. Performance assessment was measured by 4 items (e.g., The performance assessment criteria are clear in this company), with a Cronbach's <unk> of 0.88 for front-office employees and 0.91 for back-office employees. Finally, compensation was analyzed by means of 5 items (e.g., In this company, the criteria for assigning the variable component of the salary are clear), with a Cronbach's <unk> of 0.88 and 0.90 for front-office and back-office employees, respectively. High scores in these dimensions indicate that employees had a more positive perception of the HR practices in place. Work-life conflict. The work-life conflict was measured through the Portuguese version of the scale of Keeney, Boyd and Sinha [91], used by Chambel, Carvalho, and Cesário [92]. It considers 8 work-related domains: Health, family, home management, friendship, education, love relationships, leisure, and community involvement. However, the latter domain was not considered for this study since people in Portugal do not have a high and systematic involvement in community activities [92]. The interference of work in one's personal life may occur in two distinct dimensions, namely time (e.g., Work takes the time that I would like to spend with my family away from me) and stress (e.g., Due to all the pressures of work, I am sometimes too stressed to engage in family activities). Each was measured by 7 items and participants had to assess the extent to which they agreed with each statement using a Likert scale of 1 (I totally disagree) to 5 (I totally agree). Cronbach's <unk> for time was 0.93 for front-office and back-office employees, and for stress 0.94 and 0.95 for front-office and back-office employees, respectively. Affective organizational commitment. Affective organizational commitment was measured through the Portuguese version of Meyer, Allen and Smith's [68] scale used in the study of Chambel and Castanheira [15]. The scale is composed of 6 items (e.g., This company has a high personal meaning to me) that were answered using a Likert scale of 1 (I totally disagree) to 7 (I totally agree). This scale also presented good internal consistency, since Cronbach's <unk> was 0.88 for front-office employees and 0.90 for back-office employees. Well-being at work. Well-being at work was measured by work engagement and burnout. Work engagement was analyzed using the Portuguese version of the Schaufeli, Bakker and Salanova [93] scale, used previously by Chambel et al. [87]. This version consisted of 3 items to measure vigor (e.g., In my work I feel full of energy); 3 items for dedication (e.g., I am enthusiastic about my work) and 3 items for absorption (e.g., I am immersed in my work). Participants' responses were measured using a Likert scale of 1 (Never) to 7 (Everyday) and high scores indicate high levels of work engagement. Considering the front-office and back-office employees, Cronbach's <unk> were 0.94 and 0.95 respectively. Burnout was measured by means of the Portuguese version of the Maslach, Jackson and Leiter [94] scale, used previously by Chambel and Castanheira [15]. This scale is composed of 5 items that analyze exhaustion at work (e.g., I feel exhausted by my work) and 5 items related to cynicism (e.g., I have lost enthusiasm for my work), both measured on a Likert scale ranging from 1 (Never) to 7 (Everyday). As with engagement, high scores indicate high burnout levels. For exhaustion, Cronbach's <unk> was 0.92 for both front-office and back-office employees, while for cynicism, Cronbach's <unk> was 0.85 and 0.84, respectively. General well-being. General well-being was measured through an adapted version of the General Health Questionnaire (GHQ-12 [95]). This questionnaire is composed of 12 items (e.g., Have you been feeling sad and depressed?) and the participants responded using a Likert scale of 1 (Not at all) to 4 (Much more than usual). The scale was subdivided into two dimensions in order to analyze employees' stress and well-being, not in a professional context, but on a general level (i.e., in a free context). Regarding internal consistency, Cronbach's <unk> showed no differences between front-office and back-office employees, standing at 0.84 in the stress sub-scale and 0.87 in the well-being sub-scale. --- Data Analysis The data analysis was performed through the IBM Statistical Package for the Social Sciences (SPSS 25.0, IBM, New York, NY, USA) program. In order to characterize the sample, a descriptive analysis of variables such as gender, age, marital status, qualifications, work shift and tenure of the respondents, was conducted for the whole sample and for both front-office and back-office duties. A descriptive analysis of the instruments used was then carried out, which made it possible to calculate the main measures of central tendency and dispersion of each of the studied variables. The Student t-test was performed to verify whether the means of the two groups, both front-office and back-office, were statistically different. The Cronbach's alpha of each scale was also calculated to analyze the internal consistency. Finally, the intra-class correlation coefficient (ICC) was calculated to evaluate the amount of variation in the responses at the individual level for each scale that can be explained by the variability among the 15 CC companies. The intraclass correlation (ICC) was calculated to assess the amount of variance in individual-level responses for each variable that can be explained by variability among the fifteen organizations: ICC = (msb -msw)/(msb + ((ng -1) msw)) (1 ) where msb is the between-group mean square, msw is the within-group mean square, and ng is the group size, [96]. The higher the ICC value, the higher the proportion of total variance in a subscale is explained by organizational membership. When evaluating the ICC, values exceeding 0.05 are considered relevant for aggregation of individual-level data to a higher organizational level, and 0.20 is considered to be a high level. Thus, it is possible to identify which of the organization's characteristics influence the psychosocial work environment and the stress and well-being of its employees. --- Results Table 2 shows the mean (M) and standard-deviation (SD) of the variables in the sample under study and by means of the Student t-test, the comparison of means between the back-office and front-office groups may be observed. On the basis of this comparison, it was possible to verify that the employees of these two groups have a similar perception in several of the factors considered, showing that the latter are independent of the duties. Thus, it was possible to observe that employees in both groups have a moderately high perception of job demands and feel that there is moderate supervisor and peer support. As far as HR practices are concerned, employees have a slightly positive perception of the integration and recruitment processes, as well as the training and assessment carried out by the organization. Employees show a weak affective commitment to the organization, relatively low work engagement, relatively high cynicism in the exercise of their professional activity and weak general well-being. However, it was possible to observe some significant differences between the two groups. Front-office employees show lower values in the perception of autonomy, higher values in compensation, work-life conflict (stress dimension) and exhaustion, but lower for general stress. Thus, and although the conditions are similar, the results appear to indicate that front-office and back-office duties influence the perception of some job characteristics and the environment and, consequently, their own well-being. The Intra-class Correlation Coefficient (ICC) values of the variables analyzed in this study may be observed in Table 3, through which the proportion of variance explained by the organization for each duties, front-office and back-office, may be verified. As the number of participants in each organization differed, the Bonferroni test was conducted for each variance analyzed. As far as back-office employees are concerned, some of the job characteristics (job demands, autonomy, supervisor and peer support) of the work environment (recruitment, integration, training, performance assessment, compensation, and work-life conflict-stress dimension) and employees' attitudes and stress and wellbeing (affective organizational commitment, work engagement, cynicism and general well-being-stress dimension), are observed to present significant differences among the various companies (ICC values <unk> 0.05). Thus, these dimensions are dependent on the company, suggesting that back-office work may vary according to the company in which the employee works. On the other hand, exhaustion, work-life conflict (time dimension) and general well-being (well-being dimension) appear to be common to all companies, since they do not present significant differences in variance (ICC values <unk> 0.05). As regards front-office duties, significant differences in variance in job characteristics (job demands, autonomy, peer support), work environment (recruitment, integration, training, performance assessment, compensation, and work-life conflict-stress and time dimension) and attitudes and stress and well-being (affective organizational commitment, cynicism and general well-being-stress dimension) may be observed. However, this is not the case for supervisor support, work engagement, exhaustion and general well-being, -well-being dimension), since no significant differences in variance are observed and, therefore they are common to all companies. Thus, it may be concluded that feelings of exhaustion and general well-being in CC appear to be independent of the duties performed or of the companies in which employees develop their professional activity. On the other hand, the remaining job characteristics, namely those related to environment, attitudes and stress and well-being depend either on the duties performed or the company's characteristics. --- Discussion This study sought to ascertain whether the characteristics of CC work are inevitable or whether they depend on the duties performed, namely front-office or back-office, and the company's characteristics. It was possible to observe significant differences between the two functional groups: Front-office employees appear to have a more negative perception of autonomy and a greater perception of work-life conflict (stress dimension), consequently presenting worse levels of exhaustion. On the other hand, when comparing the results of fifteen different companies, job characteristics, environment, and levels of stress and wellbeing of the employees show significant differences, indicating that these characteristics are not inevitable in CC but rather depend on each company's management strategy. Regarding job characteristics, as expected, the CC context was found to be characterized by high demands and low control, resulting in high stress and low well-being levels [19][20][21]. However, front-office employees perceived less autonomy compared to back-office employees, in line with the assumption that the use of scripts that organize and structure the making of calls in the case of front-office duties [12] has negative repercussions for control in terms of the planning and organization of the tasks performed [8,11]. When comparing the professionals of these two groups, front-office employees presented higher levels of stress [67], namely exhaustion, which, in line with the role conflict theory [60], had an impact on a higher perception of work-family conflict (stress dimension). On the other hand, this study managed to demonstrate innovatively that stressful characteristics [21] (high demands and low control) are not inevitable in the context of CC, as the data suggests variability among the companies and both back-office and front-office duties, showing that it is possible to reduce the workload and increase the autonomy of employees by redesigning these duties. In fact, the data of this study suggest that employees' stress and well-being levels may differ [2,19] as the quantitative requirements that are associated with monitoring and performance assessment [12,54] and employees' autonomy, giving them some freedom to plan and organize the tasks they perform [8,11], may also vary depending on their occupation and company. Regarding social support on the part of the supervisor, this study identified a similarity among all the companies for employees with front-office duties. It suggests that the need to monitor and assess customer service may favor the standardization of supervisory duties among different companies. Given the knowledge that social support increases the resources required by employees to deal with high demand situations [47], this study highlights the need to promote the ability of supervisors to offer adequate social support in the context of CC [14]. On the other hand, the HR practices analyzed were considered to depend on the company. If these practices are considered fundamental to explain the results obtained in the context of CC [8], but also to explain the perception of the job characteristics themselves [15], then the following measures are sorely needed: Investment in appropriate recruitment and selection processes adapted to the duties [40]; investment in welcoming and integration programs that foster the creation of positive interpersonal relationships among employees [43]; provision of specific and planned training in order to increase employees' resources [47]; adoption of a remuneration model that is compatible with the HIM model [49]; and the use of performance assessment as a diagnostic tool which aims to identify features requiring improvement [59]. As for work-life conflict, stress was considered to vary depending on the company and is, therefore, an avoidable variable. This suggests that a discrepancy between the demands of the domain in which the employee participates and the resources to which this professional has access [63] is not observed in all companies. Hence, and although employees have a relatively neutral perception of the work-life conflict, organizations should take measures to promote a balance between these two domains [60]. Regarding attitudes at work, affective organizational commitment was considered to vary depending on the company, and significant variance was observed in both frontoffice and back-office groups. Affective organizational commitment reflects the bond employees experience with their organization [66] and may be used as a tool to combat their stress [68]. As this positive attitude depends on the organizational context, namely the human resources management practices in place [32], and as the perception of these practices differs depending on the company in question, differences in attitude were also expected. However, and since the data suggests that participants have a neutral perception of affective commitment, it is imperative to focus on developing a positive emotional relationship between employees and their organizations [68] in order to promote their well-being [71]. As for well-being at work, differences in relation to burnout were observed between the two core dimensions of this chronic stress at work syndrome: Exhaustion appears to be cross-cutting and independent of the company and back-office or front-office duties; cynicism, conversely, appears to be dependent on the company for both back-office and front-office duties. This difference may be justified if the development of the burnout syndrome, as posited by the Conservation of Resources theory (COR, [97]), is taken into account. According to this theory, employees invest strongly in the acquisition of resources to meet the excessive demands with which they are confronted during their professional activity, resulting in a feeling of high exhaustion which characterizes a stress situation. Thus, working in a CC may be considered a highly demanding situation conducive to stress (i.e., exhaustion), as employees tend to perceive a loss of resources, the threat of resource loss or to invest in resources to face these demands. However, in order to cope with these same demands, employees use coping strategies, which may or may not trigger distancing responses, i.e., cynicism. If the context does not provide resources to protect employees from this sense of loss or threat of loss, they tend to drain their energy resources, and consequently, to protect themselves they will adopt an attitude of de

It is broadly acknowledged that contact center employees are subject to high levels of stress. In this profession, there is a distinction between back-office and front-office employees. In addition, employees may perform duties in various companies with different characteristics (i.e., human resources practices, job characteristics, social support, work-personal life relationship, among others). Thus, this study focuses on the analysis of the contact centers' (CC) psychosocial work environment and employees' levels of stress and well-being, seeking to understand whether they change due to the specific nature of the duties they perform and the characteristics of the company. This study involved 1440 participants from 15 companies. The results indicate that front-office and back-office duties influence the perception of some job characteristics and their environment and, consequently, the stress and well-being of these employees. Furthermore, the exhaustion and general well-being of employees are seemingly independent of the duties performed and common to all companies. However, the job characteristics, psychosocial environment and employees' levels of cynicism, work engagement and general stress were found to change according to the company in which they worked, thus highlighting the need for action in the psychosocial environment of these work duties.

; cynicism, conversely, appears to be dependent on the company for both back-office and front-office duties. This difference may be justified if the development of the burnout syndrome, as posited by the Conservation of Resources theory (COR, [97]), is taken into account. According to this theory, employees invest strongly in the acquisition of resources to meet the excessive demands with which they are confronted during their professional activity, resulting in a feeling of high exhaustion which characterizes a stress situation. Thus, working in a CC may be considered a highly demanding situation conducive to stress (i.e., exhaustion), as employees tend to perceive a loss of resources, the threat of resource loss or to invest in resources to face these demands. However, in order to cope with these same demands, employees use coping strategies, which may or may not trigger distancing responses, i.e., cynicism. If the context does not provide resources to protect employees from this sense of loss or threat of loss, they tend to drain their energy resources, and consequently, to protect themselves they will adopt an attitude of detachment which will result in cynicism, a characteristic of the burnout syndrome. However, if the context provides resources (e.g., control, social support, human resource practices that respond to employees' needs), this stress situation may not become a burnout situation, as individuals do not need to adopt this cynical distance to deal with such situations of loss or threat of loss of resources. With regard to work engagement, this positive psychological state appears to depend on the company for employees with back-office duties; however, it is cross-cutting and independent of the organization for front-office duties. Considering that work engagement is mainly dependent on the resources available to employees in the accomplishment of their tasks [76], in the case of back-office duties, there appear to be situations where the availability of resources varies, thus leading to variable degrees of work engagement, while in front-office duties the differences in resources are not sufficient to reflect differences in this indicator of well-being at work. In line with the results obtained, supervisor support was also found to be a resource that appeared not to differ among companies for employees with front-office duties. Since this resource has an extrinsic motivational role for being instrumental in the acquisition of work objectives, but also intrinsic for being able to satisfy the basic psychological need for relationships [98], it can play a central role in the development of these employees' work engagement. Finally, and analyzing general well-being, it is possible to observe that general stress is dependent on the company and for both back-office and front-office duties. This result shows that organizations can implement a number of strategies to mitigate employees' stress. Furthermore, according to the burnout literature [99] and the health impairment process proposed by the Job Demand-Resource Model [76], burnout leads to different health problems outside the work context. As previously mentioned, although the levels of exhaustion (i.e., stress) are common to the different companies, their association with cynicism (i.e., burnout) is dependent on the company, hence the levels of stress outside the work context may also be dependent on the company. In conclusion, this study appears to corroborate the idea that working in a CC implies experiencing stress in the accomplishment of one's work, as the levels of exhaustion are independent of back-office and front-office duties and of the company in which one works. However, the development of burnout (i.e., cynicism) and general ill-being does not appear to be inevitable as the emergence of these conditions depends on the company in which the duties are performed. Moreover, job characteristics, peer support, HR practices, work-life conflict (in the stress dimension) and affective organizational commitment also appear to depend on the company. Thus, this study suggests that through a structured and planned intervention at the organizational level, it is possible to promote a healthier work environment that will foster the well-being of employees. --- Limitations and Future Implications A number of limitations need to be addressed regarding the present study. First, the method used for data collection may have skewed the results, since a self-assessment questionnaire was used, which was disclosed internally by each company's HR department. Thus, and despite the anonymity of the responses, they may contain some level of social desirability. Secondly, as this is a cross-sectional study, the data refers to a single point in time, which does not allow for the establishment of cause-effect relationships. Therefore, it was only possible to make inferences with regard to the assessment of positive or negative relationships between the studied variables. It would be interesting to conduct a longitudinal study in order to analyze any further developments in the perception of employees regarding the factors studied in this research, as well as to monitor intervention plans that may have been applied. Thirdly, the sample consisted only of employees from Portuguese CC, and it was not possible to generalize the results to other countries or other sectors of activity. In the future, it would be important to compare the results obtained in this study with those of other countries and/or other sectors that are dominated by customer service, such as the hotel industry or trade employees. Finally, the quantitative analysis used in this study does not allow for a comprehensive vision of employees' experience in a contact center context. In the future, it would be interesting to conduct a qualitative study that could examine the meaning of the employees' history in this context, analyzing their experiences (e.g., peer support climate, supervisor support strategies, learning of day-to-day problem-solving strategies) and respective repercussions for their well-being and health. --- Conclusions This study provides evidence that the specific nature of the duties performed by Contact Center employees, i.e., front-office and back-office duties, has an impact on how they perceive their job characteristics and environment and, consequently, on their wellbeing. In addition, it highlights that although the exhaustion and general well-being of CC are independent of their duties and common to all employees regardless of the company in which they work, job characteristics, psychosocial environment, and the levels of affective organizational commitment, cynicism, and general stress of CC employees depend on the company in which they work. --- Data Availability Statement: The data presented in this study are available on request from the corresponding author. --- Author Contributions: D.G.-C. was involved in the design, writing and original draft preparation of this paper. M.J.C. was involved in the data collection, design supervision, writing, methodology and formal analysis. V.S.C. was involved in the reviewing and editing process. All authors have read and agreed to the published version of the manuscript. --- Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Faculty of Psychology, University of Lisbon (protocol "The management of the work-personal (family) life border in the case of temporary workers" approved on 20 November 2019). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. --- Conflicts of Interest: The authors declare no conflict of interest.

It is broadly acknowledged that contact center employees are subject to high levels of stress. In this profession, there is a distinction between back-office and front-office employees. In addition, employees may perform duties in various companies with different characteristics (i.e., human resources practices, job characteristics, social support, work-personal life relationship, among others). Thus, this study focuses on the analysis of the contact centers' (CC) psychosocial work environment and employees' levels of stress and well-being, seeking to understand whether they change due to the specific nature of the duties they perform and the characteristics of the company. This study involved 1440 participants from 15 companies. The results indicate that front-office and back-office duties influence the perception of some job characteristics and their environment and, consequently, the stress and well-being of these employees. Furthermore, the exhaustion and general well-being of employees are seemingly independent of the duties performed and common to all companies. However, the job characteristics, psychosocial environment and employees' levels of cynicism, work engagement and general stress were found to change according to the company in which they worked, thus highlighting the need for action in the psychosocial environment of these work duties.

Introduction Due to its long history of war, Afghanistan has endured extreme poverty, lack of employment, a large refugee population, extreme division of social class, inadequate educational infrastructure, horrific violence, and millions of civilian casualties since 1979. [1][2][3][4][5] While living in a war-torn country takes its toll on the entire population, the situation in Afghanistan is especially harsh on Afghan women, as they are often considered unequal to their male counterparts and in many cases are subsequently abused, denied education, health care, employment, and other basic human rights. 6,7 Denial of human rights, gender inequality, and living in a war zone have been linked with severe depression and poor social functioning. [7][8][9] Research has shown that such mental health problems are often present among female drug abusers. 10,11 Further, because Afghanistan is a major source of the global supply of opium and hashish, 12 and because opium is commonly used by Afghans for medicinal purposes, 12,13 Afghan women face a unique and significant reality in which they have easier access to illegal and psychotropic drugs than they do to drug abuse treatment, education, or health care. While studies have begun to look at this important issue by reporting prevalence of human rights violations and mental health problems among Afghan women, [6][7][8][9] this study builds on prior research by assessing the prevalence of these issues among women with serious substance use problems in drug abuse treatment (DAT) centers in Afghanistan, and examining risk factors for human rights violations among this population. --- Women's human rights in Afghanistan The United Nations (UN) Universal Declaration of Human Rights states that everyone is entitled to the rights set forth in the declaration "without distinction of any kind, such as race, color, sex, language, religion, political or other opinion, national or social origin, property, birth, or other status." And, those rights include: the right to not be subjected to torture, degrading, or inhumane treatment; the right to move about freely within your own country; the right to education and work; the right to marry only with full consent of both parties; and the right to a standard of living for the health and well-being of oneself and family. 14 A UN report on the challenges of human rights and cultural diversity further states, "The right to culture is limited at the point at which it infringes on another human right. No right can be used at the expense or destruction of another, in accordance with international law." 15 The rights set forth in the UN declaration helped guide this study and were referenced when developing survey instruments in order to adequately assess the human rights violations experienced by Afghan women. The plight of Afghan women became a highly publicized international concern after the Taliban takeover, which imposed extreme gender-based restrictions on women's human rights punishable by harassment, torture, and death. Although Afghanistan has been characteristically a patriarchal society due to religious and cultural beliefs, as well as the societal structure, Taliban rule brought violent enforcement of laws which were more restrictive and oppressive than traditional life was for Afghan women. 16 Rasekh and colleagues 7 found that during this time women reported poor access (77%) or no access (20%) to health care services, occasions in which they were ill and unable to seek medical care (53%), extreme fear when in public (39%), and extreme restricted social activities (68%). --- Human rights violations and mental health problems The harsh realities of living in Afghanistan, including severe gender-based inequalities and human rights violations, have been linked to serious mental health problems among Afghan women. 8,9 For instance, Rasekh and colleagues 7 reported that when Afghan women were asked about the 2 years post-Taliban occupation, most reported a decline in physical (71%) and mental health status (81%), and a decline in access to health care (62%). Many women reported symptoms of posttraumatic stress disorder (42%), major depression (97%), and significant anxiety (86%). Amowitz et al 8 found that the majority (70%) of Afghan women in their study from Taliban controlled areas attributed their mental health problems to Taliban official policies toward women. Results showed that severe depression was far more prevalent among Afghan women exposed to Taliban policies (78%) than those living in non-Taliban controlled areas (28%). There were also higher occurrences of suicide ideation (65% vs 18%) and suicide attempts (16% vs 9%), and a higher number of those in Taliban controlled areas indicated that the Taliban contributed "quite a bit" or "extremely" to their suicide ideation or attempts (22% vs 2%). Four years after the Taliban rule officially came to an end, Cardozo et al 9 reported results which showed that Afghan women, when compared to Afghan men, were still suffering from poor mental health, including greater symptoms of depression (73% vs 59%), anxiety (84% vs 59%), posttraumatic stress disorder (48% vs 32%), and lower social functioning (M = 2.77 vs 3.92). Most of the sample (62%) had experienced four or more traumatic events, and the most commonly reported traumatic events for females included denial of basic human rights such as no access to health care (54%), lack of food and water (52%), and lack of shelter (46%). --- The current study The results of these studies allude to the potential impact that human rights violations have on Afghan women. Although Taliban rule officially ended in late 2001, the aftermath of its influence is still present in Afghanistan. The effects of intense trauma experienced by Afghan women will subsist for far longer, especially because many of them have turned to substance use in an effort to cope with feelings of depression and hopelessness, to ease pain, and cope with repressive lifestyles. 12,13 Although documented estimates are believed to be low, in 2005 it was estimated that 2% of all Afghan women (120,000) were addicted to drugs. 17 Research shows that women with a history of emotional, physical, and sexual abuse are more likely to engage in substance use as a coping mechanism. 13,18,19 Maguet and Majeed 20 posit that addiction is a direct consequence of physical and psychological violence related to conflict-affected countries. In order to better understand the challenges and problems faced by drug-addicted Afghan women, this study examines (a) the extent to which women in DAT have experienced human rights violations and mental health problems prior to entering the DAT centers, and (b) whether there are specific risk factors for human rights violations among these women. The current paper utilizes data from three of these DAT centers that are for women and their children only. At the time this study began, these centers were located in provinces considered to be in the "safe zone" of the country and thus least likely to have been disrupted by the ongoing war. It should be noted, however, that there have been reported conflicts in these areas within the past 2 years. The names and locations of these DAT centers are confidential. --- Study participants A total of 176 in-person interviews were conducted with female participants between 0-5 days after completing their detoxification process in the DAT centers. One-hundred seventy-two cases were used for analyses after removing four cases that provided more than one inconsistent response, which is a common quality assurance standard in large national datasets (Monitoring the Future investigators, written communication, December 2007). --- Measures Human rights violations Participants responded yes or no when asked whether they had experienced 10 different human rights violations prior to entering the DAT centers. The authors developed these items, based on numerous online news reports, 21,22 the UN's Universal Declaration of Human Rights, 14,15 research articles, [6][7][8] and consultation with an expert panel of four Afghan DAT center directors. Whereas we do not claim this list is exhaustive, the expert panel felt it was representative of the types of human rights violations most commonly experienced by Afghan women. To better understand various types of human rights violations, we conducted a principal component analysis (sometimes referred to as exploratory factor analysis) with varimax rotations on the ten items. According to Pedhazur and Schmelkin, 23 principal component analysis attempts to characterize the majority of the variability in all items with smaller groupings of items, while factor analysis attempts to group items into scales based on those items with the strongest interrelationships. Computationally, the only difference is distinguished by whether the diagonal elements in the correlation matrix analyzed are set to one (ie, principal component analysis, which assumes all variance is explained) or whether the communalities (ie, the variance explained in other items by a particular item) are used as the matrix diagonal. We examined two-, three-, and four-component solutions based on eigenvalues greater than one (2.91, 1.80, 1.26, and 1.05). We chose a three-component solution, as it was the most substantively interpretable solution and adding a fourth component resulted only in a one-item component that only accounted for an additional 11% of the variance. The first component accounted for 29% of the variance in the data and the five items (denied food as punishment, threatened with denial of food or shelter by husband/family, forced to live in social isolation, forced to consume drugs, and forced to work in poppy cultivation) were grouped together and termed Maltreatment. The scale had acceptable internal consistency reliability (<unk> = 0.68), and component loadings ranged between 0.83 and 0.31. The second component accounted for 18% of the variance and the three items (forbidden to drive a car, denied education, and forbidden to go out in public without a male relative) represented activities that we termed Gender-based Inequality. This scale also had acceptable internal consistency reliability (<unk> = 0.73), and component loadings ranged between 0.82 and 0.78. The final component accounted for 13% of the variance and the two items (physical abuse and sexual assault) represented severe forms of abuse, and had an acceptable internal consistency reliability (<unk> = 0.74), and component loadings ranged between 0.91 and 0.89. We ultimately calculated scale scores as an occurrence of any of the behaviors comprising the scale (ie, a dichotomous scale score), as the distributions of the items were heavily left-censored. As a summary measure, we also examined the occurrence of any of the ten human rights violations. Considering the occurrence of all ten behaviors, alpha was still acceptable (<unk> = 0.66). --- Mental health factors Social functioning measures the extent to which physical and emotional health interferes with or limits normal social activities. Four items (developed by Hays et al 24 ) measured the extent and frequency that participants experienced these physical and emotional problems 30 days prior to entering treatment (eg, To what extent did your physical health or emotional problems interfere with your normal social activities with your family, friends, neighbors, or groups?) on two scales, 0 (not at all) to 3 (quite a bit) and 0 (never) to 3 (all the time). The four scale items were internally consistent (<unk> = 0.69). Suicide ideation measures the extent to which the participants had suicidal thoughts 30 days prior to treatment using a single item adapted from Lung and Lee (2008). 25 Participants responded on a 0 (not at all) to 4 (extremely) scale. Suicide attempts was also measured by asking participants if they had attempted suicide (yes/no) 30 days prior to entering treatment. --- Substance use In order to determine the frequency and types of drugs being used among this population, participants were asked how often they used 14 substances in the past 6 months and past 30 days prior to entering treatment. As one goal is to examine the prevalence of substance use among the women in these DAT centers, all items were dichotomized to use (1) or no use (0). Here, we focused only on opium, hashish, heroin, and "crystal" (a potent form of heroin). (The additional substances included in the questionnaire were used by less than 2% of the sample in the past 6 months). Alcohol use was nearly nonexistent in this population with only one participant reporting drinking three out of 30 days prior to entering treatment. This low prevalence of alcohol use is likely because alcohol use is strictly against religious and government law and highly stigmatized. --- Individual characteristics Participants indicated their age (ie, date of birth, which is estimated, because Afghans do not typically keep track of their age), marital status (single, married, separated, divorced, widowed), ethnicity (Pashtun, Tajik, Uzbek, Turkmen, Hazara, Baloch, Kirghiz, Nuristani, Aimak, Arab), living situation prior to treatment (with spouse, with other family), and whether they had children. Participants indicated whether they were literate, their level of education (no school, primary school, secondary school, high school/vocational training, college, or higher), whether they were employed 6 months prior to entering treatment, and whether they had experienced a loss of a close family member due to war/conflict within the last 2 years. Participants rated 18 reasons for entering treatment using a 0 (not important at all) to 3 (very important) scale (eg, "Your family wanted you to go."). --- Data collection PIRE partnered with D3 Systems, Inc, a social and market research company based in Washington DC, and the Afghan Center for Socio-Economic and Opinion Research (ACSOR) in Afghanistan to collect the data for this study. ACSOR was founded in 2003 by D3 Systems and is a registered market and opinion research agency in Afghanistan with trained Afghan researchers and interviewers. ACSOR interviewers for this study included trained medical students residing in Afghanistan. An in-person structured interview that included questions previously validated in other treatment outcome studies was developed by PIRE, and carefully reviewed by ACSOR field staff and an expert panel of four Afghan DAT center directors or associate directors. Revisions were made to clarify meanings and to ensure cultural appropriateness of phrasing and terminology. The final instrument was translated into Pashto and Dari (the primary languages of our sample) and back-translated by ACSOR field managers to ensure linguistic equivalency. The translated instrument was also reviewed independently by two bilingual Afghan nationals contracted by PIRE. Data used in this paper are drawn from the in-person baseline structured interviews that were conducted with female patients in three Afghan DAT centers by ACSOR trained female interviewers from February 2010 to September 2010. Consent forms and interview questions were read aloud by the ACSOR interviewers to the patients due to low literacy levels. Patients were provided with a consent form that was read aloud to describe the study, procedures, risks, and safeguards. Patients first consented to the release of their names and admission dates to the research team, and then to participate in the study. Patients were informed that their participation was voluntary and that their decision on whether to participate would not result in any penalties or benefits to them or to the centers. Patients were informed that their responses would be confidential and that they could decline to answer any questions. If the patient signed the consent form, the baseline interview was conducted at that time in a private room in the DAT centers to ensure confidentiality. The interviews took about 1 hour to complete. Participants were given a US$5 incentive for completing the interview. --- Data analyses Simple percentages and means were calculated to address research questions one and two. For research question two, we rounded scale scores to the nearest integer, such that they could be interpreted according to the actual response scale labels in the questionnaire. Inferential analysis examined relationships with human rights violations for research questions three and four. Research question three was answered by examining the relationship between individual characteristics and human rights violations using logistic regression. The two individual characteristics that were not dichotomous (age and education) were split at the median response to yield 40 years of age and younger versus over 40 years old, and no education versus some education. All four measures of human rights violations (gender-based inequality, maltreatment, abuse, and "any") were regressed on all of the individual characteristics in four analyses. Research question four was answered by examining the relationship of mental health problems and human rights violations using logistic regression. We regressed all four measures of human rights violations on all mental health factors. --- Results --- Individual characteristics Of the 172 female participants with usable data, the mean age was 39 years with a range of 18-70 years. Participants were mainly Turkmen (30%), Pashtun (29%), Tajik (21%), or Hazara (12%). Seventy-two percent of participants were married, 15% were widowed, and 11% were single. Of those who were married, 96% of married participants had an arranged marriage, while 11% reported being forced to marry someone they did not want to marry. Most of the sample had children (70%), and, of those, the average number of children was two per participant with an average of two children living with them at the DAT centers. Before entering the DAT centers, 52% of participants lived with family and 45% lived with their spouse. The majority (88%) did not have any formal schooling. Most of them could not read before entering the DAT centers (91%). Most participants (86%) had not been employed 6 months prior to entering the centers. Thirteen percent had lost a close family member in the past 2 years due to war. When given a list of 18 reasons for entering the DAT center, 63% said that family (including their spouse) wanting them to go into treatment was "very important." Other reasons considered "very important" were: "You decided for yourself that you wanted treatment" (50%); "You had medical or physical problems" (45%); "Your friends wanted you to go" (44%); and "Spouse or family was shamed/stigmatized by drug addiction" (27%). The most commonly abused substances by participants were opium, crystal, hashish, and heroin (see Table 1). A total of 91% had used one or more of these four substances in the 6 months prior to entering treatment, while 88% had used one or more substances 30 days prior to entering treatment. Some participants reported no use in the past 6 months (8%) or past 30 days (10%), which is likely due to some participants trying to stop using drugs while on a waiting list prior to entering treatment. The most abused substance by participants was opium with 68% using in the 6 months before treatment and 66% using in the 30 days before treatment. The second most abused substance by participants was "crystal" (a potent form of heroin) with 28% of participants using in the past 6 months and 22% using in the past 30 days. These four substances account for the majority of substance use reported by these participants. Only three participants (2%) reported using other drugs asked about on the questionnaire (including hallucinogens, cocaine, sosigan, mandrax and heroin, other opiates, methamphetamines, other amphetamines, tranquilizers, barbiturates, other sedatives, or other substances) in the past 6 months and the past 30 days. --- Research question one: extent of human rights violations Considering base rates, 50% of participants reported experiencing at least one form of human rights violations prior to entering the DAT centers; 35% reported gender-based inequality, 36% reported maltreatment, and 4% reported abuse (see Table 2). We realize that 4% is an extremely low base rate, so all correlations listed for abuse must be interpreted with extreme caution. We felt it was necessary to keep abuse in the analyses since physical and sexual abuse are issues Afghan women often face, but rarely self-report due to various reasons such as stigmatization and cultural norms. 12 Research question two: extent of mental health problems As Table 3 shows, almost the entire sample (91%) of women entering the DAT centers experienced some limitations with social functioning (ie, the extent to which physical and emotional health prevented them from engaging in normal social activities) 30 days prior to entering the centers, when defining "slightly/rarely" through "quite a bit/all the time" responses as having limited social functioning. Additionally, a large proportion of the sample (41%) had experienced some suicidal ideation 30 days prior to entering the centers, when defining "rarely" through "extreme" responses as having had suicidal thoughts. Additionally, 27% of participants attempted suicide at least once 30 days prior to entering the centers due to feelings of sadness or hopelessness. --- Research question three: individual characteristics as risk factors for human rights violations As can be seen in Table 4, women who were married (38% vs 29%) were 5.08 times more likely to report indicators of maltreatment, and women of Pashtun ethnicity (62% vs 45%) were 5.80 times more likely to experience at least one of the human rights violations about which we asked. Women who were literate (80% vs 47%) were 14.74 times more likely to report at least one of the human rights violations about which we asked. Women who were employed were 4.76 times (inverse of odds ratio [OR] = 0.21) less likely to report gender-based inequality (17% vs 37%). Women who entered DAT because of their own desire were 2.30 times more likely to report being victims of maltreatment (43% vs 28%), 3.34 times more likely to report gender-based inequality (44% vs 26%), and 2.91 times more likely to report at least one human rights violation (58% vs 42%). 5). --- Research question four: mental health problems as risk factors for human rights violations --- Discussion --- Profile of women entering drug abuse treatment The majority of women in the DAT centers had not received formal schooling, could not read, and had not been employed. This is consistent with the profile of a typical female Afghan drug user who has even less education and is more than twice as likely to not have a job than other females. 12 Lack of education and employment among Afghan females is also due to the fact that they are expected (sometimes forced) to marry young, typically around the age of 15 years old, and are pressured to have children immediately. 26 Most women in Notes: a 1 = strongly agree, 2 = agree, 3 = disagree, 4 = strongly disagree; a 0 = not at all/never, 1 = slightly/rarely, 2 = moderately/sometimes, 3 = quite a bit/all the time; b 0 = not at all, 1 = slightly, 2 = moderately, 3 = quite a bit, 4 = extremely; c 0 = no, 1 = yes. our sample reported an "arranged" marriage, while a smaller number reported being "forced" to marry someone they did not want to marry. The most commonly abused substances by women entering the DAT centers were opium, crystal, hashish, and heroin. Most of the study participants had used one or more of these substances 30 days prior to treatment. Our results are similar to those reported by the United Nations Office on Drug Use and Crime, 12 which found that Afghan citizens are increasingly dependent on cannabis, opium, heroin, opioids, and tranquilizers and suffering severe mental, physical, and social problems as a result of their addiction. --- Human rights violations Half of the women in our sample experienced at least one type of human rights violation prior to entering the DAT centers. About one-third of these women experienced some form of maltreatment, including denial of food or shelter, forced to live in social isolation, forced to consume drugs, and forced to work in poppy cultivation. This is consistent with findings from similar studies that indicate "lack of food and water" is an extremely traumatic event, 9 social isolation is an ongoing problem, 7 and husbands sometimes force their wives to consume drugs to keep them from fighting back or from leaving. 13 While other research has not reported the rates of women being forced to work in poppy cultivation, news reports indicate this is a reality for some Afghan women. 27,28 About one-third of the women experienced gender-based inequality prior to entering the DAT centers, including being forbidden to go in public without a male relative, forbidden to drive a car, and denied education. These findings are not surprising given the strict enforcement of laws forbidding females to attend school or work outside the home. 3 Although the Taliban regime has ended, some maintain that women's rights have yet to be restored to pre-Taliban traditions, 16,26 and the majority of females are still not attending school. 29 Physical and sexual abuse violations had a relatively low base-rate phenomenon with a very small percentage of women reporting these types of abuse prior to entering the DAT centers. This is consistent with other reports finding this to be a rare -and likely underreported -event. 12 In an attempt to address the issue of underreporting, a 2008 United Nations Office on Drugs and Crime (UNODC) study 30 asked Afghan women whether they knew any women who were addicted to drugs and had been raped, and 54% said yes. This suggests that sexual abuse does occur among this population but that it is not reported due to stigma or shame. Alternatively, it is also possible that some women may not feel this type of behavior is a violation of their rights, but rather a tolerated behavior between husband and wife in Afghan society. 31 As evidence of this, a law was passed in 2009 permitting "marital rape" in Afghanistan. 32 --- Mental health problems Most of the women in this sample reported limited social functioning, which is likely due to the fact that women are denied rights and freedoms that would encourage social functioning, such as walking freely in public, going to school and work, and talking openly in social settings. 33 Many of the women had experienced suicide ideation before entering the centers, which is common among this population. 8 And, nearly one-third of the women (27%) had attempted suicide 30 days prior to entering the DAT centers, which is a higher percentage than those reported from a sample of Afghan women in Taliban controlled areas (16%) and non-Taliban controlled areas (9%). 8 Further, this percentage is extremely high when compared to the worldwide rate of suicide attempts. 34 --- Risk factors for human rights violations Married women were more likely to report maltreatment prior to entering the DAT centers. This is consistent with research that indicates that human rights violations among Afghan women are inflicted mostly by husbands. 3,13,35 Also, women of Pashtun ethnicity were more likely to have experienced at least one form of human rights violation prior to entering the DAT centers, which follows research that indicates that Pashtun communities still impose Taliban-like conditions where women are treated dishonorably and denied basic rights. 36 Women who reported being able to read also reported greater human rights violations prior to entering the DAT centers, which might be explained by the fact that more educated Afghan women are more aware of their basic rights and thus more likely to report violations. 37 Employed women were less likely to have experienced gender-based inequality, which is likely because women who are working outside the home have fewer social restrictions conferred upon them. Women who entered treatment because of their own desire were more likely to have experienced gender-based inequality. Because these DAT centers offered women residential treatment, women may have seen the centers as an alternative form of shelter. Women with limited social functioning were more likely to have experienced all forms of human rights violations except for gender-based inequality. Women who are mistreated and abused likely experience inadequate social functioning due to the restrictions placed on them by their spouses or family members, and possibly by the shame and terror they feel from experiencing such abuse. Although no significant relationships were found for women who had suicide ideation, women who had attempted suicide were more likely to report experiencing all forms of human rights violations. These findings emphasize that human rights violations have a severe impact on the mental health of Afghan women, and suggest that reducing such violations might also help reduce suicide attempts among these women. The results from this study related to the relationships between mental health and human rights violations are supported by Cardozo et al's research 9 that indicates the most traumatic events reported by Afghan women were related to denial of human rights, and other research that shows that women experiencing these violations also report serious mental health problems. 8 Several limitations of this study need to be addressed. First, this was not a generalizable sample of drug-addicted women in Afghanistan, but rather a purposive sample of women who entered three female DAT centers in Afghanistan, which were the only addiction services available for women at the time of this study. The majority of Afghan women live in rural areas, where little has changed for decades, and daily conditions are especially harsh in western Afghanistan. 3,35 Because the women sampled had access to treatment, it could indicate that these women had greater means, resources, or support than other drug-addicted women in Afghanistan. Second, we have very low base rates for physical and sexual abuse in the human rights violations scale. Future research with this population needs to be conducted to help us better understand the prevalence of sexual and physical abuse as well as related attitudes around this issue. Third, we had small effect sizes for our results in general. However, we still feel that these results provide a first step toward better understanding the severe impact of human rights violations on Afghan women. Finally, due to the fact that research on the denial of women's human rights in Afghanistan is in its early stages, much work needs to be done with regard to scale development. For example, there is a lack of formative work determining what Afghan women (of various ages and ethnicity) perceive as violations to their rights versus moral or cultural expectations. Qualitative research could help enrich the results reported from this study and would better our understanding of the perceptions and impact of human rights violations among Afghan women. --- Conclusions and implications This study found that women entering DAT centers had experienced a high rate of human rights violations prior to entering the centers. To be effective, DAT centers need to add services that help female patients cope with the human rights violations they have experienced. Also, services should be provided that may help prevent future human rights violations. For example, this study showed that women who were employed were less likely to have experienced human rights violations. Therefore, offering educational classes and vocational skills training may provide Afghan women with opportunities that could prevent this type of abuse. It is important that treatment programming take into consideration the cultural and contextual factors of women living in a war-torn country where denial of women's basic human rights is the norm. The study showed that female patients who reported coming to the DAT centers due to their own desire were also more likely to have experienced human rights violations prior to entering the centers. Further, patients who were married, literate, unemployed, or of Pashtun ethnicity were more likely to have experienced human rights violations. These individual characteristics should be used as a guide to identify patients at treatment entry who are most in need of special services that help women deal with human rights violations. Significant relationships between mental health problems and human rights violations were also found, including limited social functioning and suicide attempts. More emphasis should be placed on helping women in treatment cope with their experiences of human rights violations which, in turn, may strengthen their mental health. Because women in the DAT centers in this study come from rural areas with minimal professional services for them to use after residential treatment, it is important that these specialized services be provided at the DAT centers. This study found that many drug addicted women in Afghanistan who enter DAT have mental health problems. As such, many of the women in Afghan DAT programs suffer from co-occurring substance abuse and mental disorders. Understanding the importance of integrating treatment services for substance abuse and mental disorders is critical for Afghan residential DAT providers to assist their female patients in achieving effective and healthier lives. Further, despite its prevalence, few DAT programs assess, treat, or educate patients and/or staff about trauma. 38 Failure to address this issue has serious consequences, including relapse, greater violence that leads to increased substance abuse, and higher rates of depression. 39,40 Given the connection between spirituality and recovery from trauma such as the mental health problems found in this study, 41 treatment approaches that include a spiritual component offer survivors an opportunity to explore and reestablish their own personal sense of spirituality as they reflect on the lessons learned from their trauma experience. Given the prominence of the Islamic faith in Afghan culture, incorporating spirituality into the counseling regimen of Afghan women's DAT programs represents a promising practice that could assist in their recovery from addiction and trauma. In conclusion, more research about human rights violations and mental health problems among Afghan women is needed. Further, an evaluation of the recommended changes in Afghanistan DAT centers should be conducted in a larger study in order to determine the effectiveness of these services in increasing mental health and coping among Afghan women. --- Disclosure The authors report no conflicts of interest in this work. --- International Journal of Women's Health --- Publish your work in this journal Submit your manuscript here: http://www.dovepress.com/international-journal-of-womens-health-journal The International Journal of Women's Health is an international, peerreviewed open-access journal publishing original research, reports, reviews and commentaries on all aspects of women's healthcare including gynecology, obstetrics, and breast cancer. Subject areas include: Chronic conditions (migraine headaches, arthritis, osteoporosis); Endocrine and autoimmune syndromes; Sexual and reproductive health; Psychological and psychosocial conditions. The manuscript management system is completely online and includes a very quick and fair peer-review system. Visit http://www.dovepress.com/ testimonials.php to read real quotes from published authors. International Journal of Women's Health 2012:4

Denial of human rights, gender disparities, and living in a war zone can be associated with severe depression and poor social functioning, especially for female drug abusers. This study of Afghan women in drug abuse treatment (DAT) centers assesses (a) the extent to which these women have experienced human rights violations and mental health problems prior to entering the DAT centers, and (b) whether there are specific risk factors for human rights violations among this population. A total of 176 in-person interviews were conducted with female patients admitted to three drug abuse treatment centers in Afghanistan in 2010. Nearly all women (91%) reported limitations with social functioning. Further, 41% of the women indicated they had suicide ideation and 27% of the women had attempted suicide at least once 30 days prior to entering the DAT centers due to feelings of sadness or hopelessness. Half of the women (50%) experienced at least one human rights violation in the past year prior to entering the DAT centers. Risk factors for human rights violations among this population include marital status, ethnicity, literacy, employment status, entering treatment based on one's own desire, limited social functioning, and suicide attempts. Conclusions stemming from the results are discussed.

InTROduCTIOn Health-related quality of life (HRQOL) is a multidimensional concept that provides a broader perspective of health through conveying an individual's ability to function in physical, mental and social domains of life. 1 HRQOL is thus an essential patient-centred outcome measure that is useful to guide health policies. 2 3 HRQOL is preferred over other health indicators (life expectancy, mortality, morbidity) for measuring chronic disease burden as it incorporates both length and quality of life. 4 In South Asia, chronic conditions (diabetes, hypertension and heart diseases) occur at an early age, 5 with detrimental effects on length and quality of life. 6 7 In addition, episodes and fear of hypoglycaemia, recurrent heart attacks, stroke and other long-term complications (kidney diseases, diabetic retinopathy) are not always measured as such, although they have a substantial adverse impact on an individual's overall health status. 8 Therefore, it is important to quantify the effect of chronic conditions on individuals' HRQOL. --- Strengths and limitations of this study <unk> This is the first population-level health-related quality of life (HRQOL) data from South Asia using the European Quality of Life Five Dimension-Visual Analogue Scale including three large metropolitan cities in India and Pakistan with a large sample size (16 284 adults aged <unk>20 years). <unk> Our data provide the first baseline values to be used for monitoring population health status and analysed the relationships between selected chronic conditions and HRQOL. <unk> HRQOL data presented in this article could be used to complement national health targets by providing a measure of chronic disease burden based on perceived health status rather than solely on mortality and disease prevalence. <unk> Due to the cross-sectional nature of the data, the causal relationship between socioeconomic parameters or chronic conditions and HRQOL cannot be determined. <unk> Many chronic conditions (respiratory, locomotor, cancer and others) were not included in the survey. Therefore, the ranking of most severe health conditions and associated HRQOL is not complete. --- Open Access There are several disease-specific (Chronic Respiratory Distress Questionnaire, Arthritis Impact Measurement Scale) and generic instruments (Short Form 36, WHO-Quality of Life Questionnaire and the European Quality of Life Five Dimension-Visual Analogue Scale (EQ5D-VAS)) available to measure population HRQOL. 4 9-18 However, the EQ5D-VAS is favoured because it is generic, not specific to a particular disease, and it includes multidimensional measures of health profile in five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) and the single-dimensional measure VAS to summarise overall health status. 1 Also, EQ5D-VAS has been applied and validated for its use in many population surveys across the world; therefore, it makes the comparison of health status across populations easier. Data on population HRQOL across socioeconomic status (SES) from South Asia are scarce, and little is known about the relative associations between different chronic conditions and individual HRQOL. The Centre for Cardiometabolic Risk Reduction in South Asia (CARRS) study 19 had collected data on both EQ5D-VAS and selected chronic conditions from a large representative population of adults in urban South Asia. We used this opportunity to examine population HRQOL in this region. In this paper, we describe the variations in HRQOL by age, gender and SES, and explore the relationships between selected chronic conditions and HRQOL in a representative sample of adults aged <unk>20 years from three metropolitan cities in India and Pakistan. We also analysed the relationship between multidimensional EQ5D measures and single-dimensional VAS across major subgroups. --- MeThOdS --- Study design and setting We obtained data from the baseline cross-sectional survey of the CARRS cohort (2010-2011), which recruited a representative sample of non-pregnant adults aged <unk>20 years from three urban cities: Chennai, Delhi and Karachi. These metropolitan cities with large and heterogeneous populations in terms of demographic profile and economic transitions offer unique opportunities to assess variations in health status across different socioeconomic groups. The detailed CARRS study design has been published elsewhere. 19 Briefly, a multistage cluster random sampling strategy was used with wards (in Delhi and Chennai) or clusters (in Karachi) as the primary sampling units. Using the WHO STEPS (STEPwise approach to surveillance) survey 'Kish method', two participants, one male and one female, aged <unk>20 years (non-pregnant) and meeting the study eligibility criteria, were selected from each randomly selected household. 19 --- Study measures Comprehensive and uniform data collection instruments were used to capture measurements in all three sites. A summary of all surveillance measures, methods and instruments used in the study has been published in detail. 19 Briefly, a questionnaire was administered to collect information regarding demographic, socioeconomic, behavioural, and past and present health status of the participant. Trained study staff measured anthropometric parameters (height, weight) using standardised techniques and blood pressure (BP) twice at each participant's home or at a medical camp organised in the community, after 5 min in a seated position using an electronic BP measuring device (Omron Dalian, Liaoning Sheng, China). If the difference between the first two systolic or diastolic BP readings was more than 10 mm Hg or 5 mm Hg, respectively, a third reading was taken. Average BP readings of the two/three readings were recorded in the study database. Additionally, fasting blood glucose (FBG) and glycated haemoglobin (HbA1c) were measured. The overall response rates were 94.7% for questionnaire completion and 84.3% for blood tests. Population health status was measured using the EQ5D-VAS questionnaire, which consisted of two components: health state description and self-rated health status on VAS. Health state description (profile) includes five dimensions (5D): mobility (walking ability), self-care (ability to wash or dress by oneself), usual activities (ability to work, study, housework), pain/discomfort and anxiety/depression. The respondents self-rate their level of severity for each dimension using three levels (EQ5D-3L): having no problems, having some or moderate problems, or being unable to do/having extreme problems. The respondents were asked to choose one of the statements that best described their health status on the surveyed day. For example, three levels of'mobility' dimension were phrased as 'I have no problems in walking', 'I have some problems in walking' and 'I am confined to bed'. Given the possible permutations of different domains and response types, there are potentially 243 (=3 5 ) different health profiles. For overall health status, the respondents evaluated their health status using the VAS. The VAS asks respondents to mark health status on the day of the interview on a scale of 0 (worst health status) to 100 (best imaginable health status). --- Covariates Self-reported age at baseline in completed years was used and categorised into 20-24, 25-34, 35-44, 45-54, 55-64, 65-74 and <unk>75. Based on participant responses, we categorised employment status into employed, student, housewife, retired and unemployed. Income class was grouped into three categories based on household monthly income: low-income, less than 10 000 Indian rupees (equivalent to US$200); middle-income, 10 000-20 000 Indian rupees (US$200-400); and high-income strata, greater than 20 000 Indian rupees (US$400). We categorised highest education level attained into three categories: up to primary, secondary schooling and graduates. The marital status was classified as single, married, widowed and divorced. Body Open Access mass index (kg/m 2 ) international classification of <unk>17.9 was used to define underweight, 18.0-24.9=normal weight, 25.0-29.9=overweight and <unk>30.0= obese. Lifestyle habits like tobacco use were classified based on self-reports as never, former and current user. Data on chronic conditions consisted of self-reported hypertension, diabetes, heart disease, stroke and kidney disease. In addition, diabetes was categorised into self-reported, newly diagnosed (defined by no self-reported diabetes and FBG of <unk>126 mg/dL or HbA1c <unk>6.5%), pre-diabetes (no self-reported diabetes and FBG <unk>100-125 mg/dL or HbA1c <unk>5.7%-6.4%) and normoglycaemia (no self-reported diabetes and FBG <unk>100 mg/dL and HbA1c <unk>5.7%). Similarly, we classified hypertension as self-reported, newly diagnosed (no self-reported hypertension and BP <unk>140/90 mm Hg), prehypertension (no self-reported hypertension and BP 120-139/80-89 mm Hg) and normotensive (no history of hypertension and BP <unk>120/80 mm Hg). --- ethical considerations Study participants provided written informed consent before participation in the study. --- Analysis We used Stata V.14.0 for data analysis. We used the'svy' command for all analysis to account for the complex survey design. 20 Before any of the survey estimation commands were used, the svyset command was used to specify the variables that described the stratification, sampling weight and primary sampling unit variables. This analysis included data obtained from 16 284 study participants. All the responses coded as refused, unknown or missing were treated as missing data. The frequency (percentages) and mean were reported to display the level of population health status and the sample characteristics. Percentages of those reporting any problems in EQ5D domains and mean VAS were stratified by respondents' demographic characteristics-age, gender, marital status and SES-education, income and employment status; and health-related indicators-presence of chronic conditions-were reported. Additionally, prevalence ratios of moderate or severe health problems in people with and without chronic conditions were estimated using log binomial regression. Where the model did not reach convergence, Poisson regression model was used. The model was adjusted for sociodemographic covariates (age, gender, marital status, education level and household income) and city. Linear regression analysis was performed to explore the relationship between the VAS and the EQ5D measures across major subgroups. In the regression model, VAS was used as a dependent variable, and EQ5D measures were treated as independent variables. --- STudy ReSulTS --- Characteristics of the study population A total of 17 274 individuals in 10 002 households were approached in the three study sites (7596 participants in Chennai, 5420 in Delhi, 4258 in Karachi). From these, a total of 16 287 participants were recruited (the overall response rate was 94.3% at the participant level: 6906 Chennai (90.9%), 5364 Delhi (98.9%) and 4017 Karachi (94.3%)). Detailed baseline characteristics of the CARRS cohort are published elsewhere. [21][22][23][24] Briefly, the mean age was 42.4 (<unk>13.3), 52.4% were female, 61% completed secondary schooling and the majority of respondents (72.5%) reported household income level <unk>10 000 Indian rupees (US$200). Two-thirds (66%) of the study population had BMI <unk>25, one-fifth (20%) of the respondents reported current tobacco use, and 37.5% had self-reported chronic conditions (hypertension, diabetes, heart disease, stroke or chronic kidney disease). --- Overall hRQOl by age and gender A total of 16 284 study participants completed the EQ5D-VAS (99.9%). Overall, the percentage of respondents reporting any problems in mobility and pain/ discomfort (14% each) was higher than for other domains. Greater health problems were observed with higher age for both men and women (p<unk>0.001) (table 1). Problems with mobility were higher with advancing age. However, problems with anxiety/depression did not show such trend. Average health status (VAS) reported by the CARRS cohort was 74.5 (95% CI 73.7 to 74.2) (figure 1). Women reported lower health status than men (71.6 vs 79.0; p<unk>0.001). Of the respondents 74% rated a perfect health profile with no difficulties in any EQ-5D domain, and 0.06% rated the worst health profile whereby they had difficulties with every EQ-5D domain. The distribution of the VAS scores was skewed in the direction of best-imagined health state. Only 0.5% respondents rated their health status on VAS under 10, and 10% rated it under 50 (online supplementary appendix 1). hRQOl and SeS Table 2 and Figure 2 depict the mean VAS, percentage and prevalence ratios of respondents reporting moderate or severe problems in the five dimensions, across various subgroups, respectively. Employed adults and students reported better health status than home makers, retired or unemployed participants. We observed almost equal health status in home makers and retired people. Health status was also similar in the middle-income and high-income groups, while it was significantly lower in the low-income group. Individuals with higher education (graduate and above) and high income had higher HRQOL than those with secondary or primary schooling and low-income class. Also, individuals with BMI <unk>18-24 kg/m 2 reported better health status than those with BMI <unk>25 kg/ m 2. Current tobacco users reported better health status than former tobacco users or non-users. However, in a stratified analysis of HRQOL in tobacco users by presence or absence of chronic conditions, tobacco users with chronic conditions reported worse health status than non-users. Open Access hRQOl and chronic conditions Overall, individuals with chronic conditions reported lower health status than those without chronic conditions. About half of the respondents with self-reported diabetes, hypertension, stroke, heart disease or chronic kidney disease reported moderate or severe problems in all five domains (table 2). Table 3 presents the adjusted prevalence ratio of moderate or severe problems among people with versus without chronic conditions, stratified by sex and cities. Individuals with chronic conditions reported two times greater problems in mobility, usual activities domains, pain/discomfort and anxiety/depression, than in individuals without chronic conditions. Further, a small proportion of individuals with chronic conditions, mostly those with hypertension (10.5%) or diabetes (8.3%), reported having a perfect health state. --- Relationship between VAS and eQ5d measures across major subgroups We expected that each EQ5D dimension would have an independent relationship with VAS since each of them represents a different aspect of HRQOL. Online supplementary appendix 2 provides the beta coefficients of the weighted regression models (ie, with the application of the population sampling weights). In the overall population, having any problems in mobility, self-care, pain/ discomfort and anxiety/depression were associated with VAS scores that were 10-12 points lower. This inverse relationship of lower VAS with higher domain difficulties was larger in men, elderly (>60 years), low-income, less educated, divorced and high BMI individuals, compared with their respective counterparts. Tobacco users who reported difficulties in all domains of EQ5D had lower VAS scores (indicating lower quality of life). Kidney disease and stroke were the most disabling conditions on all measures. dISCuSSIOn Comparative assessments of HRQOL variations by sociodemographic factors and chronic conditions aid in prioritising public health targets for intervention. Results from this study indicate that less than 10% of the respondents rated their health status as 100 (ie, best-imagined health state) on VAS. Mobility, pain/discomfort and anxiety/ depression were the most commonly reported problems, with the extent of these problems differing across population subgroup. Elderly (>60 years) and women reported significantly greater problems in the mobility, pain/ discomfort and anxiety/depression domains. The mean VAS in our study was 74.5, which is lower than reported by most Western countries (82.5), but comparable to the results from other low-income and middle-income countries (LMIC) (71.1-77.8) (online supplementary appendix 3). [25][26][27][28] Lower health status reported by urban South Asians can be interpreted in a number of ways. The lower scores may be related --- Open Access to generally lower reporting of HRQOL among this group. Alternatively, these scores may reflect morbidity and suboptimal access to healthcare facilities to address health concerns. South Asians experience chronic conditions at relatively younger ages than other race/ethnic groups, and the associated reductions in productivity and income levels may be manifested in these self-reported EQ5D-VAS scores. A higher percentage of individuals reported problems in mobility dimension (14%-17%) in South Asia, which is comparable to results from other LMICs. 29 However, problems in anxiety/depression are pretty low in South Asia when compared with the rest of the world; this could be due to underdiagnoses of depression-related problems or stigma attached to these health conditions. These patterns could also reflect differences in how adults in different parts of the world self-rate their health. Worse health status in retired or home makers, compared with employed persons, may be related to being homebound or reflect underlying illnesses that may be the factor driving these participants to be homebound and not employed. In terms of modifiable risk factors, maintaining a healthy BMI cut-off (18-25 kg/m 2 ) is favoured because individuals with BMI <unk>18 kg/m 2 and <unk>25 kg/m 2 reported greater problems in all five domains. Although previous studies have shown that lower levels (intensity) of tobacco use are linked with higher HRQOL and regular tobacco users with worse health status, 30 31 in our study, former tobacco users reported lower HRQOL than current users. This finding may indicate reverse causality, that is, former tobacco users after experiencing an illness would have quit smoking/tobacco. Further, supported by the fact that tobacco users with chronic conditions or greater difficulties in EQ5D domains had lower VAS scores is suggestive that morbidity and not the habit of tobacco use per se is more closely related to participants' perception of health. However, a causal link between tobacco use and HRQOL cannot be confirmed in this cross-sectional study. Longitudinal analyses of the independent associations between the smoking/tobacco with HRQOL may provide a better understanding of this relationship. Notably, one in five individuals living with known hypertension or diabetes (average disease duration 4 years) still reported a perfect health state, indicating that these individuals may feel asymptomatic until they experience a clinical event. Also, very small proportions of patients with heart disease and stroke (with longer duration of illness; average 9 years) reported perfect health states, suggesting that these individuals may have adapted to their conditions over time and may be benefiting from treatment and self-care that improve their self-rated quality of life. However, we did not investigate whether these other factors like adherence influence quality of life in those living with chronic conditions. Due to the differences in statistical analyses, HRQOL measures, sociodemographic characteristics of the sample and medical conditions selected, the results of this study may not be directly comparable to reports from other countries. 32 Nevertheless, a few differences and common findings are noteworthy. Individuals with stroke or chronic kidney disease rated the lowest health status, Figure 1 Mean self-rated health status using the European Quality of Life Five Dimension-Visual Analogue Scale (EQ5D-VAS) of respondents by age groups and gender. This figure presents the mean self-rated health status for overall study population by age groups and gender. The EQ5D-VAS measures health status on a scale of 0 (worst health status) to 100 (best imaginable health status). *p Value for difference between mean EQ5D-VAS between men and women at each age group is statistically significant, p<unk>0.01. Open Access which is consistent with results reported from other studies done in China, Thailand and Western populations. 28 33-36 Since the respondent's health status could be affected by how well the condition was managed, caution is needed in interpreting study results regarding the relative effect of chronic conditions on HRQOL. [37][38][39][40][41] A more Open Access recent Canadian study conducted by Mo et al 42 indicated a strong relationship between low Health Utility Index scores and certain chronic conditions. The authors found that arthritis/rheumatism, heart disease, hypertension, cataracts and diabetes had a negative impact on HRQOL. In the USA, Medical Expenditure Panel Survey data-based study reported that, after adjusting for sociodemographic variables, all of the selected chronic conditions were associated with lower EQ5D scores, with effects greatest for emphysema, followed by heart disease, stroke, high BP, diabetes and asthma. 43 44 Strengths and limitations of this study To our knowledge, this is the first population-level HRQOL data from South Asia using EQ5D-VAS including three large metropolitan cities in India and Pakistan with a large sample size that has used multistage cluster random sampling strategy and standardised protocols and measurement tools across sites. Our data provide the first baseline values to be used for monitoring population health status and analysed the relationships between selected chronic conditions and HRQOL. This information could be used to complement national targets by providing a measure of chronic disease burden based on perceived health status rather than solely on mortality and disease prevalence. In our secondary data analysis, EQ5D and VAS measures correlated well, which confirms the convergent and discriminate validity of the EQ5D instrument. There are several limitations to this study. First, due to the cross-sectional nature of the data, the causal relationship between socioeconomic parameters/chronic conditions and HRQOL cannot be determined and is not implied. Second, many chronic conditions (respiratory, locomotor, cancer and others) were not included in the survey. Therefore, the ranking of most severe health conditions and associated HRQOL is not complete. Third, the selected chronic conditions were self-reported, and the study investigators did not examine the accuracy of information. However, this poses less of a threat to validity because self-reporting of heart diseases, stroke and kidney diseases is pretty accurate in community surveys. [45][46][47][48] Further, hypertension and diabetes were measured in this study using standardised methods. Lastly, EQ5D data were self-reported and the variation in how individuals perceive disability varies widely. However, this should be less of a problem given the large sample size in this study. Fourth, the findings of this study may not be replicable if researchers use a different HRQOL instrument, [49][50][51][52][53][54] which can be tested in a future study. Public health relevance and policy implications HRQOL data from this study provide baseline values for monitoring variations in health for specific population groups on the basis of gender, education, employment, income, presence of chronic conditions and place of residence. These data are also relevant to assess the overall burden of physical and mental health problems that are not disease-specific. In aggregate form, such information could be used to complement national health targets by providing a measure based on health status (quality of life) rather than mortality or disease prevalence alone. Therefore, the policy makers can use the HRQOL measures and the resulting data from this study to minimise health disparities and allocate resources among competing health programme based on burden of physical or mental health problems in a specific group. 55 The lower health status reported by female, less educated, unemployed and low-income groups may indicate higher levels of stress in these groups. 17 Other potential contributing factors that are known to influence health status are living conditions, gross domestic product per capita, inequities in income distributions and access to healthcare. [56][57][58][59][60][61] Therefore, public health initiatives should focus on intersectoral approaches to address issues of education, generating more avenues for employment and improving the quality and access of primary healthcare. Lastly, the issue of 'clinical' or 'policy' relevance of the difference in EQ5D measures needs much discourse. For example, if the VAS in two groups of the population is 5 or 10 points different from each other, we cannot make a clinical judgement on how much these two groups would differ in their actual health status. These issues relate to determining a minimally significant difference/change in HRQOL and needs investigation in future studies. However, because of HRQOL sensitivity to time trends as shown in previous studies, [62][63][64] these measures are also likely to be useful in determining the effect of major population-based policies or interventions. COnCluSIOn HRQOL appears to be lower with higher age and among women in South Asia. Our data demonstrate significantly lower HRQOL in key demographic groups and those with chronic conditions, which are consistent with previous studies. These data provide insights on inequalities in population health status, and potentially reveal unmet needs in the community to guide health policies. --- Open Access significantly in the revision of the manuscript. All authors have approved the submission of this version of the manuscript. --- Open Access

Objectives Health-related quality of life (HRQOL) is a key indicator of health. However, HRQOL data from representative populations in South Asia are lacking. This study aims to describe HRQOL overall, by age, gender and socioeconomic status, and examine the associations between selected chronic conditions and HRQOL in adults from three urban cities in South Asia. Methods We used data from 16 287 adults aged ≥20 years from the baseline survey of the Centre for Cardiometabolic Risk Reduction in South Asia cohort (2010-2011). HRQOL was measured using the European Quality of Life Five Dimension-Visual Analogue Scale (EQ5D-VAS), which measures health status on a scale of 0 (worst health status) to 100 (best possible health status). Results 16 284 participants completed the EQ5D-VAS. Mean age was 42.4 (±13.3) years and 52.4% were women. 14% of the respondents reported problems in mobility and pain/discomfort domains. Mean VAS score was 74 (95% CI 73.7 to 74.2). Significantly lower health status was found in elderly (64.1), women (71.6), unemployed (68.4), less educated (71.2) and low-income group (73.4). Individuals with chronic conditions reported worse health status than those without (67.4 vs 76.2): prevalence ratio, 1.8 (95% CI 1.61 to 2.04). Conclusions Our data demonstrate significantly lower HRQOL in key demographic groups and those with chronic conditions, which is consistent with previous studies. These data provide insights on inequalities in population health status, and potentially reveal unmet needs in the community to guide health policies.

Plain English text Although most women would want to wait for more than two years before having another baby, their fertility behaviours during the first year after birth may decrease or increase the length of the birth interval. The objectives of this study were to: assess how protected postpartum women were against pregnancy, based on their patterns of amenorrhoea and sexual abstinence; determine the timing of postpartum contraception in relation to amenorrhoea and sexual abstinence; and determine the predictors of postpartum contraceptive use. This was a prospective study carried out in the Mfantseman Municipality of the Central region of Ghana. Out of 1914 women attending antenatal clinics in the municipal within the study period, 1350 agreed to be part of the study to ascertain their postpartum fertility and contraceptive behaviours a year following delivery. These women were traced to their communities using telephone and house numbers provided and only 1003 of the women were finally traced and interviewed. The women were asked about their breastfeeding behaviour, postpartum sexual abstinence, duration of amenorrhoea and postpartum contraceptive use. The mean age of the respondents was 29.9 <unk> 6.5 years; adolescents constituted the least proportion (3.3%) of the women. More than half (54.1%) of the women had Middle, Junior secondary school or Junior high school education. Most (43.3%) of the women were married by means of traditional rites and more than half (51.4%) of them were petty traders. The mean durations of breastfeeding, amenorrhoea and sexual abstinence were 6.6 <unk> 2.8 months, 7.8 <unk> 3.8 months and 4.4 <unk> 3.1 months respectively, whilst the mean time of first contraceptive uptake was 3.5 <unk> 2.7 months postpartum. The time to first use of modern contraceptive method during the postpartum period indicates that about 50% of the women had started use of modern contraceptive methods by 2.7 months postpartum, and occured at 0.6 and 3.6 months before sexual relations and resumption of menses respectively. Occupation (likelihood ratio p = 0.013), area of residence (likelihood ratio p = 0.004), mode of delivery (likelihood ratio p <unk> 0.001), breastfeeding (p = 0.024), period since delivery (p <unk> 0.001), preferred number of children (p <unk> 0.001) and parity (p <unk> 0.001) were found to be predictors of postpartum contraceptive use. Postpartum women in the Mfantseman municipal who did not use contraceptives or delayed in the use of contraceptives after birth were least likely to be protected against pregnancy in the postpartum period, whilst those who adopted postpartum family planning were likely to be better protected because they were likely to adopt it within the first three months after birth and before the onset of sexual relations and first menses. The predictors of postpartum contraceptive use were breastfeeding pattern, occupation, parity, preferred number of children, period since delivery, place of residence and mode of delivery. --- Background The 2006 World Health Report proposed a 2-3 year birth interval and a six month conception interval following miscarriage or abortion, because they ensured good maternal and child health outcomes [1]. Evidence exists that if couples could space their pregnancies by at least two years, up to 35% of maternal deaths and up to 13% of child mortalities could be averted [2][3][4], whilst 25% of underfive mortalities could be averted if birth intervals were at least three years [3]. Although most women would want to wait for more than two years before having another baby, their fertility behaviours during the first year following birth may decrease or increase the length of the birth interval [5]. These behaviours include breastfeeding and its influence on return of menses (postpartum amenorrhoea), return to sexual activity (postpartum sexual abstinence) and use of maternal health services. Some of these behaviours biologically reduce fertility while some affect decisions to use family planning during the postpartum period [5]. Women are not at risk of pregnancy following a birth if they are amenorrhoeic as a result of intensive exclusive breastfeeding or abstaining from sex [6]. The period of insusceptibility, which is influenced by sexual abstinence and exclusive breastfeeding, lengthens the time until the next conception. Use of contraception within this period offers extra protection against the risk of pregnancy [7]. The length and intensity of breastfeeding and the length of amenorrhea and sexual abstinence vary among women and within societies. In this respect, postpartum behaviours need to be understood in relation to the contraception behaviours of women [7] if unintended pregnancies are to be avoided. The objectives of this study were to: assess how protected postpartum women in the Mfantseman municipal were against pregnancy, based on their patterns of amenorrhoea and sexual abstinence; determine the timing of postpartum contraception in relation to amenorrhoea and sexual abstinence; and determine the predictors of postpartum contraceptive use. --- Methods --- Study area The study was carried out in the Mfantseman Municipal area of the Central Region of Ghana. This is a coastal and predominantly rural district in the southern part of Ghana. The main ethnic group is Fante. The main occupation of the people in the district are farming, fishing and trading. The area was selected because of the high levels of teen pregnancies (13.6% of all pregnancies) high abortion rates amongst the teens and very low family planning uptake which had been consistently below 10% over a three year period from 2007 to 2009 according to the 2010 annual report of the Municipal Health Directorate [8]. --- Study design and data source This was a prospective study of postpartum women who had been encountered earlier at antenatal clinic in four selected health facilities in the Mfantseman Municipal area (Saltpond hospital, Mankessim health centre, Anomabo and Biriwa health centres) where their intention to use postpartum family planning was ascertained. These women agreed to be followed up to their communities within a year following delivery to ascertain their postpartum reproductive and contraceptive behaviours. Data on duration of breastfeeding, amenorrhoea and sexual abstinence, and time of first contraceptive use, continuation or discontinuation, were obtained within one calendar year after last birth using a reproductive event sheet adapted from DHS contraceptive calendar. The data sheets captured month-by-month data on contraceptive use or non-use, continuation or discontinuation of contraception, breastfeeding patterns, sexual behaviour and menstrual resumption to cover period since last birth. The data sheets were reviewed one year postpartum at the time of interview; however, periodic contacts with the respondents were done through telephone calls, home visits and personal contacts to ensure they were capturing data consistently and to clarify any issues that may have arisen. Reasons for use, non-use and discontinuation if any, were obtained from participants at the time of the interview. Information on socio-demographic and socio-economic characteristics was also obtained. --- Sample size estimation and sampling Based on an estimated target population of 4218 (Average total number of women visiting ANC at all selected health facilities per quarter, per year from 2008 to 2010) and the assumption that 50% of pregnant women intended to adopt postpartum family planning, within a margin of error of 3%, a minimum sample size ST, was estimated as follows: For a finite population, the sample size ST, was estimated by the formula: ST = A / [1 + (A-1)/T] [9], where A is given by [Z 2 *P*(1-P)] / C 2 ; T = estimated target population; Z = Z value (1.96 for 95% confidence); P = Proportion of pregnant women who intended to adopt postpartum family planning; and C = margin of error. This implies, A = [1.96 2 (0.5) (0.5)]/0.03 2 = 1067 and ST =1067 / [1+ (1067-1)/4218] =852; (approximated to 900). To take care of defaults and late ANC registrations (each constituting 50% of computed sample size) respectively [10], the minimum sample size (ST = 900) was doubled to 1800 with an additional 10% mark-up for women who would decline to be interviewed. The estimated total sample size was 1980. Within the survey period (2nd January to 30th April 2012), each of the selected health facilities was visited during the days designated for antenatal clinic. At the selected health facilities, all antenatal registrants (pregnant women), irrespective of the period of gestation, who lived in the Mfantseman Municipal area and who were aged 15 to 49 years, were targeted to be part of the study. During the prenatal phase of the study, 1914 antenatal registrants were encountered. Sampling of these was by total enumeration technique. Each of the antenatal registrants was asked if they would wish to be followed up a year after they had delivered to find out about contraceptive use and other reproductive health behaviours. Only 71% (1359) of them agreed to be followed up after delivery. The other 29% declined follow up for religious, socio-cultural and personal reasons considering the sensitive nature of family planning within such communities. Several strategies to further reduce non-response and attrition and improve retention rates were employed in this study. These included the following: first, giving detailed explanations about the study objectives and its possible impacts on the individual, family and society and allaying any anxieties and fears about participating in the study, whilst ensuring that emotional support was on hand to deal with extreme cases; second, obtaining detailed personal information including names, telephone numbers, house addresses (where available) and detailed descriptions of directions to houses of respondents; third, providing adequate motivation for participation by ensuring that the research assistants were friendly, showed respect and courtesy to the respondents and provided adequate privacy at the environment of the interview; fourth, improving rapport between research team and respondents by making periodic contacts with the respondents who agreed to be followed up through telephone calls, home visits and personal contacts; fifth, providing learning opportunities to the research assistants by the comprehensive training given. This motivated the research assistants to engage the respondents in ways that improved rapport and encouraged participation; sixth, providing research assistants in Phase 2 additional incentives for transportation to help them access all the respondents assigned to them, especially to the remotest parts of the municipality and seventh, regular sensitization about the study carried out by the Municipal Health Directorate through their health centres, outreach points and home visits. These strategies notwithstanding, only 1003 (74%) of the 1359 who agreed to be followed up were encountered. Twenty-six percent (26%) could not be traced mainly because of poor address systems in the communities and non-functional telephone numbers. Follow up interviews were carried out from 2nd January to 30th May 2014. These were done via face-to-face and telephone interviews using home addresses and telephone numbers provided. --- Data management and analysis The data obtained were double-entered using EPI-DATA, verified and cleaned. The cleaned data were exported into STATA (version 11) for analysis. Descriptive, chi square and logistic regression statistics were carried out to describe the socio-demographic and socio-economic characteristics of respondents, determine associations between contraceptive use and breastfeeding, amenorrhoea and sexual abstinence and determine the predictors of postpartum contraceptive use, respectively. Survival analysis techniques were used to assess the time to resumption of menstrual flow, resumption of sexual activities and time to first contraceptive use. --- Results --- Socio-demographic characteristics of respondents Table 1 presents the distribution of socio-demographic characteristics of respondents in this study. The mean age of the respondents was 29.9 <unk> 6.5 years. Majority (73.9%) of the women were in their prime age (20-34 years) whilst adolescents constituted the least proportion (3.3%) of the women. More than half (54.1%) of the women had attained Middle, Junior secondary school (JSS) or Junior high school(JSS) education and a little above 8% (8.3%) had no formal education. The ethnic group with the highest proportion was Fante (88.7%). Christians were the majority (90.3%) religious group among the women, followed by the Muslims (7.6%). Most (43.3%) of the women had married by means of traditional rites and more than half (51.4%) of them were petty traders. --- Breastfeeding, amenorrhoea, sexual abstinence and contraceptive use postpartum The mean durations of breastfeeding, amenorrhoea and sexual abstinence were 6.6 <unk> 2.8 months, 7.8 <unk> 3.8 months and 4.4 <unk> 3.1 months respectively. The mean time of first contraceptive uptake following the last birth among women who had used contraceptives was 3.5 <unk> 2.7 months. Table 2 shows the length of breastfeeding, amenorrhoea and sexual abstinence and time of first contraceptive uptake since the last birth. Among those who breastfed, majority (71.6%) had breastfed for six months whilst majority (62.5%) had remained amenorrheic for seven months up to a year. The period of sexual abstinence among the postpartum women was relatively shorter. Majority (66.2%) had abstained for periods of one to four months following birth. Out of the 1003 postpartum women interviewed, 505 (50.3%) of them reported using contraceptives following birth, however only 422 provided correct information regarding time of contraceptive use following birth. None of these women reported any discontinuation once they started using contraceptives of their choices. The 83 women dropped had incorrect or inconsistent data or had missing information on record sheets. Nearly 66% had used contraceptives between the first and third month after birth. The most common methods used were male condoms (33.7%), injectables (30.3%) and pills (20.6%) respectively (Fig. 1). Figure 2 shows the survival curves for first sexual resumption, first menstrual resumption and first contraceptive use by ordinal postpartum month. The survival curves show that the time at which 50% of the women had resumed their menses was about 6.3 months postpartum. Also, the time that 50% of them had resumed sex was 3.3 months postpartum, which is 3 months before resumption of first menses. This implies that women who were not using any postpartum contraceptive methods for any reason could be at risk of unintended pregnancy following resumption of sexual relations, especially when the intensity of breastfeeding could not be guaranteed (Reasons for non-use have been provided in Fig. 3). The time to first use of a modern contraceptive method during postpartum period indicates that about 50% of the women had used modern contraceptive methods by 2.7 months postpartum, and occured 0.6 and 3.6 months prior to resumption of sexual relations and menses respectively. --- Predictors of postpartum contraceptive use Tables 3 and4 summarize the association between postpartum contraceptive use among women and their socio-demographic and reproductive characteristics respectively. Women's age (p <unk> 0.001), educational level (p <unk> 0.001), occupation (p <unk> 0.001), area of residence (p <unk> 0.001) and marital status (p <unk> 0.001) were found to be associated with postpartum contraceptive use. However, ethnicity (p = 0.061) and religion (p = 0.287) of women were not associated with postpartum contraceptive use. There was an association between postpartum contraceptive use and resumption of sexual relationship (p <unk> 0.001), amenorrhoea (p <unk> 0.001), and breastfeeding (p <unk> 0.001). Table 5 presents the univariate and multivariate logistic regression analyses depicting the unadjusted (OR) and adjusted odds ratios (aOR) of factors that influence contraceptive use. All the variables that were associated with postpartum contraceptive use among women (Tables 3 and4) were used to construct the univariate and the multivariate models. In the univariate logistic regression, women within the age group 35-39 years were 2.62 times more likely to use postpartum contraceptive as compared to those within the age group 15-19 years (OR = 2.62, 95% CI: 1.20-5.72, p = 0.015). Women with ages 40 years or above also had higher odds of using postpartum contraceptive as compared to adolescents (15-19 years). Women with Middle, Junior secondary school or Junior high school (OR = 1.69, 95% CI: 1.06-2.69, p = 0.028) also had increased likelihood of using postpartum contraceptive as compared with women without formal education. Civil or public servants were 48% less likely to use postpartum contraceptive as compared to fish mongers (OR = 0.52, 95% CI: 0.28-0.95, p = 0.034). Women living in Anomabo (OR = 2.69, 95% CI: 1.80-3.99, p <unk> 0.001) and Mankessim (OR = 0.71, 95% CI: 0.51-0.99, p = Various postpartum contraceptive methods were used by the women These included oral contraceptive pills (pills), injectables, implants, male condoms, female sterilization, lactational amenorrhoea method (LAM), periodic abstinence and withdrawal method. Of all these methods, the ones that were used by most women were the male condom (33.8%), injectables (30.3%) and the oral contraceptive pills. The least used methods were the withdrawal (2.1%) and lactational amenorrhoea (2.1%) methods 0.045) had higher and lower odds respectively of using postpartum contraceptive as compared to women living in Saltpond. Moreover, women who were married through traditional rites (OR = 0.66, 95% CI: 0.47-0.92, p = 0.013), engaged (OR = 0.42, 95% CI: 0.29-0.61, p <unk> 0.001) and were single, separated, widowed or divorced (OR = 0.47, 95% CI: 0.24-0.91, p = 0.026) were less likely to use postpartum contraceptive. Women with assisted vaginal delivery were 87% less likely to use postpartum contraceptive as compared to women who had spontaneous vaginal delivery (SVD) (OR = 0.13, 95% CI: 0.07-0.25, p <unk> 0.001). Compared to women who had complications, those who did not have complications had decreased likelihood of using postpartum contraceptive (OR = 0.33, 95% CI: 0.33-0.65, p <unk> 0.001). In addition, women who were not breastfeeding had two-fold of using postpartum contraceptive (OR = 2.04, 95% CI: 1.50-2.78, p <unk> 0.001). Amenorrhoeic women were 52% less likely to use postpartum contraceptive (OR = 0.48, 95% CI: 0.32-0.72, p <unk> 0.001). For a unit increase in the women's preferred number of children, their odds of using postpartum contraceptive increased more than two-fold (OR = 2.16, 95% CI: 1.78-2.63, p <unk> 0.001). For a unit increase in parity, the likelihood of women using postpartum contraceptive increased by 1.9 (OR = 1.90, 95% CI: 1.71-2.10, p <unk> 0.001). --- Multivariate regression In the multivariate logistic regression, occupation (likelihood ratio p = 0.013), area of residence (likelihood ratio p = 0.004), mode of delivery (likelihood ratio p <unk> 0.001), breastfeeding (p = 0.024), period since delivery (p <unk> 0.001), preferred number of children (p <unk> 0.001) and parity (p <unk> 0.001) were found to be predictors of postpartum contraceptive use. Women who were students (aOR = 5.00, 95% CI: 1.12-22.40, p = 0.035) had higher odds of using postpartum contraceptives. Women living in Mankessim (aOR = 0.58, 95% CI: 0.34-0.99, p = 0.049) were less likely to use postpartum contraceptives as compared with those living in Saltpond. Compared with those who had SVD, women who had assisted vaginal delivery were 80% less likely (aOR = 0.20, 95% CI: 0.06-0.70, p = 0.012) to use postpartum contraceptive, whereas women who had caesarean section had 5.12 times (aOR = 5.12, 95% CI: 2.24-11.68, p = 0.049) likelihood of using postpartum contraceptive. Women who did not breastfeed had more than double likelihood of using postpartum contraceptive (aOR The women who had not used any family planning methods within the first year after delivery gave several reasons why they had not done so. These included the fact that some got pregnant, some partners were against the idea to do family planning, some women wanted to get pregnant soon, some were not interested in family planning, whilst others thought it was too early to decide. Lack of interest in family planning (29%), too early to take a decision about family planning (19.9%) and women wanting to get pregnant soon (15%) were reasons given by most of the women = 2.16, 95% CI: 1.11-4.21, p = 0.024) compared to those who breastfed. For a unit increase in the period since delivery, the women were 0.89 times less likely to use postpartum contraceptive (aOR = 0.89, 95% CI: 0.84-0.95, p <unk> 0.001). For a unit increase in the preferred number of children, the odds of using postpartum contraceptive increased among women (aOR = 3.95, 95% CI: 2.77-5.62, p <unk> 0.001). For a unit increase in parity, the likelihood of women using postpartum contraceptive increased by 2.3 (aOR = 2.26, 95% CI: 1.83-2.78, p <unk> 0.001). --- Discussion The objectives of the study were to: assess how protected postpartum women in the Mfantseman municipal were against pregnancy, based on their patterns of amenorrhoea and sexual abstinence; determine the timing of postpartum contraception in relation to amenorrhoea and sexual abstinence; and determine the predictors of postpartum contraceptive use. The intensity and duration of breastfeeding, length of amenorrhoea and sexual abstinence, individually or in combination affect women's risk of getting pregnant within the extended postpartum period and before the two-year recommended spacing period after the last birth [1], especially in the absence of contraceptive use. Traditionally, women in sub-Saharan Africa relied mainly on breastfeeding, amenorrhoea and sexual abstinence to space their births. Periods of breastfeeding for as long as two years and its attendant long duration of amenorrhoea in some settings and long periods of sexual abstinence in middle and West Africa, were the norm [11]. Increasing modernization, urbanization and social change have gradually reduced the effectiveness of these traditional birth spacing mechanisms and therefore women have increasingly been at risk of unintended pregnancies in the postpartum period [12]. The mean duration of breastfeeding among the study participants was 6.6 <unk> 2.8. This is greater than the mean duration of exclusive breastfeeding in Ghana of 3.9 months and less than the mean duration of any breastfeeding of 21.2 months in Ghana and 20.4 months in the central region respectively [13]. The mean duration of any breastfeeding in the municipality was about 14% lower than that of the region and was among the lowest in the region. This may be due to the increase in the use of bottle feeding (which some women consider as modern) which also increased in the country between 2008 and 2014 [13]. Postpartum protection from conception depends upon the intensity and duration of breastfeeding. In the absence of reasonably intensive breastfeeding, women are likely to ovulate before the end of the second postpartum month and hence become susceptible to pregnancy [14]. A woman is considered insusceptible if she is not exposed to the risk of pregnancy, either because she is amenorrhoeic as a result of reasonably intensive breastfeeding exclusively for the first six months or because she is abstaining from sexual intercourse following birth [13]. The median duration of amenorrhoea and sexual abstinence among the respondents were 7.8 and 4.4 months respectively. These were shorter than the median durations of amenorrhoea and sexual abstinence for central region (9.9 and 6.6 months) and Ghana (8.4 and 5.9 months) respectively [13]. The relatively shorter durations of amenorrhoea may be due to the shorter duration and lower intensity of breastfeeding among the women in the municipality. The relatively shorter durations of sexual abstinence and amenorrhoea and the lower intensity of breastfeeding may expose the women in the municipality to the risk of unintended pregnancy, especially in the absence of contraceptive use. This may explain the high rates of unintended pregnancy among pregnant women in the municipality [15]. Among the women who used contraceptives, time of initiation was earlier than initiation of sexual relations and first menses following their last birth. Mean time of initiation of [12]. The relationship observed between postpartum contraceptives use, the initiation of sexual relations and first menses in the Mfantseman municipal is not very common in sub-Saharan Africa [16][17][18]. The finding that women who adopted contraceptives did so early and prior to resumption of sexual relations and first menses is generally encouraging; however, it must be borne in mind that it does not guarantee the prevention of unintended pregnancies and adequate spacing of births, unless the methods adopted were not discontinued or unless permanent methods were adopted. In this study, the women after initiation of contraceptives did not report discontinuation within the first year postpartum. The most widely used methods adopted among the women in the Mfantseman municipal were male condoms, followed by injectables and pills. In Ghana, the most widely used methods among married women were injectables, implants and pills, whilst male condoms, pills and injectables were the most widely used among unmarried women [14]. The methods used among the women were different because there were several methods available and the women had the freedom to choose which methods were suitable for them. Among postpartum women in a study among urban slum dwellers in Nairobi, the most widely used methods were injectables and pills [12]. It is not easy to predict how early the return of ovulation and menses for a particular woman who did not breastfeed would be. However, evidence exist that women who do not breastfeed after birth usually have early return of ovulation and menses and are thus at greater risk of unintended pregnancies. Education about this fact is provided by midwives during antenatal care in Ghana and other settings. In this study women who did not breastfeed after birth were found to be more likely to use postpartum contraceptives compared to those who breastfed. This is consistent with a study in Egypt where women who exclusively breastfed were less likely to use modern methods of contraception postpartum [19]. In this study, women who underwent caesarean section were five times likely to use postpartum contraceptives compared to those who delivered via spontaneous vaginal delivery. In contrast, those who had assisted vaginal delivery were less likely to use postpartum contraceptives. The reasons for this difference are not very clear in this study setting; however it is known that women who underwent caesarean section were often inclined to use some form of postpartum contraception and preferred methods like bilateral tubal ligation, intrauterine device and injectables [20,21] in order to avoid repeat caesarean section from another pregnancy too soon. Consistent with a study in Iran [22], place of residence influences contraceptive use among postpartum. Women living in Mankessim were less likely to use postpartum contraceptives compared to those resident in Saltpond. This was unexpected in the sense that women from Mankessim had the highest awareness (83%) about postpartum contraceptives [15] yet were the least likely to use them. Myths, fear of side effects and opposition by significant others may likely influence this observation. Further inquiry may be required to ascertain these facts. The finding in this study that students were more likely than fishmongers to adopt postpartum contraceptives is encouraging in the sense that girls may need to go back to school to complete and advance their educational status in line with the Ghana Education Service's policy and also to avoid any future unintended pregnancies. High Parity has been found in several studies to be associated with contraceptive use [23][24][25]. The finding in this study is similar. Women of high parity may have satisfied the number and sex of their children and may wish to limit their births. --- Strengths and limitations of study The prospective design was appropriate for the study considering the fact that accurate timing of postpartum reproductive health events was important to establish the relationship between them and contraceptive use. Although some strategies were put in place to minimize non-response and attrition (see methodology), the rate nevertheless was still very high (wave I response rate of 97% versus wave II response rate of 51%) and lead to sampling bias, reduced generalizability of study results and increased variance of study estimates. The threat of selection bias existed, but was highly mitigated by ensuring that the data collectors explained the study objectives and their implications very well to the respondents. The study findings may not be generalizable to the national and regional populations but to coastal and predominantly rural populations within the central region and country. The absence of data on duration of family planning use, seriously limits any conclusions about pregnancy risk in the first postpartum year. Some of the data collectors abandoned the study because of inadequate remuneration. New data collectors had to be trained to continue data collection. This brought about some delays in data analysis and reporting. --- Conclusion The objectives of this study were to assess how protected postpartum women in the Mfantseman municipal were against pregnancy, based on their patterns of amenorrhoea and sexual abstinence; determine the timing of postpartum contraception in relation to amenorrhoea and sexual abstinence; and determine the predictors of postpartum contraceptive use. Postpartum women in the Mfantseman municipal who did not use contraceptives or delayed in the use of contraceptives after birth were least likely to be protected against pregnancy in the post partum period because of early initiation of postpartum sexual relations, poor intensity and relatively shorter duration of breastfeeding. Women who adopted postpartum family planning were likely to be better protected because they were likely to adopt it within the first three months after birth and before the onset of sexual relations and first menses. The predictors of postpartum contraceptive use were breastfeeding pattern, occupation, parity, preferred number of children, place of residence and mode of delivery. --- Availability of data and materials The data that supports the study has individual identifiers and researcher is reserved to provide it. --- Abbreviations FP: Family Planning; GSS: Ghana Statistical Service; PPFP: Postpartum family planning; SVD: Spontaneous vaginal dellivery; WHO: World Health Orgaization Authors' contributions SE: Conceptualized the study was involved in data collection, entry and analysis and developed the paper; CE was involved in data collection, analysis and reviewed the paper; ASB was involved in data collection, entry and analysis and reviewed the final paper. All authors read and approved the final manuscript. --- Ethics approval and consent to participate Ethical and administrative approvals were obtained from the Ethics Review Committee of the Ghana Health Service and the Municipal Health Directorate (GHS-ERC: 14/09/11). Written informed consent was obtained from each participant and witnessed. Before each interview, the aims, objectives and benefits of the study were explained to respondents; they were then allowed to ask any questions for clarifications. Once they consented, they were made to sign with a witness of their choice. Interviews were conducted in places that guaranteed maximum privacy. For individuals who could have emotional problems from interviews, a Clinical Psychologist from University of Cape Coast was recruited on stand-by to handle cases that were beyond the control of the Principal Investigator. --- Consent for publication Consent was obtained from all the participants to publish information obtained from them but without the personal identifiers. These were also witnessed accordingly. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Although most women would want to wait for more than two years before having another baby, their fertility behaviours during the first year following birth may decrease or increase the length of the birth interval. The objectives of this study were to: assess how protected postpartum women in the Mfantseman municipal were against pregnancy, based on their patterns of amenorrhoea and sexual abstinence; determine the timing of postpartum contraception in relation to amenorrhoea and sexual abstinence; and determine the predictors of postpartum contraceptive use. Methods: This was a prospective study carried out in the Mfantseman Municipality of the Central region of Ghana. Out of 1914 women attending antenatal clinic in the municipal within the study period, 1350 agreed to be part of the study to ascertain their postpartum fertility and contraceptive behaviours a year following delivery. These women were traced to their communities using telephone and house numbers provided and only 1003 of the women were finally traced and interviewed. The women were asked about their breastfeeding behaviour, postpartum sexual abstinence, duration of amenorrhoea and postpartum contraceptive use. Results: The mean age of the respondents was 29.9 ± 6.5 years; adolescents constituted the least proportion (3.3%) of the women. More than half (54.1%) of the women had Middle, Junior secondary school or Junior high school education. Most (43.3%) of the women were married by means of traditional rites and more than half (51.4%) of them were petty traders. The mean durations of breastfeeding, amenorrhoea and sexual abstinence were 6.6 ± 2. 8 months, 7.8 ± 3.8 months and 4.4 ± 3.1 months respectively, whilst mean time of first contraceptive uptake was 3. 5 ± 2.7 months postpartum. The time to first use of modern contraceptive method during the postpartum period indicates that about 50% of the women had started use of modern contraceptive methods by 2.7 months postpartum, and occured 0.6 and 3.6 months before sexual relations and resumption of menses respectively. Occupation (likelihood ratio p = 0.013), area of residence (likelihood ratio p = 0.004), mode of delivery (likelihood ratio p < 0.001), breastfeeding (p = 0.024), period since delivery (p < 0.001), preferred number of children (p < 0.001) and parity (p < 0.001) were found to be predictors of postpartum contraceptive use. Conclusion: Postpartum women in the Mfantseman municipal who did not use contraceptives or delayed in the use of contraceptives after birth were least likely to be protected against pregnancy in the post partum period, whilst those who adopted postpartum family planning were likely to be better protected because they were likely to adopt it within the first three months after birth and before the onset of sexual relations and first menses. The predictors of postpartum contraceptive use were breastfeeding pattern, occupation, parity, preferred number of children, period since delivery, place of residence and mode of delivery.

social class (eg, quality of education, job prestige), allows for consideration of rarely assessed socioeconomic components (ie, wealth), includes experiences of societal inequities (eg, discrimination), and incorporates consideration of future prospects. 1,4,[8][9][10] Together, these aspects of SSS may uniquely affect health-related outcomes through their associations with the psychological and physiological variables that influence health and health-related behaviors [eg, depression and negative affect (cf. 7,11 )]. Most research on the relations of SSS with health outcomes has included the use of a single SSS scale, which asks individuals to rank their standing relative to that of others in the United States (or society in some international studies). Known as the SSS-US (or SSS-SES) ladder, directions specify the consideration of relative standing in the United States with regard to, for example, money, education, and occupation. 12 A lesser used SSS scale is the SSS-community ladder, which asks individuals to rank how they stand relative to others in their self-defined communities, without reference to SES. 12 Research indicates that respondents tend to primarily consider material wealth, occupational status, and education in providing SSS-US rankings, whereas everyday altruistic practices (eg, participation in giving activities such as volunteering or being a good neighbor) were among the highest considerations in providing SSS-community rankings. 12 Accordingly, the SSS-US ladder typically demonstrates stronger associations with objective SES indicators than does the SSS-community ladder. 5,12 For this reason, the SSS-community ladder may be particularly relevant for individuals of lower SES, who might not rate themselves highly on the SSS-US ladder, but who have important and influential roles in their communities that would elevate their SSS-community rankings. Because the referents and determinants of SSS-US and the SSS-community ladder rankings differ, so too may their associations with health outcomes and behaviors. A greater understanding of the relative associations of these ladders with health behaviors would help inform future research, aid in the identification of individuals or groups at risk for negative health outcomes, and expand our knowledge of how different dimensions of social standing affect health outcomes. The first published study that examined both ladders in their relation to health behaviors was by Ghaed and Gallo, who focused on cardiovascular risk behaviors among a sample of (largely) white women. Results indicated that higher SSS-US endorsements were significantly associated with greater consumption of fruits and vegetables in analyses adjusting for sociodemographics, but this association was not significant with the SSScommunity ladder. 5 In addition, there were no independently significant relations between either of the ladders and physical activity, body mass index (BMI), or smoking status in adjusted analyses. 5 Only a couple of papers published following the Ghaed and Gallo study used both the SSS-US and the SSS-community ladders, but with a focus on different health outcomes or nonadult populations. For example, although not the main objective, one study found that leisure-time exercise among adults was significantly and negatively correlated with the SSS-US ladder, and not associated at all with the SSS-community ladder. 13 Another study among Mexican adolescents indicated differing associations with smoking and drinking outcomes for an SSS ladder focused on comparisons with the larger Mexican society (a modified version of the SSS-US ladder) versus the SSS-community ladder. Specifically, this study found negative associations between the SSS-society ladder and smoking and drinking behaviors, but positive associations between the SSS-community ladder and smoking and drinking behaviors. 14 Together, these studies seem to suggest that the SSS-US and the SSS-community ladder are distinct from one another in their association with health behaviors, and mixed results suggest the need for additional research in this area. It is also important to note that the influence of SSS on health outcomes may not be the same among different racial/ethnic groups. For example, previous studies found that race/ ethnicity moderated relationships between SSS and sleep quality, 15 self-rated health, 2 and hypertension. 10 Specifically, one study found that the SSS-society ladder was related to greater impairment in sleep quality among Asian and African Americans, but was unrelated among whites. 15 Another found that, after accounting for the effects of objective SES, ratings on the SSS-US ladder were positively associated with self-rated health among white and Chinese American women, but not African American and Latina women. Another found that ratings on the SSS-US ladder were significantly associated with hypertension among white men and women and African American women in fully adjusted analyses, but there was no relationship between SSS and hypertension among African American men. 10 In addition, research suggests that the determinants of SSS may vary by racial/ethnic group. For example, one study found that objective SES measures were not associated with SSS-US ladder rankings as strongly among African American as among white participants. 10 Therefore, it seems important to examine the effects of SSS on health outcomes within racial/ethnic groups. To the best of our knowledge, there have been no previous studies comparing the relative associations of the ladders with multiple health behaviors among African American adults. The purpose of the current work was to address this gap in the literature. The current study examined the relative associations of the SSS-US and the SSS-community ladders with multiple health-related behaviors (ie, smoking status, at-risk drinking, fruit and vegetable intake, physical activity, and BMI) in a sample of African American adults enrolled in a church-based longitudinal cohort study. This research builds upon the Ghaed and Gallo study, which examined the relative influence of the SSS ladders on these outcomes while controlling for objective SES indicators, 5 and extends it to a large African American sample of men and women. We hypothesized that the SSS-US and the SSScommunity ladders would be associated with health-related behaviors, over and above the influence of sociodemographics, but were not more specific in our predictions given the exploratory nature of this study within an African American sample. --- METHODS --- Participants and Procedures Data were collected as part of a longitudinal cohort study designed to investigate associations of behavioral, social, and environmental factors with health behaviors among African American adults. Participants were recruited from a large mega-church in Houston, Texas. Recruitment strategies included printed and televised media within the church and inperson solicitation during church services and at a church health fair. Individuals were eligible to participate if they were <unk>18 years of age, reported residence in the Houston area, had a functional telephone number, and attended church (though they were not required to be a member of the church). Participants were 1467 African Americans who were enrolled December 2008 through July 2009. Surveys were completed in person at the church. Participants were compensated with a $30 Visa debit card following survey completion. Study procedures were approved by the Institutional Review Board at The University of Texas MD Anderson Cancer Center, and informed consent was obtained from all participants. The current study reflects data from the first year of the cohort data collection (ie, baseline). --- Measures Participants viewed questionnaire items on a computer screen and entered responses into the computer using the keyboard. Sociodemographics-Sociodemographics included age, gender, partner status, total annual household income, educational level, employment status, and insurance status. Sociodemographics were treated as covariates in the analyses due to known associations with health-related behaviors. Subjective social status-SSS was measured with 2 versions of the MacArthur Scale of Subjective Social Status: the US ladder and the community ladder. 12 The SSS-US ladder presents a 10-rung ladder to represent where people stand in the United States, with higher rungs indicating higher status (ie, more money, more education, and better jobs). 12 Participants select the rung that best represents where they think they stand relative to others in the United States, resulting in a ranked indicator variable with 10 possible levels. The SSS-community ladder presents a 10-rung ladder to represent where people stand in their communities, with higher rungs indicating higher status. Participants select the rung that best represents where they think they stand relative to others in their (self-defined) community. In this study, the SSS-community ladder was administered prior to the SSS-US ladder per recommendations in the literature. 12 The SSS ladders have been used in several studies with racially/ethnically diverse participants and have demonstrated adequate reliability and validity [eg, 2,3,10,12 ]. The correlation between the SSS-US ladder and the SSS-community ladder in this sample was 0.52. Health-related behaviors-Data were collected on smoking status, at-risk drinking, fruit and vegetable intake, physical activity, and BMI. Smoking status was assessed with a single survey item resulting in classification as a current smoker (smoked <unk>100 cigarettes in lifetime and currently smoke), former smoker (smoked <unk>100 cigarettes in lifetime but quit), or never smoker (smoked <unk>100 cigarettes in lifetime). At-risk drinking was assessed with the Alcohol Quantity and Frequency Questionnaire, a self-report measure of the average alcohol consumption on each day of the week over the last 30 days. 16 Males were classified as at-risk drinkers if they consumed an average of >14 drinks per week, and females were classified as at-risk drinkers if they consumed an average of >7 drinks per week. Alcohol quantity and frequency measures have been used extensively in research, 16 including among African American samples (eg, 17 ). Fruit and vegetable intake was assessed with the NCI Five-A-Day fruit and vegetable questionnaire. 18 This questionnaire yielded a continuous variable of daily fruit and vegetable servings that was highly skewed. Because of this, we chose to focus on a binary outcome whereby participants were classified as meeting recommendations for daily intake (<unk>5 servings of fruits and vegetables a day) or not meeting recommendations for daily intake (<unk>5 servings of fruits and vegetables a day) in our main analyses. This measure has demonstrated adequate convergent validity with more comprehensive dietary intake measures 19,20 and has been used previously among African American church-based samples (eg, 21,22 ). Physical activity was assessed with the International Physical Activity Questionnaire -Short Format (IPAQ), which is a self-report questionnaire used to measure the amount of time spent in moderate activity, vigorous activity, and walking during the past 7 days. 23 Weekly minutes spent engaging in each type of activity were multiplied by the corresponding metabolic equivalent (MET) value, which is a metric used to quantify energy expenditure (ie, the ratio of energy expended during an activity to the energy expended during rest). 24 Then, MET minutes were summed to arrive at the total weekly MET minutes spent in physical activity. Again, the resulting data were highly skewed. Thus, we chose to classify participants as engaging in low, moderate, or high rates of physical activity during the previous week based on total weekly MET minutes, the number of days per week engaged in PA, and the amount of time spent in each type of PA for our main analyses (see guidelines for data processing and analysis of the IPAQ, 2005). The short version of the IPAQ has good test-retest reliability and acceptable criterion validity against the Computer Science Applications, Inc accelerometer. 23 Finally, BMI (kg/ m 2 ) was calculated using staff-administered height and weight measurements. Participants with a BMI 16 to <unk>18.5 were considered underweight, those with a BMI 18.5 to <unk>25 were considered normal, those with a BMI of 25 to 29.9 were considered overweight, and those with a BMI <unk>30 were considered obese. Because the number of underweight participants (BMI=16-18.5) in the sample was very small (N=10, <unk>1% of total sample), these individuals were combined with the normal BMI group for analysis. There were no severely underweight individuals in the sample. --- Data Analysis Participant characteristics were examined using descriptive statistics, and relations of the SSS ladders with objective SES variables were examined using Pearson and Spearman correlations. For the main analyses, a series of logistic regression models were used to examine the associations between the SSS scales and health behaviors. Models 1-4 tested the relations of the SSS-US ladder with smoking status, at-risk drinking, fruit and vegetable intake, and physical activity (respectively) controlling for sociodemographics (ie, age, gender, partner status, total annual household income, educational level, employment status, and insurance status). Model 5 tested the relations of the SSS-US ladder with BMI while controlling for sociodemographics, fruit and vegetable intake, and physical activity. Models 6-9 tested the relations of the SSS-community ladder with smoking status, at-risk drinking, fruit and vegetable intake, and physical activity (respectively) controlling for sociodemographics. Finally, Model 10 tested the relations of the SSS-community ladder with BMI while controlling for sociodemographics, fruit and vegetable intake, and physical activity. If the SSS-US and the SSS-community ladders emerged as independently predictive of the same health-related behavior, follow-up analyses were planned to examine their relative significance within a single adjusted model. Analyses were performed using Statistical Analysis Software version 9.2 (SAS Institute, Cary, NC). Significant models were checked for adherence to underlying assumptions, and no violations were found. --- RESULTS --- Participant Characteristics Participants (N=1467; 75% female) were 45 years old on average (<unk>12.9), and slightly less than half reported being married or living with a significant other. Three quarters of participants reported an annual household income of <unk>$40,000 year, with the same proportion reporting employment. Almost half of the participants had a bachelor's or advanced degree. See Table 1 for all participant characteristics. Compared with populationbased estimates of African American adults, our sample had a lower percentage of current smokers 25 and individuals meeting fruit and vegetable intake recommendations. 26 In contrast, our sample had a comparatively higher prevalence of obesity, as well as reported rates of physical activity [cf. 26 ]. --- Relations of SSS Ladders With SES The SSS-US ladder was significantly associated with total annual household income (r =. 23, p <unk>.0001) and educational level (r =.21, p <unk>.0001), but not employment status (r =.03, p =.28). As expected, associations of the SSS-community ladder and these variables were more attenuated than with the SSS-US ladder, though still significant in the case of income (r =.11, p <unk>.0001) and education (r =.11, p <unk>.0001), but not employment status (r = -.05, p =.09). --- Main Analyses The SSS-US ladder was significantly associated with fruit and vegetable intake [<unk> = 0.13, SE=.05; <unk> 2 = 7.34, p =.007; OR=1.13 (95% CI=1.04-1.24)] and physical activity [<unk> =.09, SE =.03; <unk> 2 = 8.22; p =.004; OR=1.09 (95% CI=1.03-1.16)]. Specifically, participants with higher SSS-US endorsements were more likely to meet recommended fruit and vegetable intake guidelines and were more likely to engage in high versus low or moderate rates of physical activity. The SSS-US ladder was not significantly associated with smoking status (p =.20), at-risk drinking (p =.52), or BMI (p =.84). The SSS-community ladder was not significantly associated with smoking status (p =.21), at-risk drinking (p =.42), fruit and vegetable intake (p =.09), physical activity (p =.11), or BMI (p =.17). See Table 2 for detailed results of these models. --- Exploratory Analyses A series of exploratory analyses were conducted following the main analyses. First, we were interested in how results might differ with an alternative conceptualization of fruit and vegetable consumption. Therefore, we conducted post hoc analyses to determine associations of the ladders with the number of servings of fruits and vegetables per day (a continuous variable). A log transformation was applied to the dependent variable due to the skewness of the distribution. Such transformations reduce the spread of values in the upper range of data points and are helpful when the skewness of the data distribution represents a threat to the underlying assumption of normality necessary for planned analyses. 27 A failure to transform highly skewed data can distort associations and lead to erroneous conclusions. 28 In addition to the log transformation, regression diagnostics were performed to identify unusual and influential data points. Observations with large residual and high leverage were carefully checked and excluded from analysis (<unk> 3% of observations). Results of these adjusted analyses indicated that both the SSS-US and the SSS-community ladders were significantly associated with the number of daily fruit and vegetable servings in this sample [SSS-US = <unk> =.02, SE =.01, p <unk>.001; SSS-community = <unk> =.01, SE =.01, p =. 029]. However, when both SSS ladders were included simultaneously in an adjusted model, only the SSS-US ladder emerged as independently associated with the number of daily fruit and vegetable servings [<unk> =.02, SE=.01, p =.007]. Next, we examined the extent to which the ladders were associated with physical activity as measured by the total minutes engaged in all levels of physical activity (a continuous variable). In this case, a square-root transformation was applied to the dependent variable due to the skewness of the distribution. Again, outlying observations were checked and removed from the database as indicated previously (in this case, <unk>1% of observations). Results of these adjusted analyses indicated that only the SSS-US ladder was significantly associated with the total minutes of physical activity [<unk> =.74, SE=.22, p <unk>.001]. --- DISCUSSION This study was the first to examine the associations of the SSS-US and the SSS-community ladders with multiple health behaviors among an African American sample of adults. Results of our main analyses indicated that only the SSS-US ladder was uniquely associated with fruit and vegetable intake and physical activity over and above the influence of sociodemographics. Specifically, every 1 unit (rung) increase in the SSS-US ladder was associated with a 13% increase in the odds of meeting recommended fruit and vegetable intake guidelines and a 9% increase in the odds of being in the high (versus low or moderate) physical activity group. These findings contribute to a growing research literature linking perceived social status with health behaviors [eg, [5][6][7] ], even after accounting for the effects of income, education, and other SES variables, and extend those findings to a large African American sample of adults. Assuming that these data represent a long-term pattern of behavior, results suggest that African American individuals endorsing low social status on the US ladder may be at increased risk of incurring cancer, cardiovascular, and other diseases that are influenced by less than desirable fruit and vegetable intake and lower levels of physical activity. An additional aim of this study was to compare the associations of the SSS-US ladder versus the SSS-community ladder with health-related behaviors. The SSS-community ladder was unrelated to any of the health-related behaviors examined in our main analyses. Moreover, even though the SSS-community ladder was associated with fruit and vegetable consumption in our exploratory analyses, which treated servings as a continuous variable, it was not incrementally significant in a model that also included the SSS-US ladder. Thus, results suggest the relative value of the SSS-US ladder versus the SSS-community ladder among African Americans in the prediction of at least some health behaviors. These findings were consistent with those of a similar study conducted among (largely) white women, which found significant associations between the SSS-US ladder rankings and fruit and vegetable consumption, but no significant associations between the SSS-community ladder and fruit and vegetable consumption. 5 Our results are also consistent with recent studies finding stronger associations between health outcomes and the SSS-society ladder rankings versus other SSS scales using more proximal referents, including neighbors. 29,30 It may be that perceived social status relative to others in a community setting may be less associated with health outcomes and health-related behaviors and instead more associated with psychosocial variables, as suggested by a previous study. 5 This possibility might be explored further among an African American sample in future research. It was important to rule out that the significant associations found in this study were not influenced by residence in low SES neighborhoods, which are known to be associated with reduced access to fresh fruits and vegetables [eg, 31 ] and greater barriers to physical activity [eg, fewer physical activity resources, lower neighborhood safety 32,33 ]. In order to examine this possibility, we conducted post hoc multilevel analyses that further adjusted our significant models for area-level SES (median household income at the US Census tract level from 2000). Associations between the SSS-US ladder rankings and dietary and physical activity behaviors remained significant in these analyses (p values =.006 and.014), and area-level SES did not account for a significant amount of additional variance in the outcome (p values =.395 and.352). Thus, results suggest that the SSS-US ladder offers value added beyond objective SES measures, at both the individual and area levels, in its association with these dietary and physical activity behaviors. In addition, we also wanted to rule out any influence of depression on our results, as depressed mood may negatively affect dietary and physical activity behaviors. However, post hoc analyses that additionally controlled for participants' scores on the Center for Epidemiologic Studies Depression Scale 34 did not alter the pattern of significant results (p values <unk>.001). Results suggest that social standing as measured by the SSS-US ladder may have an important influence on some health-related behaviors, and perhaps ultimately health outcomes among African Americans, which rivals or surpasses that of objective SES indicators. Previous research has suggested that SSS is an incremental predictor of healthrelated outcomes over and above objective measures of SES because it captures unique nuances of social class that play an important role in how individuals act and feel (eg, societal inequities related to race/ethnicity). Our results suggest that this may be the case for fruit and vegetable consumption and physical activity among African Americans. Thus, social standing might be an important component to include within conceptual models focused on the socioeconomic predictors of (at least some) health behaviors [eg, 35 ]. Results also suggest that the SSS-US ladder might be a useful single-item screener to identify at-risk African American individuals who might benefit from targeted interventions to increase fruit and vegetable consumption and physical activity. Future research should incorporate longitudinal designs to examine the effects of the SSS-US ladder on these health behaviors over time and explore the mechanisms responsible for these associations in order to inform such interventions. In this study, neither the SSS-US nor the SSS-community ladders were significantly associated with smoking status, at-risk drinking, or BMI in adjusted analyses. These results are similar to the Ghaed and Gallo study, 5 but extend findings from a largely white sample to an African American sample. Thus, it may be that perceived social standing is unrelated to these health behaviors, whether social status is measured in relation to one's community or the nation as a whole; or it may be that these behaviors are adequately assessed with traditional, objective indicators of SES and that the inclusion of SSS offers no added benefit. However, it is likely that the low base rates of these behaviors contributed to nonsignificant results in this study (eg, only 5% were at-risk drinkers, and 9% were current smokers). In addition, despite the confidentiality of the survey process, some behaviors (eg, at-risk alcohol use, smoking) may have been underreported in this church-based sample due to associated stigma among a religious population. Associations between the SSS ladders and smoking, drinking, and BMI should be explored among a more diverse population of African Americans. Limitations of this work include the cross-sectional design, which precludes assumptions of causality in the relations between SSS and health behaviors. In addition, we focused on a convenience sample of church-based African American adults from a large metropolitan city in the South. The sample was largely female and generally well educated. Thus, these results may not generalize to other populations inasmuch as those populations might differ from the one examined in this study. Finally, although we attempted to control for a number of potential confounders, it is possible that unknown and unmeasured confounders might have influenced these results. Future research should seek to include large, racially diverse samples in order to examine whether relations of SSS with health behaviors varies by race/ ethnicity. Barring adequate diversity, however, future research in this area may be best conducted within racial/ethnic groups so as to mitigate the influence of potential confounders. In summary, this study was the first to examine associations of the US and community SSS ladders with smoking status, at-risk drinking, fruit and vegetable intake, physical activity, and body mass index among an African American sample. Results support the incremental associations of the SSS-US ladder, but not the SSS-community ladder, with fruit and vegetable consumption and physical activity beyond objective SES indicators. Results suggest that the SSS-US ladder may better capture the role of social disadvantage in these health behaviors than do measures of objective SES status alone, at least among this sample of predominately female, well-educated African American church attendees. Health through MD Anderson's Cancer Center Support Grant (CA016672). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the project supporters. All logistic regression models controlled for age, gender, partner status, total annual household income, educational level, employment status, and insurance status. Models 5 and 10 additionally controlled for fruit and vegetable intake and physical activity. At-risk drinking was assessed with the Alcohol Quantity and Frequency Questionnaire. Fruit and vegetable intake was assessed with the NCI Five-A-Day Fruit and Vegetable Questionnaire. Physical activity level was assessed with the International Physical Activity Questionnaire -Short Format. Reference groups for the dependent variables are as follows: current smoker (smoking status), not at-risk drinker (at-risk drinking), not meeting daily fruit and vegetable intake recommendations (fruit and vegetable intake), low and moderate physical activity (physical activity), and obese (body mass index).

Objectives-To examine associations of the US and community subjective social status (SSS) ladders with smoking status, at-risk drinking, fruit and vegetable intake, physical activity, and body mass index among 1467 church-going African American adults from a larger cohort study.Results-The SSS-US ladder was significantly associated with fruit and vegetable consumption (p = .007) and physical activity (p = .005). The SSS-community ladder was not significantly associated with any health behaviors. Conclusions-Among this sample of African Americans, the SSS-US ladder is more predictive of some health behaviors than is the SSS-community ladder.African Americans; social status; cancer risk behavior; physical activity; fruit and vegetable intake Subjective social status (SSS), or an individual's perception of his or her relative position in the social hierarchy, has been associated with health status [eg,1,2-4] and health behaviors [eg,5-7] independent of objective socioeconomic indicators such as income, education, and employment status. It has been argued that SSS functions as a unique predictor of health-related outcomes over and above objective measures of socioeconomic status (SES) for several reasons.

Introduction Youth substance use is common in Latin American countries, and Argentina is no exception (SEDRONAR, 2012;WHO, 2010WHO,, 2011)). Youth substance use results from a complex combination of socio-cultural, biological, interpersonal and socio-environmental factors (Hawkins, Catalano, & Miller, 1992;Kokkevi et al., 2007;Newcomb & Bentler, 1989). Several studies in Europe and the United States have highlighted the importance of family and parenting influences on youth substance use (Baumrind, 1991;Calafat, Garcia, Juan, Beco<unk>a, & Fernández-Hermida, 2014a;Jackson, Henriksen, Dickinson, & Levine, 1997;Piko & Balázs, 2012;Thomas, McLellan, & Perera, 2013). Parenting practices involve the grouping of attitudes and behaviors of parents towards their children. Studies examining the relationship between parenting behavior and youth outcomes have traditionally distinguished between two parenting dimensions: demandingness and responsiveness (Calafat et al, 2014a;Jackson, Henriksen, & Foshee, 1998;Piko & Balázs, 2012). Demandingness refers to parental control of children's behavior. Indicators of parental demandingness include setting and enforcing clear standards of behavior, actively monitoring and supervising a child's activities, and maintaining structure and regimen in a child's daily life (Baumrind, 1991;Beco<unk>a et al., 2013;Calafat et al., 2014a;Jackson et al., 1998). Responsiveness refers to emotional warmth of the parent, involvement in their children's lives, and acceptance and acknowledgment of the child's point of view. Indicators of parental responsiveness include communicating affection, providing comfort, and being involved in social and academic development and recognizing achievement (Baumrind, 1991;Beco<unk>a et al., 2013;Calafat et al., 2014a;Jackson et al., 1998). Low parental responsiveness has been associated with lower achievement and higher delinquency and substance use in adolescence (Beco<unk>a et al., 2013;Calafat et al., 2014b). With respect to how parenting dimensions are grouped, and whether mothers and fathers were studied, one recent meta-analysis of studies assessing parenting behavior and delinquency found that only a minority of 160 studies assessed both responsiveness and demandingness, with only 20% of these assessing parenting of both mothers and fathers (Hoeve et al., 2009). --- Parenting Behavior and Adolescent Substance Use Outcomes Authoritative parents are those who succeed in being demanding and responsive at the same time. Several studies have found that authoritative parenting is particularly important as a protective factor for adolescent substance use among European and U.S. adolescents (Benchaya et al., 2011;Beco<unk>a et al., 2013;Choquet et al., 2008;Jackson, 2002), with authoritative parents having adolescents with the lowest levels of substance and alcohol use (Jackson, Bee-Gates, & Henriksen, 1994;Jackson, 2002;Huver, Engels, Vermulst, & de Vries, 2007;Patock-Peckham, King, Morgan-Lopez, Ulloa, & Moses, 2011). However, parenting behavior and its impact on substance use may vary according to the cultural context in which parents and children interact (Lorenzo-Blanco, Bares, & Delva, 2013;Calafat et al., 2014a;Calafat et al., 2014b). Indeed, compared to non-Hispanic families in the U.S., Hispanic families have been described as adhering to collectivistic values that influence parenting practices (Romero & Ruiz, 2007;Lorenzo-Blanco et al., in press;Santisteban et al., 2012). Specifically, in research studies, collectivistic values related with more parental demandingness (Romero & Ruiz, 2007) and responsiveness (Lorenzo-Blanco et al., in press;Santisteban et al., 2012). Additionally, in studies with Hispanic adolescents, parental responsiveness appeared to be more or equally protective against substance use than authoritative parenting behavior, highlighting the need to investigate the influence of parenting on adolescent substance use across cultural contexts (Lila, Garcia, & Gracia, 2007;Villalobos, Cruz & Sánchez, 2004). In surveying studies of Hispanic youth, we identified three studies that addressed parenting and substance use in Latin American youth (Benchaya et al., 2011;Lila, Garcia, & Gracia, 2007;Lorenzo-Blanco et al., 2012). One cross-sectional study of 860 adolescents in Chile examined demandingness for mothers and fathers combined, and found a multivariate negative (protective) association with tobacco use. This study did not report other substance use outcomes (Lorenzo-Blanco et al., 2012). Another study of 230 adolescents in Colombia studied parental responsiveness for mothers and fathers separately (but not demandingness) and found a negative association between maternal but not paternal responsiveness on externalizing behaviors. This study did not measure substance use specifically (Lila, Garcia, & Gracia, 2007). Finally, a study of 231 Brazilian adolescents used an authoritative parenting measure (combining responsiveness and demandingness) for mothers and fathers separately, and found this to be negatively associated (for both fathers and mothers) with general substance and tobacco use, but not alcohol or illicit drug use (Benchaya et al., 2011). Thus, the literature on parenting behavior among Hispanic youth leaves us with an incomplete understanding of how both domains of parenting (i.e., responsiveness and demandingness) for mothers and fathers is associated with substance use outcomes among youth in South America. --- The Current Study The purpose of this cross-sectional study is to investigate Argentinian adolescents' perceptions of responsiveness and demandingness separately for mothers and fathers and their association with multiple substance use outcomes. Strengths of the study include a large sample size, which gives the power to detect significant associations among multiple parenting covariates, which could be expected to be correlated. The inconsistent findings of published studies on parenting dimensions did not permit us to propose a specific set of hypotheses across substance use outcomes. --- Methods --- Study sample and procedure A convenience sample of 33 schools from three large cities in Argentina (Buenos Aires, Córdoba, Tucumán) participated in the study (n=15, 8, 10, respectively), with public schools identified by the Ministry of Health and Ministry of Education (n=18) and private schools identified through personal contacts (n=15). Private schools were included in the sample because 26% of students attend these schools in Argentina (Bottinelli, 2013). Surveys were administered between May and July of 2014, with attempts to recruit every enrolled 8th grade student. Passive consent was requested from parents or caretakers, and students signed an active consent form, allowing for both participations in the current survey and follow-up contact for subsequent surveys. The questionnaire used anonymous link procedures to allow follow-up (Galanti et al., 2007). The research protocol was approved by an NIH-certified human subjects research board in Buenos Aires based at Centro de Educacion Medica e Investigaciones Cl<unk>nicas (CEMIC). --- Development of Questionnaire Measures The questionnaire included translation of items used in surveys for adolescents previously implemented in Argentina, Mexico, and in the US (Alderete et al., 2009; Ministerio de Salud y Ambiente de la Nación, 2013; Sargent et al., 2005;Thrasher, Jackson, Arillo-Santillán, & Sargent, 2008). Items in English were translated and reviewed by Argentinean Spanishspeaking research staff and pilot tested with students in Buenos Aires to ensure students' understanding of questions, instructions and confidentiality statements. The self-administered questionnaires were completed in the classroom under the supervision of trained research staff. Questionnaires included questions about demographics, family and school characteristics, school performance, alcohol and illicit drug consumption, smoking, peer and family smoking and drinking, sensation seeking, and parenting behavior. --- Measures Substance Use Outcomes-We assessed substance use using measures commonly employed to assess these behaviors in national surveys of adolescents (Eaton et al., 2012). A respondent was considered a current smoker if he or she responded positively to "During the past 30 days, on how many days did you smoke cigarettes?". We created similar outcomes for alcohol consumption: current drinking (within the past 30days). For illicit drug use, participants were asked about life-time use of marijuana, and, separately of life-time cocaine or crack use. Participants who used either type of drug in his/her life were considered lifetime users. --- Independent Variables Parenting Behavior-Parenting behavior was assessed with questions on responsiveness and demandingness from Jackson's Authoritative Parenting Index (Jackson et al., 1998), using three items for each parenting dimension. Students used a 5-point response scale to indicate how well certain statements described the parenting behavior of their mother and father, assessed separately. The items used to assess parenting responsiveness were: "She/he makes me feel better when I am upset," "She/he listens to what I have to say" and "She/he wants to hear about my problems." For parenting demandingness, the statements were: "She/he tells me what time I have to be home," "She/he asks me what I do with my friends," and "She/he knows where I am after school". Responses from each of the items were summed and divide by the number of items to create an index (range: 1-5 for the responsiveness and demandingness) for the mother and another for the father (alpha = 0.82, 0.70, 0.89, and 0.92 for mother responsiveness, mother demandingness, father responsiveness, and father demandingness, respectively). --- Covariates We included are range of covariates found in previous studies to be predictors of adolescent substance use, and which also could be associated with parenting behaviors. The covariates included: Demographics and school function (sex, age, grade retention "Have you in your life repeated a grade in school? [yes, no]), socioeconomic indicators (has a job "Do you have a regular job for which you get paid money" [yes, no], attends public or private school, parent education "Please check the highest education that your parents have obtained [responses ranged from incompletion of primary school, to completion of a university degree]), smoking amongst network members (mother, father, siblings [family smoking assessed though a table in which students were asked to check yes if mother father or siblings smoked] or friend smoking "How many of your five best friends smoke? [0/5 to 5/5]), peer drinking [similar to peer smoking], and sensation seeking (Castrucci & Gerlach, 2006;Kilpatrick, Sutker, & Smith, 1976;Laible & Carlo, 2004;Sargent, Tanski, Stoolmiller, & Hanewinkel, 2010). Sensation seeking was measured using four items: "I like to do scary things"; "I like to explore strange places"; "I like new and exciting experiences, even if I have to break the rules"; "Sometimes I do "crazy" just for fun" (Stephenson, Hoyle, Palmgreen, & Slater, 2003), with responses on a 5-point Likert scale. Responses were summed to create an index (range: 1-5) (alpha = 0.79). --- Statistical Analysis We conducted descriptive analyses on all study variables, with all substance use outcomes treated as dichotomous variables. Bivariate relationships between continuous variables were examined using Pearson correlation coefficients. Multilevel logistic regression models (with random intercepts for school) were used to assess the associations of parenting behavior and each of the four substance use outcomes, assessed separately, providing unadjusted and adjusted estimates and confidence intervals for the odds ratios. All adjusted models included age, sex, parent education, work status and sensation seeking. The adjusted models for current smokers also adjusted for smoking by mother, father, siblings and peers. Alcohol use models were adjusted for peer alcohol use. We also ran fully adjusted models, after which we introduced an additional interaction term for sex by each of the four parenting variables, each entered separately. Overall Wald chi-square test was used to test if all the fixed effects parameters (excluding the intercept) were simultaneously zero. Missing data was less than 1% for all variables except for parental education, which was 11.9%. To control for respondent bias we imputed parental education using multinomial logistic regression (Van Buuren, 2007). Since the pattern of findings was similar in direction, strength and statistical significance for primary study variables, we present the results from the analyses that used list-wise deletion of missing cases. All data analyses were conducted by one author (AP) using Stata V.13.0 (Stata Corp, College Station, TX, USA). --- Results A total of 3826 first-year students from 33 schools in three large Argentinean cities were invited to participate. Of these, 436 (11%) were absent when the survey was implemented, 45 (1%) had parents who refused their participation, and 173 of students (4%) refused to participate. Therefore, 3172 (83%) students completed the survey. The mean age of participants was 12.8 years (SD=0.95), 42% were female, 46% of their parents had more than eight years of education, 32% attended private school, and 24% had repeated a grade in the past. Means (SD) for the parental responsiveness index was 4.22 (0.88) for mothers and 4.08 (0.91) for fathers; the means for demandingness were 3.91 (1.08) for mother and 3.66 (1.18) for fathers (Table 1). Table 2 shows bivariate correlations between parenting behavior and the covariates. As shown, demandingness and responsiveness were associated for both mothers (r=0.49, p<unk>0.001) and fathers (r=0.62, p<unk>0.001). A negative correlation existed between sensation seeking and mother demandingness (r=-0.18, p<unk>0.001) and father demandingness (r=-0.14, p<unk>0.001). Finally, a negative correlation was observed between age and mother demandingness (r=-0.12, p<unk>0.001) and father demandingness (r=-0.11, p<unk>0.001). Multilevel logistic regression models were estimated, regressing each of the youth substance use outcomes on parenting style and other study variables (Table 3). After adjusting for key covariates, mother demandingness was significantly associated with a decreased risk of: current smoking (AOR=0.77; 95% CI 0.64 -0.92); current drinking (AOR=0.81; 95% CI 0.71 -0.92); binge drinking (AOR=0.77; 95% CI 0.66 -0.99); and ever drug use (AOR=0.71; 95% CI 0.61-0.83). Father demandingness was significantly associated only with lower likelihood of binge drinking (AOR=0.84; 95% CI 0.74-0.97). There were no statistically significant interactions between demandingness and responsiveness indices for mother and father separately for any of the outcomes. The overall Wald chi-square test was significant for all estimated models (p <unk> 0.0001). --- Discussion In this study of Argentinian adolescents, maternal demandingness showed an independent association with use of multiple substances, after controlling for a number of other established risk factors. Regarding the gender of the child, maternal demandingness was an important and consistent protective risk factor against use of substances for both male and female adolescents. These associations were large enough to account for a 3-8 percent reduction in the prevalence of use across the range of maternal demandingness. On the face of it, these findings add further conflict to the parenting literature in Latin American adolescents (Villalobos, Cruz & Sánchez, 2004;Lila, Garcia, & Gracia, 2007;Benchaya et al., 2011). Our results contrast with studies of authoritative parenting from Spain (Garcia & Gracia, 2009) in which responsiveness (authoritative and indulgent) seemed to be the most important domain in terms of promoting better youth outcomes there. In a subsequent editorial, Garcia & Gracia (Garc<unk>a & Gracia, 2014) speculated that "in the South European and Latin American cultures, considered as horizontal collectivist, even if the children are very connected with their families, the relationship among different generations is expected to be more egalitarian than in vertical collectivist cultures (such as the Asian or Arabic) or individualistic (North American)". In contrast, our findings suggest that, at least for preventing onset of substance use, demandingness is an important element of parenting style, even in horizontal collectivist cultures. Among Mexican children, Villalobos (Villalobos, Cruz & Sánchez, 2004) reported a stronger association between responsiveness and multiple outcomes. On the face of it, this would be consistent with the findings of Garcia & Gracia. However, a closer look at their findings shows higher (better) scores for more responsive (authoritative and indulgent) parents on academic outcomes and lower (better) scores for more demanding (authoritative and authoritarian) parents on substance use outcomes; this finding is consistent with the results of the present study. Another study of racial/ethnic differences among US adolescents found that Hispanic children of authoritarian parents were more engaged in academic achievement, but that study did not report on substance use (Steinberg et al., 1992). Finally, a Brazilian study (Benchaya et al., 2011) used a parenting measure that combined responsiveness and demandingness so was unable to discriminate between the two. Overall, it appears that there is more support for rule setting as a parenting strategy to prevent substance use among Latino families in the Western Hemisphere, and this applies to multiple substance use outcomes in Argentina. Additionally, the evidence to date suggests that different parenting strategies may be indicated for different outcomes in Latin American adolescents. Some of the differences among studies could result the way we chose to analyze our data, using continuous values for responsiveness, independent from demandingness. We chose this approach because the data supported a dose-response relation across the continuum of each parenting dimension and we were uncomfortable with the four-typology model (Maccoby & Martin, 1983), which eliminates about half of the data by discarding adolescents who score in the middle tercile. Moreover, we felt that the authoritative construct would be verified if both responsiveness and demandingness showed an independent relation with behavioral outcomes (indeed, higher scores in both domains predicted less substance use, but the relation for responsiveness was not statistically significant). However, it is worth also noting that the correlation between responsiveness and demandingness within parent is moderately strong (0.49 for mothers and 0.62 for fathers), such that more demanding parents also tend to be more responsive on average. Once maternal parenting was accounted for, paternal demandingness was associated only with the prevention of binge drinking in this sample. These findings are consistent with other studies in which maternal demandingness had more influence on substance use than paternal demandingness (Baker et al., 1999;Garcia & Garcia, 2009;King & Chassin, 2004). The present study extends this to adolescents in Argentina. The Argentinian cities we surveyed have populations that are derived from European immigrants and emphasize many elements of this culture. It would be very interesting to see how parenting by mothers and fathers relates to substance use in Northwestern Argentina, or in countries like Bolivia, where the predominant cultural influence is more oriented toward Native Americans. Unexpectedly, the current study did not find that the effects of parenting were any different for boys compared to girls. This finding is interesting in light of the evidence that adolescent relationships with fathers and mothers may vary depending on the gender of the adolescent (Choquet et al., 2008;Patock-Peckham et al., 2011;Scalese et al., 2014). This finding also bodes well for the development of interventions, because it suggests that a focus on the mother would be effective regardless of the gender of the child. An interesting area of research for interventions could be research to determine whether training aimed at fathers could allow them to be influential over-and-above the effects of maternal parenting, or how important consistency in behavior across parent matters with respect to adolescent substance use outcomes. Although in Latin America, the family centered in the patriarchal authority is less common than previously (Burin & Meler, 1998), women still play a central role in raising early adolescents and have a better insight of and control over daily activities as compared to fathers (Ackard et al., 2006). In addition, maternal parenting dimensions may be more likely in predicting adolescent outcomes because mothers typically spend more time with their adolescents (Larson, Richards, & Perry-Jenkins, 1994). This study has several limitations which should be acknowledged. As mentioned above, the sample may not be representative of the entire Argentinean youth population, especially compared to cities in the Northwest part to the country, where the Native American culture predominates. However, the schools were selected from three main cities that represent 37% of the population and we included schools with students from different economic backgrounds. The prevalence of tobacco, alcohol and drug use in the sample is similar to the prevalence reported from national surveys carried out by the government (SEDRONAR, 2012), suggesting that the results may be broadly generalizable to urban Argentinean populations. The present study did not assess parental and sibling drinking which can influence adolescent alcohol use, and future research on the role of parenting behavior on alcohol use among youth in Argentina should assess parental and sibling drinking and drug use. In addition, although parenting behavior was not the primary focus of the survey, the questions we used have been used in many studies on youth substance use (Dalton, Ahrens, Sargent, Mott, & Beach, 2002;Sargent et al., 2004). Besides, the questions regarding parenting behavior have not been validated in the Argentinean population. However, the questions were determined to be face valid by experts in tobacco research in Argentina, had acceptable reliability (see Methods section), and have been used in a Chilean study (Lorenzo-Blanco et al., 2012). Nevertheless, the skewed nature of the responses suggests that further measurement development may be necessary to better capture variability in parenting behavior in the Argentine context. Such efforts may benefit from including a gender perspective that more completely captures the complex relationship between parenting behavior and substance use in young adolescents. Finally, this is a cross-sectional study, therefore, we cannot provide information on the temporal sequence of events, and longitudinal research may be necessary to confirm our results. In summary, the current study contributes to the Latin American literature because it is the first to assess the impact of parenting behavior on the substance use behaviors of early adolescents in Argentina. This study suggests that there is a protective association between maternal demandingness and adolescent tobacco, alcohol and illicit drug use. The findings from this study could inform the development of public policies aimed at preventing substance abuse among Argentinean adolescents. Interventions may target the mother-teen relationship as a foundation for change and research may benefit from investigating whether involving consistent support from fathers could increase the beneficial impact of the motherteen relationship. Relationship between mother demandingness index and predicted probabilities of current smoking, current drinking and drugs use, adjusted by age, sex, parent education, work status and sensation seeking. Current smoker model adjusts also for mother, father, siblings & peer smoking. Drinker models adjust for peer alcohol use. --- Unadjusted and adjusted associations between parenting behavior and youth substance use outcomes.

Background-In Europe and the United States, family relationships and parenting behavior can influence youth substance use, but less is known about their influence in Latin American countries. Objective-To explore whether parenting behavior is associated with substance use among Latin American youth. Methods-A cross-sectional, school-based survey of middle-school youth (n=3,172) in three Argentinian cities queried tobacco, alcohol, and drug use using items adapted from global youth surveys. Parenting behavior was assessed with previously validated items that tapped into demandingness and responsiveness, separately for mothers and fathers. Multilevel logistic regression models assessed associations between parenting behavior and substance use after adjusting for student characteristics, socioeconomic indicators, sensation seeking, and smoking amongst peers and family members. Results-Substance use prevalence was 10% for current smoking, 32% for current drinking alcohol, 17% for past 30-day binge drinking (≥5 drinks), and 8% for previous year illicit drug use (marijuana or cocaine). Greater maternal demandingness was independently associated with lower CONTACT Raúl Mejia

A C C E P T E D M A N U S C R I P T Page 5 --- Highlights <unk> First smoking cessation trial in a community based social service setting <unk> Participants were socioeconomically disadvantaged smokers with comorbidity <unk> Case-worker delivered intervention were not effective at aiding abstinence <unk> Important gains were made in reductions in cigarettes smoked and quit attempts. --- Introduction In high-income countries, tobacco smoking rates are highest amongst people with mental illness and substance use disorders, the long term unemployed and homeless populations, and Indigenous peoples. 1 Rates of tobacco-related diseases such as cardiovascular disease, cancer and chronic respiratory diseases are subsequently much higher in these groups. 2 Smokers from these disadvantaged, low socioeconomic groups find it harder to quit than more socioeconomically advantaged smokers. 1,3 Existing evidence for the effectiveness of smoking cessation interventions for disadvantaged groups is inconsistent and inconclusive. Two systematic reviews of smoking cessation interventions for six disadvantaged groups known to have high smoking rates in high-income countries suggest that multicomponent interventions incorporating behavioural counselling either face-to-face or via telephone, motivational interviewing, and NRT hold the greatest promise of successfully achieving abstinence amongst some disadvantaged groups but not all. 4,5 Delivering comprehensive smoking cessation interventions to smokers who experience disadvantage is challenging as these smokers are often hard-to-reach and as a result sample sizes are small. [4][5][6] In high-income countries including the UK, US and Australia, community social service organisations (CSSO) provide support to the most socially disadvantaged groups 7 with high smoking rates. 8 Small pilot smoking cessation trials suggest that the CSSO setting might be acceptable and feasible, 8,9 however, the effectiveness of this approach has not been evaluated in an adequately powered trial. --- Objectives The primary aim of this study was to examine the effectiveness of a CSSO case-worker delivered intervention (Call it Quits) for a diverse population of severely disadvantaged smokers on verified continuous abstinence at six month follow-up. --- ACCEPTED MANUSCRIPT --- A C C E P T E D M A N U S C R I P T --- Methods --- Study design and setting: Call it Quits was a parallel randomised trial of a case-worker delivered smoking cessation intervention. 10 The study was conducted in a large Community Care Centre, managed by a national non-government organisation located in New South Wales (NSW) Australia providing counselling, emergency housing and financial aid. Participants: Participants were adult clients of the Community Care Centre, who selfreported smoking daily or occasionally, with sufficient English language to give informed consent. Clients who presented to the centre in an inebriated or agitated state or were too distressed (distress related to factors contributing to accessing emergency relief) to participate were excluded. As clients arrived at the Centre, eligibility was assessed by a research assistant who obtained written consent. First, the research assistant asked participants to complete a general health survey on a touchscreen laptop computer. Second, participants who reported smoking tobacco daily or occasionally were asked by the research assistant to participate in a study where they may or may not receive a smoking cessation program requiring them to return to the Centre. Participant sociodemographic and smoking characteristics were collected during the computer-administered general health survey (Supplementary File 1). --- Randomisation and masking: A computer generated randomisation schedule which was embedded into the computer survey software allocated trial participants in a 1:1 ratio to intervention or control group. The randomisation schedule was developed by an independent computer programmer, incorporated into the Digivey survey software, 11 and tested prior to the trial commencing. At enrolment, the sequence was concealed from the research assistant who gained consent into the trial and conducted follow-up assessments. Participants were A C C E P T E D M A N U S C R I P T Page 8 made aware of their group following allocation with a paper print-out after they completed the computer survey. --- Interventions: All participants received on-screen advice to quit smoking, the state Quitline telephone number, and a "gift bag" with Call it Quits branded gifts. All participants were asked to return to the centre at 1 month and 6 month follow-up for data collection. No further intervention was offered to control group participants. The smoking cessation intervention which was drawn from existing evidence, the PRIME theory of motivation, 12 and the taxonomy of Behaviour Change Techniques (BCTs), 13 used brief advice and motivational interviewing techniques to encourage setting a quit date and maximise use of NRT, 14 and provide social support. 15 Free NRT was offered to all participants in the intervention group. Combination use of fast acting and sustained release NRT was encouraged based on evidence of increased effectiveness compared with single NRT type use. 14 The schedule of counselling sessions for intervention delivery included three face-to-face sessions and two telephone sessions. The counselling sessions were delivered by trained volunteer case-workers to mirror usual counselling practice at the Centre. and followed a written intervention manual (Supplementary File 2) which incorporated 46 BCTs. The emphasis was on setting a quit date, encouraging use of NRT, managing withdrawal symptoms and urges to smoke, enhancing self-efficacy, social support and prevention of relapse. Evidence-based strategies were employed to minimise attrition 6 including collection of comprehensive contact information for the participant and a significant other, flexible scheduling of follow-up assessments with reminder text messages and calls, and project branded gift bags. All participants received up to $120AUD grocery voucher for completion of the surveys. --- ACCEPTED MANUSCRIPT --- A C C E P T E D M A N U S C R I P T months follow-up, with abstinence defined according to the Russell Standard (modifiedregarding treatment of missing cases, see below). 16 Prior to un-blinding and data analysis, this was changed from the original protocol outcomes of 24-hour CO verified self-reported abstinence and 7-day point prevalence self-reported abstinence based on recommendations that six months continuous abstinence is the more relevant outcome for evaluating longerterm cessation and health impacts. 17 At the same time the 12 month follow-up was abandoned due to concerns regarding attrition and resourcing. To be classified as abstinent, participants had to report that they had smoked fewer than five cigarettes in each of the previous six months, from two weeks after the baseline (grace period) at the six-month follow-up visit and that they had not smoked any cigarettes in the week before the follow-up visit. As explained in the protocol paper, 10 although cotinine is the recommended gold standard measure for the verification of smoking status, it was impractical and invasive in this study and abstinence was verified by the concentration of exhaled CO of less than 10 ppm. 18 All participants were asked to return to the centre to provide a CO reading, regardless of whether they reported abstinence. Secondary outcomes were self-reported continuous abstinence at 1 month follow-up, and at both 1 month and 6 month follow-up self-reported and verified 7 day point prevalence abstinence, cigarettes smoked per day, and number of serious attempts to quit in the last month. To assess adherence to the intervention, participation in face-to-face and telephone sessions was recorded by counselling case-workers and participants were asked about use of NRT. Audio-recordings of 67 counselling sessions were coded for manual-specified BCT delivery. blinding (available upon request). Primary analysis was performed using all the available data from participants under the assumption the data is missing completely at random, followed by a range of sensitivity analyses to investigate the impact of departures from the missing data assumptions. 19 As recommended by the CONSORT statement, statistical analyses were conducted on all primary and secondary outcomes, and not on baseline data. The primary outcome measure -continuous abstinence from baseline (with a two week grace period)requires that participants be abstinent at both one month and six month follow-up. 16 Thus participants who are missing outcome data at six months, but are followed up at one month and are not abstainers at this time are by definition not continuous abstainers at six months and were classified as such in the analyses. All other participants with missing outcome data were excluded from the primary analysis. The primary analysis of CO verified continued abstinence from baseline involved a logistic regression model of all available observations (according to the definition above). SAS 9.4 and Stata 13 were used for all analyses, and statistical significance was defined a priori as p<unk>0.05. Due to the very small number of participants with the outcome, we did not adjust for the a priori covariates (age, gender, marital status, housing status, income, education, postcode, nicotine dependence, quit attempts, use of cessation aids, partner smoking behaviour depression and financial stress) specified for inclusion in the analyses. 10,20 Part way through the study recruitment we discovered a breach in protocol in that some participants were informed of their allocated intervention group prior to obtaining consent. This problem was rectified after 25 participants had been recruited, and blinding to allocation was maintained prior to seeking consent for the remainder of recruitment. However, due to the potential for this to introduce some participation bias, the regression model for the --- A C C E P T E D M A N U S C R I P T Page 11 primary analysis initially included a variable indicating whether individuals had been informed of their intervention status prior to consent. As this variable had no impact on the intervention effect, to preserve power it was excluded from the final model. For the secondary outcomes a logistic regression model for abstinence outcomes was used (validated continuous abstinence at 1 month, self-reported continuous abstinence, CO confirmed and self-reported 7-day point prevalence abstinence at one and six months), unadjusted for covariates. Linear regression models were used for the number of cigarettes smoked per day, and a negative binomial model for number of quit attempts adjusted for whether or not participants had been informed of their intervention status prior to consent, Heaviness of Smoking Index, depression and anxiety (PHQ4), and self-efficacy, and for number of cigarettes, quit method at baseline was included as a covariate due to possible imbalance between treatment groups at baseline. Robust standard errors were applied to the linear regression to account for slight deviation in the heteroscedasticity of residuals. For all outcomes, three types of sensitivity analyses were undertaken to include all participants, consistent with the intention-to-treat (ITT) principle: 1) multiple imputation (MI) 21,22 with chained equations to allow for appropriate estimates of variance; 2) analysis considering individuals with missing outcome as worst case outcomes (not continuous abstainers, no change in number of cigarettes smoked and no quit attempts), consistent with common methods of analysis of smoking cessation trials; 16 and 3) using pattern mixture models (PMM), 23 consistent with a Missing Not at Random mechanism. Based on previous pilot studies with similar populations, 24 we estimated that the control group quit rate would be 5%. On the basis that an 8% absolute difference in abstinence would be clinically important (i.e., 13% vs 5%), we calculated that a study size of 400 participants A C C E P T E D M A N U S C R I P T Page 12 (200 per group), allowing for a 30% loss to follow-up, would have 80% chance of detecting this difference with a 5% (two-sided) significance level. --- Results Figure 1 shows the recruitment and follow-up of participants from 7 The control group had a higher completed follow-up rate (return for both 1 and 6 month follow-up visits) than the intervention group. --- Figure 1. CONSORT Flow diagram Treatment groups were well balanced with respect to baseline characteristics, with the possible exceptions of fewer participants in the control group who were separated/divorced (Table 1), or last tried to quit smoking by gradually cutting down on cigarettes (Table 2). Figure 2 shows no statistically significant difference between intervention groups in the primary outcome of verified continuous abstinence at six months follow-up (1.4% and 1.0% for intervention and control groups; OR 0.77, 95% CI 0.07-8.53, p=0.828). There were also no statistically significant differences between groups in the secondary outcomes at six months of self-reported continuous abstinence, verified 7-day point prevalence abstinence, and self-reported 7-day point prevalence abstinence. Participants in follow-up. Lack of follow-up was predominantly monotone with 5.6% returning for six month follow-up not having attended one month follow-up. Age was associated with being missing only in the intervention group (younger in missing: mean 34 years, SD 10 than present: mean 41 years, SD 12; p <unk> 0.001) and income was associated with being missing only in the control group (lower income in missing than present, p=0.035). --- ACCEPTED MANUSCRIPT --- A C C E P T E D M A N U S C R I P T Sensitivity analyses (MI, worst case and PMM) showed similar results to the available case analysis for the majority of outcomes indicating the results were robust to the treatment of missing data. Estimates of the intervention effect for number of cigarettes smoked per day at 6 months varied between analysis approaches: under worst case the effect had diminished but was still significant (a difference of 2 cigarettes per day, p=0.02; and for MI we estimated a non-significant difference of 3 cigarettes per day (p=0.12). Pattern mixture modelling estimated a nullified treatment effect occurred if the missing participants were smoking at least 16 cigarettes a day more than those without missing data (an unlikely scenario). --- Intervention adherence: Of those randomised to the intervention group (n=187), 43 (23%) did not attend any counselling sessions. Of those who did attend, 22 (15%) attended one session, 30 (21%) attended two sessions, 22 (15%) attended three sessions, 25 (17%) attended four sessions and 45 (31%) attended all five sessions. Based on the audio-recording of 67 counselling sessions, the face-to-face sessions (sessions 1-3) averaged 18.09 minutes in length (range = 7.16 -46.44 minutes) and the average length of phone sessions (sessions 4-5) was 5.88 minutes (range = 1.53 -14.53 minutes). In total, 128 intervention group participants accepted an offer of NRT. Analysis of 67 counselling session audio-recordings show that, on average, fidelity to the treatment manual varied from 46% in the initial sessions, 31% in second face to face sessions and 39% in telephone follow ups. --- ACCEPTED MANUSCRIPT --- A C C E P T E D M A N U S C R I P T Discussion --- Principal findings A smoking cessation intervention incorporating behavioural counselling with the option of NRT delivered to highly disadvantaged smokers through a community social service by trained case-workers resulted in no higher abstinence rates than no intervention. Abstinence rates at six months follow-up were low for both groups. Participants in the intervention group reported more quit attempts and fewer cigarettes smoked at both one and six month followup. This trial is similar to previous trials with homogeneous groups of disadvantaged smokers (such as people with a mental illness, Indigenous Australians and prisoners) which have found low cessation rates and null outcomes for the behavioural and pharmacotherapy interventions. 4,5 The body of evidence emerging implies that these smokers find it difficult to quit, even when provided with current best practice smoking cessation aids. 4,5 The current intervention was evidence based with brief advice, behavioural counselling and BCTs, offer of combination NRT, social support and follow-up. One obvious explanation for the lack of effect may be low adherence to the intervention. The process measures collected suggest that only about a third of smokers attended all five counselling sessions and almost a quarter did not attend any at all. Not all participants in the intervention group took up the offer of free NRT (128 of 187), and of those who did take up the offer, many did not persist with the full recommended course. The outcomes suggest that more effort at increasing adherence to treatment is required. Contingency management, with even small financial and non-financial rewards for attendance to counselling sessions and adherence to treatment, is an approach with evidence of effectiveness with samples of people who use substances. 25 best practice for the general population of smokers, it may be insufficient to address the complex needs of smokers from highly disadvantaged groups experiencing comorbidities. There are a number of modifications that could be made to the intervention to strengthen it. Firstly, the provision of brief advice and motivational interviewing across five counselling sessions by minimally trained volunteers appears not to be effective for this group of smokers. Other research since this trial has shown that smokers receiving support in specialist stop smoking centres tend to have a higher short-term quit rate, compared with those receiving support in other settings from professionals for whom smoking cessation is only a part of their work. 26 The current sample reported high scores for anxiety, depression, healthrisk alcohol use and financial stress. Referring disadvantaged smokers to more qualified counsellors with experience in managing comorbidities is likely to strengthen the behavioural component of the intervention. Secondly, it is likely that the intervention would be substantially strengthened through the use of best-practice NRT, with or without other forms of pharmacotherapy for smoking cessation. For example, Cochrane reviews have shown that varenicline and bupropion result in higher cessation rates than NRT alone 27 and are safe for people with mental illness. 28 Furthermore, there are recent suggestions that for smokers who can't quit or who don't want to quit, the harm of tobacco can be substantially reduced by switching to alternative vaporised forms of nicotine delivery. 29 This is an area that deserves more attention for heavy smokers in disadvantaged groups who have tried quitting using other cessation treatments and have failed. Finally, lack of secure housing, employment, and high prevalence of Indigenous status were also characteristics of the sample. Smoking behaviours are part of the social, cultural and improving the material and social capital of groups in our society who are disadvantaged is likely to lead to improvement in health behaviours such as smoking. The CSSO setting that this trial used is an ideal vehicle for approaching this issue, however greater investment by government in providing other forms of support to these smokers is likely to improve their chances in quitting smoking and leading healthier lives. This trial recruited all smokers regardless of motivation level. The study found a significant increase in quit attempts due to the intervention. Furthermore, the intervention resulted in fewer cigarettes smoked over the six-month follow-up period, again implying that the intervention encouraged some action, which was insufficient in itself to achieve longer term cessation. Testing interventions that are applied for longer and with prolonged use of pharmacotherapy support is warranted. --- Strengths and weaknesses The most significant limitations relate to participant consent and attrition. While 49% of participants were randomly allocated to the intervention group (n=300) and 51% to the control group (n=318), a higher proportion consented in the control group (n=244; 77%) than the intervention group (n=187; 62%). The recruitment procedure involved assessing individuals' smoking status during the health survey was completed on a touchscreen laptop computer; all individuals who reported being current smokers were randomised by the software to either intervention group or control group and consent for participation in the study was then sought. The study protocol specified that consent was sought by the research assistant prior to checking the allocation of the individual. However, due to a breach in protocol, the first 25 participants were made aware of their allocation prior to consent-be required to return to the centre for three face-to-face counselling sessions if randomised to the intervention group may have been a disincentive for some. For the remainder of the recruitment, the research assistant and participants were blinded to allocation prior to obtaining consent. The impact of the lack of blinding of allocation for the initial participants had some impact on the group numbers, which was exacerbated by more participants in the intervention group withdrawing consent following allocation to group. The possible bias is likely to be small as baseline characteristics were similar between the two intervention groups. The attrition rates were high but not unusual for studies of this type 11 and were reasonably similar for the two intervention groups, 47% in the intervention group and 41% in the control group at 6 month follow-up. The imbalance in numbers highlights the difficulties in recruiting smokers in disadvantaged groups into smoking cessation trials. Evidence-based strategies were employed to boost retention, and attrition may have been greater had these strategies not been used. The consistency in the results from primary and sensitivity analysis to account for missing data points provide confidence that the impact of attrition is minimal. Also, the generalisability of the study is limited to similar CSSOs within high-income countries and their clients. The study has a number of strengths. This trial is one of the first to include a large and diverse sample of highly disadvantaged smokers, recruiting 431 participants, regardless of motivation to quit. Abstinence was verified using carbon monoxide readings. Self-reported abstinence rates were higher than confirmed abstinence, particularly in the intervention group highlighting the importance of objective verification of abstinence self-report. Our sensitivity analysis used multiple imputation as described in the methods in addition to traditional approaches to missing data in smoking cessation studies because evidence shows that the assumptions underpinning multiple imputation are more defensible than are those assumed when using other approaches to missing data. The results of the sensitivity analyses were consistent with the primary analyses indicating robustness of these analyses. --- Implications from this research New smoking cessation interventions for smokers from socially disadvantaged groups need to be developed and tested. The current study was conducted in the context of a high-income country with strong tobacco control measures and low general population smoking prevalence rate. Smokers who are highly socially disadvantaged appear to require more intensive smoking cessation interventions and possibly longer term, than smokers from the general population or more affluent groups. This has implications for resourcing and timeframes. Referral to specialist services or additional training for counsellors and care providers may be appropriate. Furthermore, new stop smoking medicines have become more readily available since this trial, including varenicline which has strong evidence of effectiveness and should be offered to smokers from disadvantaged groups. --- Conclusions In this study, evidence that a case-worker delivered smoking cessation intervention was effective at aiding abstinence was lacking. The secondary outcomes suggest that the intervention influenced processes towards abstinence such as increasing the number of attempts to quit and reducing the number of cigarettes smoked daily. For this population of highly disadvantaged smokers, with comorbidities, high proportion of Indigenous Australians and financial concerns, these are important outcomes. Because of the exceptionally high smoking rates amongst socially disadvantaged groups worldwide it is imperative that research continues to examine strategies for promoting smoking cessation. --- Acknowledgements --- ACCEPTED MANUSCRIPT --- Data statement De-identified data can be obtained from the corresponding author upon request. --- Contributors BB, CP, CD'E, RW, MS conceived and designed the study. LT implemented the study and collected the data. CD'E, CO, KP conducted data analysis. BB wrote the first draft of the article and affirms that the manuscript is an honest, accurate, and transparent account of the study reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned have been explained. All authors met the criteria for authorship, had full access to all of the data (including statistical reports and tables) in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. --- Ethics approval Human research ethics approval was gained from the University of Newcastle Human Research Ethics Committee (HREC-2010-1002). --- Conflict of interests BB has received Investigator Initiated Research grants from Pfizer and Boehringer Ingelheim.

Objectives: There remains a need to identify effective smoking cessation interventions in severely disadvantaged populations. This trial aimed to examine the effectiveness of an intervention (Call it Quits) developed to promote smoking cessation and delivered by community social service case-workers. Methods: Call it Quits was a pragmatic, parallel randomised trial of a case-worker delivered smoking cessation intervention conducted in a non-government community social service organisation in New South Wales (NSW), Australia. Adult smokers requiring financial assistance were randomly assigned to the five-session Call it Quits intervention or usual care control group. Of the 618 eligible individuals, 300 were randomised to the intervention group, of whom 187 (62%) consented and 318 were randomised to the control group, of whom 244 (77%) consented, resulting in 431 participants. The primary outcome measure was self-reported continuous abstinence up to 6-month follow-up with biochemical verification. Primary analysis was performed using all the available data from participants under the assumption the data is missing completely at random , followed by sensitivity analyses. Results: No statistically significant differences in the primary outcome were found (1.4% in the control group versus 1.0% in the intervention group, OR=0.77, p=0.828). Conclusions: A multi-component smoking cessation intervention delivering motivational interviewing-based counselling and free NRT by a trained case-worker within a community social service setting was not effective at achieving abstinence in a highly disadvantaged sample of smokers but increased attempts to stop and led to a reduction in number of cigarettes smoked daily.

Background Dementia is an age-related multifactorial disorder, and a growing body of evidence reveals that the risk of developing dementia later in life is determined by the cooccurrence of non-modifiable risk factors (e.g., apolipoprotein e4, family history) and modifiable risk factors across one's lifespan [1][2][3]. Over the last decade, evidence of modifiable risk factors for dementia has been mounting [1][2][3]. The Lancet Commission on Dementia Prevention, Intervention and Care demonstrated that 40% of dementia cases are attributable to twelve modifiable risk factors (i.e., less education, hearing loss, midlife hypertension, midlife obesity, smoking, depression, physical inactivity, diabetes, low social contact, excessive alcohol consumption, traumatic brain injury, and air pollution) [3][4][5]. Nevertheless, it is a challenge to enable individuals to change their health behaviour to tackle modifiable risk factors. Several behaviour change theories explain the determinants of health behaviour change, of which the health belief model (HBM) is believed to be the best suited model for dementia risk reduction [6][7][8]. Nonetheless, there is a consensus on four major constructs to measure the motivation to change one's lifestyle and health behaviours, including 1) knowledge of the disease and its risk factors; 2) perceived severity of the disease; 3) perceived susceptibility of the disease; and 4) motivation, including perceived benefits or barriers to performing risk-reducing behaviour [6,8]. Given that descendants of people with dementia have experience with dementia, they might be particularly eager to receive information and obtain more knowledge regarding dementia risk reduction. Moreover, descendants of people with dementia might be receptive to adopting a healthier lifestyle to reduce their dementia risk. A recently updated review by Cations et al. (2018) summarized the evidence of previous surveys on the knowledge of dementia and dementia risk reduction [9,10]. The included studies were conducted in the general population in Europe, the US, Eastern Asia, Israel, and Australia and found that knowledge about the opportunity for dementia risk reduction is poor but may be improving over time [9,10]. However, these studies' data were often collected through surveys, whereas qualitative data collection through focus groups might be more useful to obtain insight into the beliefs and attitudes towards dementia and dementia risk reduction. The open structure of focus group discussions provides the ability to identify unanticipated themes [11]. Kim et al. (2015) conducted a focus group study to investigate the knowledge, beliefs and attitudes towards dementia and dementia risk reduction in the general population aged 50 years and older [12]. They found that both fear of developing dementia and the need to improve dementia knowledge are important motivators for adopting and maintaining a healthier lifestyle for dementia risk reduction [12]. To our knowledge, none of the previous studies were aimed at a selected sample of descendants of people with dementia who have an increased risk of developing dementia [13]. To contribute to the development of a dementia risk reduction programme for descendants of people with dementia, the aim of the current study is to obtain insight into the knowledge, beliefs and attitudes towards dementia and dementia risk reduction among descendants of people with dementia. Fundamental elements can be captured to improve the willingness of middle-aged descendants of people with dementia to participate in a dementia risk reduction programme and adopt a healthier lifestyle. Moreover, by revealing areas for improvement, insight can be obtained on what factors a dementia risk reduction programme should focus on to enable health behaviour change. --- Method --- Participants In this qualitative study, focus group discussions were used. The study population consisted of descendants of people with Alzheimer's disease (AD), vascular dementia (VD) or mixed dementia diagnosed at hospital memory clinics in the northern part of the Netherlands. Twentyfour eligible participants of all adult ages and different educational levels were approached between February and June 2017 by medical specialists at the hospital memory clinic after diagnosing the individual's parents with dementia. Subsequently, these individuals were invited to participate in a focus group discussion shortly after the diagnostic consultation (1-2 months after diagnosis) and received a flyer with more information about the study. In total, nineteen eligible participants were interested in participation of which four could not participate due to practical reasons. Eventually, fifteen participants participated in one of the focus group discussions. The aim was to include four to six participants in each focus group, which were also referred to as'mini groups'. This type of focus group gives the moderator the opportunity to gain more information from each individual and give more attention to the participants on this sensitive topic [11]. --- Data collection and procedure We applied a narrative interview approach with a topic guide specifically designed for this study that enabled discussion, clarification and verification of unanticipated themes [14]. A semi-structured topic guide based on the HBM [6] was used, aiming to identify the factors influencing health behaviour change for dementia risk reduction in adult children of patients with AD, VD or mixed dementia. The topic guide included questions on knowledge, beliefs and attitudes towards dementia risk assessment and dementia risk reduction (see Supplementary file 1). The moderator used open-ended questions to facilitate discussions and to provide the opportunity to the participants to talk freely. All the focus group discussions were held in a private room at the Medical Faculty of the University of Groningen (Groningen, The Netherlands). The focus group discussions were facilitated by an experienced female moderator (EM, also an ethicist) assisted by a trained female researcher (JV) who observed and took notes during the focus group discussions. Each focus group session had a duration of sixty to ninety minutes and was audio recorded with the permission of the participants for later analyses. Before each focus group, all participants were asked to complete a short questionnaire, including questions on age, gender and educational level. Drinks and snacks were provided during the focus group discussions. Afterwards, all participants received a voucher of twenty euros. The Medical Ethics Commission of the University Medical Centre Groningen (UMCG) concluded that this study was not subject to the Medical Research Involving Human Subjects Act. --- Analysis Qualitative content analysis was used to analyse the focus group data. Audio recordings from the focus group discussions were transcribed verbatim and analysed using Atlas-Ti version 8.1. Each transcript was analysed thoroughly, and when appropriate, a code was generated and assigned by two researchers independently (JV and RB). Moreover, a coding protocol was developed based on the analysis of the first transcript. Consensus was reached regarding the content of the codes by four researchers (JV, EM, RB and NS), which were used for the two remaining transcripts. Then, similar codes were grouped together and subsequently categorized into themes. --- Results Three focus group discussions were conducted in April, May and June 2017 with four to six participants each to achieve data saturation. The participant characteristics are provided in Table 1. The age of the participants ranged from 26 to 61 years (mean 48.8, standard deviation (SD) 12.0), and 80% of the participants were female. The majority (80%) of the participants had a high educational level (see Supplementary file 2 for the definitions of the educational levels). Four themes were identified in the analysis, of which three themes emerged directly from the topic guide: 1) knowledge on dementia and dementia risk reduction, 2) beliefs and attitudes towards dementia risk assessment and dementia risk reduction and 3) the requirements for a dementia risk reduction programme. One theme was not foreseen in the topic guide but instead featured prominently in the analysis, namely, the experiences of having a parent with dementia, including the related practical and emotional consequences for oneself. Although the research aim was to evaluate the knowledge, beliefs and attitudes towards dementia and dementia risk reduction, participants first exchanged their experiences of having a parent with dementia before we were able to discuss the topics in our topic guide. Each theme is described in the following sections. --- Experiences of having a parent with dementia Participants underlined the need to talk and share their experiences of having a parent with dementia with people who have similar experiences: 'For example, the participants perceived dementia as a severe disease and saw their parent becoming a different person: 'I feel that it is a demeaning illness because, as a person, you are so different after getting sick' (female, 50-60 years). Nevertheless, the participants were relieved when their parent finally got diagnosed and finally they knew what their parent was suffering from. Having a diagnosis also improved the understanding of their parent. For instance, they can now accept that their parent is not able to do the things anymore the way they did before: 'Yes, that is why I was ultimately relieved that it had been diagnosed, that I knew then, and then, I kind of resigned myself to it because certainly in the beginning, years ago, I thought, mum, please hurry up, what do you mean, you can't find the way anymore?' (female, 50-60 years). Having a parent with dementia has practical and emotional consequences. A practical consequence is taking care of their parent, which requires time: 'Well, of course, I'm a busybody, I mean, as an informal carer. I visit on average two to three times a week, so yes, that is rather intense' (female, 60-70 years). An emotional consequence of having a parent with dementia is the anxiety to develop dementia. When a parent was diagnosed with dementia at a particular age, participants were afraid to be confronted with dementia at the same age: 'My mother was diagnosed with Alzheimer's when she was 57, and she died of it when she was 67 (...) me and my brothers, we sometimes talk about it; we are simply afraid that we may be confronted with it at the same age' (female, 40-50 years). Nevertheless, learning to cope with having a parent with dementia was more pressing than thinking about their own risk of developing dementia: 'I am more concerned about my parents than about myself' (male, 50-60 years). --- Knowledge on dementia and dementia risk reduction The general knowledge of dementia varied between participants, regardless of their age, gender and educational level. Several participants explained the use of dementia as an umbrella term: 'Well, I think that dementia is an umbrella term, covering all those [types of dementia]' (female, 60-70 years). Some of the participants thought that Alzheimer's disease is worse than "normal" dementia. A small number of participants even explained the pathology of dementia, although hesitantly: 'It's to do with proteins in the brain, that the transmission of signals is poorer, and so on' (female, 50-60 years). Most participants were uncertain about the heritability of dementia: 'I'm not sure whether it is hereditary or not, or perhaps early-onset dementia is, I really don't know' (female, 60-70 years). One participant was even hesitant to obtain information about the heritability of dementia, since she was afraid to find information she did not want to know. Regarding their knowledge of dementia risk reduction, participants were initially uncertain whether the development of dementia later in life could be prevented or delayed. Therefore, non-modifiable risk factors were often mentioned first, such as age, genetics and family history. After encouraging them, participants also correctly guessed the majority of the currently known modifiable risk factors for dementia, such as poor diet and lack of cognitive activities. Participants also had suspicions and questions about other possible risk factors for dementia, such as sleeping behaviour, stress, traumatic experiences and mental wellbeing. Furthermore, several participants believed that a regular check of cholesterol, blood pressure and diabetes could also contribute to dementia risk reduction: 'Yes, and what we can do about it? Well, be watchful and check often' (male, 50-60 years). All the identified risk factors by the group are presented in Table 2. The majority of the participants indicated that most of their knowledge was gained from the internet, family and friends or a caregiver in healthcare. Participants indicated that their general practitioner only provided minimal information about heritability: 'I have discussed it with my GP, who gave me very little information. He said we can do a test or something (...), but otherwise he didn't give me much information' (female, 40-50 years). Overall, the participants were eager to receive more information on dementia and dementia risk reduction. --- Health beliefs and attitudes towards dementia risk assessment and dementia risk reduction Initially, most participants believed that a dementia risk assessment is a genetic test that shows the chance of developing dementia later in life. Given that they were unaware or uncertain about the opportunity to reduce their risk of developing dementia, most participants were also uncertain whether they would want to have their dementia risk assessed. Some participants indicated that they would like to have their risk assessed and subsequently reduce their dementia risk, but they were uncertain about whether this was possible. Their beliefs and attitudes towards dementia risk assessment and dementia risk reduction are reflected in their motives to assess dementia risk and reduce their dementia risk, which are shown below. The most frequently mentioned motive to assess dementia risk was the possibility of acting upon the outcome of a risk assessment: 'I would only want it if you know you can do something about it because otherwise it's just a dark cloud hanging over your head' (female, 20-30 years). Another motive was the optimism of having a treatment available in the future, so if necessary, this treatment could cure their dementia in the future. Several other motives to assess and reduce dementia risk were mentioned after providing the participants with information regarding dementia risk reduction. One of these motives was to adopt healthy behaviour for dementia risk reduction to age healthily. Some said they would do anything to turn the tide of the development of dementia and grow old in good health. Another participant added that there is no harm in trying and considered to take the information more seriously: 'Yes, I feel like this can't really hurt. Maybe there is something in what they say. I don't know, do something with your life, drink less alcohol. I don't know, but well, it doesn't hurt to try' (female, 20-30 years). Some participants found it already valuable to obtain insight into their health and lifestyle and just wanted to know everything about their health, even when it was not positive. Another motive was 'to have self-control', for instance by anticipating the results of a dementia risk assessment: 'I very much want to stay in charge (...) that is most important to me. That's why I would like to know (...). I am the kind of person who would opt for euthanasia at the final stage' (female, 60-70 years). Finally, one participant also mentioned their current cognitive health as a motive to adopt a healthy behaviour for dementia risk reduction: 'I'd say yes because I forget a lot of things even now; I sometimes wonder what I did this morning' (female, 60-70 years). Nevertheless, participants also mentioned several motives not to assess and reduce their dementia risk. A frequently mentioned motive not to assess dementia risk was that they are still young, and this would be something to consider in the future. However, one of the participants noticed that it could be possible to suffer from dementia already at her current age: 'But, yes, that's a bit funny. I think I'm [only] 60, but that's nonsense, of course because there were 60-year-olds in my mother's nursing home' (female, 60-70 years). Furthermore, the participants indicated that the outcome of a dementia risk assessment would cause restless feelings or anxiety given that it is unknown when symptoms will appear and how severe the symptoms will be. They also indicated that the outcome of the dementia risk assessment does not provide certainty that they will or will not develop dementia and that a healthy lifestyle is no guarantee to prevent dementia: 'no matter how busy you are, those very active people, they get it too' (female, 60-70 years). Finally, the participants believed that the ultimate choice is a balance of interest between enjoying moments in life and having a healthy lifestyle: 'You have to weigh up the interests, I think. So I think that I would consider something like, I am enjoying myself so much now, I will have a drink now and then maybe have a week less [to live] later on' (female, 50-60 years). --- Requirements for a dementia risk reduction programme The participants expressed their need for more information on dementia and dementia risk reduction and would like to receive this information in a dementia risk reduction programme. However, their choice to participate in a dementia risk programme also depends on the content of the programme, intensity of the programme, type of advice given in the programme, outcome measure of the programme and specific functions of the programme. They mentioned several requirements for a dementia risk reduction programme. First, the programme should be a central point of reliable, clear and up to date information about dementia and dementia risk reduction. Second, the programme should offer regular health check-ups, which should reveal room for improvement in relevant lifestyle factors they can act upon: 'if you really get the result like, "yes you will get it", that is different from "maybe you can do something about this"' (female, 20-30 years). Subsequently, participants would like to receive personalized lifestyle advice and not general information that is applicable to everyone: 'I think if it's about general things like I just heard about (...), then I think that's not something new for me, you know, (...), so I think then it must really be a specific thing for me, like this is your individual chance and you really have to do this very differently' (female, 30-40 years). Nevertheless, participation in the programme should not be too time consuming, since they also have a job, a parent to take care of and other activities in their lives. Further, participants would like to have the possibility to share information with their siblings. Finally, in order to increase motivation to stay in the programme, participants mentioned that it should enable participation without the help of healthcare providers, should be easily accessible, should provide regular reminders and it should not cause guilt feelings when not adhering to the advice. See Table 3 for an overview of the requirements. --- Discussion Although the research aim was to evaluate the knowledge, beliefs and attitudes towards dementia and dementia risk reduction among descendants of people with dementia, our findings demonstrate that individuals with a parent with dementia feel the need to share their experiences on how to cope with a parent with dementia with their peers and that their worry about their own risk of developing dementia was inferior to this need. Furthermore, initially, the participants were unaware or uncertain about the possibility of reducing the risk of developing dementia, resulting in uncertainty regarding whether they would like to assess their dementia risk. Although the participants identified a large number of modifiable risk factors as a group, they were eager to receive more information on dementia and dementia risk reduction. By sharing their experiences of having a parent with dementia and their knowledge of dementia, the participants adopted a more positive attitude towards participation in a dementia risk reduction programme and provided important elements for future dementia risk reduction programmes. --- Sharing experiences of having a parent with dementia In the current study, the participants underlined the importance of sharing their experiences of having a parent with dementia with individuals who have had similar experiences. Sharing experiences of having a parent with dementia seemed to be a prerequisite to thinking about their own health and dementia risk and facilitated movement between the pre-contemplation phase and the contemplation phase of behaviour change [15]. Therefore, it is important to incorporate interactions between peers, for example, group-based interventions. This setting might encourage individuals to participate and adhere to the programme. To the best of our knowledge, the FING ER (Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability) trial is the only dementia risk reduction trial to date that provided groupbased as well as individual interventions that significantly reduced dementia risk by improving or maintaining cognitive functioning [16]. In light of our results, the group-based interventions of this trial potentially primarily contributed to the effectiveness of the multidomain intervention. --- Knowledge on dementia and dementia risk reduction We found that knowledge on dementia and dementia risk reduction was limited, even among descendants of people with dementia. Most participants believed that a dementia risk assessment is a genetic test that shows the chance of developing dementia later in life. At first, the participants were uncertain whether it was possible to modify their risk for developing dementia later in life. Nevertheless, the participants in the current study eventually identified several modifiable risk factors for dementia as a group, which included the majority of the currently known modifiable risk factors for dementia, such as cardiovascular diseases [1,2]. The risk factors loneliness, obesity and renal dysfunction were not mentioned by the group. In the Netherlands, approximately 11% of the general population identified renal dysfunction as a risk factor for dementia, indicating that the majority are unaware of renal dysfunction being a risk factor for dementia [17]. The participants in the current study also had suspicions about whether sleeping behaviour, stress, traumatic experiences and mental wellbeing were modifiable risk factors for dementia. Although strong and sufficient evidence for these factors is still lacking, some studies support that these factors might play a role in the development of dementia [18][19][20]. Furthermore, at first, the participants were hesitant about assessing dementia risk without a treatment in sight due to their unawareness of the possibility of Table 3 Requirements for a dementia risk reduction programme Requirements Central point of reliable, clear and up to date information about dementia and dementia risk reduction Regular check-ups with an easy interpretable outcome measure and amenable for acting on Personalized lifestyle advice, including the benefits of adhering to the advice Not too time consuming. The intensity of the programme should not avert enjoying life next to a job and care of parent(s) Possibility to share information with siblings Easily accessible (e.g., without having to ask the general practitioner) Regular reminders by for example text messages Adopting a healthy behaviour is their own responsibility and independently performed without the help of healthcare providers Should cause no guilt feelings when not adhering to the lifestyle advice reducing dementia risk. This lack of knowledge forms a barrier towards lifestyle changes for dementia risk reduction. Individuals with more knowledge about dementia and dementia risk reduction might be more likely to adopt healthy behaviour. Therefore, promoting dementia awareness should especially be considered for descendants of people with dementia since this group at risk for dementia might be more receptive to health behaviour change. Improved knowledge about dementia and dementia risk reduction is not only helpful for reducing dementia risk, but could also be helpful for dealing with dementia related needs of the parent. Therefore, it is helpful for both the descendant and the parent with dementia. --- Health beliefs and attitudes towards dementia risk assessment and dementia risk reduction The participants in the current study perceived dementia as a severe disease and worried about developing dementia themselves. Previous literature has shown that individuals with a parental family history have a higher perceived risk of developing dementia than individuals without a parental family history [21][22][23]. According to the HBM, perceived risk is one of the determinants influencing the probability of adopting healthy behaviour [6]. Therefore, our hypothesis was that descendants of people with dementia are more receptive to adopting healthy behaviour for dementia risk reduction. However, despite their increased motivation to adopt healthy behaviour, our findings show that having a parent with dementia causes anxiety and might form a barrier to assess their risk and adopt healthy behaviour. Previous findings about whether having a family history has a positive effect on the motivation to adopt a healthy lifestyle appear to be contradictory [21,[24][25][26]. Two studies did not identify a difference in risk-reducing behaviour (e.g., trying to stop smoking, increasing physical activity) between individuals with and without a family history of cardiovascular disease [24,26]. However, two other studies demonstrated that a family history of diabetes, anxiety, depression and high blood pressure is positively associated with risk awareness and risk-reducing behaviour [21,25]. The self-perceived risk (e.g., perceived severity and perceived susceptibility) of developing a certain disease might mediate the association between having a family history and interest in health education to adopt a healthy behaviour [21]. Further, participants were afraid that the outcome of a dementia risk assessment might cause restless feelings or anxiety, since it does not provide certainty that they will or will not develop dementia later in life and a healthy lifestyle is not a guarantee that they will not develop dementia. This may suggest that focussing on maintaining optimal cognitive health instead of reducing dementia risk is preferred. --- Strengths and limitations To our knowledge, this was the first study that explored the knowledge, beliefs and attitudes towards dementia and dementia risk reduction among a selected sample of descendants of people with dementia. A major strength of this study is that it explored not only the knowledge but also the beliefs and attitudes of these individuals towards dementia and dementia risk reduction. Adequate knowledge is not sufficient for health behaviour change. Also positive health beliefs and attitudes towards dementia and dementia risk reduction are needed. With this study, we provided insight in what health beliefs and attitudes towards dementia and dementia risk reduction need to be improved in order for dementia risk reduction programmes to be effective. Another major strength is that we used focus group discussions, which are recommended to explore beliefs about health and disease [27]. Due to this study design, we were able to identify a finding that we did not anticipate in the topic guide. In addition, participants could share their opinion and react on each other's comments, leading to a discussion. This provided us with insightful information that we might not have collected using individual interviews. However, this study had certain limitations. The recruitment of participants was difficult. Not surprisingly, mainly highly educated individuals and females were included. Moreover, the recruitment setting may have led to selection bias for several reasons. First, females are more often informal caregivers and therefore accompany their parent more often to the hospital memory clinic [28]. Second, mainly patients with complex types of dementia visit the hospital memory clinic since patients need to be referred by their general practitioner [29]. Therefore, the study sample might not be representative of all descendants of people with dementia in the Netherlands. Most of the participants are highly educated. Knowledge about dementia and dementia risk reduction might be worse in lower educated individuals, resulting in different beliefs and attitudes towards dementia and dementia risk reduction. --- Implications These findings can be used in the development of dementia risk reduction programmes for descendants of people with dementia. Our findings strongly point to the importance of incorporating the possibility of exchanging experiences related to having a parent with dementia with individuals who have had similar experiences in a dementia risk reduction programme. It seemed that sharing experiences of having a parent with dementia is a prerequisite for offspring to think about their own health and dementia risk. Additionally, descendants of people with dementia made several recommendations about which other elements should be included in a dementia risk reduction programme (see Table 3). Based on these recommendations, the online lifestyle programme for the Demin study was developed [30]. This programme consisted of: 1) a dementia risk assessment on five measurement moments during 1 year follow-up (online questionnaires, physical examination and blood sample) and 2) an online tailor-made lifestyle advice regarding protective (Mediterranean diet, low/ moderate alcohol consumption, high cognitive activity) and risk factors (physical inactivity, smoking, loneliness, cardiovascular disease, hypertension, high cholesterol, diabetes, obesity, renal dysfunction, depression) for dementia. The outcome of the dementia risk assessment was indicated by the Lifestyle for Brain Health (LIBRA) score in which each protective and risk factors for dementia was categorized into one of the following categories: 1) keep this up, 2) room for improvement or 3) remember to manage well [31]. This type of outcome measure is easy interpretable and amenable for acting on. To improve the knowledge of the potential participants, we provided general information about dementia and dementia risk reduction on the Demin website (www.demin.nl), in plain text and spoken animations. Furthermore, the participants received tailor-made lifestyle advice, including information about the protective and risk factors for dementia, its association with dementia and recommendations how to improve their lifestyle with regard to that specific protective or risk factor. Unfortunately, it was not possible within the Demin study to incorporate the possibility for social contact between participants due to its construct (online lifestyle advice). However, participants had the opportunity to invite their siblings to participate in the study too. In the Demin study the uptake and effectiveness of this online lifestyle programme was investigated among individuals with a parental family history of dementia [30]. When the opportunity to share experiences of having a parent with dementia is also included in future dementia risk reduction programmes, the willingness to participate in a dementia risk reduction programme and the effectiveness in adopting and maintaining healthy behaviour among descendants of people with dementia might be further improved. Our findings also support reinforcing knowledge about dementia and increasing the awareness of the opportunity to reduce dementia risk through a healthy lifestyle. More knowledge and awareness can contribute to more positive health beliefs and attitudes towards dementia risk reduction. For example, this increase in knowledge could be achieved through a targeted national mass media campaign aiming to motivate individuals to address their personal risk factors. As general practitioners are often the first point of contact for people who are concerned about their health and dementia risk, they should be educated about the opportunity to reduce the risk of developing dementia and methods to use this information to inform descendants of people with dementia properly. --- Recommendations for future research First, evaluating the knowledge, beliefs and attitudes of less educated descendants of people with dementia would be a valuable addition for future research since there is more room for improvement regarding lifestyle changes for dementia risk reduction. Subsequently, we encourage the development of dementia risk reduction trials for descendants of people with dementia, including the possibility of exchanging experiences with individuals who have had similar experiences to improve recruitment and to be effective in adopting healthy behaviour for dementia risk reduction. --- Conclusion Sharing experiences of having a parent with dementia seemed a prerequisite for thinking about one's own risk of developing dementia and participating in a dementia risk reduction programme. Knowledge of dementia and dementia risk reduction is limited. Due to the unawareness of the possibility of reducing dementia risk, the participants were hesitant about assessing their own dementia risk without a treatment in sight. Sharing information about risk factors for dementia and the importance of a healthy lifestyle could change people's perception of dementia risk assessment and their willingness to participate in a health behaviour programme for dementia risk reduction. Therefore, education on dementia and dementia risk reduction is needed. --- Availability of data and materials The data collected during this study will be available from the corresponding author upon reasonable request. --- Supplementary Information The online version contains supplementary material available at https://doi. org/10.1186/s12889-021-11415-2. Additional file 1: Supplementary Table 1. Focus group topics and illustrative questions. Supplementary file 2. Definitions of low, middle, and high level of education based on the International Standard Classification of Education. Authors' contributions JV, EM and NS contributed to the design of the study. EM and JV collected the data. JV and RB conducted the analyses. JV wrote the manuscript. EM, RB, AAH, PPDD, RCOV, FER, EB, SEdR and NS revised the manuscript. All the authors read and approved the final manuscript. --- Declarations --- Ethics approval and consent to participate The Medical Ethics Commission of the University Medical Centre Groningen (UMCG) concluded that this study was not subject to the Medical Research Involving Human Subjects Act. Participants provided written informed consent. --- Consent for publication Not required. --- Competing interests None declared. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Individuals with a parental family history of dementia have an increased risk of developing dementia because they share their genes as well as their psychosocial behaviour. Due to this increased risk and their experience with dementia, they may be particularly eager to receive information regarding dementia risk reduction (DRR). This study evaluated the knowledge, beliefs and attitudes towards dementia and DRR among descendants of people with dementia. Method: Using a semi-structured topic guide, three focus group discussions were conducted consisting of 12 female (80%) and 3 male (20%) descendants of people with dementia with a mean (± SD) age of 48.8 (± 12) years. Focus group discussions were audio recorded and transcribed. Each transcript was analysed thoroughly, and where appropriate, a code was generated and assigned by two researchers independently. Then, similar codes were grouped together and categorized into themes. Results: The items in the topic guide could only be addressed after participants had been given the opportunity to share their experiences of having a parent with dementia. Participants were unaware or uncertain about the possibility of reducing the risk of developing dementia and therefore hesitant to assess their dementia risk without treatment options in sight. Moreover, participants indicated that their general practitioner only gave some information on heritability, not on DRR. Although participants identified a large number of modifiable risk factors as a group during the group discussions, they were eager to receive more information on dementia and DRR. In the end, participants adopted a more positive attitude towards a DRR programme and provided suggestions for the development of future DRR programmes.

INTRODUCTION Family planning is a complicated subject with many facets that touch on different cultural and religious beliefs. One such intersection is that which exists between Islamic tradition and Somali culture. Somalia, a country with a large Muslim population, has a rich cultural background that affects how it views family planning. Nevertheless, Islam has a big influence on how the Somali community views and behaves in terms of reproductive health. The idea of family planning is frequently regarded in Somali culture through the prism of societal standards and expectations. The importance of the extended family structure cannot be overstated, and for people and their families, having children is a joy and a source of pride. According to a study by (Straus et al., 2009), Somali women see having big families as a sign of social prestige and motherhood as an integral component of their identity. In order to assure the continuance of the family line, there may be social pressure on couples to have more children, especially sons. It is crucial to remember, though, that Islam also encourages responsible parenting and places a strong emphasis on the health of both parents and children. Islamic scholars have interpreted a number of verses from the Quran and Hadiths (prophet Muhammad's sayings) to justify the use of contraception when used within specified parameters. For example, a renowned Islamic Philosopher (Al-Qardawi, 2013) has said that contraception is acceptable as long as it does not endanger the woman's health or require permanent sterilization. Overall, Islamic teachings and societal expectations both have an impact on family planning in the Somali culture. Islam provides a framework that enables ethical decision-making on reproductive health, notwithstanding the widespread desire for big families. --- FAMILY PLANNING AND THE ISLAMIC TRADITION Islamic academics and religious authorities have engaged in extensive discussion and debate on the subject of family planning. Marriage, reproduction, and family life are only a few of the topics that are covered in detail in Islam as a holistic way of life. The Quran, the Hadith (sayings and deeds of Prophet Muhammad), and the consensus of Islamic scholars are the sources of Islamic ideas and teachings on family planning. We must take a closer look at the big picture in order to completely comprehend the Islamic perspective on family planning. Naturally, it starts with promoting the life principle. Thus, the Prophet's advice to procreate and the Qur'anic ban on infanticide, a common pre-Islamic practice involving born children that was primarily driven by economic and gender factors, were both given (Azizah, 1993). The major source of Islamic doctrine, the Quran, does not specifically mention family planning. However, it places a strong emphasis on the value of good parenting and the welfare of kids. Muslims are urged by the Quran to fulfill their obligations to their families and to provide for their children's material, spiritual, and emotional needs. Additionally, it encourages moderation in all facets of life, including conception. Surah Al-Isra (17:31) is one scripture that is frequently brought up in conversations on family planning. It reads, "And do not kill your offspring for dread of poverty. We take care of both them and you. Indeed, it is a serious sin to murder them. Many academics understand this text as forbidding the purposeful termination of pregnancy only because of financial worries or fear of poverty. It emphasizes the value of life and the obligation to raise one's children. Muslims who favor family planning frequently claim that the Qur'an is silent on the subject, and they interpret this silence as an affirmation rather than a condemnation. For example, Fazlur Rahman has noted that "one finds nothing which opposes the view that we should control our population, for a time, to remedy our present situation (Rahman 1972, 94)" in the verses of the Holy Qur'an. Conservative Muslims, however, like Abul eAl Maududi, maintain that "the Qur'an is not silent" (Maududi 1974, 83) regarding the matter. They cite the Qur'anic prohibition of the practice of burying live female infants, the "killing" of infants is forbidden or condemned in the Qur'anic verses (Surah 6: Al-Adam: 137, 140, 151; Surah 17: Al-isaii:31; Surah 60: Al-Mumtahanah:12), which were common in pre-Islamic Arabia (Surah 81: At-Takwir: 8-9; Surah 16: An-Nab!: 57-59). In addition, they cite the following verses to back up their claim that procreation is a gift from God (Hassan, 2000). From the afore discussion, it is clear that when dealing with the family planning issue, Muslims are not in agreement in its permissibility or lack of it. The polarity of opinions on this topic among the Muslims paves the way for many to adopt a flexible approach mainly based on their own personal understanding and choice, while others follow people whom they think are in a better poison when dealing with Islamic knowledge. Additional information about family planning in an Islamic setting can be found in the Hadith, Contraception and birth control methods are not explicitly mentioned in the Prophet's sayings, however there are narrations that touch on related subjects. The story of Jabir ibn Abdullah asking the Prophet Muhammad(Saw) about the coitus interruptus (withdrawal method) as a method of contraception is one Hadith that is frequently cited. Although the Prophet did not expressly forbid or support this strategy, he did encourage Jabir to respect his wife's rights while simultaneously Volume: 10 | Issue: 9| September 2023 ---------------------------------------------------------------------------------------------------------------------------------------------------------- pursuing his own ambitions. Although this narrative suggests that the withdrawal procedure was known and used throughout the time of the Prophet, it does not make a firm determination about its legality. Another Hadith that is frequently quoted is one in which the Prophet Muhammad gave advice to a man to wed a childbearer. Some academics interpret this story as an exhortation to reproduce and preserve the human race's continuance. It should be understood that this advice was given to a specific person and might not be appropriate in all situations. Diverse viewpoints on family planning practices have been offered by Islamic thinkers. According to the majority of academics, family planning is acceptable as long as it doesn't result in permanent sterility or harm to the person's health. They contend that Islamic teachings support responsible parenthood and permit the use of contraceptives that are secure, reversible, and do not obstruct the normal reproductive cycle. Neither the topic of family planning is novel to Islamic societies nor is it a Western plot. Family planning is defined clearly in Islam, and it is also suggested how much of it is acceptable. Although there are many various ways to interpret these teachings, many Muslim nations have historically used them to guide their population strategies(El Hamri, 2010). From the above analysis, it is safe to seek knowledge from authentic and reliable sources and at the same time behave responsibly and fear Allah, keeping always in mind that we are all accountable to Allah for all our intentions, practices and deeds. --- FAMILY PLANNING AND THE SOMALI CULTURE Family planning is a subject that covers a number of reproductive health issues, such as contraception, knowledge of one's own fertility, and the spacing and timing of pregnancies. It is impacted by social, economic, cultural, and religious aspects. Family planning procedures and attitudes in Somali society are influenced by a blend of conventional principles, Islamic teachings, and modern trends. The idea of family planning has existed for millennia in Somali society, even though it may not have been known by that name. In the past, Somali groups used a variety of techniques to regulate the size of their families and the spacing between births. Long-term nursing, abstention during specific times, and herbal cures were some of these techniques. Rather than population control, mother and child health was frequently the primary driver of these measures. (Deyo, 2013) reported that, the practice of solely breastfeeding a child for two years is known as "child spacing," and it is done for the mother's health as well as the health of any future children she may have.Among the different methods used in family planning, child spacing, using long -term nursing is the preferred procedure regarding Somali culture. This is because both parents believe, this practice is good for the health of the mother and the child as mentioned earlier. --- Islamic Teachings Somalis' views on family planning are significantly influenced by Islamic teachings. Islam promotes responsible parenting and places a strong emphasis on taking good care of one's children. Although the use of contraception is not expressly forbidden in Islam, different scholars have different views on whether it is acceptable. According to certain Islamic scholars, contraception is acceptable as long as it doesn't endanger a woman's health or completely prevent her from having children. Others feel that contraception should only be used under legitimate medical circumstances or with both partners' permission. Religious Islamic leaders who were contacted stated that certain practice suggestions for the use of contraception are allowed in connection to birth spacing in order to promote the health of the mother and child. The term "birth spacing" is preferred over "family planning" when providing Muslim women with expert contraception guidance (Egeh et al., 2019). As explained elsewhere in the paper, the Somali culture gives priority to the health and well-being of the mother and her child over population control when considering family planning issues. Family planning strategies in Somali culture have been touched by contemporary influences as well. Social standards and values have changed as a result of urbanization, globalization, and more access to education. The attitudes about family planning have changed as more Somalis are exposed to contemporary ideas and concepts through media and encounters with other cultures. ---------------------------------------------------------------------------------------------------------------------------------------------------------- -------34 The Intersection of Somali Culture and Islamic Tradition It can be difficult to understand how Somali culture and Islamic family planning customs interact. While larger families may have been valued in traditional Somali culture, responsible parenting, and children's welfare are stressed in Islamic teachings. --- Volume: 10 | Issue: 9| September 2023 Due to both cultural shifts and Islamic beliefs, there has been an increase in the acceptance of family planning within the Somali community in recent years. Numerous religious figures in Somalia have acknowledged the value of family planning in advancing the health and welfare of households. Couples have been urged to think carefully about how far apart to have their kids, keeping in mind things like the mother's health, their financial situation, and their capacity to care for each child properly. It is crucial to remember that some Somali populations still cling more strictly to traditional cultural traditions and can be less tolerant of contemporary family planning techniques. (Gele et al., 2022) noted that more stringent and successful programs should be the current approach to assist minimize the risk of unsafe abortions in Somalia, where pervasive unfavorable attitudes regarding abortion, as mandated by the culture, obstruct access to safe abortions. Lack of education, poor access to healthcare, and deeply set cultural views are just a few of the things that can make people reluctant to use family planning methods. --- THE SOMALI GOVERNMENT The government of Somalia is aware of the value of family planning in advancing both sustainable development and mother and child health. There have been initiatives to broaden national access to reproductive healthcare services. The framework for managing the nation's supply chain for sexual and reproductive health has been enhanced(United Nations Development Programme Somalia, 2008). Nevertheless, problems such as a lack of resources, substandard infrastructure, cultural obstacles, and low awareness persist. --- CONCLUSION There are many facets and complexities to the discussion of family planning between Islamic religion and Somali culture. Large families are highly valued in Somali culture, which also sees kids as a blessing. Islamic teachings, however, also stress the value of good parenting and offer recommendations for family planning strategies. Islam provides for flexibility in family planning decisions as long as they are undertaken within the bounds established by Islamic teachings, notwithstanding potential conflicts between cultural standards and religious views. It is essential to handle this subject with consideration for both cultural and religious viewpoints. Programs for education and awareness can be quite effective in bridging the gap between religious teachings and cultural customs and assisting people in making family planning decisions. Involving community elders and religious leaders in family planning meetings might also aid in resolving any misunderstandings or potential disputes. Overall, when discussing family planning within the Somali community, striking a balance between cultural values and religious convictions is crucial. It is possible to assist people in making decisions that are consistent with their cultural heritage while still respecting their religious values by encouraging open communication, education, and understanding. --- RECOMMENDATIONS <unk> Understanding Somali Cultural Perspectives: To effectively address family planning in the Somali community, it is crucial to comprehend the cultural perspectives and beliefs that shape attitudes towards reproductive health. This section should discuss the significance of extended families, the value placed on children, and the role of traditional gender roles in Somali culture. Emphasize the importance of approaching family planning discussions with cultural sensitivity and respect for these values. <unk> Exploring Islamic Teachings on Family Planning: Islamic teachings offer guidance on various aspects of life, including family planning. In this section, provide an overview of the Islamic principles related to reproductive health, contraception, and responsible parenthood. ---------------------------------------------------------------------------------------------------------------------------------------------------- Misconceptions and myths surrounding family planning often exist within cultural and religious contexts. This section should focus on debunking common misconceptions about contraception and family planning methods that may hinder their acceptance within the Somali community. Provide evidence-based information and cite reputable Islamic scholars to reinforce the compatibility of certain methods with Islamic principles. <unk> Promoting Education and Awareness: Education and awareness campaigns play a vital role in promoting family planning practices. Discuss the importance of culturally sensitive educational initiatives that empower individuals, particularly women, with knowledge about reproductive health, contraception options, and responsible family planning. Highlight the significance of involving trusted community leaders, religious scholars, and healthcare providers in disseminating accurate information. <unk> Encouraging Open Dialogue and Collaboration: Facilitating open and inclusive dialogue among community members, religious leaders, and healthcare providers is crucial for fostering a supportive environment for family planning. Encourage the establishment of community-based forums, workshops, and counseling services that provide a safe space for discussions, address concerns, and offer guidance on family planning matters. Collaborate with local religious institutions to integrate family planning education within religious teachings. <unk> Strengthening Healthcare Infrastructure: Efforts to promote family planning within the Somali community must be supported by accessible and culturally sensitive healthcare services. Advocate for the expansion of healthcare infrastructure, including reproductive health clinics, that offer a wide range of family planning options while respecting cultural preferences. Encourage healthcare providers to receive cultural competency training to better serve the Somali population.

This essay examines the intricate connections between the Islamic faith, Somali culture, and family planning. Large families are viewed as a sign of social prestige and pride in Somali culture, which explores the societal norms and expectations surrounding family size. Examined are how maternal responsibilities and the extended family structure affect reproductive choices. The influence of Islamic teachings on family planning techniques in the Somali community is also investigated. Islam favors procreation and regards children as a heavenly blessing, but it also advocates responsible parenting and places a strong emphasis on the welfare of parents and children. The paper explores the ways in which Islamic scholars have interpreted sacred texts to advocate for the use of contraception within specific parameters. A sophisticated approach to family planning is produced by the fusion of Islamic tradition and Somali culture, balancing societal expectations with the welfare of the individual and the family. The abstract ends by emphasizing how crucial it is to comprehend these cultural and religious aspects in order to create successful tactics that respect the ideals and principles of the Somali community while advancing all-inclusive reproductive health care.

INTRODUCTION To my mind inequality is the main roadblock in our journey toward social justice, and we need an innovative approach to uprooting it that won't produce the same negligible incremental change we've seen in recent years (Willis, 2014). This paper examines the subtle drifts from the male domination in educational leadership that characterized much of the previous centuries. The paper believes that these drifts are manifesting themselves in many different ways such as the feminization of leadership styles, the decline of hierarchy and the rise of soft power, and especially the change in roles and attitudes towards work and family life, especially from men. So, educational organisations that wish to thrive in the future would have do well to develop a culture along these trends. This is imperative for future organisational life because the gender power shift that is occurring in twenty-first-century organisations will be an indisputable triumph for educational and for that matter socio-economic progress as it seeks to enable women at every level to realize their true potential. Subsequently, organisations that ignore these trends do so at their own peril. This is the second decade of the twenty-first century, and it is significant that the vexed subject of gender in organisational life is commanding more serious attention than ever before. Much of this attention focuses on the many years of equality debate which has sought to make women at ease and comfortable in their leadership positions due to the support and recognition being given to them by men, and even the achievement of parity with men at the top of organisational life, especially in educational organisations. The signs of these shifts however tend to be obscured when the spotlight is put on the top of organisational hierarchies because the signs manifest mostly at the middle management level. Besides, there are other less discussed aspects to the gender debate that signal that a shift is under way. These are signaled by the very urgency that this issue commands, as well as the coalitions that have joined the calls for change. This notion of power residing only at the top of organisations is under challenge in the 21 st century. Technological and social changes are breaking down traditional hierarchies and distributing power more widely both within organisations and between them and their networks of external partners. Thus, in the 21 st Century public administration, organisations depend on a growing "contingent" workforce of external contractors and freelancers. New and smaller organisations such as schools are beginning to consist mainly of peers collaborating with each other, with very little if any hierarchy. These trends are set to continue as work becomes increasingly portable and people experiment with more autonomous ways of working. Aron (2013) has predicted the rapid growth of clusters. These are self-governed groups of professionals with a range of skills who will be hired by organisations to work long-term on projects and operations. Thus, this shift away from massive standalone organisational structures will undermine the link between status and position in a hierarchy. Status will instead rely increasingly on unique skills and talents, or the ability to connect people, or being an inspiring leader. The whole approach to the new world of work will require workers to become personal leaders. Maitland and Thompson (2014) have said that "... you need to lead your own life rather than being controlled by the boss as in the past, and if you are able to do that, you are able to lead others." The Edelman Trust Barometer (2014) shows that the democratization of work environments runs parallel to a decline in trust in traditional authority figures, and that people are more likely to trust experts, or to put their faith in their peers. The trust barometer indicates that trust in the CEO is at 43%, while trust in academics stands at 67% and trust in "a person like yourself" at 62%. This poses a challenge for leaders of traditional organisations, and leaders in some organisations are beginning to acknowledge that they need to encourage a greater diversity of leadership styles that seek to achieve gender balance as part of the solution. Furthermore, in the 20 th century, the exercise of hard power was much lauded at the workplace. However, Nye (2005) has developed the concept of soft power which relates to persuading people to do what you want by attracting and co-opting them, rather than coercing them. He developed this concept in relation to world politics but the concept is now being used in public administration, with a growing emphasis in leadership development on the ability to listen, seek out other perspectives, and earn followers, alongside standard leadership traits such as resilience and decisiveness. The rise of soft power seeks to make leaders more effective, and is tending to open the door more widely to women's leadership. As a sequel to this view, Tost et al. (2013) have argued that teams perform less well on assigned tasks when led by people who equate leadership with power, dominate discussion, and discount the contributions of other team members. Their experiments found that the psychological effect of power on a team leader had a negative impact on team performance. This connotes the feminization of leadership in the public sector and this is manifested in the 21 st century as demand grows for skills that are suited to flatter more open and democratic organizations, and the preferences of younger generations. These skills are expressed in concepts like 'emotional intelligence', 'people skills', and 'flexibility' which are traditionally deemed to be more feminine leadership skills, and are highly valued in 21 st century public administration, and coincidentally expresses the outlook for educational leadership futures. These exigencies suggest that women are likely to lead the way in reshaping how jobs are done, including at senior levels in the educational sector. These participative concepts for leadership, (which of course is not the preserve of females) is linked to more productive teamwork. However, women on average are known to score higher than men on social perceptiveness (i.e., correctly reading emotions). Wooley et al. (2010) studied the "collective intelligence" of teams regarding their ability to solve puzzles and problems and found that, gender mix played a big part. The researchers were looking to see if a team's collective intelligence equated to the average of the IQs of its members. IQ turned out not to be a key factor. However, the teams with more women had higher scores, demonstrating a stronger collective ability to solve problems and achieve goals. On another hand, technology is also putting more power into the hands of knowledge workers in the public sector, many of whom have greater choice than ever before about how, where and when they work. Besides, people with skills that are in demand also have more choices about who to work for, or with. Organisations can therefore no longer assume that workers will just accept standard conditions of employment or traditional working patterns. This scenario has been created in the 21 st century as many responsibilities at the workplace are technology-enabled and organisations are able to staff online moderators and community managers across the world, most of whom work virtually from their homes, and have developed a reputation for high-quality customer service and a supportive working culture. From this background, the paper sees both female and male professionals leading a shift in the way that organisations are run and do business. This implies a changing roles of public sector leadership with increasing emphasis on empathy for people and a more focus on communication, especially in terms of accommodating varied work patterns that enable females to scathe the otherwise brush culture in their organisations. This is the result of the changes in the digital revolution and the increasing use of social media which has created a sea of change towards getting the right work culture. The assumption is that, the enhanced focus on people, communication and culture will create the environment where women will thrive. The evidence shows that women are leading the way in reshaping how jobs are done, including at senior levels with technology breaking down the barriers. As the future looks more promising than the past for women professionals, it creates a parallel opportunity for men too, which adds to the advantage for women. This means that the greater power-sharing opportunities at work comes with greater sharing of responsibility for children and the home. Organisations have typically regarded gender issues as women's issues. That was always a mistake, but now the pressure is increasing on organisations to acknowledge that men have caring responsibilities too. The model of male breadwinner and female caregiver is fast disappearing. So, the merging, or reversal, of traditional gender roles has major implications for work and society. Work-life conflict used to be seen primarily as a problem for women. But research reveals that employed fathers in dual-earning couples are now more likely than mothers to experience such conflict. There are ways for employers to reduce this, enabling both men and women to manage work and home life more effectively. As men seek to play their full role as fathers, it creates opportunities for enhancing women's progress at work. Moreover, shared parenting and earning creates the best advantage for the family. The idea here is that, the participation of fathers at home takes pressure off mothers, helping them to perform better at work. Such is the outlook for educational leadership futures that has created space for women to excel. The issues raised so far are salient to a convention in Ghana where boys' senior high schools are headed by males whilst girls' senior high schools are headed by females as observed by Preko (2012). He says also that in Ghana, managerial positions have been coded as masculine and this view runs through homes to workplaces. However, this trend seems to be changing as some boys' senior high schools (SHSs), now have females as their heads. On the face value, this does not appear to be an issue but given the long-standing tradition of males heading boys' SHSs, this is a critical difficulty in lieu of the challenges faced by the female leaders of such boys' only SHSs. Odei-Tettey and Mensah (2015) have said that such challenges can be surmounted if society will understand the cosmic logic of the concept of co-existence of opposites. The idea is that the world and existence are fully packed with mythic opposition, and that has implications for male-female complementarity in educational leadership. This view notwithstanding, Helterbran and Rieg (2004) have said long ago that parents and community members believe that males can effectively discipline male students than can women, and this has created a perception that men can handle boys' SHSs better than women. This perception tends to results in gender stereotype and intimidation which affect the effective functioning of the female leader of boys' SHSs. Contrary to this view however, the paper argues that the success of schools among other things depends on the effectiveness of the school head as a leader. This claim acknowledges the complexities associated with role of the school head. Thus, in performing their duties, female school leaders face myriad of challenges that range from gender discrimination to disciplinary issues amongst staff and students. These challenges are very profound in patriarchal societies for women in leadership positions due to the socialisation process that result in gender discrimination, as attitudes manifest and give much preference to males over females in many aspects of organisational life. This phenomenon tends to create patriarchal cultures that hinder women from being seen and heard (Mutabai, 2016), and renders the workplace environment non-welcoming for female leaders. The emotive issues discussed so far brings interesting and crucial twist to the gender debate which most societies have only recently begun to address. This makes for a renewed focus on continuous and structural gendered power relations, discrimination, institutional and structural inequalities, and the impact this has on everyday lives, and at the same time, also affords a discussion of organisational roles, optimism, and collaboration, as well as the increasing realisation of men's role in making major and significant improvement in creating gender spaces for leadership. So, in this paper, we place great emphasis on intersectionality and how gender issues link to other categories, such as tribes and family roles to informs the debate. 'Intersectionality' is an emerging concept in policy analysis and academic contexts, and so it is important to clearly define and explain the concept for its usage as the analytical framework for this paperthat is, in its application to policies for reinventing gender spaces for educational leadership futures. As a concept, intersectionality has been interpreted in the literature as a theory, methodology, paradigm, lens, tool and framework (Hankivsky, 2014;Collins, 2015;Cho et al., 2013;Sigle-Rushton, 2013). This paper does not seek to distinguish between these interpretations, but rather to introduce the key aspects of what it means to take an intersectional approach to understand structural inequality to aid policymaking for creating gender spaces for educational leadership futures. The intersectionality theory is traced to Crenshaw (1989) as she used the concept to refer to the double discrimination of racism and sexism faced by Black women, as a critique of the dominant single-axis framework that is used to frame antidiscrimination laws, with its focus on the experiences of the most privileged members of subordinate groups (Crenshaw, 1989). Crenshaw provided the following definition of intersectionality: "Intersectionality is a metaphor for understanding the ways that multiple forms of inequality or disadvantage sometimes compound themselves and create obstacles that often are not understood among conventional ways of thinking" (Crenshaw, 1989). However, while Crenshaw was the first to use the term intersectionality, the concept did not represent a new way of thinking. Later, Loden and Rosener (1990) developed the 'Diversity wheel' concept as a framework for thinking about the different dimensions of diversity within individuals and institutions. Intersectionality has since evolved to provide a useful tool for thinking about how different characteristics intersect with systems and structures to shape a person's experience. It is important to point out at this point that due to the rapid increase in the application of intersectionality across many fields, there exists a variety of definitions of intersectionality in the literature. However, for the purposes of this paper, we point to some common themes and assumptions embedding the theory. Consequently, we submit the foundational elements of intersectionality to encompass: a. the acknowledgement that individuals are shaped by their simultaneous involvement with multiple interconnected social entities. b. The interaction between multiple social entities takes place within a framework of linked systems and power structures involving laws, policies, and governments. This suggests that the appreciation of inequality of power is key to understanding intersectionality. c. Systemic inequalities, that reflect relative privileges and disadvantages, are the consequences of the interaction between social entities, power relations and circumstances. Hence, people's experiences of inequality can be enduring or temporary, and therefore creates unique lived experiences. d. Intersectionality is primarily a tool for understanding invisible power relations and how they shape inequality. Intersectionality looks at 'interlocking' systems of oppression and how this plays out in people's lives (see, Collins, 2015;Else-Quest & Hyde, 2016;Hankivsky, 2014;Scott & Siltanen, 2016). In order to understand the central tenets of intersectionality theory, we find it helpful to also point out what intersectionality is not. Intersectionality is not: a synonym for diversity; about adding up different kinds of inequality, and does not look to simply add up the sum of different oppressions; about pitting different people or groups against each other to assess who is most marginalised or disadvantaged; and looking to construct a hierarchy of inequality (Poverty and Inequality Commission, 2021). These imperatives of intersectionality in practice are best explained with illustrations that are found in the literature. Crenshaw (1989) has illustrated the concept thus: Black women sometimes experience discrimination in ways similar to white women's experiences; sometimes they share very similar experiences with Black men. Yet often they experience double discrimination -the combined effects of practices which discriminate on the basis of race, and on the basis of sex. And sometimes, they experience discrimination as Black women -not the sum of race and sex discrimination, but as Black women. Christoffersen (2017) has recently given an illustration of intersectionality as: "A person is not, for example, a woman on one hand and disabled on the other; rather she is the combination of these at the same time, that is, a disabled woman. In this example her identity as a woman is shaped by her identity as disabled, and vice versa as the elements of identity are not lived or experienced separately". So, 'intersectionality' refers to the interactions between the characteristics of inequality together with the wider characteristics that shape lived experiences of discrimination, inequality and privilege such as socio-economic disadvantage, occupation and care-experience (Christoffersen, 2019). These are identified to eliminate unlawful discrimination, harassment and victimisation, and to advance equality of opportunity, for individuals who are faced with discrimination (Mügge et al., 2018). So, as part of its contribution to the gender debate, this paper applies intersectionality theory to the examination of different facets of inequality associated with gender issues in school leadership in all boys' senior high schools, and advocate for policies to re-invent genders spaces and futures for educational leadership futures. The paper therefore does not seek to examine all areas of gendered life and inequality as its scope. Rather, it discusses arguments that have been put forward by the data, and contextualizing these for future possibilities regarding gender co-existence in educational leadershiptermed as, educational leadership futures. These arguments suggest that gendered inequality is visible in both public and private spheres. However, the validity of the arguments that gender inequalities are still entrenched and persist over time, place, and culture is contrasted with alternative claims from the data that gendered power relations, and for that matter inequalities, are gradually being eroded. Furthermore, the current academic focus on the concept of intersectionality regarding how variables like gender, tribe, family roles, social status, and culture intersect to define people's gendered experiences, is included in the discourse. Therefore, the study of women's educational leadership experiences will provide a framework to further examine these issues and address some of the debates. --- Statement of the problem The problem that underpinned the study is that: it is uncertain how long the leadership successes made by female heads of all boys' SHSs and enabled by the support and collaboration from their male colleagues will persist and perpetuate gender spaces for educational leadership futures. This problem is borne from an environment in Ghana where women in educational leadership positions have experienced discrimination from their male counterparts (see Mensah, Odei-Tettey & Osaebo, 2014). Even when women have equal qualifications and achievements to that of their male counterparts, and in spite of their abilities and accomplishments, they are less favourably preferred. This stereotyping tends to reflect in evaluations and promotions, and consequently puts women at a tactical disadvantage for advancement (see Eagly & Karau, 2002). Despite these challenges, some women in Ghana have accepted to lead all-male SHSs with remarkable success. --- Purpose of the study The study sought to investigate the sociocultural experiences of female managers and to explore how they have survived the patriarchal culture in the selected male SHSs in order to make policy recommendations for re-inventing genders spaces for educational leadership futures. --- Objectives of the study The study was guided by the following objectives: a. To find out the socio-cultural experiences of female managers in the selected male SHSs in Ghana. b. To explore the support services and strategies used by female leaders to deal with patriarchal challenges in the selected male SHSs as a basis for making policy recommendations towards creating gender spaces for educational leadership futures. --- Research questions The study sought to answer the following questions: a. What are the socio-cultural experiences of female managers in the selected male SHSs in Ghana? b. What support services and strategies do female managers use to deal with the patriarchal challenges in the selected male SHSs as a basis for making policy recommendations towards creating gender spaces for educational leadership futures? --- Theoretical Framework The fundamental assumption of this paper is that, 'female managers can overturn the patriarchal challenges they face in the line of duty if they adopt appropriate strategies to use the support systems available to them in male senior high schools. This is founded on two fundamental concepts that are gleaned from research questions: socio-cultural experiences of female managers in male SHSs, and support services and strategies used by females to deal with patriarchal challenges in male SHSs. Socio-cultural experiences of female managers in senior high schools Atta (2015) has said that social and cultural beliefs, practices and attitudes of society continue to perpetuate discrimination against women. So, male dominance, and the spectre of male interference, continues to be high and eminent even in this modern Ghana (Anewu, 2010). Gender discrimination in Africa is defined by the way culture has been shaped and this has created the perception that women cannot hold high leadership positions. Many researchers have indicated that women encounter huge challenges in assuming senior positions in organisations and this phenomenon is attributed to cultural perceptive that influences female participation. For example, women in Africa are perceived as domestic workers who should take care of family duties. Nukunya (1998) has asserted that in many Ghanaian societies, the traditional position is that a woman is never wholly independent. He postulated that a woman must always be under the guardianship of a man, and when she marries, her original guardian hands over some or all of his responsibilities for her to her husband. Also, traditional gender roles that have been socially constructed affect women's leadership in Africa. Policy makers have also been cited as previously lacking efforts to promote women's progress in taking up high leadership positions (Mutabai, 2016). So, traditional gender roles and socio-cultural norms together with the selection process lower women's self-confidence and career aspirations. Promoting women's progress therefore calls for cultural change which will require the efforts and cooperation of many authorities, organisations, and policymakers (Arkorful, Doe & Agyemang, 2014). Literature shows that much mistrust in women leadership is caused by a number of factors which include tradition and culture (Mensah et al, 2014, Moorosi, 2010); perception of school administration (Maseko, 2013); and stereotyped notions about women's leadership (Kitele, 2013). Literature has shown further that dual and conflicting roles are a major hindrance to women in leadership positions. Kitele (2013) has found out from the female heads that there is a conflict between domestic and professional roles with the reasons that schoolwork is so involving and interferes with domestic issues which give rise to conflicts. For example, as a child-bearer, the woman is expected to take care of the child and put away her career to meet the needs of her family (Mutabai, 2016;Davidson & Cooper, 1992;Burke & McKeen, 1994). Furthermore, some African cultures define women in terms of what they should be or do for men. For instance, a married woman's major role is to enhance her husband's career goals, by providing him with moral and emotional support, and therefore women can only pursue their professional dreams after fulfilling their culturally accepted roles (Malovi, 2014). Beeghly (2014) has also noted that women's gender identities were embedded in responsibilities for care that extended beyond mothering young children and included the care of husband and adult children and grandchildren. In some cases, women intentionally avoided work outside the home to have time to maintain family relationships. The concept of gender also connotes that the expectation held about the characteristics, aptitudes and likely behaviours of both women and men, and gender roles are learned through the process of gender role socialisation. Boys are generally taught to be masculine (interpreted as active, aggressive, tough, daring and dominant), whereas girls must be feminine, soft, emotional, sweet, and submissive. These traditional gender-role patterns have been influential in the socialization of children (Cornwell & Legersky, 2010). It is a gender role socialisation that leads females and males to behave differently, and the application of the traditional gender roles leads to many forms of differentiation between women and men. Apart from work, these gender roles are so pervasive in society and therefore leads to stereotyping which makes it difficult for males to work under the leadership of females (Alabi, Bahah & Alabi, 2013). --- Support services and strategies used by females to deal with patriarchal challenges in SHSs. In the face of patriarchal challenges to female heads in leading all boys' SHSs, studies have indicated that women's way of leading enables them to deal with some of the challenges they face as female heads. Given equal opportunities and co-operation, women's leadership styles tend to be more transformative and inclusive than that of their male counterparts. This makes females more capable to adopt a collaborative management approach to leadership than men (Aburdene & Naisbitt, 1992) which is preferable for today's schools (Anewu, 2010). This assertion is confirmed by Le (2011) when she said that, in spite of the difficulties with overwhelming huge workloads and domestic duties, women could be proud of their leadership because of the democratic and transformational leadership styles they embrace. This is one way that female heads use to deal with patriarchal challenges to enhance their performance. Mostly, female school leaders use various leadership styles and qualities to navigate their duties. These may include task-oriented, visionary, and participative leadership styles. These are types of leadership where the leader consults with the subordinates and embodies their suggestions in decision making (Okumbe, 1998). Female leaders are often too aware that when they use these types of leadership styles, they help to reduce the subordinates' stereotypes. This view is supported by Atta's (2015) argument that if a female leaders try to be more autocratic, the subordinates are more likely to complain because they expect women to be participative. Delegation of some duties goes a long way to break subordinates' stereotypes of women leadership. This may connote sharing responsibility with work colleagues. Through these types of leadership styles, the female leader can remain afloat in a masculine leadership world. Using their intuitive power, female leaders can transform the SHSs into centres of academic excellence (Kitele, 2013). --- METHODOLOGY The study is underpinned by the nominalist ontological view of reality which argues that social reality has no external existence such that it can be objectively and dispassionately accessed, but rather it is the result of human thinking (Burrel & Morgan, 1979). The study is also influenced by the constructivist epistemological view of reality which states that knowledge is constructed by subjects in their own different ways but not discovered. In line with these theoretical positions, the study adopted the interpretive paradigm which argues that social reality is created jointly through meaningful interaction between the researcher and the participants and in the socio-cultural context of the participants (Rugg & Petre, 2007). Consequently, a phenomenological research design was used to carry out the study. According to Pathak (2017), phenomenological research is a study that attempts to understand people's perceptions, perspectives and understanding of a particular phenomenon which has the primary objective to explicate the meaning, structure, and essence of the lived experiences of a person, or a group of people, around a specific phenomenon (Christensen et al., 2010). The research was conducted in 2 public male SHSs which are headed by females in Ghana where patriarchal culture is rife (the names of the schools and their locations are withheld in order to adhere to the ethical guideline of anonymity). The population of the study comprised of: all females in leadership positions in the 2 all boys' SHSs headed by females; all teachers in the selected boys' schools; and all final year students of the schools. The final year classes were used because they have more experience with the female heads than the first-and second-year students. The population of females in leadership positions in school A was three and school B were also three making a total of six. The population of teachers in school A was 134 and school B was 168 making a total of 302. The population of students in school A was 688 and school B was 946 adding up to 1634 students. The total population for the study was 1942. Out of this, maximum variation type of purposive sampling was used to select 30 participants from the categories due to their rich knowledge about the phenomenon of the study. This comprised 3 females in leadership positions (i.e., Headmistress, Assistant Headmistress and Female Head of Department), 5 teachers and 7 students respectively from each of the schools. The qualification for the selection of teachers was that she or he should have worked under the administration of the current female head for at least three years and above, and have knowledge of the heads' leadership experiences. The third category was students who were school prefects from the final-year class. They were so chosen for their experiences of their status as school prefects working with the female leaders. These students were put into two focus groups of seven students each. The six females in leadership positions and the 10 teachers were interviewed. So, the qualitative data was collected through semi-structured interviews and focus group discussions. Two separate interview guides were developed for each of the two categories of participants, and a focus group schedule for the students. The four components of trustworthiness: credibility, transferability, dependability, and conformability were observed in the conduct of this research. These components respectively ensured: a. accurate identification and description of the problem investigated in the study and also constructive feedback from respondents, b. the presentation of conclusive and vivid information concerning the research sites and rich descriptions of perceptions of selected participants, c. a detailed description of the research process including data collection and analysis procedures, d. presentation of the findings in a way that would facilitate their corroboration and confirmation by others. The data was analysed thematically. The coding was in four levels whereby the salient points were sifted out from the raw data. Cohen et al. (2011) described qualitative data analysis is an activity that involves organising, accounting for and explaining data. The data sets were triangulated to ensure authenticity of the data and to capture different dimensions of the same phenomenon (see Maree, 2010). All participants were told that taking part was voluntary. The researchers ensured the participants of anonymity and confidentiality (by referring to them with pseudonyms like: FT = Female Teacher; ST = Student; FH = Female Head; MT = Male Teacher), and fully informed them about the purpose and context of the study. --- Data Analysis and Discussion The data obtained for this study are analysed and discussed under two main themes: i. Socio-cultural barriers to female heads managing male senior high school The discussion in this section relates to data that was gathered to answer the first research question: "what are the socio-cultural experiences of female managers in the selected male SHSs?" The data revealed that, males are perceived to be in a better position to head boys' SHSs and that female heads are not very welcome in boys' SHSs. Indeed, the data showed that there is discontentedness among some staff concerning a female heading male dominated SHSs and the relevant role that female staff are supposed to play in boys' school. The Interview Respondent FH1, for example, said that, "there are factions within the school that detest females becoming head of the school. This breeds gender bias. However, the head promotes open administration and team work". The data revealed on the other hand that, students of male dominated SHSs are glued to female teachers' more than male teachers, and the boys have much confidence in their female teachers. They see female heads and teachers as surrogate mothers who empathise with their problems. From the FGD, Respondent ST1a said that: personally, I will prefer a female to head a male school than a male because with the female we feel safe. We can get closer to her to share our problems and she is always ready to listen to us because she is a female. With a male head, we feel a bit scared to approach them. We are not sure what the male head will think of us if went to them with our problems -what is he going to think? will he insult or judge me concerning what I am saying? With the female head, we actually feel freer and more open to tell her what worries us. Respondent ST3b concurred with this view: "I prefer a female head because our headmistress is caring. I am imagining that if a male comes, we will be treated like some military people. Even though the female head gives out punishments for offending students, I presume that a male head's kind of punishments will be very hash". These views means that, male students do relate better with female teachers which include female heads. The Interview Respondent FH5 said that, the boys are more comfortable and open up to female teachers especially the older ones than the male teachers. The focus group discussion data corroborated the interview data and indicated from FGD 'A' that, students of male SHSs prefer female to head their school rather than male heads. This suggests that, although female heads may need male assistance, boys cannot do without females, neither at home or in school, and that boys feel at ease in the presence of female heads than male heads. FGD 'B' also showed that boys prefer females to head male SHSs. The Focus Group Respondent ST5b for example said that, "having experienced females as heads of the school right from the basic school prove that females are helpful in leading male dominated schools". The Focus Group Respondent ST3a also said that,... just as boys feel the presence of our mothers at home and share our problems with them. Our headmistress is like a mother to us and I feel comfortable talking to her. Not like our fathers who are so busy and have no time for us. So, I prefer females to head boys' schools in order to feel the presence of motherhood even at school. Interview respondent ST2b said that, In terms of nutrition, at first, we used to eat little quantity of everything but now it has improved. Because she is a mother, she understands and has contributed to the improvement of our diet in school. Therefore, I will prefer a female headmistress in a boys' school and she can be supported by the male teachers. The data is consistent with a view from Ramsook et al. (2016) that women have a critical role to play in preparing boys for life in the 21st century. They see themselves as counsellors, and also believe that as surrogate mothers they have a responsibility to teach boys the importance of respecting women. They want boys to develop a good sense of self, leadership skills, strong moral and ethical values, and a keen sense of responsibility to others. Many of the participants also regard themselves as positive role models for boys. Then again, the data from two male teacher respondents suggested that, male teachers prefer females to head boys' SHSs because of their nice way of addressing boys' issues and their meticulousness in raising students from boys' SHSs. Interview Respondent MT6 said that, "unlike the male heads, females are meticulous in grooming boys. They do not overlook certain things like dressing, combing of hair, use of dormitories etc." Two female teacher participants and one male teacher participant agreed that, male and female teachers prefer either male or female who have the requisite skills to head boys' schools. Interview Respondent FT2 believes that, "the personality of the person in question is of utmost concern because experiences prove that some females can head male schools. However, in boys' schools where the students are known to have strong character, a male head would be preferred". The data further suggested that, majority of male teachers and few female teachers prefer a male to head boys SHSs. The later believe that due to their boisterous nature, male heads can intrude the privacy of male students without asking for permission. The Interview Respondent FT1 for example said that, "a female cannot head male institution in the sense that, unlike females, males are able to deal aggressively with their comales". This view is affirmed by Helterbran and Rieg (2004) that parents and community members believe that males can effectively discipline male students than women. Contrary to the views of male students, the data showed that, issues of gender stereotypes are prevalent among male teachers of boys SHSs., and they behave in gender stereotypical ways that confines the potentials of

This paper examines the advances made on gender issues in Ghana, particularly regarding how far misogyny is being watered down in educational leadership. The article thence reflects on how society might resolve the question of the effects of misogyny on educational leadership futures and the associated possibilities, and decisively address this enervating question in the decades ahead. The paper discusses how individuals and groups are resisting and challenging their experiences of gender inequalities, as well as attempting to intervene and correct the causes and consequences of gendered power imbalances. The paper does not intend to assess all the facets of gendered life and inequality. Instead, it discusses arguments that have been put forward regarding re-engineering gender spaces for educational leadership futures. Data obtained from two all-boys' schools suggest that gendered inequality is visible in both public and private spheres. However, the validity of the arguments that gender inequalities are still entrenched and persisting over time, place, and culture is contrasted with alternative claims from the data that gendered power relations, and for that matter inequalities, are gradually being eroded. Furthermore, the current academic focus on the concept of intersectionality regarding how variables like gender, tribe, religion, and culture intersect to define people's gendered experiences, is used as the analytical framework. The study of women's educational leadership experiences in this paper is used as a framework to further examine these issues and address some of the debates. The paper concludes with an argument that the study of inequality in relation to gendered identities, relations, and experiences must continue with an exploration of the study of men and masculinities if the theoretical analysis of gender is to be enriched to re-invent genders spaces for educational leadership futures. We also argue the view that to set the agenda for a more equal educational leadership future, society must see the need to link gender closely to other forms of social inequalities.

male and female teachers prefer either male or female who have the requisite skills to head boys' schools. Interview Respondent FT2 believes that, "the personality of the person in question is of utmost concern because experiences prove that some females can head male schools. However, in boys' schools where the students are known to have strong character, a male head would be preferred". The data further suggested that, majority of male teachers and few female teachers prefer a male to head boys SHSs. The later believe that due to their boisterous nature, male heads can intrude the privacy of male students without asking for permission. The Interview Respondent FT1 for example said that, "a female cannot head male institution in the sense that, unlike females, males are able to deal aggressively with their comales". This view is affirmed by Helterbran and Rieg (2004) that parents and community members believe that males can effectively discipline male students than women. Contrary to the views of male students, the data showed that, issues of gender stereotypes are prevalent among male teachers of boys SHSs., and they behave in gender stereotypical ways that confines the potentials of female heads. Some staff and (about 2% of) students perceived women to be weak for leadership in an all boys' SHS and therefore refuse to take directions from them. The interview Respondent FH5 said that "students do not portray any attitude of gender stereotyping but male teachers do, to the extent that they sometimes challenge the headmistress". The Interview Respondent FH3 added that, "most of the staff and some students perceive that women are weak. For that matter they refuse to heed to instructions from female leaders". The idea that male teachers find it difficult to work with female heads stems from the cultural backgrounds of such male teachers. According to the Interview Respondent FH1 "some of the male staff find it difficult to work with female leaders such that they, will indirectly refuse to do their work or do it haphazardly". The Interview Respondent FH6 said that, "some male teachers have the perception that the female head is just like their wives at home therefore, there is no need to be subordinates to them". Anewu (2010) believes that these attitudes are borne out of males' perception of women as arrogant. The data catalogued a number of labels used to describe women who enter occupations considered to be the preserve of men, such as: "woman tiger", "woman-man" and "iron woman". In spite of these views however, the data presents an idea that gender stereotypes in all boys' SHSs have no sufficient negative effect on the leadership of female heads in such schools because a female head who has the capabilities to head boys' school does not concern herself with issues of gender stereotypes but rather focuses on using every necessary leadership skill to prove her competence. Eddy ( 2009) has argued that female leaders thence have to be tough to overcome these perceived male dominant positions. Interview Respondent FH5 adds that "it is up to the female head to draw close and work with members of staff who are willing to help instead of allowing issues of gender stereotypes to influence the smooth running of her administration" because if left unchecked, these gender stereotypical issues will keep creating socio-cultural challenges for the female leader. The data has revealed that, there are socio-cultural barriers that prevented female heads from taking certain decisions and actions in boys' schools. Consequently, the female head may need the help of male staff to execute certain actions which are socially unethical for a female to do. Such actions indeed may tend to intrude the privacy of the boys. These socio-cultural barriers also inhibit female teachers from teaching some specific subjects. The Interview Respondent FH3 said that such socio-cultural issues tend to create performance gaps for the female head: There are decisions and actions you cannot take, not because you are not capable but because you are a woman. Women cannot do certain things in a male institution as entering the boys' dormitory unannounced. So, when it comes to inspection for example, the female head would fall on the male teachers to do the inspection. On a routine basis, these students think the female teachers including the head should not go into their bathrooms and rooms. However, this is a false notion because, even when we go round on inspection, the men will walk by and will not see some important details that the women will see. Once, I saw rubbish stacked under beds but the male teachers could not see it. Another thing that I noticed when I came to this school was, they preferred men teaching certain subjects like Mathematics, Science and even English Language which is a female dominated area. This wave has died down now because we have more women in this school than before. So also, Interview Respondent FH5 claimed that, Female teachers are not allowed to enter the boy's dormitories. When it comes to the home situation, the boys are closer to their mothers and we have a role to play in their lives. A point in time we were having assistant house mistresses but then the males were having problems with it because they felt it is a male thing so there is no need for us females to take part, so we were left out in the appointments. In spite of these socio-cultural barriers, the data also revealed male teachers' contentment for working under female leaders, contrary to the previous view. According to the data, this is further enabled by the statutes of most public institutions that determines who should be the head of an organisation. Five male teachers said that, once a male staff is able to live with a woman, be it a mother or a wife, he should be comfortable with a female leader. This was supported by three female teachers' assertion that male teachers have no repulsion about working with female leaders. What counts is equality and equity -considering shared responsibility rather than gender in leadership. The data indicated further that some religious beliefs also prevent women from holding leadership positions as they do not allow females to lead them. The data from the focus group discussion refuted the conventional view that, women are vulnerable and incapable of taking leadership positions, and that female heads could head boys' SHSs better than male heads. Experiences prove that women have the innate ability to head and manage institutions such as schools so long as they are able to manage their homes: In Ghana, women are perceived to be vulnerable, soft or second-class citizens. Meanwhile they are those who manage homes. It is without doubt that, so long as they are able to manage homes, they are capable of managing institutions such as schools and ensure development (Respondent ST3a). This affirms the notion that female leaders are very competent to lead boys' schools. They are strict in ensuring discipline and punishing students when necessary. The data showed that, "there is nothing like discrimination against women in the field of education" (Interview Respondent FH1). The embedding idea is that when it comes to the abilities and competencies for leading schools, gender issues do not become a factor. In education, quality and competence is what counts in choosing who should be a leader of a school including boys' schools. The Interview Respondent FH2 said that, "in educational leadership there is nothing like discrimination so long as the person in question qualifies. Rather, the discriminations emanate from socio-cultural backgrounds". Furthermore, the data showed that even where there are traces of discrimination, the heads have used team building or teamwork as a strategy to overcome it. This team effort involves engaging both male and female staff in all activities so that none will feel discriminated. The Interview Respondent FH1 for example said that, "the female head ensures all-inclusive type of leadership such that leadership roles are delegated to the men as well. By so doing, discrimination is reduced". The data further showed that the traditional roles of female heads affected their leadership roles positively than has been previously believed. The findings from four female head participants showed that, the traditional motherly role of the female head is beneficial to them and the students rather than the previously held believe that these traditional roles interfere with their leadership. Their argument is that their traditional roles and experiences as mothers enable them to give the students holistic care, that is -providing needs in terms of academics, extra-curricular activities and socio-cultural skills which are necessary for the upbringing of the male students. The Interview Respondent FH2 said that, "the motherly attribute of the female head enables the students to confide in us. It places no hindrance to our work as a head". This subject was however perceived differently by two female head respondents who indicated that, the motherly role of the female head at home interferes with their administrative work. The Interview Respondent FH6 for example said that, Traditional role as a mother of female leaders somehow interferes with their work in the sense that some of us heads who have children sometimes absent ourselves from school to attend to family issues. This is an issue with those of us who have little ones. Mothers who have grown up children do not have much problem. This assertion is affirmed by Mutabai (2016) who has said that conflicting roles are a major hindrance to women in leadership positions. This is because as the child-bearer, the woman is expected to take care of the child and put away her career to meet the needs of her family. This perspective notwithstanding, three male and one teacher participants said that, female heads do not place family demands over administrative work. Interview Respondent MT1 argued: "the assertion that female heads place family demands over administrative work is untrue. This is because sometimes the female head is able to work till midnight before they attend to any other business" while Interview Respondent FT2 said that, "the assertion that female heads place family demands above administrative work does not hold. Experiences with the female head prove she is always on the job and balances school and family time". Within the context of the data presented so far, the subject of gender discrimination within the realm of school leadership is approached here in a very specific and different way. We do not intend to explore the various normative frameworks that could support equality in the workplace, such as appeals to basic human rights, social contracts, deontological duties or utilitarian concerns. Instead, we seek to bring an understanding to the tacit gender support systems that are inherent in school practices and the embodied effects of such support systems for female leadership. The data has indicated that despite an overt acknowledgement of socio-cultural barriers, women in educational leadership still experience approval and enjoy support from their male colleagues as well as the students they superintend to carry out their duties. In this context, the so-called "glass ceiling" or as the metaphor has recently been recast, the "leaking pipe-line" (Price Waterhouse Coopers 2007), has been sealed significantly by the existence of organisational statutes and policies, as well as attitudes of male employees and loyalty from male students. Consequently, we seek to interrogate these support systems that encourage female leadership and the subtle gendered prejudices and expectations about how women lead that lie at the heart of the challenges many females face in finding their leadership role in schools, with the hope that these subtle prejudices will be acknowledged and work on to re-invent and advance gender spaces for educational leadership futures. The data has shown that these tacit support systems are partly institutionalized in everyday practices and partly structured by male preference and affinity to femaleness of leadership. These eventually shape individuals' embodied existence within the schools, and also embolden some women and some men to unpack the implications that the interplay between gender and organizational practices has for leadership, and then device alternative leadership models and gender inclusive strategies of change to be used in the schools. Within the feminist literature, there have been a number of approaches to understanding the differences between men and women and addressing matters of equality and opportunity which are the key tools for reinventing gender spaces for leadership. So, this paper tries to draw out the implications that each of these approaches could have for leadership futures. This will bring clarity to our beliefs that the origins and manifestation of sexual differences have implications for the thinking that goes into leadership role(s) of women and men in organizations, and certainly the fact that this is a sine qua non for reinventing gender spaces for increased productivity in schools. We make a claim at this point that the subtle traditional gendered prejudices and expectations about how women and men lead embed the challenges that many individuals face in finding their leadership role in organizations. The first of the approaches pertains to the subject of equality. In the leadership realm, the "equality" debate often challenges women to act as men have always acted in leadership or to do better. Consequently, women have had to use leadership practices that pertained to patriarchal organizations in which they find themselves. To achieve these goals for leadership, such female leaders often unconsciously imitate the predominantly male leadership stereotypes, and they end up making a mess of their roles in the sense that the male leadership stereotypes are very uncharacteristic of feminism. So, while equality-feminists succeed in making the argument for equal rights and opportunities, their efforts do not allow women to develop their individual leadership styles, and neither do they challenge existing stereotypes about leadership. Consequently, an alternative approach in feminism is to insist on respect for the differences between men and women, and an appreciation of the unique role that women could play in schools as explicated by the data about the special qualities they bring to all boys' schools. The data concurs with feminists who have adopted this approach that women have their own unique voices or styles that should be included within leadership discourses. From the perspective of the data, it is possible to argue that the unique capacities, traits and predispositions of women are functional in terms of supplementing the gaps that are typically present within the existing school leadership corps such as the need for surrogate motherly role in all boys' schools (see also Ely & Padavic 2007). This approach may be countered as problematic however, because it tends to set up essentialist dichotomies between men and women. Per the data, it shows women to be more caring, more communicative, and more cooperative than men. Rosener (2011) has commented that the International Women's Forum in 1984 used similar data but was seen as further attempt to solidify existing gender biases in their categorization of traits that respondents identified within themselves. In much of these surveys, female traits included being excitable, gentle, emotional, submissive, sentimental, understanding, compassionate, sensitive and dependent. Whereas male traits included being dominant, aggressive, tough, assertive, autocratic, analytical, competitive and independent. Being adaptive, tactful, sincere, conscientious, reliable, predictable, systematic and efficient were considered gender-neutral traits. This essentialist view point can be extended to present an unfavourable picture of female leadership. Gmür (2006) has argued that the unfortunate consequence of this essentialist approach is that women have always been associated with the inferior characteristic of the binary opposition: women are emotional, not rational, women are impulsive, not goal-directed, etc. His empirical studies suggest that most respondents regard the various stereotypical male leadership traits as typical of the behaviour of a "good manager". In that study, out of the number of ideal managerial traits, only two "feminine" traits are considered desirable for managers, i.e., being "adept at dealing with people" and "cooperative." All the other ideal traits, like being analytical, competent, confident, convincing, decisive, efficient, foresighted, independent etc. are associated with the male stereotype (Gmür 2006). So, these prejudices have been uncritically absorbed into the leadership discourses, and have led to the claim that feminist ethics essentially pursues "care ethics." Borgerson (2007) has commented on the problematic conflation between feminist ethics and care ethics within the organisational ethics literature. She argues that ethics textbooks like that of Crane & Matten (2004) describe "care ethics" as a feminine approach that solves ethical problems through intuition and personal subjective assessment. Even though Borgerson does not deny that certain articulations of care ethics display feminist concerns, she argues that the association of care ethics with feminism tends to essentialize the gendered experience, and hence, a proper understanding of the causes of gender prejudices and marginalizing practices is never developed. She also points out that there are other "caring" ethical approaches, which are not at all feminist in orientation, such as that of Emmanuel Levinas and other philosophers working on what can be described as an 'ethics of proximity'. These debate notwithstanding, the data has been emphatic about the relevance of female traits to leadership in boys' schools and this makes a great case for creating gender spaces for educational leadership futures. Orbach (2023) has also on another hand described the masculine traits of leadership as toxic masculinity and has consequently called for more deeper thinking about the phenomenon. She has argued vehemently that toxic masculinity has a certain usefulness and punch as a phrase because it expresses what some men put out into the world but it does not address the whys deeply enough. She has explained that until recently, and still to this day in many parts of the world, men are raised to be protectors -fighters and economic providers. She has argued that conscription (which refers to men being trained to kill) ended even in Britain only recently in 1960. Women, meanwhile, were being raised to be nurturers and carersthat is, to be midwives to their needs, to support their initiatives, whether or not the women worked also outside the home. Furthermore, the social systems that emerged as a result ensured that men were to be receivers of emotional support, and women to be receivers of economic support and "protection". However, the 1970s saw the beginnings of enormous changes between women and men, which has led to today's expanding notions of gender, which the paper sees as opportunity for reinventing gender spaces for educational leadership futures. --- Support services and strategies used by females to deal with patriarchal challenges The data presented and analysed in this section sought to answer research question 2 regarding the support services and strategies that female managers use to deal with the patriarchal challenges in the selected male SHSs as contribution to efforts to reinvent gender spaces for educational leadership futures. The support systems and strategies used may be categorised into: --- External support services for female heads The data showed that the successes of female heads in the all-male SHSs have made substantial achievement as a result of the support systems available. For example, the Ghana Education Service's (GES) policies are not gender biased which means that positions come with requirements that do not recognise the gender divide. These policies are open to both males and females except for the conventional maternity leave reserved for females. Interview Respondent FH6 said that, "there is nothing like special support specifically for females in leadership. Rather, there are equal opportunities to both male and female in management". The Interview Respondent FH5 added that, "except for maternity leave, all policies from GES are open to and cater for both male and female leaders". Such systems prevent female heads from sustaining needless gaps and stress in their careers that may tend to weaken their position and slow down their effectiveness as school leaders. The teacher participants also indicated that, the best support system available to female leaders to discharge their duties comes from the school and its staff. These encompass proximal accommodation, her office, adequate support from Housemasters, Board of Directors and old boys. The Interview Respondent FH2 said that, "there is a housing facility for the female head on campus" while the Interview Respondent FH3 said "House Masters work hard to support her". The above notwithstanding, the data revealed that female leaders are not given any special training to lead boys' schools. Interview Respondents FH1, FH2 & FH3 shared that there is no special training for female leaders to facilitate their duties in all-male schools. --- Internal support services for female heads The data obtained from female head showed that, female heads feel secured around males. In the patriarchal context, female heads feel much secured when men are around. Interview Respondent FH1 for example said that, "there is security... and no threats so long as the staff and the students are concerned". Again, the data showed that, the presence of people and respect from them provides security. Another way of ensuring security for the female head was described by Interview Respondent FH2 as "some male teachers make certain vital information available to me before any major happening takes place in the school. Such practices make the female head feel secured". The data again showed that support is given from parent of students and alumni which are very crucial to female headship and the entire school. The support from parents and alumni manifests in the commissioning of supplementary projects and services to those from government in the schools. The Interview Respondent FH1 said that "though government does not demand for parents' supports, the parents and the alumni provide financial support for the beautification of the school". The data from six male teacher participants indicated their support for the female leader, which confirms the notion that, a well-organised institution unearths the zeal of the staff to support leaders. In this circumstance, most members of staff are willing to find the opportunity to render support to female leaders in a well-structured environment. According to the Interview Respondent MT1, "teachers follow the rules and the regulations in the school. They also ensure that students are at the right place, at the right time. We follow duty rosters to enable the various leaders to contribute their part to the headmistress's administration". The Interview Respondent MT3 also said that, "some staff take up duties willingly". So, staff adherence to the code of conduct of the school is a great support to female heads which is complemented by picking up various roles (both academic and non-academic) in the school. The focus group discussion data from students corroborated the interview data regarding the support services the school provide for female leaders. This data set revealed that, provision of basic needs for female heads boosts their productivity. The school provides the basic needs of the female head in terms of shelter, security, transportation, sense of belonging and respect. The Focused Group Respondent ST3a emphasised this view that, "the school has provided the headmistress with a car, driver and security to make her work much easier". --- Policy direction for support provision for female leaders The data further gave indications for policy direction regarding gaps in the support provision for female leaders. The female heads indicated that, periodic training sessions is a key gap in the support services available to them, and that such training sessions will help to reduce the leadership challenges they face. The idea is that such periodic in-service trainings will boost female leaders' morale to face the challenges and manage activities competently. Interview Respondent MT6 claimed that, "the GES must organise counselling and training programmes on how to handle patriarchal challenges for female heads in male institutions". Mutabai (2016) confirms this view with his claim that, top management support is an important factor that contributes to women's career success. These views show that organisational policy interventions have the tenacity to contribute to the success of female leadership. Interview Respondent FT2 said that, "newly appointed female heads must be taken through orientation by the previous head before they assume work. Also, the new female head ought to consult the former head for guidance. She ought to consider the view of others and improve human relations". The focus group discussion revealed that, the success of a leader lies with the staff because the female heads cannot lead in isolation of help from staff. Hence the female head within the context of organisational statutes must be encouraged to delegate responsibilities to the staff. The focus group respondent ST1a said that "... the head must be supported in making and implementing decisions" while the Focus Group Respondent ST1b said that, "Staff must support the head in enforcing and checking discipline". Again, the data indicated that, acceptance from students is vital to the administration of female heads. Inasmuch as female heads may not work directly with students, it is important she receives their acceptance. "Students make the head feel comfortable and accepted" (Focus Group Respondent ST1a). --- Strategies used by Female Heads to deal with Patriarchal Challenges The data further indicated some strategies that female heads may use to deal with the patriarchal challenges they face. The female head participants revealed that, leadership does not thrive on hearsays and therefore any information that comes to the leaders' table must be checked before it is acted on. This enjoins female heads to be discreet and remain composed about information made known to them. Interview Respondent FH1 said that, "the head must not act based on what people say. She ought to portray positive attitude towards everything she faces. She must foster team work and give no room for divisions. She ought to investigate issues properly before she acts on them". The data showed that, female heads are references for reinforcing good attitudes among staff. This means for example that, one cannot declare war and have peace, therefore in order to enjoy the peace they seek, they should be ready to harness peace. Thus, it is incumbent on female heads to create for themselves and their staff a conducive working environment devoid of tension. So, Interview Respondent FH5 said that, as female heads, we should foster cordial relationship among everyone especially the male staff who feel they are in the best position to occupy headship position due to experience and age. In such cases, I give them certain responsibilities to make them feel part of the administration. The Interview Respondent FH4 added that, "female heads must avoid tension and seek to be nice to everyone". Again, the female heads indicated that, subordinates complement female headship, implying that a leader cannot do all the work alone. Subordinates may hold the other half of the missing pair of the solution being sought. Female leaders therefore ought to delegate roles to subordinates and coordinate and monitor their timely outcomes in relation to set goals. Set targets must be communicated thoroughly, respected and adhered to by both female leaders and subordinates. Atta (2015) has argued in corroboration with this finding that if a female manager tries to be more autocratic, subordinates are more likely to complain because they expect women to be participative. Delegation of some duties goes a long way to break subordinates' stereotypes of women leadership. This will involve sharing responsibility with work colleagues. Other studies by Elisha and Edwards (2014) affirmed that the success of their schools was dependent on the female head delegating responsibilities, especially when working with males that come from a patrilineal background. They felt that by delegating tasks, men gained some legitimacy in their roles and it helped them to feel that they could contribute to some area of leadership within the school. The teachers indicated that, fun activities ease tension amongst female heads and staff, so it is important to engage in periodic fun or tension releasing activities which in the long run tend to foster healthy relationships among people in the school. Thus, the quest to live cordially with staff, motivate them and strive to get them involved in leadership by creating spaces for their uniqueness and individuality, while spending quality time with them outside work hours tends to ease challenges of patriarchal culture. According to Interview Respondent MT4 "the female head ought to ensure cordial relationship amongst staff, tolerate staff for their unique characters and not to victimize those who pose problems". The Interview Respondent MT1 added that, "the female head ought to be a motivator and a team player. There must also be a time out of work for refreshment". Again, the teacher participants indicated that, female leaders ought to shun dictatorship. This means that, subordinates prefer and are open to leaders who involve them in their leadership rather than those who do not listen but impose their decisions. The Interview Respondent MT5 claimed that, "the female head must become a servant instead of being bossy. Also, she must not create any scenario of gender discrimination but rather follow the rules". The data further showed that, staff and students need counselling and orientation on female headship in a patriarchal context so that they can render the maximum support needed. Interview Respondent FT1 argued that "there must be counselling sessions or annual retreats for students and staff where they would be educated on the ideal that females are equally capable of leading". The data further revealed that, it is important for female heads to pay attention to their professional development in order to develop skills for competency as confirmed by Interview Respondent FT3 that, "as a leader, the female head must build herself and also ensure cordial human relationship for teachers and students". The data again shows that, empowered staff are complements to female heads. A leader who involves staff in duties eventually replicates another leader who can act on their behalf when indisposed. Female heads should therefore be instruments for staff empowerment based on their potentials. Interview Respondent FT4 indicated that "the female head must empower the staff and involve them in the administration by giving them tasks apart from teaching, considering their strength, weaknesses and talents". The data indicated that, patriarchs and matriarchs are reference points for leadership so female leaders ought to consult their predecessors and local education institutions while working collaboratively with their staff. The data explains that female leadership practices should not be characterized by masculinity as some women leaders tend to behave in very masculine ways to prove their competence for the job. The focused group discussions converged with the interview data on the subject of building good working relationships. This data set agreed that most subordinates tend to feel comfortable around leaders who come down to their level to interact with them, as argued by Focus Group Respondent ST1a that, "in order to ensure cordial relationship, the female leaders ought to be meeting the students intermittently to open up issues to them and to hear from the students as well". Again, the data showed that, female leaders are to show bravery in order to lead boys' school. Women are known to be soft. However, each position comes with its own required attitude and posture. For example, boys need to be brought up in a way to prepare them to meet the rigours of life. This cannot be done by the natural soft and pampering posturing of women. Thus, female leaders need to be trained to live up to the difficulties that boyhood impose. Respondent ST1b said that, "students, especially boys need hard training to enable them to cope with difficult times in life. Therefore, the female heads must be trained to suit such task". So, female heads must be trained to be both flexible and firm to stand the pressures of patriarchy. The data consequently indicated that, diverse opinions are good for female leadership. However, when it becomes a habit, it makes one looks weak and not firm as a leader. So female heads should maintain a good balance of consulting others especially men, but they also ought to be firm in taking decisions. Respondent ST1b confirmed this view by saying that, "though the female head would have to consult others in decision making, she must however, be firm in taking decisions". From the point of view of policy directions as prescribed by the data, we want to examine whether it is possible to transform leadership theory and practice through an engagement with the variety of ways that have been recommended for both men and women to approach their leadership roles and how these will inure to reinventing gender spaces for educational leadership futures. What seems to be required is leadership models that allow individuals to lead in their own unique ways, instead of conforming to some pre-conceived gender expectations as discussed earlier. We therefore explore the theoretical models of leadership that may create a framework for gender spaces through understanding and adopting uniquely individual leadership styles. In the process, we shall recast important leadership notions, such as "authenticity" and "vision," in more genderinclusive terms for educational leadership futures. Werhane and Painter-Morland (2011) have related recent developments in relational leadership and complexity leadership to the way women lead in organizations. They emphasise that even though complexity leadership seems to describe leadership styles that are associated with the socially constructed "feminine" style of leadership, it is a model that suits many men's leadership preferences as well. These discourses make it relevant for us to explore at this point, and reconstruct a new model of leadership for future growth of educational organisations. Such a model draws on systemic leadership, which is not necessarily restricted to individuals appointed to positions of authority. It is rather an important departure from the 'great man theories' of leadership, and their associated sexist assumptions. Systemic leadership is informed and supported by a variety of discourses that range from Peter Senge's work on organizational learning and change (1999) to Karl Weick's sense-making theories (1993; 1995). The basic assumption of systemic leadership is that an organization cannot properly learn, change or create meaning without the sharing of information and cooperative agreements. In this sense, Senge and Kaufer (2000) speak about "communities of leaders," while others make reference to "distributed leadership" (Friedman 2004), or relational leadership (Maak & Pless, 2006). An influential definition of systemic leadership is provided by Collier and Esteban (2000) who describe leadership as "the systemic capability, distributed and nurtured throughout the organization, for finding organizational direction and generating renewal by harnessing creativity and innovation." Understanding leadership as an emergent, interactive and dynamic phenomenon allows one to distribute leadership responsibilities and privileges throughout an organization's workforce (Edgeman & Scherer, 1999). Systemic leadership therefore involves a number of different leadership dynamics. Uhl-Bien et al. (2007) explain these leadership dynamics as 'administrative,' 'adaptive' and 'enabling' leadership. Administrative leaders play the more formal leadership roles of planning and coordinating organizational activities. It is important to note that though systemic leadership functions are understood in more distributed terms, this does not necessarily mean that formal leadership positions and hierarchies become redundant or have to be disbanded. Indeed, it is very important that gender-sensitivity is encouraged through main managerial responsibilities like performance targets setting, conduct of performance reviews, and engaging in mentoring activities. So, it is necessary for people appointed to formal leadership positions to be gender-sensitive and actively play a role in thinking through the gender repercussions of the decisions they make in the organisation. Furthermore, the role of administrative leaders connotes real change in practices and belief systems which requires the acknowledgement and nurturing of other leadership roles. This sets the platform for 'adaptive' leadership functions that creates a 'collaborative change movement' that enables adaptive outcomes to emerge from multiple sources due to the dynamic interactions of interdependent agents. Consequently, the direction and priorities that focusses the organization's efforts and activities develop unconsciously as an unanticipated and unexpected consequence of the routine interactions between many different members of the organization rather than stemming from the top of the managerial hierarchy. This approach to leadership creates spaces that enable all members of the organization to take initiative and responsibility (show of leadership) as and when the situation calls for it. These spaces permit the staff to garner their innate strengths and abilities to lead in their own and unique ways. In essence, these features of adoptive leadership show logically that this model is opposed to stereotypical leadership behaviours. It rather calls for unique responses that are suited for specific situations and set of relationships. In this respect, it creates spaces for women leaders to grow their own unique style of leadership. The third leadership dynamic is what Uhl-Bien et al. (2007) refer to as that of 'enabling leadership', which provides the catalyst to facilitate the emergence of adaptive leadership within organizations. It often involves a complex interplay between administrative and adaptive leadership. Enabling leadership usually require some form of authority, but it entails an active involvement in the boundary situations that organizational members are faced with. Enabling leaders tend to engage in cooperative strategies, foster interaction, supports and enhances interdependency and stimulates the adaptive tension that creates the gender spaces for the emergence of new patterns of leadership. The data has shown, and this is corroborated by Uhl Bien et al. ( 2007) that all three leadership roles necessarily coexist within the two schools. The question that remains however is how adaptive and enabling leaderships can be acknowledged, recognized and rewarded within these educational organisations. Nevertheless, the systemic leadership model is important because it provokes discourses to rethink certain leadership stereotypes that are often uncritically perpetuated within organizations and are inimical to leadership development. This model provides the raison d'etre for rethinking authenticity in the context of educational leadership futures. The concept of 'authenticity' is here connected with the consistent way in which an individual acts in accordance with his or her personal traits and beliefs. However, this can lead to some measure of inflexibility that makes the individual incapable of adapting to different or emerging situations and relationships in practice. Systemic leadership in this context calls for another understanding regarding leadership roles due to the fluid nature of leadership responses. This view is very much discussed in contemporary leadership literature. Porras et al. (2007) discusses the issue so

This paper examines the advances made on gender issues in Ghana, particularly regarding how far misogyny is being watered down in educational leadership. The article thence reflects on how society might resolve the question of the effects of misogyny on educational leadership futures and the associated possibilities, and decisively address this enervating question in the decades ahead. The paper discusses how individuals and groups are resisting and challenging their experiences of gender inequalities, as well as attempting to intervene and correct the causes and consequences of gendered power imbalances. The paper does not intend to assess all the facets of gendered life and inequality. Instead, it discusses arguments that have been put forward regarding re-engineering gender spaces for educational leadership futures. Data obtained from two all-boys' schools suggest that gendered inequality is visible in both public and private spheres. However, the validity of the arguments that gender inequalities are still entrenched and persisting over time, place, and culture is contrasted with alternative claims from the data that gendered power relations, and for that matter inequalities, are gradually being eroded. Furthermore, the current academic focus on the concept of intersectionality regarding how variables like gender, tribe, religion, and culture intersect to define people's gendered experiences, is used as the analytical framework. The study of women's educational leadership experiences in this paper is used as a framework to further examine these issues and address some of the debates. The paper concludes with an argument that the study of inequality in relation to gendered identities, relations, and experiences must continue with an exploration of the study of men and masculinities if the theoretical analysis of gender is to be enriched to re-invent genders spaces for educational leadership futures. We also argue the view that to set the agenda for a more equal educational leadership future, society must see the need to link gender closely to other forms of social inequalities.

s the gender spaces for the emergence of new patterns of leadership. The data has shown, and this is corroborated by Uhl Bien et al. ( 2007) that all three leadership roles necessarily coexist within the two schools. The question that remains however is how adaptive and enabling leaderships can be acknowledged, recognized and rewarded within these educational organisations. Nevertheless, the systemic leadership model is important because it provokes discourses to rethink certain leadership stereotypes that are often uncritically perpetuated within organizations and are inimical to leadership development. This model provides the raison d'etre for rethinking authenticity in the context of educational leadership futures. The concept of 'authenticity' is here connected with the consistent way in which an individual acts in accordance with his or her personal traits and beliefs. However, this can lead to some measure of inflexibility that makes the individual incapable of adapting to different or emerging situations and relationships in practice. Systemic leadership in this context calls for another understanding regarding leadership roles due to the fluid nature of leadership responses. This view is very much discussed in contemporary leadership literature. Porras et al. (2007) discusses the issue so succinctly thus: great leaders are too aware that over time, their role may change in a manner that, the person who takes direction from you and falls under your supervision today, on another day, may become your boss. This same individual could even become your client over time. These issues make it necessary to preserve organisational relationship as a kind of 'virtual team' even as roles change (see Porras et al., 2007). This calls for authentic relational responsiveness, connoting that in order to be authentic, individuals have to respond appropriately and realistically to the substantive situation. This logically calls for the awareness and acknowledgement that reality in terms of the relational dynamics between people in an organizational context is not static. It is always complex and dynamic. The data has shown that both students and staff do not want to see their female heads mimicking stereotypical male leadership styles, or conforming to tacit expectations regarding how leaders ought to talk, walk and make decisions. Female leaders who do that are perceived to be 'inauthentic'. The irony associated with this phenomenon is that women are condemned whether they do it or not. If they conform to the male leadership stereotype, they are seen as inauthentic, and if they do not, their leadership is either not recognized at all, or considered inferior to that of men (Eagly & Carli 2007). This provides the rationale for re-examining the meaning of authenticity. The point is that women can respond quite authentically to the tacit expectations that embed one situation and resist these same expectations in a different situation. These experiences do not express a lack of authenticity. It is rather mirror's the institutionalized prejudices that constantly face women, and the strategies that individual women adopt to deal with them. So, it is imperative for organizations to pay attention to these issues in order to get the understanding of the tacit practices of discrimination that inform the relationships between their staff, and to look for ways to change them to reshape educational leadership futures. From the purview of educational leadership futures, it is necessary for individuals to draw on their own strengths, sensibilities and perspectives as well as using their own unique styles to navigate their leadership roles as tool for reinventing gender spaces. Intersectional approach to policy for reinventing gender spaces for educational leadership These exigencies for systemic leadership make a case for integrating intersectional theory in policy making that seek to reinvent gender spaces for educational leadership, and it is directed at one key question: just how interconnected are inequalities across different sites of social and organisational life in these all boys SHSs? According to Crenshaw (1991), the focus of an intersectional approach to policy making is to highlight the importance of recognizing multiple grounds of identity when considering how the social world is created. For the purposes of this paper, the 'intersectional approach' can be best understood as a way of identifying, understanding and tackling structural inequality in the schools' context that accounts for the lived experience of female leaders with intersecting identities. For example, intersectionality helps us to understand how female heads experience leadership in the 2 male schools, as a result of their identities and unequal power dynamics. Understanding these differences facilitates the development of more effective policies that tackle structural disadvantages experienced by these female school leaders, and help to reinvent spaces for female leaders' collaboration with their male staff for effective leadership. A key feature of an intersectional approach is that it does not give a higher status to any one inequality or experience of discrimination. Hankivsky et al. (2014) have claimed that: intersectionality encourages critical reflection that allows researchers and decision makers to move beyond the singular categories that are typically favoured in equity driven analyses (e.g., sex and gender in sex and gender-based analysis)... to consider the complex relationships and interactions between social locations. Scott-Villiers and Oosterom (2016) have said that adopting an intersectional approach in policy making also calls for policymakers and analysts to "analyse their own power dynamics as much as the world they wish to change". This means that the intersectional approach is not just about considering the experiences of those at the intersection of multiple characteristics but, it is about policymakers and analysts assessing their own experiences and how this impact on their ability to develop, deliver and evaluate policies in an equitable way. Furthermore, the intersectional approach in policy making requires evidence to be adduced for the context, including the historical and contemporary structures of inequality in wider society, and within local contexts (Christoffersen, 2017). The local contexts for the female school leader may include, the school, the District Education Office and the community in which they live and work. So, it is important to examine the dynamic interaction between female school leaders and institutional actors such as the Ghana Education Service (GES) staff in the district and the schools as these provide a more comprehensive examination of policy direction for reinventing gender spaces for educational leadership futures. Hankivsky (2012) has outline a number of useful approaches towards the application of intersectionality in the policymaking process. Recent years have seen criticisms from academia and gender activists that gender inequalities still persist in public spaces in Ghana, and the approaches used so far for leveraging gender inequalities in the country have been limited to the use of a separate single characteristic approach to policymaking. Hankivsky (2012) believes that such phenomenon necessitates the adoption of intersectional policy making approach which explores major cross-cutting policy agendas aimed at reducing gender inequality to reinvent gender spaces for educational leadership futures. To achieve this, policy makers will have to use the intersectional approach to analysing the genderleadership problems and developing solutions for them. Equally important in this regard are the opportunities to use the approach to also monitor and evaluate the resulting leadership futures outcomes. --- Concluding with Accelerating the Shift to Reinvent Gender Spaces This paper has argued that the origins of discriminatory practices in educational organizations lie hidden in everyday practices, habits and interactions. This emphasise that gender stereotypes are alive and well in such organizations, and that addressing the prejudices is fundamentally a difficult task. In the first place, one has to acknowledge the ingrained social practices and beliefs about the capabilities of both men and women, which play a role from a very early age and are solidified in the workplace. To address these prejudices, society has to start rethinking the feedback and advice it provides to children and students in the course of their early development and education on this subject. Furthermore, organizations have to also create new role models and look for mentors who have found their own unique leadership styles, while realizing at the same time that society and its embedded systems are constantly shaped and reshaped through thoughts and everyday workplace practices. These ideas will lead to creating spaces within which different types of leadership practices could emerge. The analysis has shown that systemic leadership models allow for a variety of leadership roles and styles to coexist in an organization. The difficulty however pertains to acknowledging these various roles, and making sure that they are duly recognized, accommodated and used. Surmounting this difficulty will result in unearthing a lot of inspiring stories about people's authentic responses to organisational challenges, and also realise how both men and women intuitively come across visionary ideas and practices in and through their engagement with others. This will thrive in organizational environments where workers are free to become the kind of leaders that infuse their organisations with creative new solutions and practices. This plausibility of constantly becoming a new kind of leader will enable both men and women to explore the full dimensions of their innate capacities, and enable them to serve their organizations, themselves and the broader society to the best of their multiple capabilities. Based on the data and the arguments put forward so far, our concluding contention is that, we can usefully focus on three aspects of gender spaces to continue to address the leadership issue of gendered inequality in innovative and more fruitful ways. These are: to further the contemporary debate and emphasis on intersectionality in relation to gender inequality; to highlight the increasing academic focus on masculinity and gender relations; and to rethink the roles males can play to achieve more successes for advancing female leadership futures. Thus, in addressing this gendered inequality issue in more innovative ways, it is useful to advance the contemporary discourse and emphasis on intersectionality in relation to gender inequality and to highlight the increasing academic focus on masculinity and gender relations and its link to feminism. This strategy for addressing the issue will tackle the challenges of gender spaces by getting to the structural and institutional causes of inequality. One key method to do this is the acknowledgement of many, interconnected inequalities, as well as having an openness to work with diverse kinds of partners in a variety of ways, especially males. Consequently, Ghana's policy makers, academics, and gender activists will be showing a positive sign towards building good gender spaces by constantly thinking through the possibilities of an intersectional approach in different contexts, despite the complex issues this raises. We have argued with evidence from the data that there is in the making, an educational leadership shift within the context of misogyny. But this is happening more slowly than is necessary due to what the data pointed out to be the embodied and institutionalized roots of the socio-cultural barriers in the workplace. There has been an unexamined assumption or perhaps a mantra that lies at the heart of this shift -the idea that females in leadership positions in the educational sector have been receiving quantum support and collaboration from their male colleagues at work, and this is a necessary condition for productive female leadership futures. Thes slow pace of this shift further shows that simply placing women in leadership positions is not enough to re-invent gender spaces for educational leadership futures in Ghana. Social attitudes matter a great deal. If female leaders are treated with social disdain, and if derision and hostility lower their status despite their title, women will be deterred from their jobs. We therefore conclude with some recommendations to speed up the culture change that will propel women into shared leadership with men and sustain the needed shift. These are: a. Speaking out together: More male leaders need to take responsibility for driving this changeand women must help by welcoming males into the debate by connecting male and female leadership networks. b. Active intervention: organisations need to engage in active interventions such as using quotas or targets to force change where resistance is greatest. c. Transparency: Transparency is a powerful stimulus to change, and this calls for clear data that no none can dispute to make arguments for change to be accepted. So, the paper concludes with a view that the study of gender spaces for leadership in relation to gender inequalities, relations, and experiences must continue with the relatively recent exploration of intersectionality if the theoretical analysis of leadership futures is to be enriched to reinvent the concomitant gender spaces. We seek to emphasise that, to set the agenda for a more equal leadership futures, society and organisations will need to link gender much more closely to other social inequalities found in religion, culture, society and organisations. In order to achieve all of the above, Kaloski-Naylor (2017) reminds us that "... we need wider visions of resistance, ways out of the 'to and fro' of arguments which seem to move us closer to disaster. This is what thinkers can offer...." Kaloski-Naylor's view enjoins academics, practitioners, and activists continue to find new ways of speaking to each other on the issue of gender inequality in the effort to reinvent gender spaces for leadership futures. In so doing, as we corroborate with Robinson (2017) to argue that, the boundaries between academia and academics, civic and political institutions, and those who construct knowledge outside of such institutions, including activists in everyday life, that have become more fissured will have to be blurred by necessity.

This paper examines the advances made on gender issues in Ghana, particularly regarding how far misogyny is being watered down in educational leadership. The article thence reflects on how society might resolve the question of the effects of misogyny on educational leadership futures and the associated possibilities, and decisively address this enervating question in the decades ahead. The paper discusses how individuals and groups are resisting and challenging their experiences of gender inequalities, as well as attempting to intervene and correct the causes and consequences of gendered power imbalances. The paper does not intend to assess all the facets of gendered life and inequality. Instead, it discusses arguments that have been put forward regarding re-engineering gender spaces for educational leadership futures. Data obtained from two all-boys' schools suggest that gendered inequality is visible in both public and private spheres. However, the validity of the arguments that gender inequalities are still entrenched and persisting over time, place, and culture is contrasted with alternative claims from the data that gendered power relations, and for that matter inequalities, are gradually being eroded. Furthermore, the current academic focus on the concept of intersectionality regarding how variables like gender, tribe, religion, and culture intersect to define people's gendered experiences, is used as the analytical framework. The study of women's educational leadership experiences in this paper is used as a framework to further examine these issues and address some of the debates. The paper concludes with an argument that the study of inequality in relation to gendered identities, relations, and experiences must continue with an exploration of the study of men and masculinities if the theoretical analysis of gender is to be enriched to re-invent genders spaces for educational leadership futures. We also argue the view that to set the agenda for a more equal educational leadership future, society must see the need to link gender closely to other forms of social inequalities.

INTRODUCTON In Mexico, abortion law is determined at the state level and access to legal abortion is available only under narrow indications in 30 of Mexico's 32 states. 1 In 2007, first-trimester abortion was decriminalised in Mexico City, followed by Oaxaca in 2019. The Mexico City Ministry of Health (Secretaria de Salud) operates a public sector first-trimester abortion programme, known as the Interrupción Legal de Embarazo (ILE) programme. Abortion services are available to anyone who presents for care at no cost or on a sliding scale for residents outside of Mexico City. 2 3 Since 2007, the ILE programme has provided over 225 000 first-trimester abortions and approximately 30% of clients have travelled from outside Mexico City, with little change in this proportion over time. 3 Previous work showed that ILE clients who were unmarried, with less than a high school education, and who resided outside of Mexico City experienced difficulties travelling to ILE facilities. 4 Evidence also shows that access to abortion under legal indications, or 'causales', remains very limited --- Key messages <unk> Clients travel for abortions from all Mexican states to Mexico City indicating unmet need for abortion services. <unk> Women from more vulnerable communities are not travelling to Mexico City to obtain public sector abortion services. <unk> Abortion law allows access to safe abortion in Mexico City but creates disparities in access to safe abortion services for women in vulnerable communities. Original research in public facilities outside of Mexico City, 5 6 making travel for services an important research focus. Little is known about the community contexts ILE clients come from. Previous work has either used very large geographical units (regions or groups of states) 7 or has focused only on the Mexico City metropolitan area 8 ; it suggested that those who could travel to access services had higher levels of education than the average population in their home geographical area. The geographical distribution of ILE clients at the community level is unknown, as are community-level characteristics of places ILE clients come from and places with no ILE clients. Community-level marginalisation influences healthcare access 9 and varies at the municipality level, with higher marginalisation observed in the south and in rural areas. 10 The purpose of this study was to describe individual and community (municipality)-level factors among those who travel from outside of Mexico City to obtain abortions in the ILE programme. We hypothesised that (1) clients who access ILE services come from geographically and socioeconomically diverse municipalities, (2) that municipalities that are home to ILE clients are different from municipalities with no ILE clients on key markers of socioeconomic status and (3) we further explore (individual and contextual) factors associated with travel for abortion services from more marginalised compared with less marginalised municipalities. --- METHODS --- Data and variables We conducted a retrospective study using publicly available individual records from the ILE programme 11 and publicly available municipality-level data from the census. Our data include all individual-level client records from all 14 ILE sites that provided services from 2016 to 2019. In this analysis, we include only individuals who reported living outside of Mexico City (32.5%; 21 629/66 462). We leverage publicly available municipality-level data to provide community contextual information for individual ILE clients. A municipality is a second-level administrative division (below a state) that varies in size and is similar to a county in the United States. 10 In 2010, the last available national census at the time of the study, there were 2457 municipalities in Mexico, including the 16 municipalities in Mexico City. Each municipality has a unique code; using the client's municipality code, we merged in municipality-level information from the most recent available census (2010 complete or 2015 intercensal survey) acquired from multiple publicly available sources (online supplemental table 1). Thus, if two clients came to the ILE programme from the same municipality, they were both assigned the same value for each municipality-level variable. Nearly 12% of clients (2882/24 511) were missing a municipality code value and were excluded from the study because we do not know where they travelled from. The majority of excluded clients travelled from the central region (87.8%) as determined by their state of residence, but otherwise there were no patterns in missingness of municipality data (online supplemental table 2). Our outcome is municipality-level marginalisation from the census, merged into individual ILE records. In Mexico, community marginalisation is measured by a marginalisation index, 'grado de marginación', a standard measure used by the Mexican government 12 that includes composite measures of education, income, household materials, and the proportion of the population that is rural (<unk>2500 inhabitants). This index is divided into quintiles where higher values indicate higher marginalisation (more vulnerability). We also created a binary variable indicating higher marginalisation (top four quintiles) versus low marginalisation (the bottom quintile). We chose these cut-offs based on the distribution in the data where the majority of clients (81.9%) came from the least marginalised municipalities (bottom quintile). At the individual level, we extracted additional data from the ILE client record. We included age (12-17, 18-24, 25-29, 30-39, 40+ years; REF 18-24); we chose to cut-off our adolescent category at 17 because women aged under 18 years must have parental or legal guardian consent. 13 We included education level (primary, secondary, high school, university); marital status (single, married/cohabitating or divorced/ widowed); occupation (ama de casa: works at home/ unemployed, employed or student); number of pregnancies (1, 2-3, 4+); and municipality and state of residence. We created a variable to classify states into regions (North, Central, South). We included the municipalitylevel adolescent fertility rate, divided into quintiles and also collapsed into a binary variable indicating high (fourth and fifth quintiles) or moderate/low (first through third quintiles). This binary variable represents values above and below the median. We also included the following commonly used indicators of municipality socioeconomic status: whether <unk>2% of the population aged 3 years or older spoke an indigenous language; if >75% of households owned a washing machine; if >30% of households had in-home internet; the proportion of the female population with at least 9 years of education; if <unk>40% of adolescents in the municipality did not attend school; and if >30% of the female population was economically active, defined as females aged 12 years and older who worked or looked for work in the reference week. We created these binary variables based on data distributions (cut-off at the median) except for 9 years of schooling for females, which represents the national minimum standard in Mexico. 14 We calculated proportions using the relevant population denominators (online supplemental table 1). We also Original research included an indicator of whether the municipality had an Adolescent Friendly Service Center, which provides adolescent-specific sexual and reproductive health services within Ministry of Health facilities. 15 Analysis First, we described the ILE clients' individual characteristics. Next, we described the binary municipality-level characteristics of socioeconomic status by whether ILE clients travelled from the municipality or not. We calculated municipality-level averages for the municipalities where ILE clients travelled from (n=491 municipalities) and did not travel from (n=1950 municipalities outside Mexico City) and used bar graphs to compare the proportions. Next, we created a heat map of the number of ILE clients that travel from each municipality collapsed into categories (0, 1, 2-100, >100). Finally, we built a logistic regression model to identify individual and contextual factors associated with presenting for abortion services from a highly marginalised municipality (top four quintiles) compared with a municipality of low marginalisation (bottom quintile). We included individual age, education, marital status, and parity as well as municipality-level adolescent fertility (dichotomised as highest two quintiles vs bottom three) and region. We used Stata version 16 (StataCorp, College Station, TX, USA) for all analyses. This study was deemed non-human subjects research by the Oregon Health & Science University IRB. --- RESULTS Of the 21 629 clients who travelled from outside Mexico City between 2016 and 2019 to access legal first-trimester abortion services in the ILE programme, the majority travelled from the least marginalised (81.9%) and most populous (more than 100 000 inhabitants; 91.3%) municipalities. The largest age group was 18-24 years old (46.5%); had a high school (45.5%) or university (20.3%) education; and reported to be single (58.1%) (table 1). The majority of ILE clients who travelled to Mexico City were experiencing at least a second pregnancy (62.3%). ILE clients primarily travelled from the central region (94.2%). They travelled from 491 of 2441 (20.1%) municipalities within all 31 states outside Mexico City. The majority of clients (91.2%) travelled from municipalities with the lowest adolescent fertility rates (lowest three quintiles; table 1). Figure 1 shows the difference in municipalitylevel socio-economic characteristics between the 491 municipalities from which ILE clients traveled, and the average levels from the other 1,950 municipalities outside Mexico City with no ILE clients. For example, 62% of the municipalities that ILE clients traveled from (orange bar) had adolescent fertility in the lowest 3 quintiles compared to 56% of those where they did not (grey bars). Figure 1 In our multivariable logistic regression model, ILE clients who had a primary or secondary education (vs high school or more) had larger adjusted odds of travelling from a highly marginalised (vs less) municipality (adjusted odds ratio (aOR) 1.46; 95% CI 1.35 to 1.58), controlling for individual factors, municipalitylevel adolescent fertility rate, and region (table 2). --- highlights that ILE clients travelling from municipalities with adolescent fertility rates in the top two quintiles (aOR 1.89; 95% CI 1.68 to 2.12) and municipalities in the southern region (aOR 2.98; 95% CI 2.58 to 3.45) had larger odds of travelling from a highly marginalised versus a less marginalised municipality (table 2). --- DISCUSSION We present novel data on the geographical distribution of ILE clients at the municipality level, and the individual and community context of women who travelled from outside of Mexico City to obtain abortion services at the ILE programme. The majority of ILE clients came from the least marginalised and most populous municipalities with lower adolescent fertility rates. Generally, the municipalities that ILE clients travelled from were on average better off than other areas of Mexico on several measures of socioeconomic status. Clients who had a primary or secondary education and those who came from a place with high adolescent fertility had larger odds of travelling from a highly marginalised municipality compared with a municipality of low marginalisation. Our findings show that ILE clients come from all corners of Mexico and support previous work which demonstrates high unmet demand for legal abortion services. 8 While our study data do not include those who needed an abortion but were unable to travel, the disparities in vulnerability we observe at the community level between communities ILE clients travel from and communities where no one travels from suggests more Original research vulnerable communities and by extension, women, may face additional barriers to travelling for services. These findings support earlier work that showed that disparities in abortion access are exacerbated by social and economic status. 4 8 13 16 Previous work has shown that simply living where abortion is illegal in the State of Mexico, adjacent to Mexico City, significantly reduced utilisation, especially for women of lower socioeconomic status, even when accounting for travel time. 8 Socioeconomic disparities persist in Mexico and other areas of Latin America, where poor and rural communities experience health worker shortages 17 ; insufficient quality care 18 ; and have fewer health facilities and less travel infrastructure. 19 Studies on travelling for abortion services tend to evaluate women's experiences 20 or measure distance as an indicator of access. 21 22 In the United States, availability of and distance from abortion services are determinants of access, 23 24 and those who travel are most often rural residents accessing abortion services in cities. 25 Our study, however, shows that most of the women who travelled also came from more populous and less marginalised municipalities, highlighting that statelevel restrictions on abortion in Mexico add to known disparities in access to healthcare including abortion services. 18 26 These findings support our hypotheses that (1) ILE clients come from geographically and socioeconomically diverse municipalities and (2) that municipalities home to ILE clients are different (less marginalised) than municipalities with no ILE clients on key markers of socioeconomic status. --- Original research We further explored factors associated with travel for abortion service from more marginalised communities. We show that most ILE clients are aged between 18 and 24 years, are single, and have a high school or university education. However, 33% of ILE clients had only a primary or secondary education, and these women are more likely to come from more highly marginalised municipalities. This is consistent with previous work that showed that less educated women from outside Mexico City may be less likely than their more educated peers to access services from the ILE programme. 7 Additionally, we show that two-thirds (62%) of ILE clients who travel for abortion services are already parents or are experiencing at least a second pregnancy. This is similar to all ILE clients (61%) (including local, Mexico City clients) 13 and abortion clients (59%) in the United States. 27 Previous work suggests that denying women an abortion may have negative socioeconomic or developmental effects on their existing children. 28 While prior studies on abortion in Mexico focused on young women and adolescent access to the ILE programme, 29 and abortion as a strategy to prevent first births, 29 30 our findings highlight that people who are already parents also need access to legal abortion services. Our findings and data source must be interpreted with the following limitations in mind. First, our data are only from those who successfully obtained an abortion from the public sector programme; we cannot therefore estimate unmet demand for abortion services throughout Mexico. Second, we are only able to observe travel to Mexico City for a public sector abortion; women who travel for a private sector abortion are likely less vulnerable. Third, we are unable to observe those who may travel across the US border. Fourth, while out-of-facility or self-managed medication abortion is increasing across Latin America, 31 32 it cannot be estimated here. Fifth, we use clients' selfreported municipalities, and we do not know if they are reporting their permanent municipality (where they grew up) or a temporary municipality. However, a key strength of our study is our ability to include the full universe of ILE clients 2016-2019; previous work has relied on samples. 2 13 --- Conclusions Women who are willing and able to travel to access legal first-trimester abortion in Mexico City's ILE programme come from geographically and socioeconomically diverse communities. Communities where ILE clients come from are, on average, less vulnerable than communities without ILE clients. There is an unmet need for access to abortion all across Mexico and lack of access to local legal abortion services increases disparities in access to care. --- Competing interests BGD receives research support from Merck. Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. --- Patient consent for publication Not required. Provenance and peer review Not commissioned; externally peer reviewed. Data availability statement Data are available in a public, open access repository. All data used in this study are publicly available and details on the sources are provided in Supplemental Table 1. --- Open access This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work noncommercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is noncommercial. See: http:// creativecommons. org/ licenses/ by-nc/ 4. 0/.

Objective To describe the community context of women who travel to access Mexico City's public sector abortion programme and identify factors associated with travelling from highly marginalised settings. Methods We used data from the Interrupción Legal de Embarazo (ILE) programme (2016-2019) and identified all abortion clients who travelled from outside Mexico City. We merged in contextual information at the municipality level and used descriptive statistics to describe ILE clients' individual characteristics and municipalities on several measures of vulnerability. We also compared municipalities that ILE clients travelled from with those where no one travelled from. We used logistic regression to identify factors associated with travelling to access ILE services from highly marginalised versus less marginalised municipalities. Results Our sample included 21 629 ILE clients who travelled to Mexico City from 491 municipalities within all 31 states outside Mexico City. The majority of clients travelled from the least marginalised (81.9%) and most populated (over 100 000 inhabitants; 91.3%) municipalities. Most (91.2%) ILE clients came from municipalities with adolescent fertility rates in the bottom three quintiles. Clients with a primary or secondary education (vs high school or more) and those from a municipality with a high adolescent fertility rate (top two quintiles) had higher odds of travelling from a highly marginalised (vs less) municipality (adjusted odds ratio (aOR) 1.46, 95% CI 1.35 to 1.58 and aOR 1.89, 95% CI 1.68 to 2.12, respectively). Conclusion ILE clients travel from geographically and socioeconomically diverse communities. There is an unmet need for legal abortion across Mexico.

Introduction Hate speech is pervasive and can have serious consequences. According to a Special Rapporteur to the UN Humans Rights Council, failure to monitor and react to hate speech in a timely manner can reinforce the subordination of targeted minorities, making them "vulnerable to attacks, but also influencing majority populations and potentially making them more indifferent to the various manifestations of such hatred" (Izsák 2015). At the individual level, people targeted by hate speech describe "living in fear" of the possibility that online threats may materialize in the "real world" (Awan and Zempi 2015). At the level of society, hate speech in social media has contributed to fuel tensions among communities, in some cases leading to violent clashes (Izsák 2015). Following one of the most severe humanitarian crises in recent history, Europe has seen a high immigration influx, including Syrian, Afghan, and Iraqi refugees. 1 In the same period, several deadly terror attacks have occured in Western nations (Wang 2017;Global Terrorism Database 2017), leading to an increasingly alarming anti-Muslim rhetoric (TellMAMA 2017) by right-wing populist movements (Greven 2016) and right-leaning media outlets (Worley 2016), often conflating refugees and Muslims with Islamic fanatics (Diène 2006). This rhetoric has also gained adoption online (UNGP and UNHCR 2017), prompting governmental agencies2 and NGOs to call on social media platforms to step up their efforts to address the problem of hate speech (Roberts 2017;TellMAMA 2017). The concern is that the increase in hateful narratives online led to an upsurge in hate crimes targeting Muslim communities (Roberts 2017). Insights into how online expressions of hate thrive and spread can help stakeholders' efforts to de-escalate existing tensions (Burnap and Williams 2014). In this paper, we explore how hate speech targeting specific groups on social media is affected by external events. Anchoring our analysis in a series of Islamophobic and Islamist terrorism attacks in Western countries, we study their impact on the prevalence and type of hate and counter-hate speech targeting Muslims and Islam on two different social media platforms: Twitter and Reddit. Our contribution. We conduct a quantitative exploration of the causal impact of specific types of external, non-platform specific events on social media phenomena. For this, we create a lexicon of hate speech terms, as well as a collection of 150M+ hate speech messages, propose a multidimensional taxonomy of online hate speech, and show that a causal inference approach contributes to understanding how online hate speech fluctuates. Among our findings, we observe that extremist violence attacks tend to lead to more messages directly advocating violence, demonstrating that concerns about a positive feedback loop between violence "offline" and hate speech online are, unfortunately, well-founded. Paper Outline and Methodology Overview. Outlined in Figure 1, our approach consists of several steps: Step 1: We create a longitudinal collection of hate speech messages in social media from Twitter and Reddit that covers a period of 19 months. This collection is based on a series of keywords that are obtained through an iterative expansion --- Themes/ Locations --- Collection of queries --- Stance [...] Muslims Immigrants Target group [...] Framing [...] Promotes violence Intimidates Severity [...] Step 2: Data & terms categorization --- Time series generation --- Impact analysis Reddit: volume of posts/ comments/users Twitter: volume of tweets/ unique tweets/users --- Results --- Query expansion (manual & automated steps) Step 3: Event selection Step 4: Impact analysis on time series ---... --- Expanded query terms Original time series Conterfactual time series --- Pre-intervention Post-intervention Figure 1: Steps in our analysis framework: (1) data acquisition and lexicon creation, <unk>3; (2) data categorization, <unk>4; (3) events selection, <unk>6; (4) impact analysis on time series of hate related terms, <unk>5 and results, <unk>6. of known hate speech terms ( <unk>3). Step 2: We categorize the data along four dimensions: 1) the group each message refers to, 2) the attitude of speakers, 3) the severity of hateful expressions, particularly whether they advocate violence, and 4) the framing of content ( <unk>4). Step 3: We select 13 extremist attacks involving Arabs and Muslims as perpetrators or victims, like the Berlin Christmas market attack on Dec. 2016, perpetrated by a follower of jihadist group ISIL, or the Quebec City mosque shooting on Jan. 2017, by a far-right white nationalist (Table 2). Step 4: As evaluating the effect of such attacks on various slices of social media is a causal question, we frame it as measuring the impact of an intervention (event) on a time series (temporal evolution of speech). Following techniques for causal inference on time series (Brodersen et al. 2015), we estimate an event's impact on various types of hate and counter-hate speech by comparing the behavior of corresponding time series after an event, with counterfactual predictions of this behavior had no event taken place ( <unk>5). The last sections present ( <unk>6) and discuss ( <unk>7) our results. --- Background & Prior Work We are interested in the relation between online hate speech and events. To ground our study, we first review work defining hate and counter-hate speech. Given our focus on anti-Muslim rhetoric in the context of extremist violence, we outline previous works on hate speech after terror attacks, and studies of hateful narratives targeting Muslims. We also cover observational studies on social media, particularly those focusing on harmful speech online. Counter-hate speech. Censoring hate speech may clash with legal protections on free speech rights. Partially due to this tension, the position of international agencies like UN-ESCO is that "the free flow of information should always be the norm. Counter-speech is generally preferable to suppression of speech" (Gagliardone et al. 2015). Thus, it is important not only to study hate speech, but also to contrast it with counter-speech efforts-a rare juxtaposition in social media research (Benesch et al. 2016;Magdy et al. 2016). Magdy et al. (2016) estimate that a majority of Islam and Muslim related tweets posted in reaction to the 2015 terrorist attacks in Paris stood in their defense; an observation also made by UNGP and UNHCR (2017) following the 2016 terrorist attack in Berlin, and supported by our own results ( <unk>6). --- Hate Speech Online and Offline Hate speech and violent events. The prevalence and severity of hate hate speech and crimes tends to increase after "trigger" events, which can be local, national, or international, often galvanizing "tensions and sentiments against the suspected perpetrators and groups associated with them" (Awan and Zempi 2015). For instance, Benesch et al. (2016) found extensive hate and counter-hate speech after events that triggered widespread emotional response like the Baltimore protests, the U.S. Supreme Court decision on same-sex marriage, and the Paris attacks during 2015; while Faris et al. (2016) found spikes in online harmful speech to be linked to political events. While these studies are related to ours, they focus on content posted during specific events, or on correlating changes in patterns (e.g., spikes) with events' occurrence. We focus on broader patterns, aiming to quantify changes across types of events and types of content by applying causal inference techniques. Islamophobia. The conflation of Muslims and Islam with terrorism-particularly developed after September 11, 2001-is a key factor behind the increase in Islamophobic attitudes (Diène 2006). A significant increase in anti-Muslim hate crimes was observed after terrorist attacks by individu-als that identify as "Muslim or acting in the name of Islam," with those having a "visible Muslim identity" being the most vulnerable to hostility, including online and offline intimidation, abuse and threats of violence (Awan and Zempi 2015). --- Observational Studies Using Social Media Hate speech on online social platforms. While social media platforms provide tools to meet new people, maintain relationships, promote ideas, and promote oneself; they have also opened up new avenues for harassment based on physical appearance, race, ethnicity, and gender (Duggan 2017). This has led to efforts to detect, understand, and quantify such harmful speech online, with goals such as modeling socially deviant behavior (Cheng et al. 2017), building better content filtering and moderation tools (Matias et al. 2015), and informing policy makers (Faris et al. 2016). The main categorization criteria for online hate speech has been based on the group being targeted (e.g., "black people," "fat people"), the basis for hate (e.g., race, religion) (Silva et al. 2016;Mohammad et al. 2016), and the speech severity (Davidson et al. 2017). For instance, Silva et al. (2016) found "soft" targets like "fat people" to be among the top target groups; yet, these groups are often not included in the documentation of offline hate crimes. Davidson et al. (2017) further discuss challenges in distinguishing between hate speech and other types of offensive speech. Observational methods applied to social data. Recent studies show that quasi-causal methods can be applied to social media data to e.g., distill the outcomes of a given situation (Olteanu, Varol, and K<unk>c<unk>man 2016), measure the impact of an intervention (Chandrasekharan et al. 2018), or estimate the effect of online social support (Cunha, Weber, and Pappa 2017). The application of these methods to social data, including propensity score matching (De Choudhury et al. 2016), difference-in-differences (Chandrasekharan et al. 2018), and instrumental variables (Zhang, Li, and Hong 2016), was found to reduce confounding biases. Chandrasekharan et al. (2018)'s work is closest to ours, as it employs techniques from the causal inference literature to quantify the impact of an intervention on hateful behavior on Reddit. Yet, the intervention they study is platform-specifica ban on an existing community on Reddit-whereas we look at the impact of external (non-platform specific) events on both Reddit and Twitter. Our focus is on the overall prevalence of hate speech, rather than on the behavior of given groups of users, and we measure the effect of given interventions (events) on various types of hate speech ( <unk>4). Operationalization of hate speech on social media. Due to lack of consensus on what constitutes hate speech and the challenges in operationalizing existing definitions at the scale of current online platforms, prior work has used a mix of manual and automated term selection strategies to identify terms that are likely to occur in hateful texts (Chandrasekharan et al. 2018;Davidson et al. 2017). While focusing on speech targeting Muslims and Islam, we similarly combine existing lexicons with terms obtained through a combination of manual and automated steps ( <unk>3). --- Data Collection Our goal is to characterize and measure online hate speech targeting Muslims and Islam in reaction to major Islamist terror attacks and Islamophobic attacks perpetrated in Western countries. Here, we describe our data collection process, which attempts to be inclusive (high-recall) and hence uses a broad definition of hate and counter-hate speech. We iteratively expand an initial query of keywords related to relevant items by identifying new keywords in the retrieved messages. Our base datasets contain messages from Twitter and Reddit, and a collection of news articles; these are not associated to any particular event, but cover messages potentially related to hate and counter-hate speech over a period of 19 months: from January 1, 2016 to August 1, 2017. --- Data Sources Twitter (https://twitter.com/) is one of the largest microblogging platforms used by hundreds of millions every month. To collect Twitter messages ("tweets") we use an archive representing 10% of the entire public stream, known as the "Decahose." Reddit (https://reddit.com/) is a large social news aggregation platform used by millions every month. Users submit "posts" and "comments" that gain or lose visibility according to up-and down-votes. We collect posts through Reddit's Search API3 (comments are not searchable via this API), retaining all comments to posts matching our queries. News. Finally, we collect news articles from GDELT (Global Data on Events, Location, and Tone, http:// gdeltproject.org/), the largest online catalog of global news events. We use these data as exogenous variables when modeling social media time series before and after a given event. --- Query Construction We collected data using keyword queries, a sampling method applicable to both Twitter's and Reddit's APIs. As our goal was to create a high-recall collection, our sampling procedure consists in formulating an initial query (bootstrapping), followed by an expansion of that query. This method is known to improve the coverage of social media data (Olteanu et al. 2014;Davidson et al. 2017). Query Bootstrapping. We bootstrapped our query selection with an initial list of terms (keywords and hashtags) used in social media campaigns related to anti-Muslim hate and counter-hate speech. This list was assembled retrospectively (as was the rest of our data) using (i) news articles and blog posts discussing social media usage during hate and counterhate campaigns,4 (ii) resources from NGOs or governmental agencies tracking or analyzing hate speech on social media (Awan and Zempi 2015;UNGP and UNHCR 2017), and (iii) research articles (Magdy et al. 2016) terms found in this process were added to the list. This step resulted in a list of 91 terms, including "'#f***quran," "#nosharia," "ban islam," and "kill all muslims." Query Expansion. We then employed a query expansion heuristic to identify further terms that may appear in messages expressing hate or counter-hate towards different groups, including, but not limited to, Arabs and Muslims. The heuristic considers terms frequently appearing in social media messages matched by the terms in our initial list. To obtain a high-recall collection, we considered any new term that may constitute hate or counter-hate speech, using an inclusive, broad definition inspired by Silva et al. (2016) and Chatzakou et al. (2017), and expanded to also cover commentary and counter-hate speech elements. We recorded all terms related to speech that could be perceived as offensive, derogatory, or in any way harmful, and that is motivated, in whole or in a part, by someone's bias against an aspect of a group of people, or related to commentary about such speech by others, or related to speech that aims to counter any type of speech that this definition covers. This expansion was independently done in two iterations for both Twitter and Reddit. First, one of the authors did an annotation pass to identify new query terms. Second, as we favored recall, at least one other author did an additional annotation pass over the terms rejected by the first annotator. External lexicon. To further expand our list of query terms, we added terms from a lexicon built using HateBase,5 a website that compiles phrases submitted and tagged by internet users as constituting hate speech. Given that only an estimated 5% of messages containing HateBase terms were actually identified as hateful; instead of directly using these terms, we used 163 unique terms extracted from Twitter messages containing HateBase terms and manually annotated as hateful or offensive by Davidson et al. (2017). 6 --- Data Acquisition Table 1 presents a summary of the data we acquired. Acquiring Twitter data. We first queried the bootstrap terms, and retrieved 958K messages posted by 413K users. We then expanded the query by manually annotating 2088 terms that appeared more frequently than an arbitrary threshold (ranging from 75 for tri-grams to 300 for uni-grams, which are typically less precise than tri-grams and noisier at lower frequencies), after removing stopwords using the Python NLTK package. We found an extra 612 terms. We queried these terms, growing our collection by 55M tweets posted by 12.5M users. The resulting dataset contains on average 4.5M tweets per month. Since we used the Twitter Decahose (a 10% sample of all Twitter content), we estimate this collection is in fact representative of a larger set of roughly 45M tweets per month. Finally, we retrieved tweets matching the 163 external hate terms (based on HateBase), resulting in an additional 51.6M tweets by 13.7M users. Altogether, we collected over 1TB of raw Twitter data. Acquiring Reddit data. We again began by querying the bootstrap terms, and retrieved 3K posts with 140K comments written by 49K users. We then expanded the query by selecting high-frequency terms (thresholds ranging from 50 to 300 as these data were sparser than Twitter) across all posts and comments, and manually annotating them. Given that the Reddit Search API normalizes terms before running a query, we did not keep different inflections of the same terms. We annotated 4272 terms, and found 1002 related to hate and counter-hate speech. We queried these terms, and retrieved an extra 300K posts with 41M comments written by 3.1M users. Finally, we queried the external hate terms. Altogether, we collected 337K posts with 45M comments written by roughly 3.3M users. Acquiring news data. We used GDELT's Global Knowledge Graph (GKG), as it provides the list of news articles covering each event in their database. This allowed us to compute the overall volume of news per day, amounting to over 130M URLs over our 19 months period of interest. --- Characterizing Hate Speech Here, we present example themes from messages posted in the aftermath of extremist events (listed in <unk>6), and characterize them along four dimensions (stance, target, severity, and framing), which we then use to analyze the data. --- Exploration of Post-Event Messages To understand how the content and themes of messages vary with respect to who is mentioned, what is said, and how the content is framed, we review messages posted after one terrorist and two Islamophobic attacks: Manchester Arena bombing, an Islamist terrorist attack in Manchester that targeted concert goers, killing 23 people and wounding 512 others; Portland train attack, carried out by a man shouting racial and anti-Muslim slurs who fatally stabbed two people and injured a third; and Quebec City mosque shooting that targeted worshipers, leaving 6 dead and 19 injured. We focus on these particular events for their overall difference in nature. Table 2 includes example messages. Who is mentioned? Naturally, many messages mentioned (directly or indirectly) Arabs, Muslims, or Islam, given how we collected our data and the focus of our study. Yet, we also found messages mentioning the victims of the attacks, the mainstream media, political and religious groups (e.g., "the left", "Christians"), immigrants in general, and high-profile individuals (e.g., politicians, journalists). What is said, and why? The content of the messages ranged from blaming Arabs and Muslims for the attack, to providing context and defending Islam. Some messages made crude generalizations or included denigrating insults, while others appeared to either intimidate or incite violence. "killing innocent people is not Islam, there were Muslims at that concert as well" (T) "#IllRideWithYou indicates one should not be scared to be a Muslim. One should be scared to be a racist" (T) --- Denigrates or intimidates "Muslim savages brainwash their kids into hating and killing non believers, as apes and pigs, since really young" (R) Incites violence "#StopIslam wipe its followers from the face of the earth" (T) Diagnoses causes "the left say, look they were not refugees; the fact is that this would never happen if we would have banned them" (R) --- Suggests a remedy "we should deport Muslim scumbags and their families" (R) Table 2: Example messages from Reddit (R) and Twitter (T) for some of the analyzed events, provided for illustration purposes. Messages have been (sometimes heavily) paraphrased for anonymity. How is the content framed? According to Entman (1993) --- Four Dimensions of Online Hate Speech Based on prior work and our exploration of post-event messages, we derive four main dimensions of hate and counterhate speech: stance, target, severity, and framing. While these are useful, we recognize these dimensions are unlikely to capture all aspects of online expressions of hate. Stance. Magdy et al. (2016) make a distinction between online speech that attacks and blames, speech that defends, and speech that is neutral towards Islam and Muslims following a terrorist attack. Benesch et al. (2016) introduce a taxonomy for spontaneous expressions of counter-hate speech on social media platforms. We adapt these categorizations to define the following stances of speech for our study: -Takes a favorable stance in support of individuals, groups, or ideas: defend, show solidarity, propose counter narratives, denounce, or comment on acts of hatred, or emphasize the positive traits of individuals, groups, or ideas (e.g., #ThisIsNotIslam, #NotInMyName); -Takes an unfavorable stance against individuals, groups, or ideas: attack, blame, denigrate, demean, discriminate, employ negative stereotypes, seek to silence, or generally emphasizes the negative traits of an individual or group (e.g., "kill all Muslims," #RefugeesNotWelcome); -Commentary on negative actions or speech against individuals, groups, or ideas: comment on or characterize acts of violence, hatred, harassment, or discrimination (e.g., "hate speech," "racial slur"); and -Neutral, factual, or unclear if it is in support or against a person or group: none of the above; report news facts or comments, describe an event, or not related to a minority or vulnerable group (e.g., "seven injured," "white van"). Target. Hate speech can target any minority or vulnerable group by singling out its identifying characteristics. In the case of Muslims or Islam, these characteristics include religion, country of origin, immigration status, ethnicity, or a conflation of several or all characteristics. We identify the following targets of hate and counter-hate speech: -Muslims and Islam; -Religious groups: unspecified, any religion except Islam; -Arabs, Middle-Easterners, or North Africans: descent without reference to religion; -Ethnic groups or groups of foreign descent: unspecified, any foreign descent, except Arab; -Immigrants/refugees/foreigners in general: without indicating a specific religion or descent; and -Other groups of non-immigrants: based on e.g., gender, sexual orientation, appearance, disability, or age. Severity. International organizations are concerned with how hate speech can lead to violent acts (Izsák 2015). Expressions of hate take many forms (Ghanea 2013;Matias et al. 2015); they can be ambiguous, and the perception of what is hateful varies between individuals (Olteanu et al. 2017). Capturing such subtleties is essential to understanding how severe the repercussions of online hate speech can be; for instance, the Jewish Anti-Defamation League defines a "Pyramid of Hate," showing how prejudice enables discrimination, which enables violence, which enables genocide. 7 We use the following levels of severity of hate speech: -Promotes violence: threaten with violence, incite violent acts, and intend to make the target fear for their safety (e.g., "attack mosque," "kill muslims"); -Intimidates: harass or intimidate the target, or invite others to do so, while actively seeking to cause distress (e.g., "deport illegals," "Muslims not welcomed"); -Offends or Discriminates: defame, insult, or ridicule the target, showing bias, prejudice, or intolerance, while actively seeking to embarrass or harm the target's reputation, (e.g., "Muslim [expletive]," "sand [n-word]"); Framing. Kuypers (2010) defines framing as the "process whereby communicators, consciously or unconsciously, act to construct a point of view that encourages the facts of a given situation to be interpreted by others in a particular manner." Benford and Snow (2000) note that framing is critical to understand social movements and collective action; it can also operate in different ways (Entman 1993). For our analysis, from test annotations we noticed that two frames were quite distinguishable in the text and complementary:8 -Diagnoses the cause or causes for a problem (or elements seen as possible causes): identifies what creates a problem, suggests a diagnose or disagrees with a diagnose (e.g., "terrorists exist because they come from a place that, socially, is centuries behind"); -Suggests a solution or solutions for a problem (or actions seen as possible solutions): proposes or defends actions seen as solving or removing the problem (e.g., "we should target the mosques and [M]uslims, this is what you need to do when at war with these [expletive]"); -Both diagnoses causes and suggests solutions: if both of the above categories apply to the message. Terms or sentences may perform multiple of these framing functions, but they may also perform none of them (Entman 1993). Thus, for annotation purposes we add a catch-all category for those cases where none of these functions apply. --- Methodological Framework To quantify how extremist violence events affect the prevalence of various types of speech, we treat these events as interventions on observed time series. Following existing techniques for causal inference on time series (Eichler 2012;Brodersen et al. 2015), we measure this effect by comparing the behavior of an observed time series (which we refer to as treated) after an event with a counterfactual time series of its behavior had the event not taken place. This synthetic unobserved counterfactual time series (which we refer to as control) is modeled from several observed time series that may be correlated to the treated time series (yet not affected by the event), as we describe below. The causal effect is then estimated based on the differences between the treated and the control time series. Broadly, since we model the counterfactual of the treated time series, this is a generalization of the application of the differences-in-differences techniques to time series (Brodersen et al. 2015). Observed Time Series. We consider time series covering our 19-month observation period with a granularity of one day. For each of the 825 terms we have for Twitter, we experiment with three time series: one for the number of tweets, one for the number of tweets excluding re-tweets (i.e., unique after removal of "RT @user" prefixes), and one for the number of unique users. Similarly, for the 1,257 terms we have for Reddit, we experiment with three time series: one for the number of posts, one for the total number of comments in these posts, and one for the total number of unique users in the post and comments. Synthetic Control Time Series. A synthetic control time series is a counterfactual that reflects behavior had the extremist violence event not taken place. For each treated time series, we build a control series for 1 week following the event based on several data sources:9 (1) the observed series in the 11 weeks leading to the event; (2) the observed series exactly 1 year before the event, for 12 weeks (corresponding to the 11 weeks before and 1 week after the event, but a year earlier); (3) the observed series 23 weeks prior to the event, similarly for 12 week;10 and (4) external information from Twitter, Reddit, and news sources. The external information includes time series whose behavior is unlikely to be affected by the events ( <unk>6). First, we use the overall volume of news on GDELT (i.e., number of daily news article URLs per day) as it does not seem to be affected by any of our events during the observation window. Second, we use the overall number of tweets containing the word "news" which we also observe is not affected by any of our events (also a proxy for the overall volume of tweets). Third, we use the overall number of Reddit posts containing the general term "people," which we also observe is not affected by the events in our list (this is not the case for the series of posts in Reddit containing, e.g., the term "breaking news" which was affected by several of our events). The methodology for synthesizing the control follows Brodersen et al. (2015), using a state space model to predict the counterfactual from the various sources we described above. However, our models are fit using maximum likelihood estimation (Fulton 2015) rather than Bayesian methods like Markov chain Monte Carlo preferred by Brodersen et al. (2015). Our implementation uses the state space model in the UnobservedComponents Python package to model and predict the series, following existing Python implementations of Brodersen et al. (2015).11 Impact Estimation. To estimate the effect of an event using the treatment and control time series, we compute the relative lift or drop as rel effect = 100 <unk> t k -c k c k, where t k is the value of the treated time series at time k, and c k that of the control time series. The summations are over the days we observe after the event, seven in our case. We focus on relative effect as it better allows for comparison across events. For each event, we rank terms based on the relative effect. Some of our time series have intervals of low volume (particularly for Reddit) that may lead to negative-valued synthetic controls and skewed estimates of the effect. To address this, we add a large constant C to all time series before synthesizing the control and estimating the effect. This transformation preserves the shape and amplitude of the impact. Table 3: Distribution of annotations along the entire 19-month observation period, done at the term level (except framing, done at the message level). The percentages may not add to 100% as we omit the cases when none of the categories apply. --- Experimental Results In this section, we present experimental results that estimate how different types of events affect various forms of online speech. First, we select 13 extremist violence attacks (Islamist terrorist and Islamophobic), that occurred during our full 19-month observation period. Next, we annotate our data at query term level for stance, target, and severity, and at message level for framing, according to the hate speech taxonomy introduced in <unk>4. Finally, we present results on various categories of hate speech across events and platforms. --- Experimental Setup Events Selection. We select a set of extremist violence attacks in Western countries involving Arabs and Muslims as perpetrators or victims, and covered by international news media. Our sources are two Wikipedia pages listing Islamist terrorist attacks and Islamophobic incidents. 13 When two events occur within the same week, we selected the one with the largest number of victims, also the most prominent in the news. The list of events is available in Figure 2, where we also display the time series of top-5 bootstrap terms ( <unk>3) on Twitter and Reddit, which shows that these events cover most of the peaks in these terms for Twitter and a majority of them for Reddit. 13 https://en.wikipedia.org/wiki/Islamophobic incidents and https://en.wikipedia.org/wiki/List of Islamist terrorist attacks Crowdsourced Annotations. Our entire list of terms contains 1890 unique terms, which we annotate by employing crowdsource workers through the Crowdflower platform. We select workers from countries having a majority of native English speakers or that were affected by the events (e.g., Germany). Except for "framing," for cost and scalability purposes, we annotate each term with the most likely category the text containing them may fall under. For framing we annotate entire messages, as annotating at the termlevel annotations does not produce reliable labels. For each hate speech dimension and category, we provide detailed definitions and extensive examples; and, for each term we annotate, we show crowd workers clickable links to corresponding search results matching our queries, as returned by both social media platforms, Twitter and Reddit, as well as by two major search engines, Bing and Google. Following standard crowdsourcing practices, we gather at least 3 annotations per term (up to 5 when consensus was not reached), using a set of unambiguous test questions provided by the authors to catch inattentive workers, and resolving disagreements by majority voting. For framing, for each event we annotate samples of 5-6 messages matching the top 100 terms by relative effect, and posted around the time of the event. 14 To obtain the dominating frame of a term, we Figure 3: Example of impact estimation with counterfactual predictions, for the term "evil muslims." Black/red are the observed series before/after the event, green the counterfactual. Top: time series of tweets containing the term after an Islamist terrorism attack (left: Orlando nightclub shooting) and an Islamophobic attack (right: Olathe Kansas shooting). Bottom: differences between observed and counterfactual. first determine the label of the messages it matches, and then assign by majority voting to each term the most prevalent frame, or if the "causes" or "solutions" frames are similarly prevalent, we assign the "causes and solutions" frame. Table 3 shows the overall distribution of annotations; the annotations for frame provide only an approximation based on top terms as impacted by the events in our list. We observed that terms marked as unfavorable represent <unk>30%-50% of our query terms, and only <unk>20%-30% of those are identified as particularly severe (i.e., promoting violence or intimidating); corresponding to 15% on Twitter and 7% on Reddit. Given the recall-oriented nature of our collection, this supports the observation of Faris et al. (2016), who, using a similar taxonomy, also observed that the incidence of the most severe cases of hate speech is also typically small. Pre-and Post-filtering. Our estimation method requires a minimum number of messages to produce a meaningful result; hence we filter out terms matching only a small number of messages, which we operationalize through arbitrary thresholds requiring a maximum of at least 30 users or messages per day during the event observation window. Figure 3 shows an example of impact estimation on the "evil muslims" term, displaying the observed series, the control

User-generated content online is shaped by many factors, including endogenous elements such as platform affordances and norms, as well as exogenous elements, in particular significant events. These impact what users say, how they say it, and when they say it. In this paper, we focus on quantifying the impact of violent events on various types of hate speech, from offensive and derogatory to intimidation and explicit calls for violence. We anchor this study in a series of attacks involving Arabs and Muslims as perpetrators or victims, occurring in Western countries, that have been covered extensively by news media. These attacks have fueled intense policy debates around immigration in various fora, including online media, which have been marred by racist prejudice and hateful speech. The focus of our research is to model the effect of the attacks on the volume and type of hateful speech on two social media platforms, Twitter and Reddit. Among other findings, we observe that extremist violence tends to lead to an increase in online hate speech, particularly on messages directly advocating violence. Our research has implications for the way in which hate speech online is monitored and suggests ways in which it could be fought.

s and solutions" frame. Table 3 shows the overall distribution of annotations; the annotations for frame provide only an approximation based on top terms as impacted by the events in our list. We observed that terms marked as unfavorable represent <unk>30%-50% of our query terms, and only <unk>20%-30% of those are identified as particularly severe (i.e., promoting violence or intimidating); corresponding to 15% on Twitter and 7% on Reddit. Given the recall-oriented nature of our collection, this supports the observation of Faris et al. (2016), who, using a similar taxonomy, also observed that the incidence of the most severe cases of hate speech is also typically small. Pre-and Post-filtering. Our estimation method requires a minimum number of messages to produce a meaningful result; hence we filter out terms matching only a small number of messages, which we operationalize through arbitrary thresholds requiring a maximum of at least 30 users or messages per day during the event observation window. Figure 3 shows an example of impact estimation on the "evil muslims" term, displaying the observed series, the control series, and their difference in two separate events. In the figure, the widening confidence interval of the forecast matches the intuition that predictions become less certain as we look further into the (counterfactual) future. In general, after applying this process, we consider there to be effect (increase or decrease) if the 90% confidence interval of the difference between treatment and control does not include zero, which means we consider there is no effect where the 90% confidence interval is too large or centered around zero. --- Results and Discussion In this section, we want to quantify the increase or decrease of various types of speech according to the type of event and platform. over 500 M tweets to locate those matching the query terms. [2.4, 3.3]). 15 Looking at the intersection of high severity categories ("promotes violence" and "intimidates") with the target categories for Muslims and Arabs, we estimate an increase in the relative effects across events in both platforms (T: +10.1, 95%CI [1.4, 18.9], R: +6.2, 95%CI [3.9, 8.4]); also higher than the less severe category (offends or discriminates), with one exception, the 2016 Istanbul Airport attack. The question is whether Islamophobic attacks elicit a similar, consistent reaction across platforms and events. The answer seems to be no: for instance, we only observe this pattern after one Islamophobic attack (the 2016 Finsbury Park attack), while after the 2017 Olathe Kansas shooting we estimate a decrease in high severity terms in both platforms. This observation is also supported at an aggregate level by Figure 4 (per-event figures omitted for brevity). Similarly, our estimates indicate an overall increase in counter-hate speech terms following Islamist terrorist attacks (T: +1.8, 95%CI [0.7, 3.0], R: +2.9, 95%CI [2.4, 3.4]), but not after Islamophobic attacks. This effect of Islamist terror attacks on counter-speech is consistent with Magdy et al. (2016) who noticed a notable number of messages defending Muslims and Islam following the 2015 Islamist terror attack in Paris. Are these events more likely to lead to an increase in a specific type of speech? Figure 5 suggests that, on average, there is a higher increase in speech that both promotes violence or intimidates and focuses on Muslims and Arabs following extremist violence events; while there is an increase in counter-hate speech related to religion but not specifically At the event-level, Figure 6 showcases an example of a complex interplay between hate and counter-hate speech terms in reference to different groups after the 2016 Orlando nightclub shooting. This was not only a deadly Islamist terrorist incident, but also the deadliest homophobic attack in the U.S., which means it was very prominently covered in media. It triggered a substantial increase in terms referring to both Islam and the gay/LGBT community. In general, our observations agree with an increase in mentions of Muslims, Islam, or Arabs, after Islamist terror attacks; but not after Islamophobic attacks (figures omitted for brevity). Are there differences in how hate speech is shaped by the events across platforms? For Twitter, Figure 4 suggests an increase for the high-severity categories ("promotes violence" and "intimidates") after Islamic terrorist attacks, but not after Islamophobic attacks. In contrast, for Reddit this distinction is absent, as we see an overall increase after both Islamist terrorist and Islamophobic attacks. Another aspect in which we see differences between Twitter and Reddit is in terms of the framing of messages, particularly with respect to messages including a "solution or something seen as a solution." In general, the terms that tend to increase the most in this frame call for banning or deporting immigrants/Muslims/Arabs, or waging war against Islam and/or Arabs. As shown in the "solution" and "both" columns in Table 4, this fraction is more prevalent for Twitter among the top 100 most impacted terms (about 34% in Islamist terrorist attacks, about 29% in Islamophobic attacks) than for Reddit (about 10% and 16% respectively). --- Conclusions Measuring the effect of external events on hate speech on social media is a challenging task, which needs to be approached with an appropriate methodology (causal inference in our case), and requires a combination of automated processes and manual annotations that balances the needs of large-scale analysis with a finite human annotation budget. We used data from two social media sites, and from two classes of events (Islamist terrorism and Islamophobic attacks), performing a comparison of observed time series for various classes of online hate speech during such events, with counterfactual series that approximate their behavior had those events not taken place. This allows us to make more precise observations about the evolution of hate (and counter-hate) speech of various classes. Our methodology and observations provide a blueprint for better monitoring hate speech online, with particular attention to the relation between calls for violence online and deadly extremist attacks. Additionally, as we estimate increases in counter-hate speech during these attacks, social media platforms could intervene by boosting its visibility. Future Work and Limitations. We hope that the evidence of variations in hate speech following certain events will lead to further research to understand why it happens, who it happens to, and what other qualities of an event may explain these variations; as well as research that delves into the source of the differences we observed across platforms. Further, while our data collection is designed to maximize recall, aiming to provide a good coverage across several categorization dimensions, our bootstrap list of terms can still lead to bias in what gets included in our collections or not. The reliance on query-level annotations may as well introduce noise and biases due to ambiguous uses of some of the terms. We focused on English and only 13 events in the "West," yet future work includes explorations into how our observations may translate to other regions, languages, and type of events. Our frame analysis is also only a first stab at how hateful content is framed after extremist attacks; more in-depth analyses are needed. Finally, our analysis is retrospective, and harmful content is actively deleted by many social media platforms when reported (Matias et al. 2015), which can result in incomplete data collections. As a result, we are more confident in results indicating an increase in certain types of speech, than on those indicating a decrease. Reproducibility. The list of our query terms, several example time series, and the detailed instructions used in the crowdsourcing tasks, are available for research purposes at https://github.com/sajao/EventsImpactOnHateSpeech.

User-generated content online is shaped by many factors, including endogenous elements such as platform affordances and norms, as well as exogenous elements, in particular significant events. These impact what users say, how they say it, and when they say it. In this paper, we focus on quantifying the impact of violent events on various types of hate speech, from offensive and derogatory to intimidation and explicit calls for violence. We anchor this study in a series of attacks involving Arabs and Muslims as perpetrators or victims, occurring in Western countries, that have been covered extensively by news media. These attacks have fueled intense policy debates around immigration in various fora, including online media, which have been marred by racist prejudice and hateful speech. The focus of our research is to model the effect of the attacks on the volume and type of hateful speech on two social media platforms, Twitter and Reddit. Among other findings, we observe that extremist violence tends to lead to an increase in online hate speech, particularly on messages directly advocating violence. Our research has implications for the way in which hate speech online is monitored and suggests ways in which it could be fought.

Introduction What are your eating and drinking habits? How different are they from a typical individual from Japan or Germany? It is impossible to answer these questions without addressing the cultural features within groups of individuals. However, culture is such a complex and interesting concept that no simple definition or measurement can capture it. Among the various aspects that define the culture of a society (or person), one may cite its arts, religious beliefs, literature, manners and scholarly pursuits. Moreover, as Counihan (Carole 1997), and Cochrane and Bal (Cochrane and Bal 1990) pointed out, eating and drinking habits are also fundamental elements in a culture and may significantly mark social differences, boundaries, bonds, and contradictions. Since eating and drinking habits have such importance for a culture, we here address the topic of investigating and analyzing life and idiosyncrasies of different societies through them. Copyright c 2014, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved. How can we analyze eating and drinking habits at a large scale? Nowadays, the study of social behavior at a large scale is possible thanks to the increasing popularity of smart phones and location sharing systems such as Foursquare. By means of these technologies, it is possible to sense human activities related to food and drink practices (e.g., restaurant visiting patterns) in large geographical areas, such as cities or entire countries. Foursquare, created in 2009, registered 5 million users in December 2010 and 45 million users in January 2014. Data generated by this popular application triggers unprecedented opportunities to measure cultural differences at a global scale and at low cost (Silva et al. 2013). In this work, we propose a new methodology for identifying cultural boundaries and similarities across populations using self-reported cultural preferences recorded in locationbased social networks (LBSNs). Our methodology, which is here demonstrated using data collected from Foursquare, consists of the following steps. First, we map food and drink check-ins extracted from Foursquare into users' cultural preferences. By exploring this mapping, we are able to identify particular individual preferences, such as the taste for barbecue or sake. Food and drink individual preferences, as shown in this paper, are good indicators of cultural similarities between users. We then show how to extract features from Foursquare data that are able to delineate and describe regions that have common cultural elements, defining signatures that represent cultural differences between distinct areas around the planet. To that end, we investigate two properties of food and drink preferences: geographical and temporal characteristics. Next, we apply a simple clustering technique, namely k-means, to show the "cultural distance" between two countries, cities or even regions of a city, allowing us to draw cultural boundaries across them. Unlike previous efforts, which used survey data, our work is based on a dynamic and publicly available Web dataset representing habits of a much larger and diverse population. Besides being globally scalable, our methodology also allows the identification of cultural dynamics more quickly than traditional methods (e.g., surveys), since one may observe how countries or cities are becoming more culturally similar or distinct over time. The correct identification of cultural boundaries is useful in many fields and applications. Rather than using traditional methods to identify cultural differences, the pro-posed method is an easier and cheaper way to perform this task across many regions of the world, because it is based on data voluntarily shared by users on Web services. Moreover, since culture is an important aspect for economic reasons (Garcia-Gavilanes, Quercia, and Jaimes 2013), our methodology is valuable for companies that have businesses in one country and want to verify the compatibility of preferences across different markets. Another application that could rely on our methodology is a place recommendation system, which is useful for visitors and residents of a city. Foursquare estimates that only 10% to 15% of searches on Foursquare are for specific places (Chaey 2012). Much more often users are searching within broader categories, such as "sushi" (Chaey 2012). Based on this information, systems like Foursquare and other location-based search engines, as the one proposed in (Shankar et al. 2012), could benefit from the introduction of new criteria and mechanisms in their recommendation systems that consider cultural differences between areas. For instance, a person who enjoyed a specific area of Manhattan could receive a recommendation of a similar area when visiting London. The rest of this paper is organized as follows. Section 2 presents the related work. Section 3 describes our dataset and the core of our methodology for extracting cultural preferences from location-based social networks. Section 4 investigates the cultural similarities between individuals, and shows that food and drink check-ins outperforms check-ins given in all types of places in this case. Section 5 shows how to extract cultural signatures for different areas of the globe and explore the similarities among them, while Section 6 applies this knowledge to analyze the implicit cultural boundaries that exist for different cultural aspects of the society. Finally, Section 7 summarizes our contributions and discusses some possibilities of future work. --- Related Work Several studies have focused on the spatial properties of data shared in location-based services such as Foursquare (Scellato et al. 2011;Cho, Myers, and Leskovec 2011;Noulas et al. 2011a). However, those prior efforts aimed at investigating user mobility patterns or social network properties and their implications. More recently, researchers have started looking at user activity as another data source that can be leveraged for studying social interactions (Sakaki, Okazaki, and Matsuo 2010). Based on this principle, there have been many studies to extract new insights about city dynamics such as, for example, their key characteristics and the behavior of their citizens. For instance, Cranshaw et al. (Cranshaw et al. 2012) presented a model to extract distinct regions of a city according to current collective activity patterns. Similarly, Noulas et al. (Noulas et al. 2011b) proposed an approach to classify areas of a city by using all venues' categories of Foursquare. Some recent studies have shown how the use of Web systems vary across countries. For example, Hochman et al. (Hochman and Schwartz 2012) investigated color preferences in pictures shared through Instagram, showing considerable differences in the preferences across countries with distinct cultures. Garcia-Gavilanes et al. (Garcia-Gavilanes, Quercia, and Jaimes 2013) and Poblete et al. (Poblete et al. 2011) studied variations of Twitter usage across countries. In particular, Garcia-Gavilanes et al. showed that cultural differences are not only visible in the real world but also observed on Twitter. Cross-cultural studies (i.e., the study of cultural differences) do not constitute a new research area. Indeed, they have been carried out by researchers working in the social sciences, particularly in cultural anthropology and psychology (Murdock 1949). Despite globalization and many other technological revolutions (Blossfeld et al. 2005), group formation might lead to the emergence of cultural boundaries that exist for millennia across populations (Barth 1998). Axelrod (Axelrod 1997) proposed a model to explain the formation and persistence of these cultural boundaries, which are basically a consequence of two key phenomena: social influence (Festinger 1967) and homophily (McPherson, Smith-Lovin, and Cook 2001). While homophily dictates that only culturally similar individuals are likely to interact, social influence makes individuals more similar as they interact. In a long run, these two phenomena lead to very culturally distinct groups of individuals, delimited by the socalled cultural boundaries. --- Extracting Cultural Preferences In this section we present our dataset and our methodology for extracting cultural preferences from LBSNs. --- Mapping User Preferences One of the biggest challenges in the analysis of cultural differences among people and regions is finding the appropriate empirical data to use. The common approach to overcome this challenge is the use of surveys based on questionnaires filled during face-to-face interviews (Valori et al. 2012), such as the Eurobarometer dataset (Schmitt et al. 2005). Through these questionnaires, individual preferences, such as the taste for coffee and fast food, can be mapped into multidimensional vectors representing (and characterizing) each interviewee. From these vectors, it is possible, for instance, to quantify how similar or different two individuals are. Although survey data are broadly used in the analysis of cultures, there are some severe constraints in its use, which are well known to researchers. First, surveys are costly and do not scale up. That is, it is hard to obtain data of millions, or even thousands of people. Second, they provide static information, i.e., they reflect the preferences of users at a specific point in time. If some of the preferences change for a significant amount of the interviewed people, such as the taste for online gaming instead of street ball playing, the data is compromised. In order to overcome the aforementioned constraints, we propose the use of publicly available data from LBSNs to map individual preferences. LBSNs can be accessed everywhere by anyone who has an Internet connection, solving the scalability problem and allowing data from (potentially) the entire world to be collected (Silva et al. 2013). Moreover, these systems are dynamic, being able to capture the behavioral changes of their users when they occur, which solve the second mentioned constraint. However, data from such systems can be used if and only if they meet the requirements: • [R1] It is possible to associate a user to its location; • [R2] It is possible to extract a finite set of preferences from the data that is generated by the system; • [R3] It is possible to map users' actions in the system into the preferences defined in [R2]. Considering that these requirements are met, a dataset containing individual activities of N users of a LBSN can be used to map preferences as follows. First, associate each user n i with a location l i, which may be a country, a city or even a region within a city. Then, define a set of m individual preferences (or features) f 1, f 2,..., f m that can be extracted from the dataset, which may represent the taste for the most varied things, such as Japanese food or a certain football team. Finally, map the activities of each individual n i into an m-dimensional vector of preferences F i = f 1 i, f 2 i,..., f m i that characterizes the person's tastes, the same type of vector that is usually created from survey data (Valori et al. 2012). Since the preference vector F i is generated from selfreported temporal data of an individual n i, we may populate and modify it in various ways. For instance, we can use a binary representation, where f k i = 0|1 represents whether user n i has or not preference f k (e.g., whether a person likes/dislikes a certain type of food), respectively. Alternatively, we may consider the intensity at which a user likes a feature, inferred from the number of times the corresponding preference is reported in the person's data, i.e., f k i = [0; <unk>). In Section 4, we adopt a binary representation. Finally, one can group individuals by their geographical regions and sum up their preference vectors to characterize their regions. We adopt this approach in Section 5 to build preference vectors for regions (instead of individuals). --- Data Description In this work, the dataset used to infer user preferences was collected from one of the currently most popular location based social networks, namely Foursquare. We collected this data from Twitter 1, since Foursquare check-ins are not publicly available by default. Approximately 4.7 million tweets containing check-ins were gathered, each one providing a URL to the Foursquare website where information about the venue, in particular its geographic location and category, was acquired. In the dataset, each check-in consists of the latitude, longitude, identifier, and category of the venue as well as the time when the check-in was done. Foursquare venues are grouped into eight categories: Arts & Entertainment; College & University; Professional & Other Places; Residences; Great Outdoors; Shops & Services; Nightlife Spots; and Food. Each category, in turn, has subcategories. For example, Rock Club and Concert Hall are subcategories of Nightlife Spots. In order to show that our methodology is able to capture cultural dynamics in short time windows, we use a dataset that spans a single week of April 2012. Moreover, since we are primarily interested in food and drink habits, we manually grouped relevant subcategories of 1 http://www.twitter.com. the Food and Nightlife Spots categories into three classes: Drink, Fast Food, and Slow Food places. We did this by excluding some subcategories that are not related to these three classes (e.g. Rock Club and Concert Hall) and moving some subcategories (e.g. Coffee Shop and Tea Room) from the Food category to the Drink class. Besides that we also disregard the category Restaurant, because it is a sort of meta category that could fit in any of the two classes of food. After this manual classification process, the Drink class ended up with 279,650 check-ins, 106,152 unique venues and 162,891 unique users; the Fast Food class with 410,592 check-ins, 193,541 unique venues, and 230,846 unique users; and the Slow Food class with 394,042 check-ins, 198,565 unique venues, and 231,651 unique users. Moreover, the Drink class has 21 subcategories (e.g., brewery, karaoke bar, and pub), whereas the Fast Food class has 27 subcategories (e.g., bakery, burger joint, and wings joint) and the Slow Food class has 53 subcategories, including Chinese restaurant, Steakhouse, and Greek restaurant. To provide an idea about the size of the user population LBSNs can reach, consider the World Values Survey2 project. That study is maybe the most comprehensive investigation of political and sociocultural change worldwide, which was conducted from 1981 to 2008 in 87 societies, with about 256,000 interviews. Observe that our one-week dataset has a population of users of the same order of magnitude of the number of interviews performed in that project in almost three decades. --- Mapping Foursquare Data into User Preferences Several characteristics of human beings are not directly observable, such as personality traits. Thus, we rely on face-toface interactions or online signals to discover the presence of those hidden qualities (Pentland 2010). In this direction, a LBSN check-in can be considered as a signal because it is a perceivable feature/action that expresses the preference of a user for a certain type of place. With that in mind, we use Foursquare check-ins to represent user preferences regarding food and drink places. Specifically, we use the three main classes defined in Section 3.2, namely, Drink, Fast Food, and Slow Food. Figures 1a,1b, and 1c show the frequency of check-ins at each subcategory of the Drink, Fast Food, and Slow Food classes, respectively, so we can have a general idea about the popularity of user preferences for different food and drink related places. These figures show the popularity of different places according to people's preferences worldwide. Note that Coffee Shop and Bar are the two most popular subcategories of Drink places, with 86,310 and 81,124 checkins, respectively. The two most popular Fast Food subcategories are Café 3 and Fast Food Restaurant, with 91,303 and 56,648 check-ins, respectively. Finally, American Restaurant (47,373 check-ins), and Mexican Restaurant (28,712 check-ins) are the two most visited subcategories of Slow Food places. In this dataset, a user is represented by a vector of m =101 features corresponding to the 101 subcategories that comprise the three classes we have defined. A feature f i <unk> F = <unk>f 1, f 2,..., f 101 <unk> is equal to 1 if a user made at least one check-in at f i, and 0 otherwise. In this way, a feature vector represents the positive and negative preferences of a user for fast food, slow food and drink subcategories. With that, a finite set of preferences is extracted (requirement [R2], see definition in Section 3.1) and users' actions are mapped into this set (requirement [R3]). To associate a user with a location (requirement [R1]), we analyzed the GPS coordinates of all check-ins performed by the user. If all check-ins performed are from the same country, according to the free reverse geocoding API offered by Yahoo 4, we assume that the user taken into consideration is from that country. Otherwise, we do not consider the user in our analysis. In this way, we minimize the wrong association of a user with a country. Following this procedure, approximately 1% of the users were disregarded from our analysis. --- Cultural Analysis of Individuals In this section, we use the map of preferences presented in Section 3.3 to analyze the individual preferences of users, showing, among other results, that food and drink preferences are good indicators of cultural similarities. In order to assess the cultural similarities among users, we construct a similarity network G s = (V s, E s ), where s is a similarity threshold used to build the network, vertices V s represent the set of users, and an edge (v i, v j ) exists in E s if users v i and v j have a similarity score above s. The similarity score s i,j between two users v i and v j is the Jaccard index (JI) between their preference vectors 5 multiplied by 100. In this way, s i,j varies from 0 to 100 and measures the 3 Like in many European countries, this term is referred as a restaurant primarily serving coffee as well as pastries. 4 http://developer.yahoo.com. 5 The Jaccard index of sets A and B is computed as A<unk>B A<unk>B. percentage of preferences shared by the users v i and v j. For example, considering a similarity threshold s = 65 (or 65%network6 ), there is an edge between vertices v 1 and v 2 if the corresponding users have, at least, 65% of preferences in common. We have built two similarities networks: G 1 s ; and G 2 s. The network G 1 s considers only food and drink preferences, i.e., only check-ins at food and drink places. On the other hand, G 2 s consider all preferences, i.e., all Foursquare subcategories, including food and drink venues. To build both networks we consider only the users who performed at least 7 check-ins in the dataset (i.e., at least one check-in per day on average). In total, 28,038 users were considered in G 1 s and 194,902 in G 2 s. Moreover, isolated nodes were disregarded. We here consider the following values of s <unk> <unk>65, 70, 75, 80, 85, 90, 95, 100<unk>. Note that G 1 s and G 2 s are undirected unweighted and symmetric graphs. We first analyze relevant properties of G 1 s and G 2 s. Figure 2a shows the percentage of vertices (i.e., users) in the two largest components of the network G 1 s, for various values of s (figure omitted for the network G 2 s due to space limitations). Figure 2a shows that the largest component of the 65%-network practically contains all nodes. The percentage of users in the largest component slowly decreases as the similarity threshold increases, until s reaches 85. For larger values of s, the number of users in the largest component drops sharply, becoming comparable to the size of the second largest component. This is explained by observing networks built using large values for s, such as the 100%network, where every component is composed of very similar users. Since users with very similar preferences are rare, the largest components tend not to have very large differences in size. We note that the results for the network G 2 s are similar to those observed for the network G 1 s, for example, the largest component of the 65%-network also contains practically all nodes. In order to verify the tendency of users from the same region to be connected, we calculate the assortativity of the similarity networks. Assortativity measures the similarity of connections in the network with respect to a given attribute, and varies from -1 to +1 (Newman 2002). In an assortative network (with positive assortativity), vertices with similar values of the given attribute (e.g., same country) tend to connect with (be similar to) each other, whereas in a disassortative network (with negative assortativity), the opposite happens. The assortativity analysis for the networks G 1 s and G 2 s formed from various values of s are shown in Figures 2b and2c, respectively. Note that the assortativity for the net- work G 1 s with respect to the geographical attributes (region Western/Eastern, continent, and country) decreases with the similarity threshold. This happens because most of the edges in the networks, formed from similarity threshold s <unk> 90, connect users who have preference vectors with a few positive features (as defined in Section 3.3). This also helps to explain why, in both figures, the degree assortativity increases with the similarity threshold: considering only very particular tastes, the network tends to be composed mostly of cliques, making the degree assortativity very close to 1. On the other hand, if we vary the value of s in the network G 2 s, the assortativity for geographical attributes remains roughly the same. It is possible to explain this behavior by looking at the size of the preference vector F for the network G 1 s, which is much smaller compared to that for the network G 2 s (101 against 435). Since the preferences are distributed over almost all the categories, a larger preference vector implies a lower probability of having preferences in common between two users, and, consequently, fewer edges in a similarity network, even for lower values of s. Note also that, in both Figures 2b and2c, all similarity networks we take into consideration are assortative. However, the assortativity values of the geographical attributes for G 1 s are most of the time higher compared to those obtained for G 2 s. When considering all preferences/features we also increase the number of features that do not discriminate cultural differences sufficiently well (e.g., venues like homes, hotels, student centers, and shoe stores), since they are essentially present in all the cities and countries in the world. This suggests that, in this case, a similarity network considering only food and drink preferences might provide better insights in the study of cultural differences. --- Extraction of Cultural Signatures Given the results discussed in Section 4, we hypothesize that it is possible to define cultural signatures of different areas around the planet. In this section, we show how to extract features from Foursquare data that are able to describe regions from their cultural elements. In particular, we investigate two properties of food and drink preferences: their geographical (Section 5.1) and temporal (Section 5.2) aspects. --- Spatial Correlations Here our goal is to define a set of features that are able to characterize the cultural preferences of a given geographical area in the planet, such as a country, a city or a neighborhood. Thus, for a given delimited area a (e.g., the city of Chicago), we sum up the values of the features in the preference vectors of the users who checked in at venues of that area. In other words, we count the number of check-ins C a = c a 1, c a 2,..., c a 101 performed in venues of each of the 101 subcategories s 1, s 2,..., s 101 of the Fast Food, Slow Food and Drink classes (Section 3.2) that are located within the perimeter of area a. Next, we represent each area a by a vector of 101 features F a = f a 1, f a 2,..., f a 101, where each feature f a i is equal to c a i / max(C a ). That is, we normalize the number of check-ins at each subcategory by the maximum number of check-ins performed in a single subcategory in area a (max(C a )). Thus, each area a is represented by a feature vector F a containing values from 0 to 1, indicating the preferences of people who visited that area, i.e., the profile of preferences for that area. From now on, we use F a drink, F a sf ood and F a f f ood to refer, respectively, to the subset of features that correspond to subcategories belonging to the Drink, Slow Food and Fast Food classes in area a. In order to verify if two areas a and b are culturally similar, we compute the Pearson's correlation coefficient between the two feature vectors F a and F b of those areas. We compute the correlation considering all features (F a and F b ) as well as a subset of them (e.g., F a drink and F b drink ). In particular, Figure 3 shows the correlations between areas corresponding to 27 different popular countries for the Drink (3a), Fast Food (3b), and Slow Food (3c) classes; the darker the color, the stronger the correlation (blue for positive correlations, red for negative correlations). The same correlations computed for city level areas (16 cities around the world) are shown in Figure 4. Analyzing the results for the Drink class (Figure 3a), we find countries with very strong correlations, such as Argentina and Chile, as well as countries with low correlation, such as Brazil and Indonesia. Moreover, although regions close geographically tend to have stronger correlations, this is not always the case. For example, the correlation between Brazil and France is stronger than the correlation between England and France, which are geographically closer. Similarly, Figure 4a 7 shows that cities in the same country tend to have very correlated drinking habits in most cases, but there are exceptions: Manaus (Brazil), for instance, has weak correlation with other cities in Brazil. This might be due to this city being located in the North region of Brazil, which is known for having a strong cultural diversity compared to other parts of the country. Turning our attention to food practices, we observe in Figures 3b and 4b the global penetration of fast food venues, at both country and city levels, explained by the diffusion of fast food places worldwide (Watson 2006). This is not observed in the same intensity for the Slow Food class (Fig- ures 3c and4c). The Slow Food class presents the highest distinction, or smaller correlation, across most of the countries and cities. This is expected, since Slow Food venues usually are representative of the local cuisine. Note, for instance, that cities from Brazil and USA have highly correlated drinking and fast food habits, but almost no correlation in slow food habits. Finally, we turn our attention to the cultural habits within city boundaries. It is known that, in many cities, there is a strong cultural diversity across different neighborhoods (Cranshaw et al. 2012), reflecting distinct activities typically performed in these areas. To analyze these local cultures, we focus on three populous cities, namely London, New York, and Tokyo. We divide each city's geographical area using a grid structure. Next, we select the most popular cells in the grid of each city and label them with a number, as shown in Figure 5. We then compute the correlation between the selected cells. Note that we here assume a grid with regular (rectangular) cells to show the potential of the proposed analysis. However, our approach can be applied to any other segmentation of the city areas (e.g., by city districts). Figure 6 shows the correlations for pairs of cells within 7 The ratio of check-ins per inhabitant is similar among all the cities taken into consideration. For example, comparing Manaus (one of the cities with fewer check-ins) with Sao Paulo (largest number of check-ins in Brazil) we find the following ratios: 0.35 <unk> 10 -3 and 0.37<unk>10 -3 (Drink class); 0.73<unk>10 -3 and 0.75<unk>10 -3 (Fast Food class); and 0.54 <unk> 10 -3 and 0.71 <unk> 10 -3 (Slow Food class). the same city and from different cities. Note that, for the Drink class, different areas within the same city tend to have very strong correlations. There are also areas from different cities with strong correlations (e.g., areas NY-5 and TKO-1). For Fast Food places, the correlations between areas within the same city are much stronger for Tokyo, although the correlations between New York and London areas are fairly moderate. In contrast, there are areas with negative correlation, e.g., NY-3 with most of Tokyo areas. Finally, for the Slow Food class, once again Tokyo areas are very strongly correlated among themselves. In comparison with the Fast Food class, there is a more clear distinction (weaker correlation) between London and New York areas as well as among distinct areas in London. This last observation is probably due to a specific characteristic of London, that has neighborhoods with a strong presence of a cuisine of a particular region of the globe. Observe also that two specific areas of New York, namely NY-7 and NY-8, are par- ticularly not correlated with the others from this city. This is probably related to the location of Chinatown in those areas (mainly NY-7). Indeed, this particular area (NY-7) has a strong correlation with a particular area of London, LND-5, where Chinatown/London is located. --- Temporal Analysis We now turn our attention to the temporal and circadian aspects of cultural habits. The time instants when check-ins are performed in food and drink places may also provide valuable insights into the cultural aspects of a particular region. For example, in a particular area, one may like to drink beer during the weekends but not during the weekdays. To that end, we first count the number of check-ins per hour during the whole week covered by our dataset in venues of each class (Drink, Fast Food and Slow Food) for different regions. Next, we group days into weekdays and weekends, summing up the check-ins performed on the same hour of the day in each group and for each region. We then normalize this number by the maximum value found in any hour for the specific region, so that we can compare the patterns obtained in different regions. For illustration purposes, we show the results for three countries (Brazil, USA, and England) and for three American cities (Chicago, Las Vegas, and New York) in Figures 7 and8, respectively. Results for each class are shown separately for weekdays and weekends. Focusing first on weekday patterns, Figure 7 shows that American and English people have similar peaks of activities, despite differences in their preferences for different categories of places, as previously shown (Figure 3). In contrast, Brazilians tend to have significantly different temporal patterns, particularly in terms of activities in Slow Food places (Figure 7c): whereas Americans and English people tend to have their main meal at dinner time, Brazilians have it at lunch time. Observe also that Brazilians have their meals later, compared to Americans and English people. Concerning the times when people go to drink venues, it is possible to note similarities among most of the cities from the same country, but also some different patterns. For example, most of the analyzed cities from USA exhibit a weekday pattern similar to New York and Chicago, shown in Figure 8a, with three distinct peaks around breakfast, lunch and happy hour (around 6pm). This behavior is consistent with the general pattern observed for the country, shown in Figure 7a. However, Las Vegas is one exception, since there is an intense activity during the dawn, besides many other peaks of activities that do not occur in other cities. Turning our attention to eating habits on weekdays, Figure 8 shows that most cities in the USA present activity patterns very similar to the general pattern identified for the country, both in terms of Slow and Fast Food places. However, as observed for drinking patterns, there are exceptions, such as Las Vegas, which exhibits distinct trends that reflect inherent idiosyncrasies of this city. We also note relevant similarities and differences in eating habits of people from cities in different countries. For example, comparing Figures 8b and8c with similar graphs produced for different Brazilian cities, we find that while all curves for the Fast Food class are very similar, the curves for Slow Food places are quite different, reflecting distinct habits for each country, as discussed previously. The curves for weekends have very distinct peaks of activities from those of weekdays, both at the country and city levels. For instance, as shown in Figure 7, English people have a very distinct drinking pattern from Americans on weekends. Moreover, the differences among the countries in terms of preferences at Slow Food places are also clear on weekends: Brazilians tend to go to Slow Food places more often at lunch time, whereas Americans and English people do it more at dinner time. We note that there is no clear (dominant) temporal checkin pattern for Fast Food places on weekends, when considering different cities of a country. However, we do note that most activities happen after noon, which was expected. In contrast, there is a dominant pattern for check-ins at Slow Food places on the weekends, and it is similar to the one observed on weekdays. This is possibly because such places (often restaurants) have well-defined opening hours, serving meals around lunch and dinner times only, which coincide with the times of check-in peaks (Figures 7c,7f,8c,and 8f). Assuming that the height of such peaks reflects the importance of that meal for a certain culture, we note once again a key distinction between Americans and Brazilians. --- Discussion In addition to temporal and spatial patterns of check-ins at different types of places, we also compute the Shannon's entropy (Shannon 1948) of preferences for each venue subcategory among all considered areas. The goal is to analyze whether the check-ins at specific subcategories are more concentrated at specific areas (low entropy) or not (high entropy). We compute the entropy for subcategories of each class (Drink, Fast Food and Slow Food) at country and city levels. The average entropy for subcategories of the Drink class is 3.23 (standard deviation <unk> = 0.93) for countries and is 3.88 (<unk> = 1.09) for cities. Sake bar is one example with low entropy (1.13 for countries and 1.89 for cities), which indicates that this subcategory is popular on very few countries and cities. Surely Japan contributes considerably to this result. On the other hand, the average entropy for subcategories of the Slow Food class is much larger

Food and drink are two of the most basic needs of human beings. However, as society evolved, food and drink became also a strong cultural aspect, being able to describe strong differences among people. Traditional methods used to analyze cross-cultural differences are mainly based on surveys and, for this reason, they are very difficult to represent a significant statistical sample at a global scale. In this paper, we propose a new methodology to identify cultural boundaries and similarities across populations at different scales based on the analysis of Foursquare check-ins. This approach might be useful not only for economic purposes, but also to support existing and novel marketing and social applications. Our methodology consists of the following steps. First, we map food and drink related check-ins extracted from Foursquare into users' cultural preferences. Second, we identify particular individual preferences, such as the taste for a certain type of food or drink, e.g., pizza or sake, as well as temporal habits, such as the time and day of the week when an individual goes to a restaurant or a bar. Third, we show how to analyze this information to assess the cultural distance between two countries, cities or even areas of a city. Fourth, we apply a simple clustering technique, using this cultural distance measure, to draw cultural boundaries across countries, cities and regions.

a certain culture, we note once again a key distinction between Americans and Brazilians. --- Discussion In addition to temporal and spatial patterns of check-ins at different types of places, we also compute the Shannon's entropy (Shannon 1948) of preferences for each venue subcategory among all considered areas. The goal is to analyze whether the check-ins at specific subcategories are more concentrated at specific areas (low entropy) or not (high entropy). We compute the entropy for subcategories of each class (Drink, Fast Food and Slow Food) at country and city levels. The average entropy for subcategories of the Drink class is 3.23 (standard deviation <unk> = 0.93) for countries and is 3.88 (<unk> = 1.09) for cities. Sake bar is one example with low entropy (1.13 for countries and 1.89 for cities), which indicates that this subcategory is popular on very few countries and cities. Surely Japan contributes considerably to this result. On the other hand, the average entropy for subcategories of the Slow Food class is much larger, 2.63 (<unk> = 0.78). This higher entropy reflects the widespread popularization of various cuisines. For example, a check-in at an Italian restaurant does not necessarily mean that it represents a behavior of an Italian, since it is a very international type of restaurant, confirmed by the high entropy (3.63). Note, however, that if the check-in at an Italian restaurant is made at lunch time it could be more likely to represent a Brazilian behavior than American, since Brazilians have their main meal at lunch time, as presented in Section 5.2. Time plays an important role in this case. Given these considerations and all the observations reported here, we propose the use of spatio-temporal correlations of check-ins as cultural signatures of regions. 6 Identifying Cultural Boundaries --- Clustering Regions In this section, we use the cultural signatures of regions described above to identify similar areas around the planet according to their cultural aspects, delineating their so-called "cultural boundaries". To that end, we first represent each area a by a high dimensional preference vector composed of 808 features, namely the normalized number of check-ins at each of the 101 subcategories in four disjoint periods of the day, on weekdays and on the weekends. We then apply the Principal Component Analysis (PCA) (Jolliffe 2002) technique to these vectors to obtain their principal components8. Finally, we use the k-means algorithm, a widely used clustering technique, to group areas in the space defined by these principal components. We perform this analysis for areas defined at the country, city and neighborhood levels. The score values for the first two principal components generated by the PCA for countries, cities, and regions are shown in Figures 9a, 9b, and9c, respectively. The variance in the data explained by these first two components is shown in each figure. Each color/symbol in those figures indicates a cluster obtained by k-means, which used the p first principal components that explain 100% of the variation in the data (p=15 for countries, p=26 for cities and p=22 for regions). The k value in the k-means varied according to the characteristics of the considered areas. For countries, we set k=7 (same number of clusters used in (Inglehart and Welzel 2010)). Following the same logic, we set k=4 for cities, since we considered cities from 4 different continents/countries, and k=3 for regions inside a city, because we considered 3 cities. We used the cosine similarity to compute the similarity between locations. It is possible to observe in Figure 9a that countries with closer geographical proximity are not necessarily associated with the same cluster. For example, Australia and Indonesia are not in the same cluster. Although they are geographically neighboring countries, they are culturally very distinct. When analyzing large cities from the considered countries, Figure 9b shows that they are well clustered by the geographical regions where they are located: Asia, Brazil, Europe and USA. Intuitively, this result makes sense, since, for instance, cosmopolitan European capitals tend to present more similar cultural habits among each other than among cities from different continents. Turning our attention to regions inside London, NY, and Tokyo, we observe in Figure 9c that all regions in the same city are in the same cluster. This result was also expected when considering all features. Besides that, when we analyze a subset of features, for example, drinking habits during weekends in all regions of London, NY, and Tokyo (result omitted), we find that some regions of London and NY are clustered together. This is corroborated by the results shown in Section 5: for certain categories, there are regions from different cities that are very similar and, thus, end up clustered together., observe that the similarities are striking, with only two major differences. First, the "Islamic" cluster dissolved, with Turkey joining Russia and Indonesia joining Malaysia and Singapore. Second, USA and Mexico left the "English Speaking" and the "Latin America" clusters, respectively, and paired up to form a new one. Note, nevertheless, that these differences might not be surprising as these new boundaries. --- Comparing with Survey Data We formally investigate the differences between boundaries given by the WVS study and by our approach. In order to do so we rank, for a given country, all the other countries according to their cosine similarity towards it. We compute the similarity using the dimensions produced by the WVS data (Inglehart and Welzel 2010) and the dimensions computed by our approach. Then, we compute the Spearman's rank correlation coefficient <unk> between these two ranks to see, for instance, if the most similar (and distinct) countries to England using the WVS data are ranked similarly when we use our approach. In our approach, we use two different datasets. In dataset 1, we use the full set of features, as done so far. In dataset 2, we use solely the features extracted from the fast food check-ins performed during the weekends9. Table 1 shows these results. We highlight in bold all the coefficients which are statistically significant, i.e., with a p-value <unk> 0.05. Observe that the correlation <unk> is significant and positive for several countries. For dataset 1 and dataset 2, 9 and 12 countries have similar ranks with the ones given by the WVS, respectively. This shows that our approach, which is based solely on one week of participatory data, has a clear potential to reproduce cultural studies performed using surveys, such as the ones relying on the WVS, which is based on 4 years of survey data. We would also like to point out the reasons for the differences between our cultural map and the WVS map, as well as for the negative correlations seen in Table 1. First, the traits of each dataset are significantly different. While the WVS looked at several cultural dimensions, from religion to politics, from economics to lifestyle, we looked only at food and drink preferences. Second, the WVS data has a distance of 4 to 7 years to our data. During this time, significant cultural changes may have happened, given that the world is getting more connected at every day. Third, the most significant differences are related to multi-ethnic, multicultural, and multilingual countries, such as Malaysia and Turkey. In these countries it is probably hard to find culturally homogeneous samples of individuals, which might be the cause of the discrepancies seen between our results and those described in (Inglehart and Welzel 2010). --- Conclusions and Future Work This work proposes a new methodology for identifying cultural boundaries and similarities across populations. For that, we map food and drink check-ins extracted from Foursquare into users' cultural preferences, considering spatio-temporal dimensions. We then apply a simple clustering technique to show the "cultural distance" among countries, cities or even regions within a city. The considered set of features allows the identification of cultural boundaries that despite often agreeing on common knowledge, is based on large-scale data. Thus, unlike other empirical work, which is based on survey data, our methodology can reach global scale much faster and at a much lower cost. It is also important to emphasize that the proposed methodology could be used to work with other types of features, which might be useful for other kind of studies. One of the obvious directions is to exploit the criteria for identifying cultural boundaries defined in this paper in order to perform social studies at large scale. Besides that, we also want to develop recommendation mechanisms considering the cultural characterization of specific urban areas. This could be useful, for instance, for location-based social networks like Foursquare to improve their current recommendation systems.

Food and drink are two of the most basic needs of human beings. However, as society evolved, food and drink became also a strong cultural aspect, being able to describe strong differences among people. Traditional methods used to analyze cross-cultural differences are mainly based on surveys and, for this reason, they are very difficult to represent a significant statistical sample at a global scale. In this paper, we propose a new methodology to identify cultural boundaries and similarities across populations at different scales based on the analysis of Foursquare check-ins. This approach might be useful not only for economic purposes, but also to support existing and novel marketing and social applications. Our methodology consists of the following steps. First, we map food and drink related check-ins extracted from Foursquare into users' cultural preferences. Second, we identify particular individual preferences, such as the taste for a certain type of food or drink, e.g., pizza or sake, as well as temporal habits, such as the time and day of the week when an individual goes to a restaurant or a bar. Third, we show how to analyze this information to assess the cultural distance between two countries, cities or even areas of a city. Fourth, we apply a simple clustering technique, using this cultural distance measure, to draw cultural boundaries across countries, cities and regions.

Introduction In 1974, the Canadian Minister Marc Lalonde published a pivotal report in public health that stressed the importance of health promotion and included aspects rarely considered before then, such as the environment and lifestyle [1]. Socioeconomic status (SES) conditions all the determinants of health defined by Lalonde, constraining or shaping our beliefs, behaviors and even our biology [2]. It is also at the root of health inequalities, since political, economic and social factors give rise to an unequal distribution of opportunities to enjoy health [3]. Social inequalities in childhood can lead to the same inequalities in adulthood, fueling a harmful legacy from generation to generation [4]. SES is therefore one of the many factors that can determine the existence of overweight in children, adolescents and adults. According to the World Health Organisation (WHO), it is low-and middle-income countries that are primarily affected, and the prevalence of overweight and obesity in preschool children living in countries with emerging economies can exceed 30% [5]. Numerous studies have reported an inversely proportional relationship between SES and overweight and associated problems in childhood, whereby the lower the SES, the higher the prevalence of health problems [6][7][8][9][10][11][12]. An association has also been found between SES and the two decisive factors in this major epidemic of overweight and obesity: diet and physical activity. For example, young people from certain ethnic groups with a low SES consume more fatty and high calorie foods [12] and in general have worse eating habits [13,14]. Meanwhile, the relationship between SES and physical exercise is directly proportional: the lower the former, the less the latter is performed [15]. In relation to the subject of the present study, adolescents' personal and individual social networks determine behavior related to food and physical exercise in this age group: it has been shown that besides adults, peers and friends are also able to modify habits related to excess weight [16][17][18][19][20][21][22][23]. Although this influence on such habits can be negative as well as positive, these social networks should nevertheless be viewed as a source of material, personal and/or institutional resources from which to obtain the necessary information, support and services [24]. Adolescents need to belong to a social network with which they identify to attain satisfactory physical, psychological, and social development [25], since their social environment provides them with the tools necessary for managing group relations [26]. SES also determines the structure and function of social networks. For example, it has been demonstrated that adults with a low SES tend to report fewer social networks and less support [27]. The configuration of adolescents' social networks might be similarly affected by this factor, compromising the support networks can provide to combat problems such as overweight and obesity, and reducing the resources available for information or support on issues related to food or physical exercise. In light of the above, our goal was to study the relationship between SES and overweight (overweight + obesity) in our sample and to relate this parameter to adolescents' centrality in their social network at school, both in general and in relation to their weight status, by conducting a social network analysis (SNA) from a sociocentric or global perspective. We believe that the information obtained would be useful for the design of strategies to combat overweight and obesity and would shed light on one of the most pressing public health problems today: obesity in the adolescent population. --- Materials and Methods --- Population and Sample We invited 776 students in their third and fourth years of compulsory secondary education at five schools in the city of Ponferrada (Spain) to participate in the study. Permission for data collection was sought from the Castile and León Education Department and the Spanish Data Protection Agency, and interviews were conducted with school heads and teachers to obtain their collaboration in the study. To participate in the study, students were asked to provide their informed consent via a form signed by their parents and designed in line with the recommendations of the University of Salamanca Bioethics Committee. This gave a detailed explanation of the purpose of the study and information on data collection and processing, pursuant to the Law on the Protection of Personal Data [28]. Participants were clearly informed that they could retract their consent once their parents had signed the form, without needing to provide a reason, and an email contact address was given should they require any further information. Participation was voluntary, and subject availability was respected at all times. To obtain a satisfactory sample, we required a minimum participation rate of between 40 and 50% of class members. We received a response from 276 students from 11 different classes (Table 1). Weight status formed an inclusion criterion: we sought the exclusive participation of individuals classified as "normal weight", "overweight" or "obese" according to WHO criteria [29]; hence, we excluded students classified as "low weight". This yielded a final sample of 235 students divided into 11 networks (Table 2). Network1A-Network5K: Representative code name for each participating network in the study. --- Data Collection Data were collected on gender, SES, anthropometric measurements (weight and height) and contacts in participants' social networks at school between March and December 2015. Nursing staff trained in this procedure, collected the study data in paper-based survey and took physical measurements of the anthropometric parameters. In line with the recommendation of school heads and teaching staff, questionnaires were administered during tutorial classes and weight and size measurements were taken during physical education classes. This latter procedure required a closed changing room, a portable Seca 700 stadiometer (Seca, Hanover, MD 21076, USA) provided by the Nursing and Physiotherapy Department of the University of León and electronic Fagor Slim scales (Fagor, Mondragón (Gipuzkoa), Spain) calibrated to zero for each measurement. --- Variables Gender was considered a dichotomous variable. SES (independent variable) was evaluated using the FAS II questionnaire [30,31], which assesses the family's purchasing power according to the everyday goods purchased. Responses to the FAS II questionnaire were cored between 0 and 3: negative responses were awarded a 0, and this value rises as the number of possibilities increases. The score for the total scale ranges from 0 to 9 and was subsequently grouped into three categories that reflect socioeconomic status. The FAS II has been validated by Boyce et al. and in line with their interpretation criteria, we classified scores 0, 1 and 2 as indicating low SES; 3, 4, and 5 as medium SES; and 6 as high SES [30]. Since an initial statistical analysis indicated that only a small number of participants presented a low SES, we aggregated the two lower levels to form a medium-low SES group, thus creating a dichotomous variable consisting of high and medium-low SES. Once each participant's weight and size data had been collected, we calculated the percentile and body mass index (BMI) according to exact age and gender using the WHO's Anthro Plus®application (World Health Organization, Cyberjaya, Selangor, Malaysia) [32]. Participants were then classified according to their weight status as "normal weight", "overweight" or "obese". Next, we generated a dichotomized variable using normal weight as the reference category and combined overweight and obesity as the second category (to represent excess weight), which we termed overweight. To obtain data about social network contacts (dependent variable), each questionnaire contained a closed list with the names and surnames of other classmates participating in the study, and the following question: "Using the list below, indicate how much time you spend with your classmates", formulated in line with the recommendations proposed by other experts in SNA [18,33]. Since the definition of "peers, classmates or friends" is complex and could seriously affect estimation of its effects when completing the questionnaire [34], we assessed contact intensity by means of time frequencies, using a 5-point Likert scale where 1 = "we never spend time together" and 5 = "we're always together" [26,35]. Please note that from the outset, all personal information that could identify any of the participants was encoded using a simulated name to ensure confidentiality. Peer contact data were used to generate an initial n <unk> n matrix (single-mode or type I network), consisting of students belonging to each class network. Since we wished to study contact intensity, each frequency was assigned a score, and three different adjacency matrices (0/1) were created from the initial matrix, based on three dichotomization criteria: (i) a "minimum contact" matrix, an adjacency matrix where the original value of 1 ("we never spend time together") represented the absence of contact (0) and the values 2, 3, 4 and 5 ("we sometimes spend time together", "we spend quite a lot of time together", "we're almost always together" and "we're always together") indicated the existence of the same (1); (ii) an "intermediate contact" matrix, where the values 1 and 2 ("we never spend time together" and "we sometimes spend time together") indicated the absence of contact (0) and the values 3, 4 and 5 ("we spend quite a lot of time together", "we're almost always together" and "we're always together") represented the existence of a tie (1); and (iii) a "maximum contact" or "friendship" matrix, where the values 1, 2 and 3 ("we never spend time together", "we sometimes spend time together" and "we spend quite a lot of time together") indicated a lack of contact (0), and 4 and 5 ("we're almost always together" and "we're always together") represented the existence of a relationship (1). For each contact intensity matrix, an analysis was conducted of the seven parameters representing social network centrality [36] from a sociocentric or global perspective: (i) outdegree (nominations emitted); (ii) indegree (nominations received); (iii) degree (number of ties that one actor has) [36]; (iv-v) closeness (in/outcloseness) (number of steps that one actor must take to reach another) [37,38]; (vi) betweenness (degree of connections that pass through an actor for one actor to reach another) [37,38]; and (vii) the eigenvector (a measure to identify the most central actors with the shortest distance to the rest of the nodes) [39]. This analysis yielded 21 normalized variables (values in which the ends were relativized) organized dichotomously according to the median for each parameter. --- Statistical Analysis The relationship between SES and the study variables was determined by unconditional logistic regression. In each case, we calculated the odds ratio (OR) with a confidence interval (CI) of 95%. Statistical significance (p-value) was established as p <unk> 0.05. Statistical analyses were performed using SPSS v.23 IBM, Armonk, NY 10504, USA) and network contact data were calculated using UCINET v.6.365 (Analytic Technologies, Inc, Collegeville, PA 19426, USA ) [40]. --- Results Descriptive data for the sample indicated that 49.4% were female (n = 116) and 50.6% were male (n = 119). Participants' ages ranged from 14.0 to 18.1 years old, with a mean age of 15.5 <unk> 0.9 years old. Mean BMI was 22.1 <unk> 2.9 kg/m 2 and the mean percentile value was 64.8 <unk> 24.1. In line with WHO criteria [29], the prevalence of overweight was 25.5% and of obesity 4.7%, indicating a total prevalence of overweight of 30.2%. Regarding SES, the mean score obtained using the FAS II questionnaire was 6.3 <unk> 1.6, corresponding to a high SES. As can be seen in Figure 1, 29% of the adolescents presented a medium-low SES and 71.1% a high SES. An analysis of the relationship between SES and overweight indicated that students with a medium-low SES presented a two-fold higher probability of being overweight (OR: 2.43; 95% CI: 1.33-4.40; p = 0.003). --- Results Descriptive data for the sample indicated that 49.4% were female (n = 116) and 50.6% were male (n = 119). Participants' ages ranged from 14.0 to 18.1 years old, with a mean age of 15.5 <unk> 0.9 years old. Mean BMI was 22.1 <unk> 2.9 kg/m 2 and the mean percentile value was 64.8 <unk> 24.1. In line with WHO criteria [29], the prevalence of overweight was 25.5% and of obesity 4.7%, indicating a total prevalence of overweight of 30.2%. Regarding SES, the mean score obtained using the FAS II questionnaire was 6.3 <unk> 1.6, corresponding to a high SES. As can be seen in Figure 1, 29% of the adolescents presented a medium-low SES and 71.1% a high SES. An analysis of the relationship between SES and overweight indicated that students with a medium-low SES presented a two-fold higher probability of being overweight (OR: 2.43; 95% CI: 1.33-4.40; p = 0.003). Regarding reticular data, several density and centralization measures were calculated for each of the studied networks (Table 3). Additionally, the analysis of the social network indicated that overweight adolescents' social ties did not reflect their SES at any of the contact intensity levels (Tables 456). In contrast, a statistical analysis of the overall sample irrespective of weight status revealed significant results at all three levels of contact intensity (Tables S1-S9). At the minimum contact level, adolescents with a high SES were almost twice as likely to present a greater capacity for intermediation (betweenness) (OR: 1.77; 95% CI: 1.001-3.148; p = 0.049), regardless of gender. By way of illustration, node size in Figure 2 represents the capacity for intermediation (betweenness) of adolescents in one of the social networks analyzed. As can be seen, larger nodes were predominantly associated with a high SES. At the intermediate contact level, we found differences by gender, more specifically in female adolescents. Thus, females with a high SES were more than twice as likely to be nominated as friends (indegree) (OR: 2.37; 95% CI: 1.022-5.518; p = 0.042) (Figure 3). At the maximum contact level, considered to represent friendship, we obtained similar results both for the overall sample and for male adolescents. We found that in general, the ease of establishing ties (outdegree) was two-fold higher in adolescents with a high SES (OR: 2.01; 95% CI: 1.126-3.588; p = 0.017). We obtained similar results for male adolescents (OR: 2.60; 95% CI: 1.139-5.962; p = 0.021) (Figure 4). Regarding reticular data, several density and centralization measures were calculated for each of the studied networks (Table 3). Additionally, the analysis of the social network indicated that overweight adolescents' social ties did not reflect their SES at any of the contact intensity levels (Tables 456). In contrast, a statistical analysis of the overall sample irrespective of weight status revealed significant results at all three levels of contact intensity (Tables S1-S9). At the minimum contact level, adolescents with a high SES were almost twice as likely to present a greater capacity for intermediation (betweenness) (OR: 1.77; 95% CI: 1.001-3.148; p = 0.049), regardless of gender. By way of illustration, node size in Figure 2 represents the capacity for intermediation (betweenness) of adolescents in one of the social networks analyzed. As can be seen, larger nodes were predominantly associated with a high SES. At the intermediate contact level, we found differences by gender, more specifically in female adolescents. Thus, females with a high SES were more than twice as likely to be nominated as friends (indegree) (OR: 2.37; 95% CI: 1.022-5.518; p = 0.042) (Figure 3). At the maximum contact level, considered to represent friendship, we obtained similar results both for the overall sample and for male adolescents. We found that in general, the ease of establishing ties (outdegree) was two-fold higher in adolescents with a high SES (OR: 2.01; 95% CI: 1.126-3.588; p = 0.017). We obtained similar results for male adolescents (OR: 2.60; 95% CI: 1.139-5.962; p = 0.021) (Figure 4). As can be seen, male adolescents with a high SES presented greatest centrality according to the degree of ties. Graphs were produced using UCINET software [40]. --- Discussion Our results showed that adolescents with a medium-low SES presented a higher probability of being overweight than those with a high SES. This agrees with several other studies that have reported an inverse relationship between SES and overweight, whereby the lower the SES, the higher the prevalence of overweight [6,8,11,[41][42][43]. One explanation for this finding may be the influence SES exerts on the purchase and/or consumption of certain food products [12,13,44], for example, the high price of healthy products [43,45], the availability of grocery stores in certain neighborhoods offering a variety of products [43], or the possibility of eating homemade food [46]. However, other indicators such as parental educational level can also influence weight status by facilitating or restricting recommended information on this subject [44,46]. It has been shown that a low educational level in parents is related to the development of obesity [10,41,42]. Family structure (separated parents, single-parent family, large family, etc.) is another cultural factor that can influence the weight status of family members [47]. Similarly, the parenting styles according to the SES condition the existence of obesity [46,48]. Furthermore, the influence of the SES on the physical activity has also been proven, finding a greater level of physical activity when the SES is higher [15,49]. Aside from the role of the parents in this fact [50], the physical structure of certain neighborhoods, including their limitations, as the lack of recreational areas or playgrounds [43], the lack of appealing low-traffic zones with green areas [51], or the physical insecurity when practicing outdoor activities [43], can condition the level of physical exercise in the adolescent, according to his socioeconomic status. Nevertheless, other studies have obtained different results to those reported here. For example, Santos found a direct relationship between SES and overweight in Brazilian adolescents, whereby the higher the SES, the higher the prevalence of overweight [52], and Zhang, Zhao and Chu obtained the same relationship in an analysis of Chinese adolescents [53]. One possible explanation that has been suggested for this association is that young people in these countries have greater access to fast food restaurants and make greater use of computers and videogames, promoting obesity and physical inactivity, maybe related to the economic growth experienced by certain countries and the corresponding sociocultural changes. circles represent individuals with a medium-low SES and diamonds a high SES. As can be seen, male adolescents with a high SES presented greatest centrality according to the degree of ties. Graphs were produced using UCINET software [40]. --- Discussion Our results showed that adolescents with a medium-low SES presented a higher probability of being overweight than those with a high SES. This agrees with several other studies that have reported an inverse relationship between SES and overweight, whereby the lower the SES, the higher the prevalence of overweight [6,8,11,[41][42][43]. One explanation for this finding may be the influence SES exerts on the purchase and/or consumption of certain food products [12,13,44], for example, the high price of healthy products [43,45], the availability of grocery stores in certain neighborhoods offering a variety of products [43], or the possibility of eating homemade food [46]. However, other indicators such as parental educational level can also influence weight status by facilitating or restricting recommended information on this subject [44,46]. It has been shown that a low educational level in parents is related to the development of obesity [10,41,42]. Family structure (separated parents, single-parent family, large family, etc.) is another cultural factor that can influence the weight status of family members [47]. Similarly, the parenting styles according to the SES condition the existence of obesity [46,48]. Furthermore, the influence of the SES on the physical activity has also been proven, finding a greater level of physical activity when the SES is higher [15,49]. Aside from the role of the parents in this fact [50], the physical structure of certain neighborhoods, including their limitations, as the lack of recreational areas or playgrounds [43], the lack of appealing low-traffic zones with green areas [51], or the physical insecurity when practicing outdoor activities [43], can condition the level of physical exercise in the adolescent, according to his socioeconomic status. Nevertheless, other studies have obtained different results to those reported here. For example, Santos found a direct relationship between SES and overweight in Brazilian adolescents, whereby the higher the SES, the higher the prevalence of overweight [52], and Zhang, Zhao and Chu obtained the same relationship in an analysis of Chinese adolescents [53]. One possible explanation that has been suggested for this association is that young people in these countries have greater access to fast food restaurants and make greater use of computers and videogames, promoting obesity and physical inactivity, maybe related to the economic growth experienced by certain countries and the corresponding sociocultural changes. With respect to the relational data, little variations have been found regarding the centrality and density data in all the studied classrooms. This fact could be explained by the sociocentric perspective of the study, where each classroom is analyzed with a single system without having into account the existing contacts with other peers outside the class. Furthermore, we also found that overweight adolescents' ties were not modified by their SES but instead were independent of their purchasing power. This fact leads us to think about the lack of homophilic characteristics in the overweight adolescent according to his SES, understanding homophily as the people's preference for interacting with those with similar characteristics [54]. A priori, these results are consistent with the literature, since overweight and obese adolescents present greater difficulty in establishing ties and tend to be more isolated [55]. In light of our results, it seems that SES did not affect overweight adolescents' willingness to establish contacts; it neither increased nor decreased their social capacity. We therefore deduce that it was their weight status which really determined their relational capacity. In contrast, other studies have found that purchasing power conditions social integration, with varying results according to the country where the study was conducted [56]. What our study has demonstrated, however, is that adolescents with a high SES present a significant trend towards socialization, irrespective of their weight status. At the minimum contact level, they occupied positions of intermediation, forming the necessary connections between the different groups established in the class. As the level of contact intensified (intermediate level), we found differences by gender. Female adolescents were more frequently nominated as friends, and thus had the opportunity to expand their social network by accepting these friendship ties. According to literature [57,58] it seems, then, that studied female adolescents are more prosocial regarding these friendship levels, being able to obtain more and greater resources from the network and protecting themselves from exclusion and isolation. At the friendship level, we found that all study participants with a high SES, but males in particular, were better positioned to form ties and therefore to establish support networks. In this regard, authors such as Nieminen et al. have already noted the power of the social environment to condition norms and attitudes that modify behaviors aimed at improving health in general and self-esteem in particular [59]. Having a more extensive support network in adolescence implies having a resource that protects various aspects of health [60]. Difficulties in forming ties can be compounded by SES: individuals with few resources in the social structure are disadvantaged by their dependence on larger networks to access social resources [61], and in turn, the configuration of social networks is conditioned by socioeconomic status. Although networks vary over the course of life according to SES [62], the presence of these deficits at an early age restricts subsequent possibilities for improvement, thus contributing to greater inequality. --- Limitations In this study, we used the FAS II questionnaire to measure participants' SES. This decision could be seen as a limitation for the study, because the most widely used indicators in this respect are the social class, based on the more highly regarded occupation of the father or mother, and the parental educational level [4,46]. The reason for choosing this questionnaire as a SES indicator was, on the one hand, the design adapted to the adolescent population and, on the other, the fact that this questionnaire avoids questions concerning the SES of the parents, which participants may not know or not wish to answer. This way, we circumvented one of the difficulties posed by a study of this nature, namely whether data should be obtained from parents or students. Other limitations found in this work are the small sample size, as well as the lack of evaluation of other variables able to modify the capacity of establishing contacts with peers, and so the difficulty to extrapolate the results to different types of population. Also, the work has not taken into account relationships outside the classroom environment. Although studying the classroom as a social system itself is something inherent to the SNA from a sociocentric perspective, it also supposes an important limitation as contacts with other, external to the classrooms, groups are unknown to the study. --- Conclusions Our study reveals the need to implement macro, meso and micro policies to combat the main problems that arise from having from a lower socioeconomic status (Table 7). In general, improving the life conditions since the early childhood stages, fighting for the equitable distribution of power and wealth, as well as recognizing the problem, measuring it and evaluating the results of the interventions are all necessary to tackle the problem effectively [63]. Equally important is the general public awareness about all the health determinants and especially this one [64]. We must, then, reconsider, and give constructive criticism, about the strategies set up at every action level, studying and analyzing what has been implemented and what has not, and keep working for an equitable community. This work highlights the social consequences of the SES at early ages, supposing this result a limitation regarding the establishment of the contact network and so a decrease in the acquisition of resources provided by the community. Having a low SES at an early age, indicates the need for actions that target children and adolescents, as well as adults, to reduce social inequalities [63,65]. It is, therefore, essential to improve the social networks of disadvantaged groups so as not to impede or hinder access to different resources precisely among those populations that need them most. Moreover, we have noticed a clear gap in the literature regarding this issue, what reveals a need for a greater amount of studies that can explore social networks and SES both in children and adolescents. Own compilation based on literature [63,66,67] --- Supplementary Materials: The following are available online at http://www.mdpi.com/1660-4601/15/9/2014/s1, Table S1: Estimation of probability of the relationship between SES in the overall sample and network parameters, at the minimum contact intensity level. Table S2: Estimation of probability of the relationship between SES in the overall sample and network parameters, at the intermediate contact intensity level. Table S3: Estimation of probability of an analysis of the relationship between SES in the overall sample and network parameters, at the maximum contact intensity level. Table S4: Estimation of probability of an analysis of the relationship between SES in the female gender of the overall sample and network parameters, at the minimum contact intensity level. Table S5: Estimation of probability of an analysis of the relationship between SES in the female gender of the overall sample and network parameters, at the intermediate contact intensity level. Table S6: Estimation of probability of an analysis of the relationship between SES in the female gender of the overall sample and network parameters, at the maximum contact intensity level. Table S7: Estimation of probability of an analysis of the relationship between SES in the male gender of the overall sample and network parameters, at the minimum contact intensity level. Table S8: Estimation of probability of an analysis of the relationship between SES in the male gender of the overall sample and network parameters, at the intermediate contact intensity level. Table S9: Estimation of probability of an analysis of the relationship between SES in the male gender of the overall sample and network parameters, at the maximum contact intensity level. Author Contributions: Conceptualization, Natalia Arias and José Alberto Ben<unk>tez-Andrades; Formal analysis, Natalia Arias, José Alberto Ben<unk>tez-Andrades and Beatriz Alonso-Cortés; Investigation, Mar<unk>a Dolores Calvo, Mar<unk>a José <unk>lvarez and Carmen Benavides; Methodology, Natalia Arias, Beatriz Alonso-Cortés and Carmen Benavides; Resources, Mar<unk>a Dolores Calvo and Mar<unk>a José <unk>lvarez; Supervision, Mar<unk>a Dolores Calvo; Validation, José Alberto Ben<unk>tez-Andrades; Writing -original draft, Natalia Arias, José Alberto Ben<unk>tez-Andrades, Mar<unk>a José <unk>lvarez, Beatriz Alonso-Cortés and Carmen Benavides. Funding: This research received no external funding. --- Conflicts of Interest: The authors declare no conflict of interest.

Socioeconomic status (SES) influences all the determinants of health, conditioning health throughout life. The aim of the present study was to explore the relationship between socioeconomic status and obesity in adolescence through an analysis of the patterns of contact between peers as a function of this parameter. A cross-sectional study was performed, analyzing a sample of 235 students aged 14 to 18 and 11 class networks. Social network analysis was used to analyze structural variables of centrality from a sociocentric perspective. We found that adolescents with a medium-low SES presented a two-fold higher probability of being overweight, but we did not detect any differences in the configuration of their social networks when compared with those of normal-weight adolescents. However, we did find significant differences in the formation of networks according to SES in the overall sample and disaggregated by gender, whereby adolescents with a high SES in general presented a higher capacity to form wider social networks. Elucidating the relationship between SES and overweight and its influence on social network formation can contribute to the design of preventative strategies against overweight and obesity in adolescents, since their social environment can provide them with several resources to combat excess weight.

Introduction The importance of a nutritious diet and an active lifestyle has been found to be central to a healthy population, which leads to lower healthcare costs for a nation and a better quality of life for its citizens. Specifically, a healthy diet and physical activity can reduce advancement of (and prevent occurrence of) chronic diseases [2], prevent vascular diseases such as cardiac disease and stroke [22,7,15], and result in overall better health [31]. Despite greater medical and scientific knowledge, North Americans continue to be more obese due to poor diet and a sedentary lifestyle. We address this problem through the design and evaluation of an online social network (OSN), grounded in theoretical models, to improve nutrition and increase physical activity. In addition to diet and physical activity, a significant factor in health outcomes is one's close and distant social networks [8,23]. Social supports can promote better health by providing a sense of belonging and building self-efficacy [6]. Social technology has been shown to increase social support for individuals afflicted with a particular condition, such as ACL (Anterior Cruciate Ligament) [25], ALS (Amyotrophic Lateral Sclerosis) [13] and menopause [36]. Furthermore, the social dynamics for peer involvement in weight management include supportive relations and passive involvement (such as social norms), and observed patterns of interaction such as (un)disclosure [24]. Studies have realized the benefits of social connection by designing technologies for health and weight management, which we review in detail in the Related Work section. Recently, social connection through websites and other web enabled technology have had an explosion of engagement and use, namely in use of OSN services such as Facebook®. We leverage this engagement as a means to build social support in the design of our OSN for health behaviour change. We also take heed of studies that have shown we need to be cautious in using large OSN services such as Facebook® for our health needs, as issues of sharing private health information on such a large social network service does not make sense [29,33]. We build on these works by designing an OSN for small social network groups rather than using existing OSN such as Facebook®. This is done to allow users to share their nutrition and physical activity data with individuals from a known social network group rather than a large social network that would broadcast their data to a diverse range of individuals known from various sources. A small social network group allows users to have control over who sees the information that is disclosed. The main contribution of this work is to apply two theories to the design, prototyping and evaluation of an OSN, VivoSpace, for health behaviour change. We combine the Social Cognitive Theory from health behaviour that underpins the importance of social aspects of healthy living with the Uses and Gratifications Theory that provides motives for participation in OSNs. The justification for this approach is that our OSN needs to be engaging in order for it to be used, so we look to the Uses and Gratification Theory to provide the factors that motivate use of OSNs [17,19]. Similarly, the factors for health behaviour change from the Social Cognitive Theory [4] are also applied to the design with an aim to change health behaviour. This theoretical approach is undertaken to create better design principles that help people to both start and maintain healthy habits. Others have used health behaviour theories to design technologies for motivating increased physical activity, improved nutrition, or both; however, the factors from the Social Cognitive Theory have not been directly used to design and evaluate technology to promote positive health behaviour. Furthermore, the Social Cognitive Theory has never been combined with the Uses and Gratification Theory to ensure engagement in the technology. Finally, OSNs have not been explicitly designed for the purpose of health behaviour change before. This work informs research in both HCI and Persuasive Technologies [12], and guides the design of the multitude of applications that are coming to market for fitness and improved diet. In our design of an OSN, VivoSpace, we chose to focus on healthy eating and physical activity as both are closely linked to health outcomes such as ideal weight, prevention of illness and management of chronic diseases [7,2,15]. We are not focusing on weight loss specifically, but rather nutritious eating habits and physical activity over sedentary lifestyle. We deployed VivoSpace to three separate social network groups in both Canada and the United States in a 4-week study with 35 participants. We use the factors from the theories to create design elements that ensure engagement in VivoSpace and changes in health behaviour. In providing this contribution, the main outcomes of this study are: 1) an understanding of different social dynamics in small social network group through use of an OSN for healthy; and 2) details of design mechanics in an OSN that influence health behaviour and motivation to continue to use the website. --- Foundational Theories Theoretical models assist us to understand human behaviour. The theories used in our approach to design, prototype and evaluate VivoSpace are the Social Cognitive Theory and the Uses and Gratification Theory. Various HCI researchers have considered the Social Cognitive Theory in designing social technologies to influence health behaviour change (as described in the Related Work Section). We build upon previous works by utilizing the Social Cognitive Theory as a foundation for our design and evaluation. We did not directly use other health behaviour theories in our approach, as we wanted to provide clarity in linking factors for health behaviour change from the theory to the design. Further, as engagement in the technology is necessary for the system to be used, we apply the Uses and Gratifications Theory to derive design strategies for motivating the use of the OSN. In other words, the OSN needs to be used in order for the design elements for health behaviour change from the Social Cognitive Theory to be effective. --- Social Cognitive Theory The Social Cognitive Theory holds that behaviour is determined through expectancies and incentives, and of key importance is self-efficacy to changing health behaviour [3,4]. The Social Cognitive Theory [4] was first drawn from the Social Learning Theory by [3], and it was further developed into five core determinants of health behaviour change: 1. Knowledge is an individual's knowledge of and expectations of outcomes related to the health risks and benefits of different health practices. 2. Perceived Self-Efficacy is an individual's own competence to perform the behaviour needed to influence outcomes and the individual's ability to exercise control over one's health habits. Perceived Self-Efficacy can be influenced vicariously through others, which allows for social modeling and the building of social norms. 3. Goals include proximal and distal goals, and they need to be concrete plans and include strategies to realize them. This factor builds accountability. 4. Perceived Facilitators include social and structural factors, and social approvals and social supports can be facilitators. They incorporate the value of the outcome of health behaviour change, which can be health, appearance, social approval, or economic gain. 5. Perceived Impediments is the opposite of perceived facilitators, and they also include both social and structural factors. --- Uses and Gratification Theory The Uses and Gratifications Theory originated from communications research and is generally recognized to be a sub-tradition of media effects research [28]. Initially, it provided a theoretical approach in defining the motivations for using traditional mass media such as newspaper, radio and television [35]. It has received renewed interest because of its applicability to telecommunications, computer-mediatedcommunication [34,11] and the Internet [30]. It has also been applied to online communities [11,19], and more recently, uses and gratifications has been studied to understand the motivations for using Facebook® [17]. We consider five key works based that present motivational factors based on the Uses and Gratification Theory on their application to the Internet, online communities, and OSNs [11,19,30,35,17]. By reviewing these works, we find that there are eight main motivational factors that influence use of OSNs based on the Uses and Gratification Theory: 1. Entertainment refers to a motivation that stems from enjoyment of playing or interacting with others [30,11,19], and also encompasses Diversion [35] and the need to Pass Time [30]. 2. Social Enhancement is a value that one ties to their status within social peers [11,17], and is also linked to Social Utility [35] and Interpersonal Utility [30]. 3. Maintaining Interpersonal Connectivity pertains to sustaining contact with one's social networks [11,19], and it also relates to Social Utility [35] and Interpersonal Utility [30], or as Joinson describes as Social Connection [17]. 4. Self-Discovery occurs when self-knowledge is obtained through social interaction [11,19], and it includes Personal Identity, which is a reinforcement of ones values, attitudes and beliefs [35]. 5. Get Information is an instrumental purpose like receiving information [11,19], and it can also overlap with Information Seeking [30] and Surveillance [35]. It also suggests social sharing, which includes factors such as Content Creation and Curation, Social Network Surfing, and Social Investigation [17]. 6. Provide Information is democratized with online communities and OSNs compared to previous media, and refers to a motivational factor that is unique from Get Information [19]. In OSNs, we see Status Updating [17] as a specific means of Providing Information. 7. Convenience provides an understanding of why people would choose to use the Internet as opposed to read the newspaper, or socialize via an OSN service as opposed to meet in person. This factor was only described by [30]. 8. Shared Identity occurs as one constructs their identities through the use of an OSN and relates to others through shared interests and values [17]. --- Related Work Although OSNs have not been studied specifically for health behaviour change, technologies with social components designed for health behaviour change have been explored by HCI and Persuasive Technology communities. We build on this work by specifically looking at the design of OSNs through the use of theoretical models. Some of the works that have incorporated social aspects in their design have considered the Social Cognitive Theory. We also review studies that investigated social aspects of health behaviour without consideration of the Social Cognitive Theory. Finally, we present some works that did not investigate social aspects for health behavour change as a central component of their study, but their results informed our work. A description of all these studies is presented in Table 1. There are no works that combined the Social Cognitive Theory with the Uses and Gratification Theory to design of an OSN for health behaviour change. Table 1. List of related works showing the name of the application developed, description of the application, and the number of participants (p) and duration of the evaluation --- Application Name Description Evaluation Stepping Up for Health (SUH) [34] Internet mediated walking program that measured the impact of adding an online community on the walking habits of individuals. --- p,16 wks VERA [5] Users take pictures at moments when they make health decisions and provide annotation about the picture's relation to health. --- p, 2 wks; 44 p, 4 wks Shakra [1] Mobile phone application infers whether the user is still, walking or traveling in a car, and provides and shares the amount of time that the user was active with some competition features. 9 p, 1 wk Fish'n'Steps [21] Links users' step counts to the growth and happiness of a virtual fish. --- p, 14 wks Houston [9] Pedometer and mobile phone application that allows users to set goals. It includes a sharing version and a personal version. 13 p, 3 wks --- Lifestyle Coaching Application [14] Smart phone and website version where users diary their food and physical activity, set goals and provide a points system for a game experience. --- p, 4 wks EatWell [16] Social mobile phone application designed to assist African American communities make better food choices, where users share audio recordings of experiences related to food. 12 p, 4 wks Chick-Clique [37] Mobile application provides sharing of step counts, visibil- ity of group averages, and text messaging capability. 7 girls, 4 days MAHI [27] Mobile application to communicate with one's diabetic educator. 49 p, 6 mo. UbiFit Garden [10] Wallpaper of the users cell phone shows a garden. Flowers appear with more physical activity, and it shows butterflies as goals are met. 12 p, 3 wks; 28 p, 3 mo --- IMPACT [20] A step-count personal informatics system that is evaluated to understand the influence of context on step count. --- p, 8 wks Fit4Life [33] A mock design of a health system that collects dietary and activity level, shares on OSN sites, and provides advice. --- N/A Since the Social Cognitive Theory postulates that health is a social matter and not just an individual one, we look at three works that have considered this theory in the design of their technologies. First, the SUH [34] study considered social support and social modeling from the Social Cognitive Theory to explain how the online community experimental group can promote increased step count and lower attrition. The study found that individuals were more engaged and participant attrition was lower with those that used the online community version. However, the Social Cognitive Theory was not central in the design and evaluation of SUH. Second, the VERA system [5] showed that the social aspects allowed participants to mimic others, and that the system built group identity and allowed for greater accountability. Although the authors did mention that the design of VERA was consistent with social modeling and accountability aspects from the Social Cognitive Theory, the use of the theory was not central in the design and evaluation. Third, the Shakra [1] system loosely based the social aspects of their design on the Social Cognitive Theory to show that community was important to health behaviour. Shakra did not show any increase in activity, but the participants enjoyed the sharing and competition aspects. We expand the use of the Social Cognitive Theory by anchoring the design and evaluation of our OSN on factors from the Social Cognitive Theory as well as the Uses and Gratification Theory. Other HCI researchers have done significant work in considering social aspects when designing health technologies without the use of the Social Cognitive Theory. The results from these studies are mixed; however, the social aspects did lead to building community empowerment and social supports. The Fish'n'Steps [21] used the Transtheoretical Model to evaluate the stage or level of health behaviour change of the study participants, but did not use the theory in the design of the system. The Houston [9] system showed those that used the sharing version met their goals more often than those that used the personal version. The Lifestyle Coaching Application [14] revealed no significant difference between mobile and web versions, or individual and team player groups. The qualitative analysis for EatWell [16] revealed that the use of the system led to a sense of empowerment for the community, as users were from the same geographic location. Chick-Clique [37] showed an increased step count for one group, but not the other, due to confounding reasons. Finally, the MAHI system allowed for social support for newly diagnosed diabetics [27], and a means to construct one's identity for those that have been living with diabetes for more than one year [26]. The variable results can be better understood with the use of theoretical models in the design and evaluation of these systems, which can provide insight into how these systems were successful in achieving health behaviour change and how they can be improved. The HCI community produced other works related to health behaviour change that inform our study, where social aspects did play a role in their work although it was not a central part of their study. First, the UbiFit Garden [10] took a theoretical approach to develop design strategies for health behaviour change; however, different theoretical models were used and the UbiFit Garden was not an OSN, so our work extends this approach to the design of an OSN to change health behaviour. Secondly, the IMPACT system was created to better understand the influence of context on personal health informatics [20], where social influence is a critical part of context. Finally, another work that deserves mention is a mock parody design called Fit4Life, which shows the importance of mindfulness in understanding our behaviour, and the privacy concerns of posting our health behaviours on large OSN services such as Facebook® [33]. Uses and Gratifications Theory, we designed VivoSpace. The VivoSpace system provides features that promote eating healthy foods and engaging in physical activity because both are important to good health behaviour. The design of VivoSpace was developed through a User-Centered Design process, initially with paper prototypes that were evaluated using one-on-one interviews with 11 people, and then with a medium-fidelity prototype that was evaluated in a laboratory with 36 people [18]. These studies revealed that the VivoSpace design allowed for greater understanding of one's health behaviour, but social gaming features were needed to build motivation to use the system. Finally, this high-fidelity prototype was tested with seven individuals using a cognitive walk-through prior to the field study described in this paper. The high fidelity prototype was developed using PHP and MySQL within the CodeIgniter application framework. Twitter's® Bootstrap was used for the user interface elements and Doctrine PHP libraries were used for the database abstraction layer. Furthermore, VivoSpace incorporates Wolfram Alpha's® Application Programming Interface (API) to obtain nutritional information for meals and caloric expenditure for physical activities logged. When the API cannot return the nutritional information, a site administrator enters the correct information based on web searches, which we anticipate will eventually be crowd-sourced. --- Fig. 2. Left: the nutritional content of a meal is displayed when the logged meal is clicked. Right: goal details show the user's and participating friend's progress towards the goal target. VivoSpace has three main components: logging, goals, and a personal dashboard. The main home page of VivoSpace is shown in Fig. 1. Users have the ability to log their meals, physical activity and weight. After users log their meals, the nutritional content of the meals is available when they click on the name of the food (Fig. 2. left). Similarly, when they log a physical activity, the caloric expenditure is available. Users have the ability to make a log private or to share it with their social network. The newsfeed shows the log entries for themselves and their friends (we will use the word "friend" to refer to social connections on VivoSpace). The shared log entries allow for commenting between the individuals in the user's social network. The historical trends for calories consumed, calories burned, and each nutrient is charted on the dashboard, which is always visible on the left side of the home page. User's can customize their targets for each of the nutrients or use the defaults, which are based on US and Canadian national standards. Users also have the ability to create concrete proximal goals from a library of goals such as walk for 60 minutes over 2 days, or eat 35 g of fiber over 3 days. They can invite friends to participate in their goals with them. The goal progress is shown on the home page, and details are shown when the user clicks on the goal (Fig. 2. right). VivoSpace also provides the following gamification features: users earn small experience points (XP) by using VivoSpace and major experience points for successfully completing goals. Users progress through 10 levels with each level requiring more points to advance to the next level. Furthermore, each level reveals a new character. These design elements are developed based on the factors from the Social Cognitive Theory and the Uses and Gratification Theory, which are mapped on Table 2. --- 5 The Study The VivoSpace OSN provides design mechanics that should promote motivation to use the system, as well as promote health behaviour change by applying factors from the Uses and Gratifications Theory and the Social Cognitive Theory (respectively). We evaluated VivoSpace in a field study with three separate social networks in both Canada and the United States. This study's contribution is the application of theories to design and evaluate an OSN for health behaviour change. Since we look at three distinct social networks, we provide a discussion on how different social dynamics within a social network contribute to healthy behaviour through the use of VivoSpace. We provide details for which design elements contribute to positive health behaviour change within VivoSpace, and how the design can be modified to further contribute to health behaviour change. Perceived Self-Efficacy Seeing historical log entries and charts of nutrients consumed on the dashboard shows one's capability to eat healthy and be physically active. Also seeing the log entries of others builds selfefficacy vicariously through others by social modeling. --- Goals Proximal goals can be created and they are always visible at the top of the homepage. Distal goals are achieved through the charts (on dashboard) displaying historical trends and targets. --- Perceived Facilitators Visibility of others' activities allows users to mimic behaviour, and the commenting on shared entries provides dialogue for encouragement. The gaming features provide motivation. Perceived Impediments The commenting feature on shared log entries can provide dia- logue with one's friends to overcome barriers. --- Uses and Gratification Theory Entertainment The gaming features include the accumulation of experience points and advancement through 10 levels. --- Social Enhancement Social enhancement is provided by: 1) visibility into the level obtained by one's friends; 2) visibility into the shared meals of one's friends; and 3) participating with a friend on a goal and seeing their progress towards the target. --- Maintaining Interpersonal Connectivity Social connection is provided through the ability to comment on friends' log entries and participate with friends on goals. --- Self-Discovery Ability to see the nutritional value of foods eaten and caloric expenditure for physical activity and the historical trends on the dashboard allow users to discover their health behaviour. --- Get Information Nutritional information is obtained for each log entry, and historical information is available on the dashboard. --- Provide Information Information is provided through logging of meals, physical activity, and weight, and through commenting on friend's entries. --- Convenience Obtaining the nutritional value of foods and caloric expenditure of physical activity, and having a historical storage of one's nutritional information provided convenience of access to information. --- Shared Identities Through the sharing of one's meals and physical activity, and allowing friends to comment on these entries. --- Methods Recruitment and Participant Overview. We recruited a total of 35 participants from three different settings including clinical and non-clinical settings. For the clinical groups, we considered patients that are associated with the same clinic, share the same geographic location, and are generally from the same socioeconomic status, as was found to be important by Grimes et al. [16]. The first group of participants from a clinical setting was recruited from a primary care clinic in suburban Chicago, which has a focus on wellness and weight loss. A total of 22 participants were recruited by sending an email through the clinic's patient portal. Half of the participants were randomly assigned to the control group and the other half to the experimental arm. The participants in the experimental group used VivoSpace during the 4-week study period, while participants in the control group did not use VivoSpace. The reason for having a control group was to control for newly acquired health behaviours from association with the clinic. The second group of participants from a clinical setting was recruited from an integrative medical center in Vancouver, Canada. This clinic provides a focus on prevention and partnership between physician and patient. Participants were recruited from their medical-group-visit patients, where patients meet with a physician as a group, so they already take advantage of social aspects to good health. A total of 3 participants were recruited from this clinic. The small number reflects the limited number of patients that take part in the medical group visits, and it means that we could not have a control group for this clinic. The participants were recruited through direct email contact from the clinic. The non-clinical group of participants was a close-knit group of colleagues at a small software gaming company in Vancouver Canada, where all the employees have good knowledge of each other. A total of 10 participants were recruited by directly asking the employees of the company. The company has a total of 35 employees. There was no control group for the non-clinical group, as they were not associated with a wellness or integrative medical clinic, and we used the results from the prequestionnaire (see Measures section below) to evaluation their health without VivoSpace. A total of 35 participants were recruited from three groups: Chicago clinic (11 in experimental group and 11 in control group), Vancouver clinic (3), and the nonclinical group from the Vancouver gaming company (10). Out of the 22 participants recruited from the Chicago clinic, 3 dropped out prior to the end of the study. One was in the experimental group, and the other two were in the control group. Table 3 shows an overview of the remaining participants from all three groups. Overall, there is a bias towards women. Furthermore, the participants from the non-clinical group were younger than those from the clinics, and this group ranked their health better than those from the clinics (health was ranked on a 6-point Likert scale, where 1=very unhealthy and 6=very healthy). Participants from each group were asked to friend each other, but there was no friending across the social networks, so that we could evaluate the social dynamic within each group. All participants received a $50 honorarium. Measures. All participants completed a pre-questionnaire prior to the start of the study and post-questionnaire after the 4-week study period. The questionnaires both had the following: 1) current health behaviour; 2) knowledge test of nutrients; and 3) 5-point Likert questions that linked back to the factors from the Social Cognitive Theory. The post-questionnaire for those participants that used VivoSpace also included the following: 1) 5-point Likert questions that linked back to the factors from the Uses and Gratification Theory; and 2) open-ended qualitative responses about their thought on VivoSpace and how it contributed to health behaviour change. Change in health behaviour was analyzed through comparison between the pre-and post-questionnaires. The first set of questions that inquired about health behaviour included: questions about height and weight to obtain a body-mass-index in order to assess if the participant was overweight or obese; the number of salads, vegetables, fruits, French fries and potato chips that they ate in the past week (no other foods were enquired about); and how often they walked, performed cardio exercise, and other physical activity in the past week. The current stage in health behaviour change from the Transtheoretical model was also inquired, as was done by [9,21,23]. The Transtheoretical Model defines the stages of change as pre-contemplation, contemplation, preparation, action, and maintenance [32]. The second set of questions was a 10-question multiple-choice test on the meaning of nutrients to assess knowledge. They included question such as, "What foods have the highest fiber content?" The answers included: a) whole wheat breads, beans, and vegetables; b) any breads and cereal grains; c) whole wheat bread and meats; and d) all of the above. The third set of questions was 5-point Likert style questions that asked for the level of concurrence on statements that mapped to the factors from the Social Cognitive Theory. For example, the statement for perceived self-efficacy for eating nutritious foods stated, "I am capable of eating highly nutritious foods and resisting unhealthy foods." The 5-point Likert questions ranged from "strongly disagree" (1) to "strongly agree" (5). The post-questionnaires for those participants in the experimental group also included 5-point Likert questions for the eight factors from the Uses and Gratification Theory. The post-questionnaire also inquired about which design features of VivoSpace contributed to the factors from the Social Cognitive Theory. For example, after the afore mentioned Likert question on perceived self-efficacy, another 5-point Likert statements inquires concurrence with, "I was able to learn about my capabilities by logging my meals on VivoSpace" The open-ended qualitative responses included general thoughts on VivoSpace and how it can be improved to motivate use and health behaviour change, and specific responses for how it can provide factors from both theories more effectively. --- Results The results cover the following: 1) whether the incorporation of the Uses and Gratification Theory into the design encouraged use of VivoSpace; 2) whether the incorporation of the Social Cognitive Theory led to an increase in the self-report on the five factors from this theory; 3) and if there was any observed change in health behaviour at the end of the 4 weeks. Analysis includes statistical comparison of pre-and postquestions as well as qualitative analysis of open-ended free form questions. --- Overview of System Use The use of VivoSpace over the 4 weeks varied substantially. All 23 participants that were using VivoSpace created an account and friended others from their network group. The participants from the non-clinical group (Vancouver gaming company) used it more than those from the Chicago clinic, which can likely be attributed to this group's ease with technology. One participant from the Vancouver clinic used the system the most. Table 4 shows an overview for usage of VivoSpace for all three groups. The nonclinical group was less inclined to share their entries than those from the clinical groups. The comments included: encouragement about the post, "good job!" and "Woot!"; additional information about the food or activity entered, "Santa Cruz organic added to Perrier"; or their feeling about the activity, "I love Thai food, yum". The questionnaire also inquired about each participant's perception of use based on the eight factors from the Uses and Gratifications Theory. The results of the 5-point Likert responses for all three groups are shown in Table 5. The strongest factors for motivating use were to Provide Information and Self-Discovery for all groups; however, interestingly Shared Identity was strong only for the loose social connections of the clinical groups. Although the Likert response for to Get Information was a bit lower with mean of 2.67 for the non-clinical group, the qualitative responses showed that they did use it to receive information; for example, "I liked watching my vitamin consumption over time" (P9). The qualitative responses for the Chicago clinic revealed sentiment for the Shared Identity motivational factor; such as, "I liked being supportive to like minded people achieving goals"(P32). The Vancouver clinic had one active user, who was the heaviest user of VivoSpace. She continued to use it on a daily basis despite the fact that the she did not have the full social experience of Vi-voSpace. This user had some mistrust in the nutritional and energy expenditure information that was provided by VivoSpace, "I'm not sure how accurate the nutritional and calorie burning info was, and I think that more detailed information would need to be made available for it to be useful" (P31). With respect to Self-Discovery, she said, "It was interesting to be accountable; it made me think more about it" (P31). Perceived Self-Efficacy. A central concept to the Social Cognitive Theory is Perceived Self-Efficacy. The paired t-tests for all groups showed a statistically significant increase between the pre-and post-questionnaires' 5-point Likert scale inquiry into one's Self-Efficacy (Fig. 3). The greatest increase was observed for the Chicago clinical experimental group and the smallest increase was observed for the non-clinical group. However, these results need to be taken with caution, as the control group from the Chicago clinic also showed an increase in Perceived Self-Efficacy, which perhaps means that there were other factors contributing to the increase in self-efficacy such as association with the Chicago clinic. For those using VivoSpace, we also inquired about the aspects of VivoSpace that contributed to their self-efficacy. Generally, the participants felt that logging their meals and physical activity can lead to increasing their understanding of their capability. --- Goals. The statistical analysis revealed no significant difference. The qualitative responses showed that users wanted to be able to customize their goals, "most of the goals currently in the system were not appropriate for me so there was nothing for me to do with other members" (P7). Making the goals more social and more central to the system also can contribute to more creation and completion of health goals, as one participant says, "putting [the goals] out there for people to see" (P32). Perceived Impediments and Facilitators. No statistically significant difference was observed for either decreasing perceived impediments or increasing perceived facilitators. Most users of VivoSpace felt that the system's ability to provide their nutritional information does assist them to reduce impediments or increase facilitators. However, as one participant points out, time constraints are the biggest impediment to healthy behaviour, and logging one's information on VivoSpace takes time (P28). This perhaps shows the need for the system to have reminders, as suggested by P1. There were some features that some participants found to be facilitators: "I like having the charts to show my progress; especially seeing my weight line go down" (P28); and "nutritional information regarding foods I was eating helped push me to keep cooking those foods or avoiding other foods" (P9). --- Changes in Health Behaviour There was health behaviour change observed between the pre-and post-questionnaires for the Chicago clinic experimental group and a strong trend toward health behaviour change for the other two experimental groups. The Chicago clinic experimental group increased the number of salads they ate in a week with a mean difference of 1.0 (t=2.65, p<unk>0.05). For the non-clinical group, the number of servings of potato chips (and similar foods) consumed in a week went down with a mean difference of -0.56 (t=1.89, p=0.095). For the clinical group (combination of both clinics), we saw a strong trend toward a positive move in the stages of change from the Transtheoretical Model [32] with a Wilcoxon Signed Ranks test (Z=1.84, p=0.066). By comparison, there was no health behaviour change observed in the Chicago clinic control group. --- Discussion An OSN can change health behaviour if it is designed with features that influence factors from theoretical models. The results for health behaviour change do not show large changes, but the small steps towards healthier living are encouraging because positive health behaviour change did occur in some way for each of the experimental groups, and did not occur for the control group. The maintenance of these new behaviours is not known at this time, but would be important to investigate further. By evaluating VivoSpace with the use of the Social Cognitive Theory and Uses and Gratification Theory, we can better understand which design elements contributed to health behaviour change; furthermore, the results reveal the factors from the theories that were not provided in the design of VivoSpace, which provide possibilities for design improvement. Logging meals and physical activities to reveal the nutritional value of meals and caloric expenditure of activities combined with charts showing trends of calories and nutrients consumed build perceived self-efficacy. The greatest increase in self-efficacy was observed in the Chicago clinical experimental group, which suggests that the process of logging meals and physical activity does build self-awareness. By logging health information, users become more accountable and aware of their capabilities. The design can be improved by also providing nutritional information for foods not eaten) and having tailored messages for how to improve health behaviour to build knowledge. The qualitative results show that the design can be modified to promote an increase in knowledge of nutritional value of food and energy expenditure of physical activities. The participants suggested that VivoSpace could assist them to make decisions about food, so it should have the ability to obtain nutritional information for foods that were not eaten. Further, VivoSpace can be more prescriptive and provide tailored messages for how the user can make healthier choices. Providing reminders for goals and allowing for customizable goals will increase user engagement. The goals feature on VivoSpace should provide reminders to the users after they are set, and goals should be more central in the design. A key change to the goals feature would be to allow users to create customized goals, as the rigidity of the goals provided on VivoSpace led to fewer people creating goals. These changes will increase Social Enhancement and make the system more Entertaining; both of which are factors for motivating use from the Uses and Gratification Theory. Email notifications should be provided when comments are made to one's log entries, and a link should be provided to the log entry and the comment. VivoSpace should provide notifications when comments are made to their

Of key importance to avoiding significant health problems such as cardiac disease and stroke is eating nutritious foods and leading an active lifestyle. However, leading a healthy lifestyle remains elusive and obesity continues to increase in North America. We investigate how online social networks (OSN) can change health behaviour by blending theories from health behaviour and participation in OSNs. We designed, prototyped and evaluated an OSN, called VivoSpace, using factors derived from the blend of theoretical models with the intention to change health behaviour. Our results from field studies with 35 participants from two clinical and one non-clinical social network groups revealed that designing these factors into the online social network lead to positive health behaviour change, namely, improved diet for the clinical group. The use of theoretical models in the evaluation provided a means to understand how the design can be improved to see further health behaviour change.

for how to improve health behaviour to build knowledge. The qualitative results show that the design can be modified to promote an increase in knowledge of nutritional value of food and energy expenditure of physical activities. The participants suggested that VivoSpace could assist them to make decisions about food, so it should have the ability to obtain nutritional information for foods that were not eaten. Further, VivoSpace can be more prescriptive and provide tailored messages for how the user can make healthier choices. Providing reminders for goals and allowing for customizable goals will increase user engagement. The goals feature on VivoSpace should provide reminders to the users after they are set, and goals should be more central in the design. A key change to the goals feature would be to allow users to create customized goals, as the rigidity of the goals provided on VivoSpace led to fewer people creating goals. These changes will increase Social Enhancement and make the system more Entertaining; both of which are factors for motivating use from the Uses and Gratification Theory. Email notifications should be provided when comments are made to one's log entries, and a link should be provided to the log entry and the comment. VivoSpace should provide notifications when comments are made to their posts, or when one responds to one's existing comment. This will increase dialogue between participants, which will build social supports to allow for an increase in the motivation to Maintain Connectivity from the Uses and Gratification Theory and Perceived Facilitators from the Social Cognitive Theory. Design should include a mobile version and other conveniences such as automatic sensing to ease the time required to use it. The time that was required to enter the data was a contributing factor for some participants to ask for easier means of entering their data. A mobile version and linking to devices such as fitbit® to automatically sense steps taken can ease the burden of data entry. The differences observed between the clinical and non-clinical groups show the potential to use an OSN such as VivoSapce with clinical groups. Overall, both clinical groups disclosed their log entries more than the non-clinical group, and the group from the Chicago clinic saw observable changes in health behaviour. Furthermore, both clinical experimental groups saw the greatest increase in self-reported selfefficacy. The Chicago clinic experimental group showed greater social motivation to use VivoSpace compared with the non-clinical group. --- Conclusions An OSN called VivoSpace was designed and prototyped based on factors derived from the Uses and Gratifications Theory for use of the OSN, and the Social Cognitive Theory for health behaviour change. Through a 4-week field trial with three social network groups, we used these theories combined with usage data and health behaviour data to evaluate VivoSpace. We were able to determine the design elements through the application of these theories that will lead to use of the system and changes in health behaviour. The study also revealed opportunities to improve the design, so more significant health behaviour change can be observed. We found that a close-knit non-clinical social network group interacts with the system differently than a group of strangers from the same clinic. An OSN such as VivoSpace can be applied to clinical groups, who are associated with the same primary care clinic to improve health behaviour. OSNs such as VivoSpace can provide a health intervention to populations with the greatest need to change health behaviour leading to a healthier population. The use of theoretical models provides an effective means to design and evaluate technologies for health behaviour change.

Of key importance to avoiding significant health problems such as cardiac disease and stroke is eating nutritious foods and leading an active lifestyle. However, leading a healthy lifestyle remains elusive and obesity continues to increase in North America. We investigate how online social networks (OSN) can change health behaviour by blending theories from health behaviour and participation in OSNs. We designed, prototyped and evaluated an OSN, called VivoSpace, using factors derived from the blend of theoretical models with the intention to change health behaviour. Our results from field studies with 35 participants from two clinical and one non-clinical social network groups revealed that designing these factors into the online social network lead to positive health behaviour change, namely, improved diet for the clinical group. The use of theoretical models in the evaluation provided a means to understand how the design can be improved to see further health behaviour change.

Introduction Many migrant physicians (MP) who arrive in their destination country intend to continue their professional careers as medical specialists. However, in some Western countries, MPs tend to fill shortages in medical specialties and sectors that are perceived as 'low status' [1][2][3][4][5]. This might suggest an inequality and imbalance in the medical specialty labour market based on the country in which a physician obtained their medical degree (cf. [4]). Further, this may indicate that MPs have fewer opportunities and decreased autonomy to choose their desired employment and specialty compared to their domestically trained peers, which can adversely impact their motivation, well-being and work-life sustainability (cf. [6,7]). Therefore, it is imperative to explore MPs' employment and specialty choices and the aspects that influence their decisions. Choosing a medical specialty is an extensive individual process that often starts during medical school [7] and is often influenced by social background and family upbringing [8], gender, demographics, lifestyle compatibility, working conditions, and salaries [9]. Interest in treating specific types of patients and medical conditions also influences the choice of specialty [8][9][10], as do personality traits [11][12][13]. In addition, a specialty's perceived status and prestige can impact the choice of specialty [8][9][10]13]. Surgical specialties, for instance, afford high status and prestige, while psychiatry often does not [8,10,14]. Moreover, low-status specialties, such as general practice (family medicine), psychiatry and geriatrics, usually suffer from staff shortages [8-10, 13, 15]. The hierarchy of medical specialties has previously been analysed and discussed through the lens of Bourdieu's concepts of social field, symbolic capital and habitus [8,10]. A social field is constructed by a group of individuals and institutions with joint interests that are struggling over resources (capital) (cf. [7,10,16]). By gaining capital (social, cultural, economic), recognized as symbolic capital, individuals are hierarchically positioned in relation to one another [8,10,16,17]. The medical labour market can be understood as a social field [8,10] or more specifically, the medical field (cf. [18]). The medical field includes many professions in addition to physicians, such as allied healthcare workers, nurses, auxiliary nurses, healthcare assistants and untrained caregivers [10]. Primary health care (PHC) and allied health professions are usually associated with lower status [10]. The specialties in which MPs choose to train vary internationally. In Norway, MPs are more likely to become specialists than domestically trained physicians; however, domestically trained physicians are more likely to become surgeons [14]. In contrast, in Finland, MPs are less likely to specialize than domestically trained physicians, but surgery is a common specialization [2]. MPs in Finland, however, often work in PHC [2]. In the UK, MPs are more likely to work in less popular specialties, such as geriatric and psychiatry [1,4]. Similarly, career advancement in underserved specialties might be easier for MPs in Ireland, where access to post-medical training for MPs is difficult [1]. Filling gaps in underserved areas is also an established way of shortening the recertification process in Canada [19]. This can also be seen in the high representation of MPs in PHC in the US and Finland [2,3,5]. Underserved areas refer to types of sector, medical specialties, and geographical locations [20,21]. Despite the societal benefits of filling such gaps, professional development and career advancement should be equally accessible to all physicians, not assigned based on the country in which a physician received their medical education. Having less freedom to pursue a desired medical specialty and location can adversely affect well-being [6]. Additionally, if the MP workforce is concentrated in particular medical specialties, the medical specialty workforce could become unbalanced. It could also be argued that the distribution of medical specialists should reflect the diversity in the society at large. In Sweden, 43% of medical specialists are women, and 57% are men [15]. In 2019, approximately 9800 physicians were in specialty training [15], almost a quarter (n = 2300) in general practice [15], making it the largest medical specialty. Physicians specializing in general practice and working as general practitioners (GPs) make up the largest employee group in PHC. In Sweden, staff shortages have been reported in PHC [8,22] and psychiatry [8,22]. There are possibilities that MPs with a medical degree from outside EU/EEA might work in medical specialties experiencing staff shortages in Sweden, such as general practice [23]. This study aims [1] to explore aspects that influence MPs', with a medical degree from outside EU/EAA, choice of employment and medical specialty in Sweden and [2] to explore and understand a potential over-representation in general practice. This study will try to contribute with knowledge to help explain aspects that influence MPs' choice of employment, why MPs may be over-represented in general practice, and if this could be interpreted as an inequality in the medical specialty labour market based on where a physician obtained their medical degree. If so, policies in order to counteract negative consequences such as decreased well-being should be developed. --- Methods The study employed a mixed-method approach, conducted within a pragmatic and interpretative research tradition, and was part of a larger research project [24]. The current data have not been presented elsewhere. Ethical approval was received from the Regional Ethical Board in Stockholm, Sweden (2017/1717-31/5). --- Context: recertification of MPs and specialist training in Sweden In Sweden, MPs with a medical degree from outside the EU/EEA who wants to continue to practice need to: [26] which is 10% of the 11.570 Swedish medical licenses that were awarded in total [27], 41% were awarded to physicians educated in another EU country of which many are Swedish citizens, and 49% to physicians educated in Sweden [26]. In 2020, approximately 41.000 physicians practised their profession [27]. --- Participants A nonprobability convenience sample of volunteer MPs with medical degrees from outside the EU/EEA were recruited via a questionnaire that was disseminated to CPP enrolees from the programme launch in 2009 and to 2017 (N = 497) with a response rate of 57% (n = 283). Respondents who indicated having a specialty training position or respondents working as specialists or as senior doctors were included in the study sample (n = 101). --- Data collection Data were gathered through an electronically disseminated questionnaire (for more information, see [18]). The Swedish Board for Health and Welfare's categorization of medical specialties was used to cluster medical specialties in the questionnaire. Specialities identified in the literature as 'low prestige' (general practice, geriatrics and psychiatry) were extracted and separated from their overarching categories. The following questionnaire items were used for the current study: (1) whether respondents had begun or completed specialist studies before migrating to Sweden, and if so, in which specialty; (2) whether general practice was considered a specialty in their education countries; (3) whether they were undergoing or had completed medical specialty training in Sweden (if so, in which specialty) and whether their specialty was their first choice (if not, why); (4) how they perceived the status of different medical specialties in Sweden; and (5) what aspects influenced their choice of employment. The questionnaire included closed-ended questions employing 3-point and 5-point Likert response and open-ended questions with free text answers. To deepen our understanding of MPs' choice of general practice and how different aspects overlap on an individual level, data were also gathered via semi-structured interviews with MPs specializing or being specialists in general practice and working as GPs. Interviewees were recruited via an email sent to MPs who had enrolled in the CPP between 2012 and 2016 (n = 278). Seven MPs in specialty training responded, four of whom were pursuing general practice and were included in the study (two men and two women). The interview guide included questions about the MPs' choice of specialty and what influenced their choice, differences in the perceived prestige of specialties, and difficulties encountered in obtaining a specialty training position. The interviews lasted approximately one hour each and were audio-recorded and transcribed verbatim. --- Data analysis The Statistical Package for the Social Sciences (SPSS) version 25.00 (IBM Corporation, Armonk, NY) was used for descriptive and inferential statistical analysis of the questionnaires. The descriptive analysis involved calculating the means and standard deviations of the quantitative variables. For the categorical variables, frequencies and percentages were generated. The data distributions were assessed visually via boxplots, by contrasting potential discrepancies among the parameters of central tendency, by evaluating the skewness and kurtosis of the distributions and by employing Shapiro-Wilk tests. Parametric statistical tests were utilized: an independent sample T-test for examining differences between groups and a one-way between-groups ANOVA for examine differences between three or more groups. The Chi-square (<unk> 2 ) test and Cramer's V (phi) effect size was used to compare proportions between two or more independent groups and to investigate associations between two nominalscale variables. The Pearson correlation coefficient was calculated to test the association between the variables. P-values were adjusted for multiple comparisons via a Bonferroni correction of primary endpoints. P-values less than 0.05 were considered statistically significant for all statistical tests. The data from responses to open-ended questions and the interview transcripts were analysed using an inductive approach to thematic analysis [28,29]. The transcripts were first read by LS for familiarization. In another round of reading, paragraphs were highlighted on the printed transcripts, and keywords related to the manifest content were noted. In addition, thoughts and interpretations of the manifest content in relation to the researcher's prior understanding of the research field and the entire dataset were noted. A spreadsheet was used for data sorting: interviewees were sorted by row; paragraphs, keywords and latent content (researcher's interpretations) were sorted by column. Themes were identified by researchers and noted in another column. The findings were discussed and subject to adjustments until consensus among all researchers was reached. Despite the fact that the aforementioned steps seem consecutively ordered, the process of analysis and search for patterns was in no way linear; rather, it was iterative and recursive. --- Results In this section, the respondents' characteristics are presented. Aspects that influenced the respondents' choice of employment are then addressed. A comparison of MPs in general practice and MPs in other medical specialties is then presented. The results section concludes with the interview findings. Mean age was 41. Of the respondents, 46% were women, and 54% were men. More than one-half (55%) were either specialists or were completing their specialty training in general practice. Of the respondents who had begun or completed specialty training before migrating to Sweden, 21% were practising in their original specialty in Sweden. Of the 79% who had changed specialty, almost 60% were now practising general practice. Lastly, 24% of the respondents reported they were not practising in their first-choice medical specialty. None of the interviewees had completed or begun specialty training before migrating to Sweden. All participants were born in the 1980s and were educated in Asia, Eastern Europe or South America. They had all migrated to Sweden between 2010 and 2013. --- Aspects that influenced choice of employment and the specialty general practice Aspects that influenced MPs' choice of employment and to specialize in general practice related to (1) work, private life and the surroundings, and to (2) medical labour market conditions. We also identified differences between respondents in general practice and respondents in other specialties that related to the respondents' characteristics as presented in Table 1. --- Work, private life and the surroundings Explored aspects that influenced the respondents <unk> choice of employment are presented in Table 2. The most influential aspect was the ability to combine work with family. Second was the ability to develop one's competence. Third was the competence of colleagues. The least influential aspects related to religion or other beliefs and being close to or having colleagues from one <unk>s country of origin or education. Gender differences related to private life and the surroundings were identified (Table 2). When comparing MPs in general practice (n = 55) with MPs in other specialties (n = 46), a significant statistical difference was found in only one aspect: 'ability to have the same patients for a long period' (see aspects presented in Table 2). This aspect influenced MPs in general practice significantly, compared to respondents in other specialties (M = 3.08 (SD = 0.93) vs. M = 2.18 (SD = 1.05), p = 0.000). A significant association was found between gender and medical specialty (general practice or other): p = 0.017, phi coefficient = 0.238 where women were more likely to specialize in general practice than men (Table 1). Female respondents regardless of specialty scored higher on some aspects related to private life (Table 1). A significant difference was found concerning respondents' perception of the status of general practice as a specialty: MPs who specialized or were training in general practice perceived the specialty to be of lower status than physicians in other specialties perceived general practice to be (M = 1.82 (SD = 0.7) vs. M = 2.15 (SD = 0.71); p = 0.028). No significant association was found between MPs in general practice and MPs in other medical specialties with regard to the variables of birth period, geographic region of origin or geographic region of medical education. However, the MPs with medical degrees from countries where general practice was not considered a specialty were more likely to specialize in general practice in Sweden than MPs with medical degrees from countries where general practice was considered a specialty: X 2 (1, n = 98) = 5.09, p = 0.024, phi coefficient = 0.228. No significant difference between MPs in general practice respondents and MPs in other medical specialties was found in relation to the item 'Was the specialty your first choice?' (Table 3). --- Conditions at the medical labour market The respondents who were not working or had specialty training in their first-choice specialty, were asked to indicate their perceived reasons for why not. The respondents had applied for, but not received their first-choice specialty. Perceived reasons were often outside the MPs' control, such as experiences of discrimination and medical labour market conditions. Ultimately, these reasons indirectly influenced MPs' choices by redirecting them towards other specialties. Some participants cited experiences of discrimination: 'I did not get it [specialty training position] even though I have a PhD, but the students who were educated in the country got it right after the medical internship. All university hospitals had the same behaviour. You get no answers despite having a PhD, post-doc, and medical Medical labour market conditions included intense market competition, which prevented respondents from getting their first-choice specialty position. Further, a lack of market contacts and references was noted as a limitation, as were a poor work environment. Perceived reasons for not working in a first-choice specialty differed between the MPs in general practice and MPs in other specialties to some extent. For MPs in general practice, the most common reasons for not working in a first-choice specialty were the length of time it would take to land a first-choice position and a lack of contacts, respectively. For MPs in other specialties, the most common reasons for not working in a first-choice specialty were the intensity of competition and a lack of contacts and references, respectively. --- Migrant physicians' choice to specialize in general practice: results from interviews In-depth information about how different aspects influenced MPs' choice to specialize in general practice was relayed through the interview stage of our study. The following themes were identified: job opportunities, positive experiences of PHC, working conditions, and family conditions. Job opportunities are plentiful in PHC, making it easier for MPs to be employed in this area. Interviewees had been offered the opportunity to participate in specialty training at the same PHC location where they had done their mandatory medical internship. The interviewees had positive experiences of PHC during temporary work conducted before or during their internship. They mentioned enjoying working at the PHC location where they had conducted their internship prior to obtaining their Swedish medical licence. This was because they had the same colleagues and had become a team. One interviewee contrasted PHC with a hospital in which no one knows anyone and one cannot make contact with others. Interviewees enjoyed the variety of work tasks, working with patients of different ages and diverse backgrounds, and the opportunity to establish a professional relationship with their patients: 'You can see a little bit of everything'. One interviewee highlighted their lack of knowledge about general practice before being exposed to it: 'I had no, not much idea about general practice before, it was more... talking, talking about cardiology, for example, neurology, and such super specialty, but not general practice as a specialty, there was no [such specialty] in [country], nor in [country]'. Working conditions, such as salaries, working hours, and not having to be on call, were also mentioned as influential aspects in MPs' choice to specialize in general practice. For instance, one interviewee had changed to general practice because dissatisfaction with the working conditions in another specialty. The working hours associated with general practice and working as a GP at a PHC location were also highlighted as convenient for physicians who were parents: 'You get home at five every day, and red-letter days you are home'. Some interviewees had considered other specialties, but when their family conditions changed, their priorities did as well: 'Then I became a mother, and all priorities changed. --- Considering that I have foreign background and therefore no relatives here, it is just my partner and me, and then I thought about on-call work, night work, weekend work and that was nothing I would [do]... I was not going to be the mother who was not with her children because my daughter is my first priority.'Some interviewees initially pursued other specialties, but different circumstances changed their direction. One interviewee did not receive their first-choice specialty position, and when their second choice was not immediately available, waiting was not an option as the MP needed an income. Like the interviewee with children, this MP faced changing priorities, illustrating how influential aspects and their importance can change over time. Other aspects of working conditions related to language: 'If I was insecure about my language [skills], I would not have chosen general practice', and to flexibility, as the general practice specialty and PHC workplace did allow participants to pursue their specific medical interests: 'I have told that this is my greatest interest, so I take --- Discussion This study aimed to explore aspects influencing MPs' choice of employment, and to explore and understand a potential over-representation in general practice. The results are discussed related to work, private life and the surroundings, and labour market conditions. Then follows a discussion that focus on over-representation in general medicine. --- Influential aspects: work, private life and the surroundings The results indicate that the respondents, regardless of specialty, valued and prioritized competence development, meaning they considered lifelong learning to be important. This is probably not specifically related to the respondents being MPs, but rather to the profession. Forty-six percent of the respondents were women, and 54% were men. This reflects the gender distribution patterns among specialists in Sweden (in 2014, 43% were women and 57% men) [15]. We found that women, regardless of specialty, scored significantly higher on some aspects that related to private life and the surroundings. We also found that women were more likely to specialize in general practice than men. The results are in line with previous research, indicating an imbalance related to gender and specialty choices [15] in which 'lifestyle factors and domestic responsibilities are much more important to women than to men' [30]. These factors and responsibilities have though been suggested to have an increasing meaning also for men [30]. Research also suggests that women's interest in certain specialties has decreased due to poor work climates [30]. Circumstances, such as working hours, can make it arduous to work in some specialties if, for example, one has few relatives nearby to help care for children. This might explain why, in the present study, the aspect 'ability to combine work with family' scored highest among all MPs, regardless of medical specialty or gender. A majority of the respondents indicated that partner and children living at home were aspects with partial to total influence when choosing employment. These aspects might have been valued differently if, for example, the respondents had not yet started a family. The respondents mean age was 41. Scientific literature has reported that MPs hold less prestigious positions [2]. In our study, the 'possibility of leadership' was not a major influential aspect for the respondents when choosing employment. This suggests that respondents' motivation to fill leadership positions is low. This may be due to competing interests, but may also relate to labour market conditions that might cause resignation [31], and decreased motivation. Attitudes might be aligned towards actual career opportunities to decrease any demotivation and frustration, thus attitudes may change over time. --- Influential aspects: medical labour market conditions Some of the study participants mentioned experiences related to discrimination as a reason for not receiving a specialty training position in their first-choice specialty. Discrimination is a barrier to entering the labour market in general for most migrants [26][27][28], and this is often true for MPs [18,32,33]. Additionally, a devaluation of MPs competence might occur during a job seeking process [18]. Intense competition could negatively impact MPs' careers. Research has suggested that domestically trained physicians are occasionally recruited to specialty training positions informally, via a recruitment process that begins during their medical programme [1,8]. This can be a difficulty for domestically trained physicians, but this aspect is probably more difficult for MPs as they receive their basic medical training abroad, and therefore do not have the same opportunities in this informal recruitment process (cf. [1]). This is reflected in our results, as a lack of contacts was cited as a reason for not working in one's first-choice specialty. Empirical studies have suggested that informal networks affect one's career advancement opportunities in Sweden, and that this is of disadvantage for migrants [18,31,34]. Returning to Bourdieu <unk>s concept of symbolic capital [16,17], our results indicate that social capital (networks) might be an advantage when competing for certain specialty training positions. This might indirectly impact choices of employment and specialty and how the physician is subsequently positioned in the medical field. The lack of sufficient social capital among MPs can be explained by them not having developed their cultural capital in the Swedish medical field, as they had their basic medical education elsewhere. --- Over-representation in general practice In Sweden, physicians in general practice constitute about a quarter of all specialists in training or working (cf. [15]). More than half (55%) of the respondents with a medical degree from outside EU/EEA were active general practitioners. We also found that some respondents had changed their specialty to general practice after migrating to Sweden. The respondents hence have a tendency to work in a specialty with staff shortages in Sweden, thus filling a gap in a low-prestige specialty which MPs usually also do in other countries [1,2,5,19,20,35]. In our study, respondents in general practice indicated that it would take too much time for them to obtain a position in their first-choice specialty. Previous research has suggested that MPs adjust their career plans to pursue general practice, psychiatry and geriatrics for the sake of job availability [2,35], rather than preferred career choice [2]. In Sweden, employers have difficulties to recruit physicians to general practice [15]. Applying for a medical specialty training position in a specialty suffering from staff shortages likely accelerates career advancement. This may be desirable as MPs with medical degrees from outside the EU/EEA in Sweden must undergo a time-consuming recertification process before they can proceed to specialist training, regardless of any specialty training they received before migrating to Sweden. In general, MPs are also older than domestically trained physicians at the same stage in their career [14,25,36]. Age in combination with labour market conditions which may include an informal recruitment process that begins in medical school, and feelings of discrimination, may help explain the tendency of MPs to work in specialties with staff shortages (cf. [1,2,5,19]). Labour market conditions might limit freedom to choose and decrease motivation to pursue a certain specialty. This can lead to stress, dissatisfaction and decreased well-being [6,7], and if change of specialty, a waste of competence. The extent of free choice can be discussed by drawing parallels to motivation (cf. [6]). Accordingly, conditions on the labour market, as well as circumstances in life, that might limit an individual's freedom of action may be internalized and consequently accepted, turned to intrinsic motivation, culminating in the perception that one's choices are free (cf. [6]). Our interpretation of the findings is that the respondents in our study did not perceive their choice to practice general practice as more involuntary than respondents in other specialities. Our findings demonstrated no differences between respondents in general practice and respondents in other medical specialties regarding whether they were practising in their first-choice specialty. We also found that respondents in general practice scored significantly higher on 'ability to have the same patients for a longer period'. Previous research has shown that an essential part of working as a GP is to develop a continuous relationship with patients [37]. Not surprisingly, this aspect also influences domestically trained physicians' specialty choice [7]. Our analysis of the interviews revealed that previous positive experiences at PHC locations and changes in family conditions were shown to influence the interviewees' choice of specialty and their attitudes towards their specialty. Their choice to practise general practice was described as closely connected to their positive experiences of working in PHC and the accompanying working conditions. Aspects that influence MPs' choice of employment and specialty, and the importance of these aspects, might thus change as time passes and life circumstances change, probably regardless of specialty. Our results show that life situation affects individuals' consideration of working conditions. As described, MPs are in general older than domestically trained physicians, and if having families, one might choose to pursue with specialties considered as lifestyle-friendly (cf. [10]), such as general practice. Other specialties that have been described as more lifestyle-friendly include public health medicine and dermatology [10]. This might also correlate with gender if, for example, women are more responsible for dropping off and picking up children from daycare. As mentioned in the beginning of the discussion, women respondents more often specialized in general practice than men, and previous research shows an imbalance regarding gender and specialty choices, which can relate to domestic responsibilities and lifestyle factors [15]. Based on these results, we suggest that labour market conditions on societal and organizational levels and circumstances of being a MP, and influential aspects on an individual level that relate to work, private life and the surroundings, and motivation are intertwined and might lead to over-representation in general practice and PHC. We found that MPs in general practice valued their specialty less than medical specialists in other specialities valued general practice. Our analysis showed that the MPs in general practice were not in the specialty for its status. These results are congruent with research that has shown that domestically trained physicians in general practice also rank the specialty as having low status [38]; that they are aware of their specialty's low status but still are satisfied with their choice [8]. Not all physicians strive for high-prestige specialties, like surgery [8]. As previous international research has suggested that specialties are ranked, with general practice considered low, we suggest that some ideas within the medical field, such as the hierarchy of medical specialties, might be transnational and thus not bound by national borders (cf. [14]). Research suggests that those who work in the PHC sector have lower prestige in the medical field [10], and that specialties with a majority of women usually have low status [13]. The present study found that more than half of the respondents had specialized in general practice, the majority in PHC. In addition, women respondents were more likely to specialize in general practice than respondent men. A negative consequence of labour gaps being filled by a certain group is an imbalanced professional community. This can create divisions in the medical field based on country of medical education, or country of origin, or gender. An imbalance might produce, maintain or reproduce differences in the statuses of specialties, thereby reinforcing other types of negative differences, such as devaluation of competence based on country of origin or education, but also based on gender. --- Strengths and limitations A limitation of the study is that only a few physicians specializing in general practice volunteered to be interviewed which may influence the qualitative results. A small sample of interviewees limited a selection based on variation regarding age, country for medical education and country of origin, and whether general practice was a specialty or not in the country of medical education. Therefore, the interesting finding revealing that respondents with medical degrees from countries where general practice was not a specialty were more likely to specialize in general practice in Sweden, was not explored in-depth. The mixed-methods design of the study was however a strength, as the interview data provide more in-depth illustration of the questionnaire data. In addition, the breadth of the questionnaire data compensated for the small sample. We also had a well-defined population for the questionnaire as all of the CPP participants in Sweden between the admission years 2009-2017 were eligible (N = 497). The questionnaire respondents (n = 283), were in comparison to the population in question, representative regarding gender, age, CPP admission year, university for the CPP, and frequency of obtained Swedish medical licence. The population mean age was 40, the questionnaire respondents 39 (n = 283), and respondents in specialist training or being specialists 41 (n = 101). The respondents in the current study were slightly overrepresented in the earlier admission years which probably reflects the process and development of becoming a medical specialist in Sweden. Therefore, we consider our sample to be representative. Another strength of the study was that the questionnaire was digitally disseminated which enabled a dynamic approach where followup questions were asked only when relevant, to reduce any survey fatigue. --- Implications for research and healthcare We recommend future studies exploring aspects that influence employment and specialty choices of domestically trained and EU-trained physicians, in order to do comparison with our study participants. Such studies would provide increased information about if and how the same medical specialty labour market may (re) produce different conditions depending on where a physician acquired their medical degree. Knowledge of similarities, on the individual level, across the different groups could be used for employers to attract physicians of diverse backgrounds to medical specialties or geographical regions where there are staff shortages, and to address inequality. Results could then be used to address needs to create sustainability in work life. Further research could be undertaken to identify advantages and disadvantages regarding the tendency for MPs to fill staff shortages in, for example, general practice, as well as the related implications for MPs, patients, and society at large. Future studies could explore how skills and knowledge of MPs specialized in general practice with a specialist training from another country can be utilized. These studies might investigate how the specialty can be enriched by MPs' additional expertise and knowledge, as such enrichment may relieve other specialist areas, even if only to a small degree. --- Conclusion Regardless of specialty, study participants valued being able to combine work with family and having opportunities to develop competences high when choosing employment. Women scored higher on some aspects related to private life and the surroundings. MPs are generally older than their domestically trained peers and have to participate in a time-consuming recertification process. Specializing in a specialty suffering from staff shortages such as general practice may reduce time to become a specialist. The present study showed an over-representation of respondents in general practice. We found that women respondents were more likely to specialize in general practice than men. When MPs encounter difficulties reaching certain specialties, they may instead choose to fill staff shortages in others. However, we did not find any indications that respondents' choice to specialize in general practice was any less voluntary than that of respondents in other specialities. MP general practitioner interviewees had positive experiences of working at PHC, and had their priorities changed which had influenced their choices. Influential aspects might evolve with life circumstances over time. The choice to specialize in general practice amongst MPs is complex, and involves different aspects that could be related to societal, organizational and individual levels, and these aspects interact. However, filling staff shortages creates a risk that the MPs' competences will be underutilized, and that the medical field becomes segmented, which have implications for both healthcare and future research. the support of TS, M<unk>, PP and GN. All authors read and approved the final manuscript. --- Availability of data and materials The dataset generated and analysed in the course of the current study is not publicly available due to ethical restrictions. The data are protected by confidentiality rules pursuant to the Swedish Public and Privacy Act, which means that no unauthorized person can access the data. This is to protect the participants' confidentiality and privacy. Any questions regarding the data can be emailed to [email protected]. --- Authors' contributions LS designed the study with the support of TS. LS developed the questionnaire with the support of TS, GN, M<unk> and PP. LS collected the data, except for some interviews that were conducted by Andreas Heiding. LS performed the analysis with the support of TS, M<unk>, PP and GN. LS wrote the paper with --- Declarations Ethics approval and consent to participate Ethical approval was obtained from the Regional Ethical Board in Stockholm, Sweden (2017/1717-31/5). Study participants gave consent before participation. --- Consent for publication Not applicable. --- Competing interests The study was part of a larger research project. Focus was placed on specialty choices. The resulting data have not been presented elsewhere. The authors have no competing interests. • fast, convenient online submission • thorough peer review by experienced researchers in your field --- • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Objective: In many countries, migrant physicians (MP) tend to fill staff shortages in medical specialties perceived as low status. The aim of this study was to explore aspects that influence MPs' , with a medical degree from outside EU/ EEA, choice of employment and medical specialty in Sweden, and to explore and understand a potential over-representation in general practice (family medicine), a specialty suffering from staff shortages in Sweden. Methods: A mixed-methods approach was applied. This included questionnaire data from 101 MPs training and working as medical specialists in Sweden and semi-structured interview data from four MPs specializing in general practice. Results: Regardless of specialty, the most influential aspects when choosing employment were the ability to combine work with family, to develop one´s competence, and to have highly competent colleagues. Women scored higher on some aspects related to private life and the surroundings. More than half (55%) of the respondents specialized in general practice, and more women than men. The MPs in general practice scored higher on the aspect 'ability to have the same patients for a longer period' than MPs specializing in other specialties. No significant difference between MP general practitioner respondents and MPs in other medical specialties was found in relation to the item 'Was the specialty your first choice?ʼ. Aspects identified in the interviews that influenced the choice to specialize in general practice related to job opportunities, positive experiences of primary health care, working conditions, and family conditions.Labour market conditions such as high competition, and the time-consuming recertification process, can influence the choice to specialize in general practice as this reduces the time to become a medical specialist. We however did not find any results indicating that MPs' decision to specialize in general practice and to work as general practitioners was any less voluntary than that of MPs who chose other specialties.

Introduction N ationally, more than one-half of all pregnancies are reported as unintended, defined as an unwanted or mistimed pregnancy at the time of conception. 1 Ineffective use of contraception, such as inconsistent condom or birth control pill use or use of the withdrawal method, is prevalent and constitutes a major risk factor for unintended pregnancies (UPs) among sexually active young women. 2 Promoting effective use of contraception is a highly economical approach to preventing UPs. For every $1.00 invested in preventing an UP, $5.68 in Medicaid expenditures would be avoided. 3 Current initiatives to reduce UP have focused on increasing access to and use of long-acting reversible contraception (LARC), such as intrauterine devices and implants, 1 but the proportion of sexually active women reporting use of condoms (14%), birth control pills (23%), or nonuse of contraception (10%) still remains high. 2,[4][5][6] In 2012, over 40% of sexually active young women reporting an UP reported inconsistent use of birth control or condoms, thus initiatives to promote effective use of contraception are needed. 3 Particular groups of women have been found to be at high risk for UP, including women who report childhood violence, live in urban areas, are in a racial/ethnic minority group, have limited education, live in poverty, and are at the youngest or oldest spectrum of reproductive years. [7][8][9][10][11][12] Exposure to violence is very prevalent among urban sexually active women. Nationally, over 6% of women reported interpersonal violence in the past year and over 4% of women reported at least one episode of childhood sexual violence and these proportions are even higher among young urban women. [13][14][15][16] In addition, violence exposure has been shown to influence reproductive health with the report of ineffective use of contraception at last sexual intercourse the highest among women reporting physical violence or history of sexual victimization. 17 Stress, depressive symptoms, and self-esteem have also been linked to use of contraception. High levels of depressive symptoms have been related to inconsistent use of contraception in cross-sectional and longitudinal cohorts. 18,19 Hall et al. found that the odds of weekly consistent contraceptive use were reduced by 47% among women reporting high depressive symptoms (odds ratio [OR] = 0.53, confidence interval [95% CI]: 0.31-0.91) and long-term consistent contraceptive use was 10%-15% lower among women reporting high levels of depressive symptoms and stress compared with women with relatively low levels of depressive symptoms and stress. 20 Recently, high levels of self-esteem have been related to reports of more consistent use of contraception among college-age women. 4 A limitation of prior research examining predictors of UP is the scant number of longitudinal studies to clarify the causal direction between violence exposure, psychological factors, and use of contraception. To our knowledge, no studies have examined the independent and interactive effects of violence exposure and psychological factors on use of contraception among high-risk women. In the current investigation, we conducted a longitudinal study among young, urban primarily minority women to test theory-driven models of the direct independent effects of both violence exposure and psychological factors on ineffective use of contraception. Based on vulnerability-stress models 21 and resilience models, 22,23 we also predicted that psychological factors would moderate the positive relationship between violence exposure and ineffective contraceptive use. Specifically, we hypothesized that (1) depressive symptoms would be a vulnerability factor and increased the odds that women with high violence exposure report ineffective contraceptive use and (2) self-esteem would be a resilience factor and decreased the odds that women with high violence exposure would report ineffective contraceptive use. By investigating violence exposure and psychological effects in the same assessment of factors related to ineffective contraceptive use, these results may inform interventions designed to promote self-esteem and decrease depressive symptoms to improve effective contraceptive use and reduce UP in this high-risk population of young women. --- Materials and Methods --- Enrollment and study eligibility The Young Women's Health Study (YWHS) recruited 315 young urban women who sought care in a high-volume, urban family planning clinic from January 2013 through November 2013 and followed each woman for 9 months. Eligible women resided in North Philadelphia, were aged 18-30 years, reported sexual activity with a man in the past 3 months, and were not currently pregnant, attending a postpartum visit, or planning to get pregnant. This project recruited women from a clinic serving North Philadelphia, a low-resource neighborhood with particularly high rates of poverty, teen pregnancy, and violence. The Temple University IRB approved the study protocol. Recruitment, consent, and baseline data collection occurred in clinic waiting rooms. Research coordinators approached each woman in the waiting room, described the purpose of the study, and screened women for study eligibility. An audio computer-assisted survey instrument (ACASI) was used to administer the baseline questionnaire. Follow-up CASI telephone interviews were conducted 9 months after baseline to assess the contraception method and consistency of use. Women were compensated ($45) for their time participating in this study, with $30 provided at baseline and $15 after completion of the follow-up interview. --- Measures Descriptive variables and covariates. At baseline, race/ ethnicity, educational status, relationship status, homelessness, sexual health history, gravidity, and substance and alcohol use were measured. Problem drinking was measured with the 5-item TWEAK (Tolerance, Worried, Eye-Opener, Amnesia, K/Cut) scale. 24 A total TWEAK score of 3 or more indicated problem drinking (yes/no). Violence. The following measures of violence exposure were collected at baseline and examined as independent variables of interest. Childhood physical and sexual violence: Women were prompted to think about their entire childhood and remember all episodes of violence before 16 years of age. Exposure to physical violence was measured with the item: ''How often, before you were 16, were you slapped, pushed, hit, punched, or beaten up by someone you know or by a stranger?'' Exposure to sexual violence was measured with the item: ''Before you were 16 years old, how often did anyone ever force you to have sex?'' Response options for both questions were never, once or twice, sometimes, often, or very often. The childhood violence measures were dichotomously classified as exposure to childhood physical violence (y/n) or exposure to childhood sexual violence (y/n). These questions have been used in previous research. 13,25 Adult physical and sexual violence: To assess violence since 16 years of age, women were asked about their experience with physical violence since 16 years of age (y/n), their experience with sexual violence since 16 years of age (y/n), and their experience with physical violence by an intimate partner in the past year (y/n) using standardized questions. 13 Community violence: Current community-level violence was measured at baseline using the City Stress Inventory (CSI), a validated 18-item scale to assess perceived neighborhood disorder and exposure to community-level violence. 26 Higher scores of CSI indicated a high perception of neighborhood violence. We also collected information on the degree of reproductive coercion reported from a sexual partner in the past year using a modified version of items that determine the amount of birth control sabotage and pregnancy coercion, as described by Miller et al. 27 Moderators. Two psychological factors were assessed to determine the role of psychological vulnerability and resilience factors on ineffective use of contraception. The 10item Center for Epidemiologic Studies depression (CES-D) scale was used to measure depressive symptoms at baseline. This scale has been shown to be reliable among urban populations. 28,29 Higher scores indicated a higher level of depressive symptoms. Scores of 10 or higher indicated high levels of depressive symptoms and scores below 10 indicated normal levels of depressive symptoms (y/n). 29 The validated 10-item Rosenberg self-esteem scale was used at baseline to measure self-esteem and higher scores indicated higher levels of global self-worth. 23,30 Scores ranged from 0 to 30, with scores below 15 indicating low self-esteem and scores over 25 indicating high self-esteem. Contraception. At the 9-month follow-up survey, women reported all methods of birth control used during the follow-up period and the consistency of use. A composite measure reflecting both the type and consistency of contraception was the outcome of interest in this assessment. First, the contraception method collected at follow-up included the self-reported use of birth control pills, the withdrawal method, condoms, Depo-Provera, intrauterine device (IUD), patch, vaginal ring, Implanon, morning-after pill, some other method, or did not use anything to prevent pregnancy. As described by others and used in the National Survey of Family Growth, the type of contraception method was then top coded and classified by the most reliable contraception method. 3,31 For example, women reporting using both birth control pills and condoms were classified as pill users since the pill has a lower failure rate than condoms. Second, the consistency of contraceptive use was defined as reporting the use of that method every time during sex in the past 9 months or the adherence to the patch, ring, or Depo-Provera. To be classified as an effective contraception user at follow-up, women had to report (1) using LARCS (IUDs or Implanon), (2) adherent use of Depo-Provera, patch, or vaginal ring during the follow-up period, or (3) consistent use of birth control pills or condoms for each sexual act during the follow-up period. Two women reported consistent and exclusive use of the morning-after pill and they were included in the effective user group. Women were classified as ineffective contraception users at follow-up if they reported (1) using the withdrawal method, (2) not using any form of contraception in the follow-up period, or (3) reported using birth control pills or condoms, but not consistently during follow-up. Thus, in this assessment, we are most interested in understanding the factors involved with consistent use of effective contraception methods to inform new strategies to prevent UP. --- Statistical methods Descriptive statistics and a series of logistic regression models were used to assess main and moderating effects. De-scriptive statistics were generated for the total sample and contraceptive user group (i.e., effective vs. ineffective contraception user). Analyses of variance (ANOVAs) or nonparametric Kruskal-Wallis tests were used to compare the contraceptive user groups on continuous variables. Fisher's exact tests or chi-square tests were used to assess the association between the contraceptive user group and categorical variables. For this assessment, the independent variables used to measure exposure to violence were childhood sexual violence, intimate partner violence, and level of community violence. Other independent variables included problem drinking and gravidity. The moderating variables were depressive symptoms and self-esteem. The outcome of interest was ineffective use of contraception over the 9-month period compared with effective use of contraception over the 9-month period. Multivariate binary logistic regression models were developed to assess the main effects and moderating effects of childhood sexual violence and ineffective use of contraception by the psychological factors (depressive symptoms or self-esteem). A p-value £0.2 in the bivariate examination was used as a cutoff to include variables to assess in the multivariate models. Given this conservative cutoff, we included variables marginally related to the outcome on the bivariate examination to assess for confounding effects in the multivariate model. Given the collinearity between depressive symptoms and self-esteem (R = -0.41, p <unk> 0.001), separate multivariate models were initially created for each moderator. The first multivariate logistic regression model included childhood sexual violence, level of self-esteem, and an interaction term (childhood sexual violence and low selfesteem compared with no childhood sexual violence and normal self-esteem). The second multivariate logistic regression model included childhood sexual violence, depressive symptoms, and an interaction term (childhood sexual violence and high depressive symptoms compared with no childhood sexual violence and normal depressive symptoms). To examine the collective effect of depressive symptoms, self-esteem, and childhood sexual violence, a final multivariate logistic model was developed. Two-tailed p-values of 0.05 were considered statistically significant in all models. All analyses were performed using SPSS 22 (IBM Corp, Armonk, NY). --- Results Of 621 women screened for study, 390 were eligible and 81% of the eligible women consented to participate. We were successfully able to recontact and reinterview 88% of enrolled women 9 months after the baseline interview (N = 276), with a 5% lost to follow-up and a 7% refusal rate. At follow-up, we found that 7 women reported trying to conceive and 11 women were not sexually active during the follow-up period; these 18 women were excluded from the analyses. Thus, the final analysis included 258 eligible women with complete baseline and follow-up data. The majority of participants were young African American women with a mean age of 22 -3.3 years (Table 1). Sixteen percent of women were Latina, 22% of women did not finish high school, and 21% of women reported ever being homeless. Over one-half reported being single and in a relationship and 11% reported >20 lifetime sexual partners. Substance use was high in this population with 15% reporting problem Violence exposure and poor psychological health were also high in this group of young women. Overall, 16% reported at least one episode of childhood sexual violence, 47% reported at least one episode of childhood physical violence, 45% reported a history of physical violence, and 34% reported violence by an intimate partner in the past year. The mean level of perceived community violence was 37.96 -11.5. Over 40% of women reported high depressive symptoms at baseline, 59% reported low self-esteem at baseline, and the mean level of self-esteem was 23.34 -5.6 (Table 1). Sixty-four percent of women reported effective use of contraception through the follow-up period, and 36% reported ineffective use of contraception. For this assessment, we were most interested in determining the role of violence and psychological factors related to ineffective use of contraception. As shown in Table 1, the experience of childhood sexual violence was significantly related to effective use of contraception. Among young women reporting childhood sexual violence, 48% reported effective use of contraception and 52% reported ineffective use of contraception ( p = 0.05). We also found that problem drinking related to ineffective use of contraception; 42% of women reporting problem drinking reported ineffective use of contraception and 58% reported effective contraceptive use ( p = 0.01). In addition, we found a higher mean number of pregnancies among women reporting ineffective contraceptive use (1.96 -2.13 and 1.72 -1.48; p = 0.01) (Table 1).The mean level of depressive symptoms was higher among young women reporting ineffective use of contraception compared with effective use of contraception (10.1 -5.8 vs. 8.7 -5.6, respectively, p = 0.09). The mean level of self-esteem was also significantly lower among women reporting ineffective use of contraception compared with women reporting effective use of contraception (22.0 -5.3 and 23.9 -5.6, respectively, p = 0.01) (Table 1). From the bivariate models, the experience of childhood sexual violence and the report of high depressive symptoms increased the likelihood of ineffective use of contraception, while the report of high self-esteem decreased the likelihood of ineffective use of contraception (Table 2). Specifically, women reporting childhood sexual violence at baseline (OR = 2.25, 95% CI: 1.15-4.39)and high depressive symp-toms at baseline (y/n) (OR = 1.71, 95% CI: 1.02-2.87) were significantly more likely to report ineffective use of contraception and women reporting higher baseline levels of selfesteem were significantly less likely to report ineffective use of contraception (OR = 0.42, 95% CI: 0.25-0.73). Including problem drinking and gravidity did not change the significant findings in these models. In the multivariate models, the experience of childhood sexual violence and low self-esteem continued to be related to ineffective use of contraception (adjusted odds ratio [aOR] = 2.69, 95% CI: 1.18-6.17, and aOR = 0.51, 95% CI: 0.28-0.93; respectively) (Table 3). The interaction term assessing the moderating effect of low self-esteem on the relationship between childhood sexual violence and ineffective contraception did not support moderation (Table 3). The multivariate logistic regression model, including high depressive symptoms and experiences with childhood sexual violence, showed that childhood sexual violence was significantly related to ineffective use of contraception, but high depressive symptoms were no longer related (aOR = 3.32, 95% CI: 1.31-8.40, and aOR = 1.27, 95% CI: 0.71-2.28; respectively). The interaction terms assessing the moderating effect of depressive symptoms on the relationship between childhood sexual violence and ineffective contraception did not support moderation by these psychological factors (Table 3). A final multivariate logistic regression model, including childhood sexual violence, high depressive symptoms, and low self-esteem, found that the experience of childhood sexual violence and low self-esteem were significantly related to ineffective use of contraception (aOR = 1.99, 95% CI: 1.00-3.97, and aOR = 0.47, 95% CI: 0.26-0.86; respectively), but high depressive symptoms were not (aOR = 1.18, 95% CI: 0.66-2.09). --- Discussion This longitudinal assessment of young, sexually active, primarily African American women aimed to examine the direct effect of violence exposure, depressive symptoms, and self-esteem on ineffective use of contraception, as well as evaluate the moderating influence of depressive symptoms and self-esteem on the relationship between violence and use of contraception. The results from this study identified both modifiable (low self-esteem) and nonmodifiable (exposure to childhood sexual violence) factors related to ineffective use of contraception, but did not find that depressive symptoms or self-esteem modified the relationship between childhood violence and use of contraception. Others have also examined the main and moderating effects of self-esteem. Morrison et al. recently reported a positive relationship between selfesteem and consistent contraceptive use, but they did not examine the role of violence exposure. 4 UP has been linked to delayed prenatal care, increased substance use and depression during pregnancy, high rates of infant mortality and preterm birth, reduced breastfeeding, and increased child and maternal violence. [32][33][34][35] Reducing the high rate of UP could have long-term individual and community effects by reduc-ing poverty, improving education and socioeconomic status for women, expanding the national workforce, and reducing rates of sexually transmitted diseases, HIV, and teen pregnancy. [36][37][38][39] These findings identified several factors to recognize when designing interventions to improve consistent contraceptive use and reduce the risk of UP among young sexually active women. A high proportion of women enrolled in this urban clinicbased study reported childhood sexual violence, nearly double the proportion reported from national surveys. 14 Prior crosssectional studies have shown a link between childhood sexual violence and later adverse health outcomes such as high-risk sexual behaviors, inconsistent contraceptive use, and UP, and our longitudinal results support a direct relationship between the experience of childhood sexual violence and ineffective contraceptive use. [40][41][42][43][44][45][46][47][48] Others have reported a role of other types of violence, such as intimate partner violence, reproductive coercion or a high level of community violence, and inconsistent use of contraception. 49,50 We found high levels of violence exposure in this sample, but did not find the experience of reproductive coercion or the report of community violence to be related to ineffective use of contraception in this study. 27,51,52 It is important to note that in this project, we used a modified version of the reproductive coercion scale developed by Miller and Silverman and did not include all the questions on pregnancy coercion. 11 These findings suggest that an experience of sexual abuse in childhood may influence long-term reproductive health behaviors and screening for a history of childhood sexual violence may be particularly important when discussing pregnancy prevention. Developing trauma-informed approaches to promote consistent and effective contraceptive use among women who have experienced childhood sexual violence could be an effective initiative to reduce UP. It should also be noted that over 15% of our sample were Latina, a group of women at particularly high risk for violence, and future studies should specifically examine the role of violence exposure and use of contraception with ethnically appropriate messages for Latina women. 53 We found that women with low self-esteem had an independent and significant increased risk for ineffective use of contraception. Limited research has examined the role of self-esteem on use of contraception while recognizing childhood violence exposure. A recent report linked high levels of condom communication assertiveness among sexually active couples and consistent condom use, but violence exposure was not assessed in this study. 54 Most research has examined the role of depressive symptoms on contraceptive use and found that high depressive symptoms among sexually active women influence contraceptive nonuse 18,19,[55][56][57] and that adolescents with higher depressive symptoms were most likely to report not using contraception, report inconsistent or incorrect use of contraception, and report early discontinuation of contraception. [18][19][20]58 We initially found a role of high depressive symptoms linked to ineffective use of contraception; however, this positive finding did not remain significant in the multivariate model after recognizing the experience of childhood sexual violence. In fact, the report of depressive symptoms in adulthood has been linked to childhood and current violence. 55 The results from this study indicate that family planning interventions that increase selfesteem among young sexually active women may improve use of contraception. Improving self-esteem may impact a young women's ability to say no to sexual advances, her assertiveness in requiring contraceptive use during sex, and her ability to discuss use of contraception with sexual partners or a healthcare professional. In fact, we found that women reporting low self-esteem in this study were less likely to report that they were certain about their ability to refuse a sexual advance by their partner, their ability to have a sexual encounter without feeling obligated to have intercourse, and their ability to promote the use of condoms with a sexual partner. In addition, these findings highlight the need to examine and better understand the interrelationship between a woman's mental health status, her prior and current exposure to violence, and the likelihood of consistent and effective use of contraception with a sexual partner. There are several study limitations that should be noted. First, the sample of women was specific to young, urban, primarily African American women reducing the generalizability of these findings. Second, although we have longitudinal data, the observed associations are subject to alternative explanations due to unmeasured variables causing a spurious association. Third, the psychological factors and contraceptive use were self-reported, a common method used in other research with community-based cohorts. 18 Fourth, methodological issues in the wording of the follow-up questions concerning consistent use (y/n) may have introduced misclassification in the assessment of consistent use of Depo-Provera. Given the limited information on the adherence of required visits for consistent Depo-Provera use, women reporting this method of contraception may have been incorrectly classified as effective contraceptive users if they reported consistent use without an additional probe regarding adherence to required provider visits for Depo-Provera injections. Fifth, we did not collect information on feelings of ambivalence surrounding UP or the role of an UP among close family members, and research has found that ambivalence and UP among family members may contribute to contraceptive choice and use among young sexually active women. 59,60,61 Finally, we did not collect information on household income or information on the experience of violence during the follow-up period. In this study, we found a strong, significant, and positive relationship between exposure to childhood sexual violence and ineffective use of contraception. In addition, low selfesteem at baseline was significantly related to ineffective use of contraception during the follow-up period. These findings suggest the importance of screening for childhood sexual violence when discussing pregnancy prevention and promoting interventions to increase self-esteem to improve consistent use of effective contraception methods to reduce UPs among young sexually active women. --- Conclusion Screening for childhood sexual violence, developing trauma-informed approaches, and designing interventions to increase self-esteem could contribute to the improvement in contraceptive use among urban minority women. --- Author Disclosure Statement No competing financial interests exist.

Objectives: Ineffective contraceptive use among young sexually active women is extremely prevalent and poses a significant risk for unintended pregnancy (UP). Ineffective contraception involves the use of the withdrawal method or the inconsistent use of other types of contraception (i.e., condoms and birth control pills). This investigation examined violence exposure and psychological factors related to ineffective contraceptive use among young sexually active women. Materials and Methods: Young, nonpregnant sexually active women (n = 315) were recruited from an urban family planning clinic in 2013 to participate in a longitudinal study. Tablet-based surveys measured childhood violence, community-level violence, intimate partner violence, depressive symptoms, and self-esteem. Followup surveys measured type and consistency of contraception used 9 months later. Multivariate logistic regression models assessed violence and psychological risk factors as main effects and moderators related to ineffective compared with effective use of contraception.The multivariate logistic regression model showed that childhood sexual violence and low self-esteem were significantly related to ineffective use of contraception (adjusted odds ratio [aOR] = 2.69, confidence interval [95% CI]: 1.18-6.17, and aOR = 0.51, 95% CI: 0.28-0.93; respectively), although self-esteem did not moderate the relationship between childhood sexual violence and ineffective use of contraception (aOR = 0.38, 95% CI: 0.08-1.84). Depressive symptoms were not related to ineffective use of contraception in the multivariate model. Conclusions: Interventions to reduce UP should recognize the long-term effects of childhood sexual violence and address the role of low self-esteem on the ability of young sexually active women to effectively and consistently use contraception to prevent UP.

KEY RESULTS: Of 855 women in the ROI, 36.8% (95% CI 33.7-40.1, unweighted) and 38.6% (95% CI 30.7-47.2, weighted) reported contraceptive use, compared to 66% of rural women in the NSFG sample. Among the ROI women, 27% had received prior 30-day SUD treatment via outpatient counseling or inpatient program and these women had increased odds of contraceptive use (aOR 1.50 [95% CI 1.08-2.06]). There was a positive association between contraception use and recent medications for opioid use disorder (aOR 1.34 [95% CI 0.95-1.88]) and prior 6-month primary care utilization (aOR 1.32 [95% CI 0.96-1.82]) that did not meet the threshold for statistical significance. CONCLUSION: WWUD in rural areas reported low contraceptive use; those who recently received SUD treatment had greater odds of contraceptive use. Improvements are needed in expanding reproductive and preventive health within SUD treatment and primary care services in rural communities. --- INTRODUCTION In the United States (U.S.), women are using drugs at increasing rates, 1 with consequent rises in overdose deaths, substance-exposed pregnancies, and other related consequences over the past two decades. 1,2 The national prevalence of maternal opioid-related diagnoses at time of hospital delivery 3,4 and neonatal opioid withdrawal syndrome (NOWS) have increased considerably, 4 with trends indicating higher rates of maternal opioid-related diagnoses and NOWS in rural compared to urban U.S. areas. 5 Reproductive-age women who use drugs (WWUD) experience unintended pregnancies at up to double the rate than the general population. 6 A contributing factor is substantially lower rates of highly effective contraceptive utilization among WWUD as compared to women who do not use drugs. 7 Nationwide household surveys including the National Survey on Family Growth (NSFG), which include no questions about substance use, likely fail to capture the reproductive health needs of WWUD, particularly those in rural communities. 8 Despite increasing awareness surrounding unintended pregnancies and associated maternal and neonatal morbidity and mortality among WWUD, little is known about disparities in utilization of reproductive health services of WWUD in rural communities. Individual, community, and environmental factors contribute to 60-90% of pregnancies being unintended among WWUD, 6 compared to 45% among the general U.S. population. 9 WWUD less often use highly effective contraceptive methods (long-acting reversible contraceptionintrauterine device or implant), and effective methods (hormonal pill, patch, injection, or ring), 7,10,11 while relying heavily on condoms for contraception. 7,12 Condoms are less effective at preventing pregnancy, given imperfect and inconsistent use. Contraceptive use and reproductive choice by WWUD is further influenced by high rates of intimate partner violence and sexual and reproductive coercion. 13,14 While not all WWUD want or need highly or moderately effective contraception, the notable differences in unintended pregnancy rates may indicate a gap in reproductive health access and utilization and necessitates further investigation and interventions. Pregnant and parenting WWUD also face considerable stigma, 15 frequently delaying or avoiding prenatal care, citing concerns around child protective service involvement and potential custodial loss. 16,17 Delays to care contribute to maternal and neonatal complications including placental rupture, premature delivery, NOWS, and intrauterine growth restriction. 18 Gender and sex-related factors unique to women contribute to their initial substance use, progression from initial use to SUD, and substance use disorder (SUD) treatment engagement and retention. 2 WWUD tend to engage in higher-risk injection and sexual behaviors, experience higher rates of intimate partner violence, and have higher risk of acquiring HIV and HCV than their male counterparts. [19][20][21] WWUD from rural communities experience additional layers of risk compared to WWUD in urban settings. 22,23 In rural communities, WWUD experience poorer health outcomes partially due to unique barriers to accessing healthcare and reproductive services (limited service availability, insurance/financial concerns, limited transportation), 22,24,25 and SUD treatment (specialist shortages, 26 drive times to specialists, 27 and ruralspecific stigma 28,29 ). Large-scale efforts are needed to expand access to family planning and reproductive services for WWUD in various care settings, particularly in rural communities where these services are generally lacking. 30 However, there are limited data on current contraceptive use in the rural U.S. by WWUD. We sought to estimate contraceptive use prevalence among WWUD in rural communities compared to a national sample of women from rural areas, one of the first studies to do so. We hypothesized that WWUD in rural communities would have lower prevalence of contraceptive use compared to a nationally representative sample of women living in rural areas. We also aimed to identify characteristics of WWUD in rural communities associated with contraceptive use and to evaluate possible associations between contraceptive use and SUD treatment utilization, health care utilization, substance use, and HIV/HCV testing. We hypothesized that recent SUD treatment and healthcare utilization would be associated with increased contraceptive utilization, those with recent substance use would have reduced contraceptive utilization, and recent HIV/HCV testing would be associated with increased contraceptive utilization. --- METHODS --- Rural Opioids Initiative -Study Design, Participants, and Data Collection This investigation is a sub-analysis of the Rural Opioids Initiative (ROI)a multi-site cross-sectional survey of people who use drugs from ten U.S. states (Illinois; Kentucky; North Carolina; New England which included Massachusetts, Vermont, and New Hampshire; Ohio; Oregon; Wisconsin; and West Virginia) with recruitment from January 2018 to March 2020. 31 The ROI collected data on demographics, drug use, consequences of use, SUD treatment, HIV/HCV screening and treatment, and healthcare utilization. Eligible participants reported use of any opioid via any administration route and/or any other drug via injection in the prior 30 days "to get high" (e.g., smoking heroin; injecting heroin and smoking cocaine; injecting methamphetamine) except for Wisconsin which limited to injection use only. Inclusion criterion for all sites was age <unk>18 years old except two states (Illinois, Wisconsin) where the age criterion was <unk>15 years old. For this analysis, our study population included all women 18 to 49 years old who were able to become pregnant determined by survey responses. The upper age limit of 49 years aligns with the NSFG. The NSFG includes women between 15 and 18 years old; however, given only two ROI sites included this younger age range, with only two participants otherwise meeting inclusion criteria, they were excluded from further analysis. All sites conducted recruitment using modified chainreferral sampling, 31,32 a strategy based on respondent-driven sampling (RDS) methods to improve sampling of "hidden" populations. 33,34 Study sites identified "seed" participants who represented local population demographics and recruited within their network. Participants were linked via referral chains when estimating weighted prevalence rates. Additional ROI data collection and management details are previously published. 31 The ROI data coordinating center collected, standardized, managed, and distributed data for analyses approved by the publication committee. All study procedures were approved by the Institutional Review Board at each site. --- National Survey of Family Growth The National Center for Health Statistics conducts the NSFG, 35,36 collecting nationally representative household estimates of family planning and reproductive health topics for men and women. 35 The NSFG does not ask about drug use and as a household survey does not include those currently incarcerated or unhoused. 8 The NSFG age limit expanded from 15-44 to 15-49 years old with the 2015-2017 cohort, and purposefully oversamples non-Hispanic Blacks, Hispanics, and teens. 35,36 The NSFG uses U.S. Census Bureau Office of Management and Budget for residence location. We included NSFG respondents residing outside of U.S. Census Metropolitan Statistical Areas to approximate ROI regions. --- Measures We determined the ROI population of reproductive-age women who were not currently pregnant but could become pregnant via survey responses. The ROI survey asked all participants their gender (male, female, transgender, other) and was designed so only participants identifying as "female" were asked their pregnancy status (yes, no, or don't know). Participants who selected "no" or "don't know" regarding pregnancy status were then asked, "Are you using any medical forms of birth control such as pills, an IUD, implant, injection, ring, or patch, or are your 'tubes tied'?" (yes, no, not applicable [no vaginal sex with a male in past 10 months], not applicable [not physically able to get pregnant right nowhysterectomy, health condition, menopause], or decline to answer). Those who answered "yes" or "no" were included in the analysis. We use the term "medical/procedural contraceptive" to encompass all medical forms of hormonal contraception (pill, injection, ring, patch) and all procedural/surgical contraceptive measures (intrauterine device, implant, tubal ligation) consistent with the ROI survey. Participants were not asked additional details about contraceptive methods, family planning preferences, or adequate information to assess use of condoms or other barrier methods as contraception. The primary variable was medical/procedural contraceptive use. Characteristics potentially associated with contraceptive use were examined on the basis of previous literature and a priori hypotheses. 9,37,38 We included categorical characteristics: age; education level; race/ethnicity; relationship status. We included yes/no characteristics: homelessness in prior 6 months; incarceration in prior 6 months; trading sex for drugs, money, or housing in prior 30 days; and having had sex without a condom <unk> 1 time in prior 30 days. We divided SUD treatment service utilization within the prior 30 days into two categories: (1) medication for opioid use disorder (MOUD) use (buprenorphine, methadone, or naltrexone), and (2) SUD treatment via outpatient counseling or inpatient/residential program. Survey participants could indicate use of more than one SUD treatment (e.g., prescribed buprenorphine and engaged in counseling). Another outcome was healthcare location in the prior 6 months with four categories: (1) ambulatory care (private clinician or community health), (2) acute care (urgent care or emergency department), (3) health department or mobile van, (4) no care received. Other outcomes included the following: prior 30-day substance use (opioids, methamphetamine, cocaine, alcohol, or tobacco) and prior year HIV/HCV testing. --- Analyses We used descriptive statistics to summarize participant characteristics, prevalence estimates, and site-specific variation of medical/procedural contraceptive use. Overall, unweighted and RDS-weighted estimates of medical/procedural contraceptive use and 95% confidence intervals were calculated using mixed-effects models. 34 Study site was the random effect in these models, and estimates were obtained by averaging over site-specific effects. We then stratified the data by site to compute site-specific estimates of medical/procedural contraceptive use and bootstrapped 95% confidence intervals. West Virginia was excluded from weighted estimates due to incomplete RDS data at time of analysis. To compare rates of medical/procedural contraceptive use between the ROI and NSFG groups, we restricted the NSFG sample of women to match the ROI inclusion criteria, excluding currently pregnant NSFG respondents, those not sexually active within the past 10 months, and those reporting infertility due to a medical condition or non-contraceptive surgery. Differences in age, race/ethnicity, and education between the ROI and NSFG samples were assessed using chi-squared test with Rao and Scott's second-order correction to account for weighting in the NSFG survey data. We fit a logistic regression model on the pooled sample and calculated marginal means to estimate the rates of medical/procedural contraceptive use in the ROI and NSFG groups after adjusting for demographic differences. Within the ROI sample, we assessed bivariate associations between contraceptive use and participant characteristics using Pearson's chi-square test. Multivariable associations were assessed using generalized linear mixed methods with sitelevel random intercepts between contraceptive use and MOUD treatment; SUD treatment with outpatient counseling or inpatient program; primary care access with ambulatory services (private clinician or community health center); recent substance use; and HIV/HCV testing. Potential confounders with p-values <unk> 0.10 in bivariate analyses were included as covariates in each model, in addition to the main independent variable of medical/procedural contraceptive use. Analyses were conducted in R v.4.0.5 using the 'lme4' package. --- RESULTS The ROI recruited 3,048 participants from eight study sites. Sixteen (0.5%) who self-identified as transgender or other gender and 30 (3.4%) women who were currently pregnant were excluded from the analytic sample. Our sample included 855 WWUD, with average age 33 years (SD 8), who were predominantly White (83%) and insured (79%) (Table 1). Overall, 50% had engaged in condom-less sex in the prior 30 days and 53% had experienced homelessness in the prior 6 months. The overall prevalence of medical/procedural contraceptive use was 36.8% unweighted (95% CI 33.7, 40.1) and 38.6% weighted (95% CI 30.7, 47.2) (Table 2). Overall and site-specific prevalence estimates of medical/procedural contraceptive use are listed in Table 2. In the prior 30 days, 85% of those in ROI sample had used opioids/heroin/fentanyl and 73% had used methamphetamine/ crystal (Table 1). Women reporting not using a medical/ procedural contraceptive were more likely to have recently used methamphetamine than those who used contraception (76% vs. 69%; p=0.021). This difference was also seen in those who used both opioids and methamphetamine (64% vs. 55%; p= 0.015) Women who reported medical/procedural contraceptive use were more likely to have received SUD treatment using MOUD in the prior 30 days (25% vs. 20%; p=0.049) and SUD treatment via outpatient counseling, or inpatient program (33% vs. 24%; p=0.006) compared to those not using contraceptives. In analyses adjusted for age, race/ethnicity, education level, relationship status, recent homelessness, and insurance status, those in the ROI sample who used SUD treatment services via outpatient counseling or inpatient program within the prior 30 days were 50% more likely to report contraceptive use compared to those who had not (aOR 1.50 [95% CI 1.08-2.06) (Table 3). There was no evidence supporting an association between contraceptive use and recent opioid/heroin/fentanyl, cocaine/crack, combined opioid and cocaine, alcohol, or tobacco use. However, those with recent methamphetamine use, whether alone or in combination with opioids, were less likely to use medical/procedural contraceptive than those without recent methamphetamine use (aOR 0.72 [95% CI 0.52, 0.99]; aOR 0.71 [95% CI 0.53, 0.95], respectively). There were 570 women included in the NSFG cohort; 46 were excluded as they were currently pregnant (7.5%). This cohort sample is representative of 7,035,913 housed women of unknown drug use status who live in rural areas. Women in the ROI compared to the NSFG sample of women had a 69% lower likelihood of medical/procedural contraceptive use (aOR 0.31 [95% CI, 0.25-0.4]) after controlling for age, race/ethnicity, and education. Women in the NSFG sample were more racially and ethnically diverse and more likely to have attained some college education or higher compared to the ROI sample (Table 4). --- DISCUSSION Medical/procedural contraceptive use by reproductive-age WWUD in rural U.S. communities was substantially lower (36.8% unweighted across all sites, 38.6% weighted all sites except West Virginia), compared to women from rural communities surveyed in the NSFG, of whom 66% reported medical/procedural contraceptive use. The NSFG likely includes some WWUD; however, given the high prevalence of recent homelessness (53%) and incarceration (40%) among the ROI WWUD, household-based surveys such as NSFG likely miss the majority of those in this highly stigmatized and marginalized population. Our findings highlight the urgency for better addressing the reproductive health and family planning needs for WWUD in rural America who experience marked socioeconomic consequences and medical complications of unintended pregnancies. We found a positive association in the ROI with prior 30day SUD treatment involving outpatient counseling or inpatient program suggesting that these interventions that often involve more time with patients are perhaps more likely to give advice regarding contraception and family planning. While recent MOUD use was greater among women using contraception than those not using contraception (25% vs. 20%), the positive association with MOUD treatment and contraception use did not meet the threshold for statistical significance (aOR 1.34 [95% CI 0.95-1.88]). This discrepancy could perhaps be explained by the high prevalence of methamphetamine use in the ROI population, and current lack of FDA-approved medications to treat stimulant use disorders. Overall, the ROI population had a low utilization of SUD treatment, with 22% recently receiving MOUD and 27% utilizing outpatient counseling, or inpatient programindicating a clear need to expand access and utilization of SUD treatment in rural communities. SUD treatment encounters are a potential missed opportunity to discuss reproductive health and offer contraceptives to those interested. 39 Among the ROI population, 44% reported they primarily utilized healthcare via ambulatory services with a positive association between using medical/procedural contraceptive and prior 6-month utilization of ambulatory services that did not meet the threshold for statistical significance (aOR 1.32 [95% CI 0.96-1.82]), after controlling for insurance status. This suggests that factors besides disparities in healthcare access and availability may be contributing to lower contraceptive use among WWUD in rural areas. Outpatient clinicians caring for WWUD are frequently addressing more acute medical and psychosocial issues and they may wait until patients are deemed more stable before discussing family planning. 40 WWUD, particularly those currently pregnant, face considerable stigmathus may be reluctant or uncomfortable accessing medical services. 41 WWUD may also have less trust in the healthcare system, 42 which, along with child custody and criminal/legal concerns, could hinder counseling on reproductive health. Some women are also equivocal about possibly getting pregnant, and WWUD more frequently report ambivalence around child-bearing or believe they cannot conceive. 43 Clinicians need to consider what constitutes patientcentered contraceptive counseling and family planning, 44 particularly given the stigma, discrimination, and historical harms WWUD have experienced and concerns around possible or perceived coercion. There is a critical need for development of women-specific integrated programs offering SUD treatment and/or harm reduction services with reproductive health services, 30,45,46 in various service delivery models. Pregnant and parenting WWUD may also benefit from these types of care models. 47,48 Rural communities, with less overall healthcare availability, have further limitations in availability of SUD treatment programs that include wraparound services (behavioral health, social services) and family-specific programs. 49 This further emphasizes the importance of primary care clinics in delivering SUD treatment and reproductive health services for WWUD in rural areas. Women with prior 30-day methamphetamine use alone or in combination with opioids had a lower likelihood of using medical/procedural contraceptive compared to those not recently using methamphetamine. Lower contraceptive use in women using methamphetamine is noteworthy given methamphetamine is frequently used for energy and sexual enhancement and is associated with higher-risk sexual behaviors and STI, HIV, and HCV transmission. 50 Of the 855 ROI participants, 50% had reported prior 30-day condom-less sex. Methamphetamine is also disproportionately used in many rural communities, 51,52 and 73% of women in our study reported recent use. Research into the most effective contraceptive options and family planning services for women who use methamphetamine is warranted. HIV/HCV testing within the last year was low, at 35%, regardless of contraceptive use. Women who inject drugs are at higher risk for acquiring HIV, HCV, and other infections, given high rates of needle sharing, high-risk sexual behaviors, transactional sex, and engagement in sex work. 21 Women who inject drugs and those engaging in transactional sex and sex work should be screened for HIV and offered pre-and postexposure prophylaxis (PrEP/PEP) to reduce HIV acquisition risk. 53 Testing and treating reproductive-age WWUD for sexually transmitted and blood-borne infections is important for their health and to reduce the likelihood of vertical transmission if they were to become pregnant. 54 Efforts are needed to expand adoption of clinical practice guidelines around prescribing PrEP/PEP, screening for infectious disease, and offering highly effective HCV curative treatment for WWUD in rural communities in various treatment settings. --- Limitations and Future Directions There are limitations to our study. When participants were asked about contraceptive use, one choice included not using contraception as they were "not physically able to get pregnant right now" with reasons including "hysterectomy, health condition, menopause." Given the incorrect assumption among many WWUD that they are unable to get pregnant, 43,55,56 our sample may underrepresent the number of reproductive-age women who could become pregnant. The cross-sectional survey design only assessed contraceptive use at one timepoint. We also did not collect more detailed information about specific contraceptive methods thus limiting assessing differences in use of highly versus moderately effective methods or use of combined methods. Further research is needed to study contraceptive and family planning preferences and how to provide patient-centered contraceptive counseling for WWUD. 44 The survey also did not ask participants' plans or desire for pregnancy or about their use of condoms or other less effective methods including rhythm or withdrawal as primary or additional form of contraception. The ROI sample also lacked racial/ethnic diversity, a major limitation, given Black, Latina, and multi-racial women face increased rates of reproductive coercion and unintended pregnancy. 57 They also experience gendered racism and reproductive harms by the healthcare system with ongoing mistrust of contraceptive counseling. 58 Finally, the survey only asked female-identifying participants to answer questions about contraceptives and pregnancy. We note that female-identified individuals are not the only people who use contraception, need reproductive services, and can get pregnant. Studies are needed on approaches to best serve people of all racial/ethnic and gender identities who use drugs and who can become pregnant. Some sites had large variation between weighted and unweighted prevalence estimatesparticularly Illinois which went from a 31.1 to 61.9% estimated prevalence. This can be explained by very short referral chain lengths in Illinois for women who reported contraceptive use, leading to these participants being weighted more heavily. The smaller the networkthe number of community connections a participant hasthe more these participants are weighted in statistical calculations to account for the possibility that those with smaller networks (shorter referral chains) represent a more difficult to reach and recruit population when calculating prevalence estimates. --- CONCLUSION Despite these limitations, our study provides important insights into the lower prevalence of contraceptive use among WWUD in rural communities compared to a general population of rural women. Interventions that expand access to and improve integration of reproductive health services within SUD treatment, primary care, and harm reduction programs for WWUD are urgently needed in rural areas. --- Conflict of Interest: The authors declare that they do not have a conflict of interest. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. Publisher's Note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

BACKGROUND: Women who use drugs (WWUD) have low rates of contraceptive use and high rates of unintended pregnancy. Drug use is common among women in rural U.S. communities, with limited data on how they utilize reproductive, substance use disorder (SUD), and healthcare services. OBJECTIVE: We determined contraceptive use prevalence among WWUD in rural communities then compared estimates to women from similar rural areas. We investigated characteristics of those using contraceptives, and associations between contraceptive use and SUD treatment, healthcare utilization, and substance use. DESIGN: Rural Opioids Initiative (ROI)crosssectional survey using respondent-driven sampling (RDS) involving eight rural U.S. regions (January 2018-March 2020); National Survey on Family Growth (NSFG)nationally-representative U.S. household reproductive health survey (2017-2019). PARTICIPANTS: Women aged 18-49 with prior 30-day non-prescribed opioid and/or non-opioid injection drug use; fecundity determined by self-reported survey responses. MAIN MEASURES: Unweighted and RDS-weighted prevalence estimates of medical/procedural contraceptive use; chi-squared tests and multi-level linear regressions to test associations.

ess are different from men's experiences since many of them are subjected to gender-based violence and persecution (Pittaway & Bartolomei, 2001;Pulvirenti & Mason, 2011;United Nations High Commissioner for Refugees et al., 2016). Women and girls who are normally protected because of their gender now lack that protection and are at risk for gender-specific human rights violations (United Nations High Commissioner for Refugees, 2014b; United Nations High Commissioner for Refugees et al., 2016). Refugee women and girls are especially vulnerable to sexual gender-based violence in refugee camps, where they are often required to walk long distances to get important resources such as firewood. Moreover, they are still at dramatically elevated risk for gender-based violence through migration routes in Europe and even after resettlement (Robbers et al., 2016). The lifetime prevalence of sexual violence against girls and women over 15 in the general population was 11%, while it reached 69.3% of migrants and refugees (Keygnaert et al., 2012). Additionally, gender-based and sexual violence has long-term impacts on refugee women and girls' mental health and well-being including post-traumatic stress disorder (PTSD), depression, anxiety, and suicidality (Robbers et al., 2016). Gender-specific human rights violations have harmful impacts on refugee women's physical and psychological states, which are further exacerbated by the challenges of resettlement (Deacon & Sullivan, 2009;Hynes & Cardozo, 2000;Miller et al., 2002). --- Mental Health Challenges Refugee women have been found to have especially high rates of psychiatric symptoms. Indeed, the traumatic experi ences and post-migration challenges refugee women endure have the potential to reinforce increased vulnerability to psychiatric symptoms and distress including symptoms of trauma, anxiety, depression, somatization, and PTSD (Schweitzer et al., 2018). Given the higher risk of experiencing mental health issues as a refugee during childbearing age (Korukcu et al., 2018), refugee mothers are especially vulnerable to developing mental health issues due to their rapidly changing living situation and circumstances (Degni et al., 2014). For instance, refugee mothers in camps face a high risk of developing a mental disorder, such as depression, and to experience suicidal ideation. Indeed, a study revealed that 36% of the mothers in camps, were diagnosed with a mental disorder, 91% percent of which reported having had suicidal thoughts in the past month ( Rahman & Hafeez, 2003). Furthermore, Schweitzer and colleagues (2018) found that the responsibilities of caring for a child as a refugee represent a risk factor for refugee mothers experiencing higher levels of trauma symptoms and contribute to the development of psychiatric problems such as anxiety and somatic symptoms (Schweitzer et al., 2018). Newly resettled mothers are less likely to seek out professional services due to various barriers including stigma, language, poor knowledge of community services, and prioritizing their children's needs to the detriment of their own well-being (Nicholson et al., 1998); refugee mothers' focus on supporting their children and coping with these barriers leave them at an elevated risk for mental health problems (Betancourt et al., 2015). Thus, refugee mothers face a unique set of challenges that require specialized evidence-based interventions. --- Language Barriers Moreover, the language barrier through the resettlement process is an impediment to refugee women's access to serv ices which in turn can negatively affect their psychological well-being (Hou & Beiser, 2006). This barrier is especially salient in the post-resettlement context where refugee women seldom have enough time and resources to address their educational needs and learn the language of the host country (Watkins et al., 2012). Furthermore, the language barrier disproportionately affects women, as immigrant men have been shown to achieve better proficiency of the host country's language than women due to inequalities in social and educational opportunities (Hou & Beiser, 2006). A Canadian study found that refugee women's English fluency was longitudinally associated with higher rates of employment and lower rates of depression (Beiser, 2009). In a study conducted in Germany, difficulties related to the language barrier were found to be a widespread concern for Syrian refugees which impeded their integration, their ability to socialize, and their access to proper healthcare (Green, 2017). Similarly, an Australian study found that having limited English proficiency increased refugee mothers' risk for marginalization, isolation, and family dysfunction, while a stronger proficiency made them more comfortable accessing mainstream services (Riggs et al., 2012). Therefore, it is critical that language education services consider the needs of refugee mothers to support their host-country language fluency. The mental health difficulties of refugee mothers such as depression, anxiety, trauma, suicidality (Rahman & Hafeez, 2003;Schweitzer et al., 2018), require specialized services because more general mental health services do not simultaneously address their cultural and linguistic backgrounds, their need for community, their mistrust and difficulties navigating healthcare systems in host countries, and the unique adverse experiences brought about by being a mother through the process of forced displacement (McLeish, 2005;Schweitzer et al., 2018;Tsai et al., 2017). Thus, the present paper will consist of a systematic review of interventions for refugee mothers while using specific inclusion and exclusion criteria, identify gaps in the literature, and provide research contributions and practical implications and recommendations for future evidence-based interventions and research. --- Method Search Strategy An electronic search was performed on PsycINFO (the year of 1806 to March 16, 2020), MEDLINE (the year of 1946 to March 16, 2020), and Web of Science Core Collection (accessed March 16, 2020). The search was limited to articles in the English language. The time periods were selected to encompass the year the database was founded to the present date. The following search was performed on all three databases: refugee* AND mother* AND interven*. The asterisk was used in order to truncate search words and include articles that use similar words such as mothers or motherhood, refugees, and intervention or intervening. A second researcher simultaneously conducted the search. The screening process, choice of articles, inclusion and exclusion criteria involved both researchers to ensure inter-rater reliability. Studies were included based on the following inclusion and exclusion criteria: (a) presents an original intervention, (b) focuses on refugee populations, (c) includes or targets mothers, (d) differentiates between data gathered with mother and father refugees, and (e) is not solely medical in scope, (see Figure 1). --- Results The study selection process comprised three databases and incorporated specific inclusion and exclusion criteria. This search resulted in 132 studies. 112 studies were excluded because they did not present an original intervention, with many being qualitative in-depth interviews, research on refugee mothers' reproductive health, vaccine research, and post-partum depression data research. Two studies were excluded because they did not focus on refugee populations, nine were excluded because they did not target or include refugee mothers, five were excluded because they did not differentiate between the needs of refugee mothers and fathers, and one study was excluded because its scope was solely medical. Three studies were included from this search, and three additional studies were found through a subsequent snowball search, which entails consulting the bibliography for relevant papers. Six relevant studies that involved a total of approximately 350 participants were included in the review. --- Interventions for Refugee Mothers --- Children and Mothers in War: An Outcome Study of a Psychosocial Intervention Program The aim of the psychosocial mother-child intervention was mainly to improve child development and well-being through better mother and child interactions, support and education (Dybdahl, 2001). The participants were 87 Bosniandisplaced mother-child dyads. They were divided into an intervention group (psychological intervention and medical care) and a non-intervention group (medical care only). The intervention was based on therapeutic discussion groups for traumatized women in the context of war and the International Child Development Program. The group discussions also involved topics related to mothers' mental health such as symptoms of PTSD, traumatic experiences, exposure to violence, and other adverse experiences lived in refugee camps. Before the start of the intervention, both the mothers and children's psychological, cognitive and physical health were monitored. The study was culturally specific, and Bosnian mental health specialists reassessed and reviewed the instruments used in the study and intervention for cultural specificity and appropriateness. Mothers' trauma and wellbeing were measured using the Impact of Event Scale (IES) and the mothers' interviews pre-intervention and post-intervention provided demographics, perceived social support, and living conditions informa tion. Mothers who were in the intervention group perceived more social support after the intervention and felt they had people to go to for support and advice. Indeed, there was a significant increase in perceived social support between pre-test and post-test in the intervention group, while there was a slight decrease in the non-intervention group (Dybdahl, 2001). There was a non-significant increase in total social support scores; pretest (M = 4.4) smaller than post-test; (M = 5.0), p >.05. There was a significant increase in social support for "advice" scores; pretest (M = 4.7) smaller than post-test; (M = 5.4), p <unk>.05. There was a non-significant increase in mother's well-being for "today", pretest (M = 4.1) smaller than post-test, (M = 4.4), p >.05, and there was a significant increase in mother's well-being for "usually", pretest (M = 3.5) smaller than post-test, (M = 4.6), p <unk>.05. There was a non-significant decrease in mother's well-being for "prefer", pretest (M = 6.2) bigger than post-test, (M = 5.8), p >.05. There was a significant decrease, pretest (M = 71.2) bigger than post-test, (M = 56.1), p <unk>.05 in total IES scores and on the hyperarousal symptoms, pretest (M = 22.7) bigger than post-test, (M = 16.7), p <unk>.05 subscale for mothers in the intervention group, which was significantly larger than the decrease observed in the control group. However, there was no significant difference between the decrease observed in the intervention group and the non-intervention group on the avoidance and intrusion symptoms subscales. Dybdahl ( 2001)'s study is a good example of an intervention program that addressed the needs and prognosis of refugee mothers. However, the data gathered focuses heavily on the children and provides limited insight into the Search Strategy Flowchart impact of the intervention on mothers. While mothers' mental health problems are taken into consideration in the conception of the intervention, it is unclear whether it is efficient in promoting their well-being. Mothers' wellbeing was only lightly discussed in relation to children's wellbeing (Dybdahl, 2001). It is important to note that though the study demonstrated a decrease in mothers' IES scores, those outcomes were not discussed sufficiently. Indeed, the researcher acknowledged that discussing mothers' mental health in detail goes beyond the scope of the paper, but that the IES measures revealed a high level of distress both before and following the intervention, and that the IES scores were related to trauma exposure. It is worthwhile to highlight that the study aimed to be culturally specific and consulted numerous Bosnian mental health workers as it has been shown that culturally specific approaches are beneficial for refugee populations (Tsai et al., 2017;Williams & Thompson, 2011) --- Collaborative Health Education for Somali Bantu Refugee Women in Kansas City The goal of this study was improving the health literacy of Somali Bantu refugee mothers. Eleven Somali Bantu refugee mothers resettled in Kansas City were recruited through a resettlement agency to participate in a health education program held weekly over 12 months consisting of a total of 42 sessions of 90-minutes each (Mulcahy et al., 2019). Rooted in community-based collaborative action research (CBCAR), the small-group sessions took into account the needs and interests of mothers, and consequently included various topics such as family health, nutrition, sexuality, prenatal health, child safety, and mental health. To gain an understanding of the health narratives of refugee mothers through the resettlement process, each mother also participated in an individual interview. Additionally, they completed questionnaires pertaining to their health behaviors, the content of the testing sessions and whether they recalled the information presented. Nutrition was the most requested and discussed topic. Pearson co-efficient analyses revealed a positive correlation between the number of times a topic was presented and the number of women who retained the topic (r =.852, p <unk>.01). Nutrition was the topic with the highest retention, as it was the topic that was presented the highest number of times. The women were interested in discussing nutrition in relation to their own health and that of their child, and nutrition in relation to prenatal health. PTSD symptoms were discussed in terms of the women's exposure to potentially traumatizing experiences, hardship, and violence, but were not endorsed by participants during individual interviews. Finally, the women expressed being generally satisfied with their healthcare access, interactions with doctors, and access to interpreters. Mulcahy and colleagues (2019) conducted the study to improve the health literacy of Somali Bantu refugee mothers. However, the mothers were asked about PTSD symptoms as "feeling sad" by culturally representative professionals with experience working with refugees. Moreover, recommendations in the literature highlight the importance of providing refugees with opportunities to cope with distress and negative affect without forcing them to relive their traumatic experiences (Hansen & Houston, 2016). In addition, women did not endorse PTSD symptoms when individually inter viewed, and the results of the study do not provide information regarding their mental health through the course of the intervention. No other symptoms of mental illness or psychological distress were discussed. Furthermore, despite holding a full session on the topic of mental health, the impact of the session on participants was not discussed in the paper. Despite claiming that some sessions focused on mental health, mental health was not discussed in the article beyond symptoms of PTSD. Additionally, it is stated that retention of the material on mental health was null, and that they found an "inconsistency between personal narrative and recognition of symptoms" (Mulcahy et al., 2019, p. 5). Unfortunately, mental health remains one of the biggest outcomes of concern for refugee mothers and this study did not properly assess nor intervene on refugee mothers' mental health symptoms. --- A Home-Based Intervention for Immigrant and Refugee Trauma Survivors Visiting Moms is a program for high-risk refugee and immigrant mothers and their infants based at the Massachusetts General Hospital Chelsea Health Care Centre (Paris & Bronson, 2006). The intervention program focused on the needs of both mothers and infants and their functioning as a dyad and adopted a community-based approach in which paraprofessionals (obstetricians, midwives, pediatricians, and psychotherapists) provided at-home visits to participating new mothers. To be eligible, due to limited resources, immigrant and refugee mothers underwent a screening process to establish existing risk factors including severe depression, isolation, trauma, risk of child abuse, health needs, and safety of the family environment. 105 mothers participated in the study. The intervention comprised multiple aspects such as education on child development, family advocacy, and social support. In its conception, the intervention program was claimed to be highly individualized, provided holistic support for mother-child dyads, and was rooted in evidence-based practice as well as local realities. However, no details were provided in the article to supplement those claims. Parent-child interactions, and additional stressors, such as poverty and language barriers, were assessed and taken into consideration. The home visitors intervened by using relationship-based model (Lieberman, 2004;Paris & Bronson, 2006). The first step of the intervention was for the home visitor to listen to the mother, which made home visitors realize that the families needed basic supplies. The paper does not specify how information was elicited from mothers. The second step was developing a working alliance. The alliance was developed as the mother and the home visitor got to know each other every visit. The home visitor engaged in modeling and self-disclosure as communicational tools to develop an alliance, to teach the mother certain practices, and to make the mothers more comfortable opening up to the home visitor about their troubles. The last step was expanding the relationship to community supports, such as giving the mothers access to English classes or helping them get a library card. No data was collected to assess outcome measures, as the paper consists of a description of the program. One positive facet of the intervention was advocacy: the home visitors advocated for the mothers and connected them to resources and to the community. The home visitors developed a working alliance with the mothers by supporting their autonomy and fostering their competence. Autonomy was fostered by the home visitor teaching the mother skills that could be useful in the host country. This is important to highlight, as the literature review previously mentioned that newly resettled refugee mothers are less likely to seek professional services because of various challenges, such as stigma and poor knowledge of community services (Nicholson et al., 1998). Thus, the Visiting Moms intervention attempted to tackle important issues refugee mothers face, such as loneliness and lack of knowledge and access to resources. However, the outcomes of Visiting Moms were not formally studied. Therefore, there is no empirical evidence of the efficacy of the program and its impact on participating refugee mothers. Finally, the paper implied different cultural practices may be dangerous for the children, and that they taught mothers to practice those safely. However, the paper did not present an empirical method to define what is safe and unsafe for the children and operated under the assumption that western child-rearing practices are superior to other cultures. --- The Moving Forward Project: Working With Refugee Children, Youth, and Their Families The Moving Forward Project was an intervention program for refugee families based in Saskatchewan, Canada (White et al., 2009). The objective of the intervention was to provide knowledge and skills to refugee parents and youth to support them in efficiently addressing issues pertaining to trauma in the context of resettlement. The second goal was to bring awareness and knowledge of the resources and programs available to refugees and immigrants. The third goal was to improve service providers' capacities, so they can respond better to refugee families' needs. The goals were to be achieved through education, group discussions, resource development and dissemination. The first intake group included seven to ten families from Sudan and Afghanistan, and the second in-take session included eighteen families from Colombia, Afghanistan, Sudan, Burma, Rwanda, Congo, Egypt, Mongolia, Bosnia, and Burundi. There were six-week sessions held with the participants. The intervention's group sessions focused on topics such as the impact of trauma on the family, positive coping skills, and problem-solving skills. It should be noted that the authors did not disclose their data collection process or method. Nevertheless, the study highlighted the importance of relationship-building and support between women. Indeed, they stated that according to theories on group principles, participants would learn they are not alone in their experiences, and they could consequently learn from each other and support each other. They found that the language barriers limited the efficacy of the intervention and that the mothers preferred to talk about sociocultural integration and making Canadian friends instead of talking about past experiences. Finally, they found that the session ended up being gatherings, a construction of a safe space for sharing, and that making the groups open to newcomers contributed to the construction of a safe space where mothers cried, talked, bonded, and laughed together. A safe space in this context refers to any physical space where refugees can feel physically and emotionally safe to express themselves and build social networks (United Nations High Commissioner for Refugees, 2014b). A problematic aspect of the study was how the researchers stated that the language barrier and the refugees' tendency to speak to each other in their mother tongue was a difficulty during the sessions. As the literature has shown, refugees have language difficulties when they arrive to their host country (Green, 2017;Watkins et al., 2012). Conducting an intervention with refugees in the host country language, in a language they either do not know or are struggling with, is a methodological problem. The authors sought to teach the refugee mothers skills (e.g., problem-solving and coping skills to deal with trauma) but claim the refugee mothers discussed their future in Canada instead. However, it is possible that the inability to communicate comfortably in their mother tongue contributed to a general discomfort resulting in hesitancy to share more personal information, although this would need to be evaluated. Certainly, evidence stresses the importance of using a professional interpreter with refugees when professionals do not speak the language, as the language barrier is a major challenge in providing accurate and adequate healthcare (Kavukcu & Alt<unk>ntaş, 2019;Williams & Thompson, 2011). Finally, though the intervention claimed to delve into trauma, the researchers failed to consider the more specific mental health challenges of refugee mothers and their willingness and psychological readiness to discuss those issues in a language they are not comfortable in. --- Sweet Mother: Evaluation of a Pilot Mental Health Service for Asylum-Seeking Mothers and Babies O'Shaughnessy and colleagues (2012) evaluated a novel pilot intervention for refugee mothers and their infants called Sweet Mother. The objective of Sweet Mother was to promote participants' mental health who have been exposed to adverse circumstances through the resettlement process during the perinatal period. A total of 13 mother-child dyads participated in the study. However, only seven dyads attended many or all of the 21 group sessions, while six dyads only attended between one and four sessions. The intervention was rooted in attachment theory and focuses on building on the mothers' strengths to foster the development of a positive mother-child relationship. Moreover, it adopted a community-building approach to mitigate the negative impact of being separated from their home communities and adopted a participatory approach where children's needs helped shape the group sessions. The intervention consisted of therapeutic infant-mother group sessions lead by specialists. Mothers also participated in individual interviews, in reflective group discussions, and completed questionnaires pertaining to their relationship with their babies at each session. Thematic analysis of the reflective exercises revealed the mothers had an overall positive experience with the intervention. They expressed appreciating a new sense of togetherness, highlighted the importance of their babies socializing with other babies, expressed feeling safe in this group, learned about motherhood and parenting, and valued discussing and strengthening their relationship with their babies. Finally, the CARE-Index, an observation measure for adult-child dyads, revealed that two participants improved the quality of their mother-infant interactions from "seriously compromised" to "of concern", while two mothers remained at the cusp of "of concern" and one mother's scores increased within the "good enough" range (O'Shaughnessy et al., 2012). It is important to note that the results of the Sweet Mother intervention provided an encouraging insight into an intervention model for at-risk refugee mothers with young children. However, as this was pilot study, the scope of the results was quite limited. Only five mothers were evaluated using the CARE-Index, limiting our understanding of the impact of the intervention on mother-child dyads. Additionally, one of the main objectives of the study was to "support maternal mental health by reducing isolation and increasing access to community resources" (O'Shaughnessy et al., 2012, p. 217). Yet, no screening for exposure to traumatic experiences or psychiatric symptoms was performed. Likewise, no identified outcomes were examined vis-a-vis the mothers' psychological well-being. The methodology of the study lacked a direct measure of participating mothers' mental health and general well-being and thus did not properly align with its outlined objectives. Nonetheless, the results of this pilot study and the positive response from participants provide important information on the needs of refugee mothers and the feasibility of such interventions. --- I Think Someone Is Walking With Me: The Use of Mobile Phone for Social Capital Development Among Women in Four Refugee Communities The goal of the intervention was the development of social capital by providing refugee mothers and women a phone. The main component of the intervention was examining the effects of acquiring and utilizing social capita. The intervention encompassed face-to-face peer support training sessions and mobile phones for the timespan of one year. 111 Afghan, Burmese, and Sudanese refugee women and mothers residing in Melbourne participated in the study, and a subset of 29 refugees was interviewed after the one-year period. The phone number provided had many call categories, such as Translating and Interpreting Service (TIS), participants from the same community, and the training facilitator researchers (Koh et al., 2018). Because many refugees struggle with the host language, groups were divided to share a same-language proficiency for oral communication and by culture of origin. The first six weeks, weekly training sessions were conducted, and the consequent five weeks, five bi-monthly training sessions were conducted. The goal of the training sessions was to improve communication skills with community interpreters. The study utilized mixed methods and interviewed a subset of the refugees regarding their perception of the intervention. Intra-community calls were calls made to people who belong to the same community, extra-community calls were calls made to people who do not belong to the same community but live in Australia, and overseas calls were calls made to people in other countries. Analysis of how the phone was used portrayed that intra-community calls represented social capital bonding. Indeed, in each community, the category of the intra-community had a higher number of calls and call durations, compared to the other call categories. The thematic analysis of the interviews found that the perceived effects of the phone were social capital on an extra-community level; the intervention was shown to be beneficial in increasing refugees' interactions with the Australian wider society. According to the refugees, skills taught in the intervention, such as confidence and communications skills, facilitated their interactions with the general host country society (Koh et al., 2018). An interesting effect of the intervention was the strengthening of the community relationships of the mothers. Refugee mothers helped each other through childcare services and driving each other when needed. Furthermore, those with better English language skills would help the ones struggling with the language. The phone improved their social network, which became a source for emotional support and information access. The training sessions happened with community interpreters, and groups were divided as a function of shared mother tongue. This is beneficial as language is an important barrier for refugees (Green, 2017;Watkins et al., 2012). The study being conducted in the refugees' mother tongue, and community interpreters being present increases the validity of the findings. Secondly, the intervention focused on community building, which seems to be a protective factor in the interventions reviewed thus far. Finally, the intervention created a network where the refugees helped and supported each other and were each other's language liaison. The creation of a bond of trust when it comes to information access is important. For example, Persian and Syrian refugees in Germany trust information provided by people in their own social network, who have successfully resettled, the most (Borkert et al., 2018). --- Discussion Refugees are at a highly elevated risk of developing mental illness due to the adverse experiences and instability brought about by the resettlement process (Agic et al., 2016;Crowley, 2009). Moreover, refugee women, and especially refugee mothers, find themselves at a greater risk than their male counterparts to develop psychiatric symptoms and mental health problems (e.g. Rahman & Hafeez, 2003;Schweitzer et al., 2018;United Nations High Commissioner for Refugees et al., 2016). Thus, the current review sought to evaluate existing interventions aimed at supporting the specific needs and well-being of refugee mothers. Despite considerable evidence in the literature demonstrating the need for such interventions, only a limited number of interventions were found. Many interventions involving refugee mothers focus on children and youth or on the family unit as a whole, seldom assess and address the mothers' needs appropriately, and include little consideration for the mother's psychological needs and well-being. The present review suggests that across different populations and methodologies a pattern emerged whereby contact with others with similar experience and culture was essential (Koh et al., 2018;O'Shaughnessy et al., 2012;White et al., 2009). The Moving Forward Project, Sweet Mother and The Use of Mobile Phone for Social Capital Development interventions portray that community building and safe community spaces are protective factors for refugee mothers (Koh et al., 2018;O'Shaughnessy et al., 2012;White et al., 2009). The Moving Forward Project sessions provided the mothers with a safe space and what was described as a gathering (White et al., 2009), to connect with fellow refugees and simply talk, laugh, and bond. Similarly, the mothers experienced a new sense of togetherness in The Sweet Mothers intervention (O'Shaughnessy et al., 2012). Furthermore, refugee mothers felt that The Use of Mobile Phone for Social Capital Development intervention enhanced their relationship with their fellow community members, which provided them a space for emotional support and information access, and a network for exchanged childcare help, and consequently made their life "easier and better" (Koh et al., 2018). This is consistent with previous literature which has demonstrated that connection with those with similar experiences, values, and cultures is an essential aspect of social support (Chester, 1992;Kim et al., 2008;Stewart et al., 2008), which in turn is known to be a protective factor when struggling with mental health issues (e.g., Hefner & Eisenberg, 2009;Kawachi & Berkman, 2001;Watkins & Hill, 2018). Social support from refugees of similar backgrounds and experiences is a critical part of positive integration for refugee women, as it helps them build social capital in the post-resettlement country, allows them to socialize and express themselves, and thus relieve stress. Moreover, group learning in interventions focusing on social support and community building provides refugee women with otherwise scarce opportunities to build new relationships and better integrate in the host society (Saksena & McMorrow, 2020). Furthermore, the refugee mothers benefited from having a safe space to discuss their shared experiences and challenges (O'Shaughnessy et al., 2012;Paris & Bronson, 2006). Having the chance to speak on the phone with members of their own community, in The Use of Mobile Phone for Social Capital Development intervention, or meeting with their community weekly, in the Moving Forward Project, was a positive experience for them (Koh et al., 2018;White et al., 2009). This safe space is especially valuable for refugees as they often struggle to discuss their difficulties in the host country because of various challenges including language barriers, complex intergroup relations, and the lack of access to a platform (Hansen & Houston, 2016). Moreover, creating safe spaces where refugee women can communicate while being physically and emotionally safe is central in helping them build a social network, receive social support, and learn important skills and information from women with similar experiences (United Nations High Commissioner for Refugees, 2014c). Similarly, the United Nation's International Organization for Migration (IOM) has organized safe spaces for women and girls to come together, express themselves safely and openly, and develop positive coping strategies (International Organization for Migration, 2018). The use of such women-centered safe spaces for refugee women and girls has been shown to have positive impacts on participants' mental health and well-being (Jahan Seema & Rahman, 2020). Thus, creating safe spaces for refugee women and girls is an effective way of supporting their post-resettlement needs and improving their mental health, well-being, and resilience through community-building. The review also suggests that having allies increased refugees' willingness to connect more to the host country society (Koh et al., 2018;O'Shaughnessy et al., 2012;Paris & Bronson, 2006). Mothers expressed feeling safer to honestly discuss their situation and experiences relating to resettlement within the context of the Sweet Mother intervention than with other professionals in different settings (O'Shaughnessy et al., 2012). This is worthwhile because newly resettled mothers do not tend to seek professional services because of barriers such as stigma, and lack of knowledge of community services and resources (Nicholson et al., 1998). In addition, having an ally to advocate for them, such as the Visiting Moms intervention working alliance, or the strengthened relationships that were formed because of The Use of Mobile Phone for Social Capital Development intervention helped improve the refugee mothers' daily life and increased their interactions with the extra-community or host country wider society (Koh et al., 2018;Paris & Bronson, 2006). Indeed, community-building in the post-resettlement context leads to better access to health services, and increased access to adequate professional response in cases of gender-based violence (Jahan Seema & Rahman, 2020). Moreover, community programs and safe spaces can be used as an entry point by healthcare professionals to reach refugee women, mothers, and girls, build trust and rapport with the community, educate them on available services and health literacy, and better understand their needs (Abbasova, 2017;Shrivastava et al., 2017). --- Limitations and Future Directions The review only includes scientific articles, published articles, and articles written in English. Plus, the review includes very few studies with great variability which makes generalization across studies difficult. In the future, there needs to be a careful consideration of the creation of safe spaces, advocacy, and community building in the conception of future interventions that aim to improve the well-being of refugee mothers. Future studies evaluating interventions for refugee mothers need to also be evaluated in more systematic ways. --- Conclusions and Practical Implications While refugee mothers are at risk because of various factors, there are not many interventions that aim to meet their specific needs and improve their prognosis and well-being. Even though there are some interventions that incorporate refugee mothers' needs and have positive impacts, they either do not calculate and assess outcomes, or the interventions' methodology does not properly align with their outlined objectives. There is a concern for validity in interventions with refugees that conduct face-to-face training or interviewing in English when refugees have language barriers. This systematic review suggests that emergent protective factors for refugee mothers are creating and providing a safe space, being a linguistic liaison, advocating for refugee mothers' needs, and community-building. Thus, it would be helpful if practitioners and professionals working with refugee mothers educated them and informed them of accessible resources through pamphlets written in their native language, especially those that could enhance community-building and provide safe spaces. Indeed, providing refugee mothers with resources that can connect them to refugees from their own community or culture, rather than general refugee populations, is more helpful for them as they can use their own language or practice their own culture. Furthermore, whatever the institution, having professionals and front-line workers who can speak refugees' native language seems to be imperative for the refugees' knowledge of their rights, the resources they have access to, and the ways in which they can start connecting with the host country's general community. --- Competing Interests: The author has declared that no competing interests exist.

Refugee mothers endure and are at risk for depression, post-traumatic stress, suicidality, and anxiety. There is a gap in the literature regarding interventions for refugee mothers' mental health and well-being. Interventions involving refugee mothers rarely provide adequate support for refugee mothers' specific mental health needs and challenges. This paper presents empirical evidence to contextualize the risks refugee mothers face, such as gender-based violence, mental health challenges, and language barriers. Then, the paper provides a critical systematic review of interventions conducted with refugee mothers. The critical systematic review suggests that creating and providing a safe space, being a linguistic liaison, community-building, and advocating for refugee mothers' needs are emergent protective factors for refugee mothers. Finally, based on the review, recommendations for future interventions followed.

are creating and providing a safe space, being a linguistic liaison, advocating for refugee mothers' needs, and community-building. Thus, it would be helpful if practitioners and professionals working with refugee mothers educated them and informed them of accessible resources through pamphlets written in their native language, especially those that could enhance community-building and provide safe spaces. Indeed, providing refugee mothers with resources that can connect them to refugees from their own community or culture, rather than general refugee populations, is more helpful for them as they can use their own language or practice their own culture. Furthermore, whatever the institution, having professionals and front-line workers who can speak refugees' native language seems to be imperative for the refugees' knowledge of their rights, the resources they have access to, and the ways in which they can start connecting with the host country's general community. --- Competing Interests: The author has declared that no competing interests exist.

Refugee mothers endure and are at risk for depression, post-traumatic stress, suicidality, and anxiety. There is a gap in the literature regarding interventions for refugee mothers' mental health and well-being. Interventions involving refugee mothers rarely provide adequate support for refugee mothers' specific mental health needs and challenges. This paper presents empirical evidence to contextualize the risks refugee mothers face, such as gender-based violence, mental health challenges, and language barriers. Then, the paper provides a critical systematic review of interventions conducted with refugee mothers. The critical systematic review suggests that creating and providing a safe space, being a linguistic liaison, community-building, and advocating for refugee mothers' needs are emergent protective factors for refugee mothers. Finally, based on the review, recommendations for future interventions followed.

Introduction One day I got a message via Facebook suggesting I should sign an online petition against the plans to demolish the old community-run (but city-owned) bathhouse two blocks away from where I lived in southern Stockholm. Since I had enjoyed the bathhouse and their different activities, I signed the petition, joined the Facebook-group, started to follow their Twitter feeds, and added many of the participants as Facebook-friends. I soon came to realise that online visibility through practices of updating on social media platforms such as Facebook and Twitter would get me closer to some of the activists. By echoing popular arguments through practices of retweeting and through posting encouraging entries on some of the core activists Facebook profiles, I was not primarily showing my sympathy for what has been understood as participatory values of activist groups semipublically displaying their communication on social media platforms (see Jenkins, 2006;Bruns, 2008;Shirky, 2009). What I did was to reinforce core positions of certain activists. It became apparent to me that activists using social media platforms are not necessarily equal as the technooptimist/determinist discourse around social media participation suggests, that social media platforms are supposed to level out power hierarchies through lowering the threshold of participation and mobilising political action (as Shirky, 2009 andCastells, 2012 have argued). In this article I challenge such assumptions by suggesting a typology for analysing power asymmetries within activist demands in network societies and discussing what role social media platforms play for upholding these. Before outlining to this typology, I will first attend to some conceptual clarifications and present the bathhouse demand that will be used to exemplify my argument. Activism is approached as a form political participation. Acknowledging the widening understanding of political participation (see Bengtsson, 2008, p. 116;Esaiasson and Westholm, 2006, p. 15), activism could be defined as participation from outside representative democratic institutions, but with an outspoken aim to influence them (Author). It is argued that activism is important for broadening political participation beyond established power elites (Bennett and Amoshaun, 2009). Indeed, participatory democratic theory has its roots in a broad understanding of politics as encompassing areas beyond the Parliament. Contemporary representative democracies cannot include all political demands, since majority decision-making always favours one over another (Mouffe, 2005). Hence, it becomes apparent already here that power cannot be left out of the equation since activist participation is understood as revolving around excluded demands (see also Carpentier, 2011;Fuchs, 2014, p. 55). Laclau's (2005, pp. 73-74) splitting of groups into smaller units of demands -as presenting claims to a certain order -further contributes to this understanding of activism. Demands include both social movement types of participation and short-time commitment to single issues, the two types of political action that has come to dominate the discussion of activist participation. A demand may be formed into a more long-term social movement, or a single-issue campaign that disperses once the demand has been met (or considered lost). For example, people in southern Stockholm started to rally around a rather temporary commitment to a single issue, saving a bathhouse. Some of the participants continued their engagement though by forming a local lobby group S<unk>F<unk> (Södra Förstaden, the Southern Suburb). In this article the focus is on participation in activist demands in network societies. But how are we to understand network societies? In Castells' (2000) famous work, he uses the network as an intersectional concept for overcoming boundaries between society and technology. He describes a tendency to organise processes and functions as networks, the network thus becoming the social morphology of society, influencing everything from processes of production to individual experiences, power and culture (2000, p. 519). Network society therefore has been defined as a social formation with an infrastructure of social and media networks enabling its prime mode of organisation at all levels, individual, organisational, societal and global (van Dijk, 2006 pp. 19-20, 27). According to van Dijk (2006, pp. 2, 23), networks are becoming the nervous system of our society and we can expect them to influence our entire life. It is important to remember when discussing network societies that complex social networks have always existed. Recent developments in communication technology have however made the social network a more dominant form of social organisation (Wellman, 2001). Social media stands out as a defining feature of network societies around which practices are organised, demands are communicated and individuality is negotiated. But how are we to understand social media? I have approached social media as communication platforms where the user him/ herself is able to contribute to the platforms content. Here I rely more on O' Reilly (2005) general definition of web 2.0 than Ellison & boyd's (2007) popular, but more specific, delineation of social network sites (SNS). In my research it has been important to include both interactive and networking functions of social media (i.e. possibilities for user-centric multipath communication and collaboration, what is often implied as the "social" in social media, which is a very narrow understanding of the social, see Fuchs, 2014, pp. 7, ch. 2) without for that sake downplay their broadcasting functions that continue to be central for social media practices (especially among groups and individuals seeking visibility and to create public opinion for a political demand). In this way my approach resembles Jenkins, Ford & Green's (2013) recent account of social media as "spreadable". Spreading I believe is a more accurate description of what is happening on social media than the often-used verb sharing. But in contrast to Jenkins and co-authors I am more critical towards supposed empowering potentials of such spreading (see Fuchs, 2014, 53-54, for an in-depth criticism of their account). To avoid a deterministic or an overtly uncritical understanding of social media platforms, I lean towards Kaplan and Haenlein's (2010, p. 60) definition of social media as internet-based applications that build on the ideological and technological foundations of web 2.0. Indeed, media platforms have a dynamic relationship between the social and the technical (Feenberg 2010). In network societies our communities become increasingly technologically mediated (Andersson and Jansson, 2012, p. 106). In activist demands using internet-based social media platforms, communication practices entangle with the demands' organisational structure and identity to the point that they are hard to separate (Breindl, 2012). Since network societies have emerged largely because of media networks, a study of activist demands in such societies will inevitably put emphasis on practices of social media use, even though these are becoming increasingly hard to separate from the offline, see Baym, 2010;Chadwick, 2013. In the bathhouse demand for example activists relied heavily on social media platforms for communication, organisation and mobilisation, in tandem with more traditional offline methods (such as rallies and demonstrations). Numerous studies have focused on how activists use the internet to mobilise support and organise themselves and their demands (see Breindl, 2012, for an overview). Some argue that internet-based organisation facilitates more horizontal and equal distribution of power, and that politics and participation become more accessible because the internet is supposed to lower the threshold, even for groups previously excluded from the political arena (Jenkins, 2006;Bruns, 2008;Shirky, 2009). While acknowledging that the landscape of power is changing in network societies, there is no reason to believe that increasing organisation in (media) networks will cause a society devoid of power relations. Elias made this argument already in 1939. When people become increasingly dependent on each other, a need to attune conducts arises (Elias 1998(Elias /1939, pp. 51-52), pp. 51-52). Thus, the interdependence of people in network societies corresponds with the increasing importance of managing behaviour (see also van Dijk, 2006;Breindl and Gustafsson, 2011). Power has to do with who can influence, and who is allowed to influence the course of events (Fuchs, 2014, p.70, 218). Hence, power concerns the interdependence between people that today has come to revolve around networking because of increased network organisation and social media use in network societies. What I propose to label networking power revolves around being in a position, or being positioned in a network to exert influence by setting the agenda and defining the reality as well as getting the attention for the information you spread. The underlying argument here being that a democratisation of opportunities for displaying information, information that may or may not spread, does not imply the levelling out of differences between users. As I hope will become apparent towards the end of this article, everyone do not have the same skills, or are in the equal position to spread information, set the agenda or define the situation/ reality and in this way influence the course of events here in an activist demand. Of pivotal importance for my argument is recognition. Who is allowed to exert influence and who is heard in the increasing information buzz in network societies depend to a large extent on recognition. Here social media may be conceived of as sites of power struggles since they have specific mechanisms (algorithms) for the generation of reputation (see van Dijck, 2013, pp. 13, 62) which in turn determines who's information will get noticed or not. Hence, networking power intersects with status, recognition, legitimation and asymmetries between users in terms of visibility and attention (here my account differs from Castells, 2009, idea of networking power as inclusion in important global networks). This is what I intend to study in this article, focusing on relations of power between people within activist demands in network societies. To do this I will exemplify with the bathhouse demand in southern Stockholm. Hence, I will start with a short presentation of this before attending to the typology for studying networking power. The Bathhouse Demand I will illustrate this largely theoretical account with examples from a (n)ethnographic study of a contemporary middle-class activist demand, saving a local bathhouse from destruction. This demand engaged (mostly) inhabitants in the southern Stockholm suburbs of Aspudden and Midsommarkransen. They are among the oldest suburbs, situated close to the waterfront, with buildings dating back to the end of the 19 th century, and located just two subway stops away from the inner city. The two suburbs are populated by an educated and politically aware middle-class. Nearby Midsommarkransen is located the University College of Arts, Crafts and Design, Konstfack, and in Midsommarkransen you also find the community run Cinema, Tellus, where members organise activities, show movies, documentaries as well as children blockbusters. Hence, the suburbs are popular both with urban middle-aged couples looking for bigger apartments without having to go too far away from the inner city, as well as with youngsters studying at, or attracted to, the creative atmosphere around Konstfack. The suburbs are a political stronghold of the Green Party with up to 23 per cent voting for them in the 2010 national elections (compared to seven per cent nationally). Inhabitants in these suburbs started to rally already in 2007, first to renew their bathhouse in Aspudden, and later to save it from destruction. Together with traditional offline activist campaigns, online social media platforms were used to call for engagement, to spread information and to gather support for keeping the bathhouse. The bathhouse was demolished despite of heavy protests, campaigns and even an occupation. Most activities took place during the couple of months leading up to the overtaking and demolition of the bathhouse late November 2009. Some of the bathhouse activists continued their participation in S<unk>F<unk> (Södra Förstaden, the Southern Suburb), a group that has continued to act in the suburbs against development plans, for the preservation green areas and playgrounds among other things. Concerning social media platforms the activists used a blog during the battle for the bathhouse through which they disseminated information, mobilised participation and mocked municipal politicians. During October and November 2009, the activists also used a Twitter-feed, mostly to spread information on activities as well as a means to mobilise participation. For quick mobilisation activist used a text-messaging list. For more lengthy comments, activists posted both on the blog as well as on a Facebook-group Rädda Aspuddsbadet (Save the Aspudden bathhouse). S<unk>F<unk> use both a Facebook-group and at Ning-community platform on which participants have their own profiles, can connect and message each other as well as start discussions, specialised groups, blogs et cetera. The study of this activist demand article was conducted within a larger project researching rationales of political participation in network societies. In previous publications on this demand I have discussed the importance of the values of reflexive connectivity and responsiveness and how these values encouraged/demanded a social negotiation of activists' selves, something that was done through practices of updating (Author). Furthermore, I concluded that these values, largely connected to using online social media platforms, seemed to push/discipline activists to participate (Author). Revisiting my interview material and field notes I also conclude that the location itself, the southern suburbs, were important around which values were negotiated. Activists talked about a unique southern suburb character consisting of old buildings among green leafy areas as well as neighbours knowing each other and doing things together, such as running the Aspudden bathhouse and cinema Tellus. The bathhouse was frequently described as a non-commercial meeting place, run by the inhabitants for the inhabitants, a symbol of a cherished value of a location bound community of neighbours. Accompanying this was a value of being active, involved, to voice concerns and show support for the demand. But rather than to be reactive, participants should be proactive, i.e. do something and engage others. The individuals that were held in high regard were the engaged ones, so-called "fire starters" and activists who succeeded in engaging many others to the cause. These values are important if trying to understand practices positionings within the activist demand which leads to the next section. --- Positions and habitus within activist demands According to Bourdieu (1985, p. 724 and1987, p. 2) agents are defined and act according to their position in social space. Here I depart from an assumption that how participants establish their position within a demand, structures the values of the demand and vice versa. Hence, the shared values (such as the ones briefly discussed above) should be analysed in intersection with how activists position themselves and others in relation to these. Bourdieu's conceptual framework of social fields, habitus and capitals are useful for such analysis. Starting with social fields, Bourdieu is somewhat unclear about the differences between the terms social space, world, field and sub-field. In one instance he talks about the political field as consisting of sub-fields of parties and unions, which could lead one to argue that activism is a subfield within a larger political field (1985, p. 737). Unfortunately Bourdieu does not guide us how to delineate between fields and sub-fields here. But his definition of a social field can be connected to Laclau's understanding of demands. Bourdieu (1993, p. 16) defines a social field as a collection of people that gather around a common belief worth fighting for. Agents within the same field can be of different opinions; it is the belief that the fight is worth the effort that binds them together (1993, p. 17). In southern Stockholm, the fight for the bathhouse was the common demand that bound activists to each other as a field in the first place. Of prime importance here, is that Bourdieu (1985, p. 724) describes a social field as a multidimensional space of positions and positionings. In this article I focus on two positions, belonging to the core and to the periphery with the assumption that core positions entails greater opportunities for setting the agenda for the demand and defining the situation in which the demand is operating in, and in this way exerting influence over activists positioned more to the periphery of the field, networking power in other words. Core/ periphery positions are also underlined by Castells (2009, pp. 26, 34) who argues that practices of positioning within a community are primarily used to determine core/ periphery positions. In online activist demands such positions have further been conceptualised by Breindl and Gustafsson (2011) as concentric circles of participation. Core activists are the leaders, setting up email lists, creating applications, being in charge of following the political process, analysing as well as orchestrating the campaign. Fuchs (2014, p.86, with reference to Gerbaudo, 2012) refers to such core-activists as soft leaders, choreographing protests by being in charge of most of the communication flow. More towards the periphery we have occasional contributors who follow what core activists (soft leaders) do and participate from time to time, and mere followers who are inscribed on discussion lists and possibly spread information, but do not actively contribute to the organisation of the campaign itself (Breindl and Gustafsson, 2011). In southern Stockholm, and resonating with the value of being proactive rather than reactive as well as with the account of soft leadership, belonging to the core or the periphery was divided along lines of who updated/ engaged others and who were updated/ engaged by others. Being updated indicated a more peripheral position, while being in charge of doing the updating, indicated a more central position. This was clearly illustrated by one (core) activist when asked about her Facebook practices; she stated she received no information online, she gave information there. It is important to underline that core positions tend to be less stable today in network societies where participants rather unite around temporary demands (so called issue engagement -see Author). Nonetheless, by taking control of information spreading and negotiating recognition and visibility, social media platforms offer spaces for some to negotiate core positions and thus having priority in setting agendas and defining the reality. The question that arises, and that I will devote the remainder of this article, is how this is done. To understand why certain activists come to be entrusted with, hold and maintain core positions, Bourdieu's concept of habitus is helpful. Habitus refers to socially learned dispositions, the sense of one's place, the luggage an agent carries with him/her, which in turn positions the agent in relation to language, culture, class and the future (Bourdieu, 1987, p. 5 and1993, pp. 12-14). Andersson and Jansson (2012, p. 38) describe habitus eloquently as acquired knowledge that give the bearer a sense of an embodied navigation skill on the field in which he/she is acting. An agents habitus both have a bearing on the field, in terms of organising, structuring and determining how field practices are conceived (Bourdieu, 1993, p. 300), as well on the agent him/herself, by being connected to his/hers position within the field, providing meaning to practices and perceptions (Bourdieu, 2010(Bourdieu, /1984, p.166, p.166). If we apply the concept of habitus to the bathhouse demand, core activists referred to experiences from solidarity and animal rights movements. Engagement in the cinema Tellus also seemed to have built both a sense of a southern suburb community feeling, as well as skills and knowledge for organising and mobilising participation (i.e. being proactive). Interviewed activists also referred to experiences from student councils, student nations and the scout movement and similar organisations. Talking to activists and asking about important skills and knowledge for their participation and how these had been acquired, many referred to previous experiences in such organisations, organisations that indicate that they have had a middle-class upbringing. Similarly, Breindl and Gustafsson (2011) argue that activists closer to the core often possess educational knowledge, social, technical and organisational skills. Habitus is thus also connected to class. Several studies have underlined that users from lower socioeconomic groups tend to be less skilled in using digital platforms, hence pre-existing inequalities are both reflected and perhaps increased online (DiMaggio et al. 2004;Hargittai 2008;Gui and Argentin 2011). A report from the OECD (2010) concluded that the digital divide in education is along lines of skills of use rather than access, i.e. between those with the competencies and skills to benefit from computer use and those without such competencies. Such skills and competencies are influenced by socio-economic factors (DiMaggio et al. 2004). Also Hargittai and Hinnant (2008) found that among American young adults (18-26 years old), those with higher levels of education and wealthier parents (i.e. habitus) used the web for more capital enhancing activities (see aslo Gui and Argentin's, 2011 study among high school classes in Italy). Hence, those coming from wealthier socio-economic backgrounds and with better education seems to be better equipped to manage and controlling their online activities to negotiate and reinforce core positions. Habitus thus influences networking power. Here it is also important to underline timeliness and circulation of information (Chadwick, 2013, pp. 87, 190) i.e. knowing when to inform, when to delay, when to spread. Hence, to successfully navigate the social fields of contemporary network societies depends on networking skills, skills that are intertwined with our social learned dispositions (i.e. habitus). Such skills have a bearing on the field in terms of how practices are organised, structured and conceived. Concepts such as online social networking skills, networking skills and digital literacy have been elaborated in order to map differences among users and groups of users in their ability to process meanings of digital content, and why some are more successful than others in negotiating core positions online (Breindl and Gustafsson, 2011;Hsieh, 2012). I have myself previously underlined that online social networking requires a new form of competence in order to manage ones visibility online at the same time avoiding being subject for surveillance (Author). Breindl and Briatte (2013, p. 34) also talk about digital protest skills as the combination of social and technical skills aimed at online collective action. Charisma and social competencies could be conceived of as resources for networking. Studying activist core positions in southern Stockholm, it became apparent that their actual and perceived networking abilities, intertwined with their habitus, in turn were intertwined with previous achievements, active participation and successful mobilisation of others. All this leads me to Bourdieu's concept of capital. An agent's habitus and composition of capital can not be understood as apart, since the manner in which an agent uses capital(s) reveals how the agent acquired it and therefore also how the agent is predisposed to use it (Bourdieu, 2010(Bourdieu, /1984, pp. 58, 61), pp. 58, 61). Bourdieu (1993, p. 269) defines capital as a social relationship, an energy that exists and produces its effects within the field it is used. Capital represents a power over the field "like aces in a game of cards that define the chances of profit in a given field at a given moment" (Bourdieu, 1985, p. 724). --- Activist Capitals in Network Societies Profit here is to be understood as positions more to the core of a demand (or field to use Bourdieu's terminology). The notion of capital is related to practices of positioning because capital use cannot be understood without reference to the agent's habitus, position within the field, and in turn, the field position is dependent on the specific capital the agent can accumulate (Bourdieu, 1993, p. 269). There have been many attempts to outline internet specific capitals, also by references to Putnam's (2000) often cited elaborations of Bourdieu's social capital. Ellison et al. ( 2011) for example discusses online social capital, or socio-technical capital, as based on technological affordances. One of the most detailed accounts is Urry's (2007) outline of network capital. He argues that contemporary societies are more and more organised around the value of circulation -mobilitiesand by investigating how social relations change from such mobilities, he discerns an ability to form and sustain networks, something he labels network capital (Urry, 2007, pp. 196-197). This is about the potential of being mobile and connected at the same time, the capacity to engender and sustain social relationships with those people who are not necessarily geographically proximate but do generate emotional, financial and practical benefits. According to Urry (2007, p. 198) network capital is a product of increasing possibilities of relations between individuals afforded by travel and communication technologies. The importance of values of connectivity and responsiveness thus resonate in Urry's reasoning. According to him, network capital is about being connected, making yourself connectable for capital enhancing purposes (Urry, 2007, p. 203), networking in other words. At the same time the aforementioned ideas of networking skills and digital literacy underline that not everyone is equally skilled/ prepared/ socially equipped in using social media platforms for accumulating such network capital. And this is concnerns how communication technologies are used rather than access to them. The suggestion here is that Bourdieu's conceptual framework could inform a typology of capitals that could be used for studying networking power within activist demands in network societies, how certain come to occupy core positions and others not. I will therefore suggest a typology of four different, but overlapping forms of capitals: participating, mobilising, connecting and engaging capital. Starting with participating capital, in southern Stockholm, the value of being active and engaged made it possible for certain activists to accumulate and use a type of participating capital when positioning themselves in the field. For example, many postings on Facebook revolved around having attended rallies and campaigns. Similarly, in Breindl and Gustafsson's (2011) study, individuals holding power in activist networks were generally the more active ones. The number of postings/actions of a user and their level of participation in a shared project has become a currency in many online environments (Bruns, 2008, p. 55). But being active was not enough to accumulate this capital. You had to be recognised for your participation. This explains the importance to update others on your participation and make it visible online. Such practices can thus be understood as acts of positioning within the activist field. This resonates in Biggar's (2010) study of crowdsourcing activities. He claims that taking part in such activities is about building one's online portfolio and leveraging one's cultural and social capital within a community (Biggar, 2010, p. 10). In southern Stockholm, to write to politicians, being their watchdog and bark as soon as they did something considered bad for the cause, was also a way to collect this participating capital. For such activities, activists needed to possess knowledge on how society and politics work (cultural capital) in order to appeal a decision as well as knowing were and how to find information (cultural capital) and a sense of knowing how to navigate the field (habitus). Here we can clearly see that there is an exchange between habitus, cultural and participating capitals. While participation was considered important, there was another thing that was perhaps even more recognised in southern Stockholm, mobilising others. As discussed previously, there is a difference between peripheral activists and core activists along lines of who updated others and who were updated. Or in other words, there is a difference between those who are expected to be mobilised and reactive and those who are proactive and mobilising others. In southern Stockholm for example, one core activist was mentioned several times for what was labelled as an "infectious" engagement. Hence, what can be discerned here is a type of mobilising capital. Similarly, Breindl and Gustafsson (2011) refer to temporal elites whose power comes from the possibility of mobilising others. Indeed, the creation of online content is of small matter without a large enthusiastic audience to use the technology to its full potential (Kaye, 2011, p. 208). Thus, recognition is also about having a wider supporting group who can spread information through social networks and rapidly mobilise. I would like to differentiate here between mobilising within the demand and connecting outsiders to the demand. According to Castells (2009, pp. 45, 430) it is along lines of who has the ability to connect networks to each other that constitutes (networking) power in network societies. Online communication has enabled individuals to act as social switchboards, centre points for multiple changing and overlapping networks of interaction. Nodes/ activists that can act as switches between networks/ demands become fundamental sources of networking power. This relates to discussions of bridging social capital. In contrast to bonding social capital, bridging social capital refers to connections with weak ties (Granovetter, 1973;Putnam, 2000). The importance of bridges can be traced back to Heidegger and Simmel (discussed in Urry, 2007, pp. 31-32) who pointed at the possibility of bridges to bring streams, banks and land into each others neighbourhood, and hence reorganising the ways people moved within an area. In network theory, researchers have underlined bridging capital as most important for networking power because they give people access to new and different resources, not the least through connections with weak ties (Baym, 2010, p. 136). Weak ties are conceived of as resources since it is through such ties that new information/ opportunities reach in-groups of users, and it is through weak ties a community/ group or demand can reach out to others (Granovetter, 1973;Wellman, 2001;Ellison et al., 2011). The activists I interviewed in southern Stockholm were all active in other demands at the same time as fighting for the bathhouse. Some activists explicitly tried to create connections between the bathhouse activists and other groups they participated in. I call this connecting capital. This capital can be accumulated if the information that a user posts goes viral. Bennett and Segerberg (2012) have discussed such easy-to-personalize action themes (memes) that travel through personal appropriation and spreading as connective action. This is clearly intertwined with creating visibility for a demand and being in a position to spread information online, things that are of pivotal importance for activists today. Connecting capital also implies that users need to gather connections in order for the information they post has the possibility to go viral. But connecting capital also has to do with stratified attention structures already existing within a network. Highly recognised users in the online attention economy determine what information gets spread or not (as Nahon and Hemsley, 2013, convincingly have argued). Their networking power makes them into gate-keepers of what information goes viral or not. Barabási (2011) has criticised the notion of week ties in favour of intermediate ties. I find this notion helpful for further delineating connecting capital to something that I suggest labelling engaging capital. It is one thing to connect activists, groups, demands together, spread information and create visibility in connective information flows online, and another thing to engage these other activists in actions. According to Barabási, engaging others (to an activist demand for example) do not come from weak ties but from intermediate ties since users rarely pay attention to the weak ties in their networks, overflowing with updates and information from ever growing social networks. Here there is a difference between spreading information from others in your networks (connecting capital) and to act on calls for arms. And for this intermediary ties (and not weak ones) are important to understand who possesses the ability to engage other groups and activists to a demand. In Southern Stockholm for example an artist managed to engage her colleagues in an art barricade. Other examples are activists who also were active politicians (in the Green Party) and thus functioning as important intermediaries between activists and the political system. This reasoning further resonates in Breindl and Gustafsson's (2011) claim of the existence intermediary elites in contemporary societies. All these capitals have ultimately to do with negotiating recognition as an activist, recognition that could be exchanged into a core position (depending on the habitus of the activist). For example, participation has to be displayed to, and recognised by, connections within the demand in order for accumulation of participating capital to take place. Mobilising capital can be accumulated if one is recognised as successful with mobilising ones network connections, i.e. getting other activists within the demand to act. Connecting capital can be accumulated if connecting other activists or groups outside of the demand to the cause and being recognised for this. And finally engaging capital can be accumulated if this connection is of that sort that these other groups and activists will be engaged in action in ways that others within the demand appreciate. Recognition thus concerns the evaluation of actions, participations and displays by others/ peers, and this is one area in which social media is important, for turning capital accumulation into recognition. Insert figure 1 --- about here The focus on recognition underlines the interdependence of individuals (here activists) in network societies. But this also has to do with an activist's skills and position in the network to begin with. In other words, an activist's habitus is of importance if being able to turn participation, mobilisation, connections and engagement over into a core position. Habitus also matters if able to use experiences and actions from other activist demands to accumulate capital within the particular demand under study (as the arrows in the figure above illustrates). Connecting and engaging capital clearly concerns putting connections from other demands to use within the demand under study. Participating capital can also be accumulated, with the help of habitus, from actions in other demands. Analysing recognition we thus have to consider previous experiences in order to understand why certain activists come to occupy core-positions. Or as Bourdieu (2010Bourdieu ( /1984, p. 105), p. 105) frames it, agents enter the social field with previously acquired capital. In another text he discusses agents entry into a field already with a kind of legitimacy as agents from previous fields (Bourdieu, 1993, p. 100). In other words to discuss these capitals it is important to consider the in-group versus the out-group in the capital accumulation. The negotiation for recognition it seems could only take place within the demand (the in-group). However, some of the capitals discussed above were dependent on involving outsiders, such as connecting, engaging and to some extent participation capitals. In southern Stockholm it was apparent that activists' previous experiences were used to negotiate core positions. Activists who had a track record were for example often referred to in the interviews as important for the bathhouse demand. Examples from southern Stockholm range from being recognised as efficient runners of the cinema Tellus, reputation as having successfully led political actions, or having organised campaigns. The boundaries of an activist demand are thus very permeable. Bourdieu (1987, p. 4) himself mostly talk about economic capital (the accumulation of money), cultural capital (the accumulation of qualification, education and knowledge), social capital (the accumulation of social relations) and symbolic capital (the accumulation of reputation). It is apparent by focusing on recognition that I am exploring the terrain of symbolic capital in Bourdieu's terminology. In information overloaded network societies, getting noticed is everything. Networking power underlines that this unequal distribution of attention and visibility influence others, actions and definitions of reality within a demand. Social media can thus be conceived of as sites for power struggles as they have specific mechanisms for the generation of recognition and visibility. Van Dijck (2013, pp. 13, 62) talks here about a popularity principle, important for users recognition, but also, underpinning a so-called attention economy of social media. This attention economy online is indeed stratified, as powerful media conglomerates are better in attracting attention than others (Fuchs, 2014, p. 82). Concerning activists demands I also argue that attention is unequally distributed among activists within the same demand. And on social media platforms, recognition is relatively easy to negotiate because here relations are counted, measured and put

Network societies are characterized by social media, media that are supposed to level out power hierarchies, making political participation more inclusive and equal. By developing a typology for studying networking power within activist demands in network societies, such technooptimistic/deterministic assumptions are questioned. This typology is based on Bourdieu's conceptual framework of social fields, habitus and capitals and revolves around participating, mobilising, connecting, and engaging capital and how these intersects, overlaps and are used for negotiating recognition which I argue is of pivotal importance in for upholding core positions and thus in todays activist demands. Such core positions are related to networking power, i.e. knowing how and being in a position to network in order to decide about courses of events in the organisation of the demand/ social field and its actions. This largely theoretical account is exemplified from a (n)ethnographic study of a middle-class activist setting in southern Stockholm using online communication platforms in tandem with more traditional offline activist participation to organise and mobilise participation.

mostly talk about economic capital (the accumulation of money), cultural capital (the accumulation of qualification, education and knowledge), social capital (the accumulation of social relations) and symbolic capital (the accumulation of reputation). It is apparent by focusing on recognition that I am exploring the terrain of symbolic capital in Bourdieu's terminology. In information overloaded network societies, getting noticed is everything. Networking power underlines that this unequal distribution of attention and visibility influence others, actions and definitions of reality within a demand. Social media can thus be conceived of as sites for power struggles as they have specific mechanisms for the generation of recognition and visibility. Van Dijck (2013, pp. 13, 62) talks here about a popularity principle, important for users recognition, but also, underpinning a so-called attention economy of social media. This attention economy online is indeed stratified, as powerful media conglomerates are better in attracting attention than others (Fuchs, 2014, p. 82). Concerning activists demands I also argue that attention is unequally distributed among activists within the same demand. And on social media platforms, recognition is relatively easy to negotiate because here relations are counted, measured and put on display together with ones ability to attract and maintain relationships with (like-minded) others (van Dijck, 2013, p. 62). This argument is perhaps better illustrated if related to the notion of fitness in network theory (Barabási, 2011). If we take the notions of nodes and links in network theory and replaces nodes for activists and links for connections between activists, fitness would refer to an activist's ability to attract other activists and connect with them. Being recognised as an activist would make the activist more fit, in the sense of being more likely to attract the attention of other activists and form relationships with them. In information overloaded network societies, knowing who to trust and who to connect to, is increasingly based on agents past achievements and others evaluation of these (Urry, 2007, p. 221). Recognised activists will become even visible since others are more likely to stumble across them online, learn about them, connect and link to them and hence contribute to their on-going accumulation of recognition as the popularity principle postulates. For example, rather soon after having joined the bathhouse campaign I come to realise which were more esteemed simply by observing who was retweeted and whose Facebook-postings received links and likes. In this way recognition could also be understood as a measure of habitus, of who possessed a sense of knowing how to navigate the field of activists demand. I have discussed here the differences the online makes for displaying and using recognition to negotiate core positions, not the least through popularity principle (van Dijck, 2013, pp. 62). But it was also apparent to me that the online was interlinked with the offline and not easy to separate. In the bathhouse demand for example activist participating in offline actions put them on display on their social media profiles. Information about, and call to arms for offline actions, were most often posted online. Hence, social media do not replace offline collective action and actions that involve spatio-temporal presence. Bourdieu (2010Bourdieu ( /1984, p. 81), p. 81) claims that each field sets it highest price on the outcomes being created within it. In activist fields it is clearly the actions themselves that are the most desired outcome. Offline actions thus mattered as apparent in the elaboration of the typology above. It is one thing to be able to connect people/ groups/ demands online in order to spread information and increasing visibility for a demand. But it is another thing to engage these others in offline collective actions. In southern Stockholm mobilising capital was accumulated mostly by engaging demands sympathisers in offline actions. Indeed, activists' actions may only take place online (as in the case of hacktivism and Anonymous), but it seems to me that most activist actions still mainly take place offline. There is a difference between spreading information through networks connections online and to act offline and in activist demands it seems that offline actions are the more valuable for accumulating participating, mobilising and engaging capital. What is happening in today's network societies though is that such capital is mostly negotiated for recognition (displayed) online on social media platforms. It thus becomes increasingly difficult to separate the online from the offline here (as also others have argued, see Baym, 2010;Chadwick, 2013). Indeed as Chadwick (2013) argues, activist actions often combine online and real-space behaviours and impacts (p. 189) and activists often operate in hybrid media ecologies (p. 193) in which both online and offline media has a role to play. --- Conclusion To understand networking power within activists demands I have outlined a typology of activist capitals out of an argument that recognition is becoming increasingly important for understanding relations of power in network societies characterised by social media platforms that in turn are characterised by an attention economy and popularity principle. This typology could be used for studying networking power in terms of who come to occupy core positions in an activist demands. In other words, holding a core position is connected with knowing how to network, gaining recognition through participating, mobilising, connecting and engaging activists to the demand. This is dependent on the habitus of the activists, their luggage of previously learned skills and sense of knowing how network in order to navigate the fields of activism in a network societies. While far from a detailed account, the aim here was to contribute to the understanding of contemporary activist participation in network societies and how relations of power are still at play in the form of networking power, even though it has been widely claimed that the internet and social media lowers the threshold for political participation. By outlining a typology of capitals, exemplified with the activist demand in southern Stockholm, the aim has been to suggest a way to study why certain activists come to occupy core positions and others more peripheral positions. I hope this typology has provided some understanding for the complexity of positions and positionings in an activist demand in a network society, and the intricate workings of how positions are gained and upheld, even in an affluent Swedish neighbourhood amongst a relatively homogenous group of people. Indeed, the typology would probably have to be modified and developped if put to use in a heterogeneous setting populated by a more unequal crowd and context. Hence, the aim at this stage is not generalize but to suggest and hopefully to inspire further exploration into the complexities that may emerge when negotiating power today. Empirical studies are needed in which this typology will be put to use, be criticised and further developed. I for one will work with this in the future in order to more empirically and carefully map the distribution and volume of the different capitals here outlined in the bathhouse demand. However, this will be another article. So what are the implications of these findings other that relations of power are still at play in activist demands today? First of all this typology has implications for the study of power in todays network societies. By conceiving power as relational and networked, evolving from exchanges between interdependent people and technologies allows us to move away from categorizing people who are powerful or not and instead focus on the diversity of mechanisms and interactions that enable power to be exercised. Indeed, as Chadwick (2013, p. 207) has argued, power is shaped by hybrid networks in which both social and technological nodes have agency in terms of their interdependence and interactive exchanges with other nodes in the network. Social media platforms are neither neutral nor a determinant of change. Second I believe this indicates the importance of practices of updating in network societies. Networking power is time bound to the participation of the user. Hence, constant participation in the form of continuous practices of updating is mandatory for negotiating the recognition that is needed to occupy of core-positions. Activists in southern Stockholm, that had not participated for some time, lost recognition and thus also their core-positions. One activist for example complained that he had to start from scratch after having been offline a longer time. He was left behind in plans, discussions and had not participated in several events. As I have argued elsewhere (see Author) the logic of social media requires users to be present, archive connections, update and be updated. And as I have argued here, time spend on social media platforms could generate recognition through participating, mobilising, connecting and engaging capital. However, time spent online is also contributing to the economic value of social media companies (van Dijck, 2013;Fuchs, 2014, p. 114). We should not forget that capitalist companies own social media platforms. There is another layer of power at play here, economic power and exploitation of visibility seeking and recognition accumulating social media users. Economic capital is accumulated for social media owners when users use these media, even if used to negotiate corepositions within a for example counter-capitalist political demand. Hence, it becomes apparent that social media practices are informed by the capitalist logic of the companies that owns these platforms and capitalise on users updating practices. A third implication thus is that being online and updating social media profiles with information, users are making themselves surveillable and exploitable for capitalist companies. The logic of updating thus feeds neatly into the logic of capital accumulation and the business models of social media companies. In this way participation and political expression, organisation and mobilisation becomes subsumed under economic capital (Boltanski & Chiapello, 2005). --- About the Author --- P. DiMaggio, and E. Hargittai, and C. Celeste, and S. Shafer, 2004. "Digital Inequality: From Unequal Access to Differentiated Use," In. K.M. Neckerman (editor). Social Inequality, pp. 355-400. New York: Russell Sage Foundation. N. Elias, 1998. "The Social Constraint towards Self-Constraint," In S. Mennell and J. Goudsblom

Network societies are characterized by social media, media that are supposed to level out power hierarchies, making political participation more inclusive and equal. By developing a typology for studying networking power within activist demands in network societies, such technooptimistic/deterministic assumptions are questioned. This typology is based on Bourdieu's conceptual framework of social fields, habitus and capitals and revolves around participating, mobilising, connecting, and engaging capital and how these intersects, overlaps and are used for negotiating recognition which I argue is of pivotal importance in for upholding core positions and thus in todays activist demands. Such core positions are related to networking power, i.e. knowing how and being in a position to network in order to decide about courses of events in the organisation of the demand/ social field and its actions. This largely theoretical account is exemplified from a (n)ethnographic study of a middle-class activist setting in southern Stockholm using online communication platforms in tandem with more traditional offline activist participation to organise and mobilise participation.

Background Single-handed vs group GP practices in primary care General practices may operate as single practices or as groups of self-employed physicians. Group practices comprise the majority in some countries such as Great Britain where 85.5% of all practices were group practices in 2010 [1]. Single-handed practices remain dominant in other countries such as Hungary where all public health insurance-funded practices had been operated singlehandedly by GP practitioners up until 2013. The present paper introduces the contribution of a new support worker called" health mediator" of the multidisciplinary GP group practice based on the experiences of a primary health care Model Programme that established group practices in primary care in Hungary since 2013. --- A novel model programme of multidisciplinary group practice in Hungary Core features of tax-funded primary care in Hungary are similar to that of the UK. Primary medical care is provided by general practitioners [2] whoas opposed to those in the UKoperate single-handed GP practices aided by one or more practice nurses. Coverage is based on a mandatory health insurance scheme with no opting-out [3]. The provision of primary health care is the responsibility of local municipalities, whereas it is financed by the national health insurance fund. The health status of the Hungarian population has long been below the European average [4], partly due to the insufficient uptake of health services in general, and preventive services in particular that has been shown among the largest (Roma) minority [5] as well as in the general population [6]. In order to improve preventive services in primary care, a Model Programme established and funded by the Swiss-Hungarian Cooperation [7] had introduced group practices in primary care in 2012 in the two economically most disadvantaged regions of the country as described in the Operations Manual of the Programme [8] and elsewhere [9]. The Swiss-Hungarian cooperation was set up to promote the reduction of economic and social inequalities within Hungary by implementing mutually agreed projects in various fields. The Model Programme was agreed upon in the field of health care services. Briefly, 24 GP practices were selected on the basis of demographic and population health data, number of vacant GP practices in the region, and willingness of general practitioners and their host municipalities to participate. Four groups of general physicians designated as GP clusters, each consisting of six general practitioners were organized and received funding to employ a range of ancillary health workers such as public health specialists, dietitian, physiotherapist, health psychologist [10]. The specific aims of the Model Programme were to extend public health services in a cost-effective manner to all population groups, particularly disadvantaged ones, to reduce health inequalities. In addition to acute and chronic care, new services were offered (not available in other GP offices) such as lifestyle counselling, nutritional, physiotherapy and psychological services, as well as health promoting community programmes as published in detail elsewhere [11]. An important new service of the group practices was the provision of an invitation-based health status assessment or general screening (investigation of cardiovascular and cancer risk factorsexcluding cervical and breast cancer screening organized by a national public health agency -, based on health examination and questionnaire survey) as mandated by law since 1999 for adults on a regular basis depending on age [12] which has not been done systematically by GPs unless specifically requested by their patients/ clients. The dismal health status of the Hungarian population, the low uptake of health status assessment available from GPs, the low uptake of preventive services in the general population and in minorities referred to above made it necessary to include health status assessment in the GP clusters' services at the beginning of the Programme. All adult clients of the GPs were individually invited in writing to the health status assessment carried out by non-medical workers at the beginning of the Programme which was followed by a medical risk assessment carried out by the GP. Based on its results, patients were directed to various paths (further medical examination, or any of the new services: individual or group lifestyle counselling, or specific nutritional, physiotherapy, etc. services, or health promoting community health programmes). Tailored lifestyle counselling and health promoting community programmes were also recommended to patients who had already been in chronic care. --- Employment of support workers called health mediators in primary health care The involvement of support workers (community workers) with no professional qualification in primary health care has a long history around the world. Chinese peasants with a few months of training ("barefoot doctors") provided basic care to large segments of the rural population from the 1930s onward. This model was taken up by other developing countries worldwide in order to address the problem of unserved population groups due to shortages of professional health care workers [13]. The World Health Organization recognized the important role of community health workers in providing essential primary health care services and thereby reducing health inequalities, and recently issued policy guidelines for creating relevant programmes [14]. The mostly Roma minority disadvantaged groups of the developed countries of Europe posed a different challenge. Roma minority groups had difficulties accessing and/or utilizing various public services, or/and expressed mistrust in these services. As a means to decrease mistrust and clarify misunderstanding between governmental agencies and minorities, mediation had been introduced in France and Finland in social services in the 1960s with the ultimate aim of increasing the access and uptake of those services [15]. Mediation in primary health care was initiated in Romania during a vaccination project in 1997 when large numbers of Roma refused vaccination due to complex reasons [16]. Since then, health mediators have worked in a number of European health projects, most of them during the Decade of Roma Inclusion [17]. The inclusion of health mediators in the Model Programme was justified based on two decades of experience with Roma communites [18]. Health mediators were recruited from the local communities and employed part-time with no requirement for professional or vocational training (Fig. 1). They were recruited by public advertisement as prescribed by law and by locally distributed leaflets facilitated by the participating GPs and practice nurses. All health mediators were required to reside in the local community; preference was given to those applicants who identified with or had experience working with the largest minority (Roma) population of the regions. Their major task was to bridge the gap between general practitioners and their socioeconomically vulnerable clients by ensuring individualized support for the latter, counterbalancing the potential increase of health inequalities inherent in the population approach of prevention uncovered by Frohlich and Potvin [19]. The Programme planned to employ 12 mediators per GP cluster or altogether 48 persons (most of them middle-age women living in the local communities and identifying at the time of employment as Roma) on part-time contracts equivalent to 20 work hours per week. Work allocation and supervision of health mediators was the responsibility of the supervisor of all non-medical personnel of the GP cluster, the so-called public health coordinator (a public health professional) who reported to the head of the GP cluster (one of the GPs elected by all GPs from themselves). --- Training for health mediators Vocational training of 800 h in assistant nursing was provided in the first year of employment to all those health mediators who had the appropriate educational qualification to enter training and who did not yet have any health-related vocational qualification. 3-day training in health mediation was provided to all health mediators in the first and third years of employment in the Programme. All expenses related to both vocational and mediator trainings were fully paid for by the Programme, and both were completed during work hours. A number of short courses of continuing education were also developed for health mediators and completed during work hours. --- Work tasks of health mediators Health mediators acted as facilitators between GP cluster workers and the serviced populations with the aim of increasing the access and uptake of health services among vulnerable groups (characterized by low income, primary education only, minority status or a combination of these) who have been known to be reluctant to attend such services [5,15]. They participated in the organization and operation of various preventive services as specified by the public health coordinator (their supervisor) and the GP, being specifically responsible for increasing attendance at the health status assessment. Health mediators received a list of those who did not show up at the assessment in spite of receiving a written letter of invitation, and they had to make house visits to persuade non-attendees to participate. They carried out the majority of fieldwork related to community-based events at which they also participated, manning various posts during such events. Health mediators were also involved in various health education activities facilitated by training, relevant material and printed leaflets for distribution. Mediators also had administrative tasks of reporting their work, including participation at the monthly meetings of the GP cluster. --- Methods Our aim was to assess the workload of health mediators and to estimate their contribution to the implementation of health status assessment and community health promoting programmes in the group practices of the Model Programme. --- Data collection and analysis The framework of the Programme along with major indicators was specified in a bilateral agreement between the Swiss and Hungarian Governments in 2012. The percent of Roma patients accessing services had been specified as one of the major indicators. Ethnic identity was reported by patients themselves by responding to two questions taken from the 2011 Census [20] in the self-filled questionnaire of the health status assessment (HSA). The questionnaire of the HSA used items and scales from the Hungarian version of the European Health Interview Survey 2009, Hungarostudy 2013, and other validated instruments in Hungarian [8,[21][22][23]. The detailed development of the Model Programme (including all indicators) was carried out by a Consortium of experts delegated by nine bodies including four national health institutions and five leading universities. The programme was managed by an expert team affiliated to the national institution responsible for health care management. There had been institutional and personnel changes in the management of the Programme during 4 years of implementation (2013-2017), therefore the names of the institutes are not reported. All administrative data, including work hours and activities of all workers of the Programme, as well as monitoring of attendance of all individual, group and community services by clients was mandatorily reported on a monthly basis according to the Programme Implementation Manual (398 pages and 22 Annexes) written by the Management and approved by the Consortium. Health mediators logged their work hours and activities daily on standard forms and reported them monthly to the public health coordinator as prescribed in the Manual (along with all other employees of the GP clusters). Heads of the GP clusters were responsible for data collection and reporting in their clusters. All data from the GP clusters were sent to Management at headquarters where data processing was carried out. Data analysed in this paper were provided by Management in aggregated electronic format so as to prevent individual identification of any worker in any GP cluster. --- Work performance of health mediators Altogether 48 part-time positions for health mediators (12 per GP cluster) had been planned during the implementation phase. However, these entry-level half-time positions provided quite low wages that led to many of the mediators leaving their jobs if they found full-time employment elsewhere. Remaining mediators repeatedly requested their work hours to be increased to that of full-time and some GP clusters did request changing the part-time contracts to that of full time (granted by the Management). In order to make the workloads among GP clusters directly comparable and to account for changes in contracts and length of employment, work time for all health mediators was calculated in work hours based on the number of positions and duration of employment between July 2013 and January 2017. Potential work hours were calculated by the number of available part-time positions; actual work hours were calculated based on the number of filled positions and by number of work hours per week per person taking into account the type of contract (part-time or full-time). --- Data on new services of GP clusters All data on service uptake and attendance of programmes used in the analysis were made available by the Management of the Programme that was responsible for data collection throughout the Programme. Descriptive statistics and Pearson correlations were calculated in MS Excel 2016. --- Results --- Job fulfillment/vacancy As Table 1 reveals, health mediators spent most of their time recruiting for and helping with health status assessment. This activity was exclusively organized and carried out by non-medical personnel of the clusters, similar to community health promoting events, hence the noninvolvement of doctors. Regarding health status assessment, health mediators had to visit and individually engage with patients who did not show up at the health status assessment after receiving written invitation. Therefore, work minutes per health mediator per patient in each cluster was calculated as independent variable to account for the different numbers of patients in the clusters for Fig. 2. 80% of clients listed with and invited by the GPs in the Programme did attend the health status assessment. This service is provided by other GPs in Hungary only by the patient's request, so attendance of this service could only be compared to participation at other, invitation-based national screening programmes of the country. Attendance of the health status assessment in the Model Programme was 1.3-1.7 times higher than that of national screening programmes. A correlation analysis between the number of mediator work minutes per client and the participation rate at health status assessment by GP clusters showed a strong positive linear correlation (r = 0.549) that was not significant due to the low number of datapoints (Fig. 2). Part of the mediators' work hours was dedicated to organizing community health promoting events (data on these events are restricted to one year between October 2014 and September 2015), not requiring individual engagement with patients, but rather, organizational and logistic activities. Since this work was not dependent on the number of patients in the GP clusters, the total number of work minutes of health mediators was calculated as independent variable for Fig. 3. These events were altogether attended by 74% of all persons listed with the 4 GP clusters (Fig. 3). Pearson's correlation coefficient was high (r = 0.713) though not significant due to the low number of datapoints for the positive association between the total number of health mediator work hours and the total number of participants at community health promoting events in the examined period in the four GP clusters. Though the number of datapoints (each representing one GP cluster) are insufficient to support significance, the relationship of work performance (work minutes) of health mediators to the number of participants can be seen both in terms of the health status assessment (Fig. 2.) and health promoting events (Fig. 3). As part of the programme evaluation, a patient attitude survey in a representative sample of 1022 persons was conducted in 2016 of whom 83.6% had attended health assessment. 20% of all, and 40% of Roma respondents of this survey mentioned that they attended health assessment on the recommendation of health mediators. As new additions of the primary care workforce, health mediators had to find their niche in the group practices that took time and effort as reflected by the relatively high proportion of mediators who left the Programme during its examined 43 months (45%, Table 1). However, the proportion of job leavers was down to 20% between January 2016 and January 2017 reflecting increased integration. Forty-eight health mediators in half-time employment entered the Programme at its start. The turnover was quite high; several new mediators were hired during 4 years. The employment status of health mediators was changed to that of full-time position if requested by the Head GPs of the clusters from the Management. By January 2017, only 66% of the mediator positions were filled (32 out of 48) but 38% of the mediators had worked full-time, reflecting their useful contribution to the services of the GP clusters. --- Discussion Our paper gives account of the contribution of health mediators as members of multidisciplinary primary health care teams that were created in Hungary in the framework of the so-called Primary Care Model Programme from 2013. The Model Programme was the first in the country to create multidisciplinary teams in primary care, necessitated by a number of problems related to single-handed practices such as lack of preventive services, inequalities in access to primary care services, and uneven distribution and shortage of general practitioners, among others [9]. Their work was evaluated in relation to two new services that are not provided by single-handed GP practices in the country. Health assessment is carried out by GPs only if it is requested by the patient, so attendance of this service in the Model Programme can be compared to participation at other, invitation-based national screening programmes in the country. Breast cancer screening has been carried out by invitation among 45-65 year-old women every 2 years; its national rate of attendance ranged from 45% in 2015 [24] to 61% in 2002-2003 [25]. Cervical screening is requested from gynecologists by 50 to 60% of women aged 25 to 65 years. A national programme of cervical screening inviting women to attend since 2003 did not significantly increase the proportion of those who had been screened by 2010 [26]. Attendance of the health status assessment in the Model Programme was 1.3-1.7 times higher than that of these national screening programmes. Correlation analysis between the relevant indicators of mediator work time and participation at health status assessment and community health promoting events by GP clusters showed positive correlations, reflecting the substantial contribution of health mediators in the uptake of these services. --- Strengths and limitations Indicators of traditional and new services of GP clusters had been collected during a relatively long time-period (43 months). However, separating the contribution of health mediators from other workers of the GP clusters is limited by the fact that they participated in services in which other workers had also been involved so their contribution can only be approximated. The quantification of the share of workers in primary care outcomes is possible only for those services that are provided by particular workers independently [27]. Access to the new services by Roma patients was definitively improved due to the tenacity and persistence of health mediators, but the proportion of Roma patients is based on self-identification during service uptake but not in the database of GPs since the latter is not allowed in the country. Since the proportion of Roma in the patient attitude survey was almost three times higher than the proportion of Roma in health assessment (20% vs 7.2%), selection bias cannot be excluded, and probably resulted in a slight overestimation of the impact of health mediators on participation. However, this does not call into question the substantial motivational effect of health mediators on the participation of ethnic minority patients at health assessment. --- Health mediators as team members in primary care The institutionalization and professionalization of health mediation had been recommended on the basis of the accumulated experiences during the Decade of Roma Inclusion [17]. The Hungarian model programme described above is the first in which these recommendations were fully implemented. Health mediators were recruited from the serviced communities, received vocational training, and became employees and equivalent members of GP group practices who facilitated the access to and uptake of services among Roma minority groups, though at the expense of putting in high numbers of work hours the sustainability of which remains to be seen. Moreover, as the unprecedentedly high participation rate in the health status assessment allows us to surmise, health mediators likely facilitated access to primary care services for many members of the community, not only to those with Roma identity. In this respect, health mediators shifted towards the role that community health workers fulfill in primary care in many countries outside of Europe [13,14,28]. It is of interest to note that though the WHO policy guideline on community health worker programmes was published in 2018, the system support of health mediators in the Hungarian Model Programme established in 2013 partially or fully met all 15 recommendations of the WHO guideline reflecting conceptual overlap between the two types of nonprofessional workers. The composition and changes of the health mediator workforce will be described in a separate paper. The Programme had planned to employ altogether 48 persons in 4 GP clusters in half-time jobs. They were able to bridge the gap between general practitioners and their vulnerable clients attested by data on attendance of various primary care services. Additionally, twice as many Roma patients mentioned health mediators as the motivators for accessing services, underlining the importance of attending to vulnerable persons [19]. The type of work carried out by health mediators required personal contact with patients in their homes so a considerable part of the work hours of health mediators was spent on travelling on foot since most of the GPs worked in villages with no mass transportation. (Bikes were provided for mediators only in the last year of the Programme.) Based on experiences of this Programme, full-time (instead of part time) employment of health mediators can be justified along with the provision of some means of transportation in primary care to which sizable disadvantaged groups belong. --- The future of primary health care There has been a continued debate in the past decades whether single-handed practices can remain long-term alternatives to group practices since the latter have lower structural costs, more ancillary staff, tend to provide a wider range of services, and their GPs are much less at risk of becoming professionally isolatedall pointing to higher quality of care [29]. There have been conflicting accounts whether single-handed practices provide lower quality of care compared to group practices [30,31], but relevant reports agree that single-handed practices tend to operate in deprived areas serving clients with higher needs [32]. The special needs of populations in low socioconomic strata were also highlighted in Hungary where the rate of non-performed preventive services was found to be highest among those with no more than primary education [6]. The future of general practice as spelled out by the World Health Organization [33] and the Royal College of General Practitioners [34] lies in multidisciplinary teams that provide integrated, comprehensive, cost-effective and patient-centered care in a way that also contributes to the reduction of health inequalities at the community level. These multidisciplinary teams must expand their workforce in order to provide a wide range of services to the local communities that are easily accessed and taken up by clients from all strata of society. Health mediatorsexpanding their work from minority persons to that of all vulnerable patients of group practicescan not only ease the workload of GPs, a key issue of primary care [35], but can also increase patient satisfaction by bridging the physical and societal distance between health professionals and their disadvantaged patients. --- Conclusions Health mediators as facilitators of the access and uptake of primary health services among Roma minority groups had been employed as support workers of primary care practices in a Model Programme that established group practices (GP clusters) in Hungary in 2013. The contribution of health mediators during 43 months of the Programme was reflected by the high participation rate (80%) of all clients registered with GPs at health status assessment that exceeded participation rates of other national screening programmes; as well as by the correlation of mediator work time and participation rates at health status assessments and at community health events. Health mediators recruited from the serviced communities can be valuable members of multidisciplinary primary healthcare teams, especially in deprived areas. --- Availability of data and materials The datasets generated and/or analysed during the current study are not publicly available due to containing personal and special data the availability and use of which is regulated by law (2011. évi CXII. törvény az információs önrendelkezési jogról és az információszabadságról). Data collected in the framework of the Programme is stored and managed by the National Public Health Institute of Hungary and available on reasonable request from the director of the Institute. --- Authors' contributions KK contributed to the conceptualization and study design, performed the literature review, analysed data, drafted the manuscript and approved its final version. CK performed the literature review, performed data collection and management, analysed data, and drafted the manuscript. MP contributed to study design, performed data collection and management, analysed data and approved the final manuscript. GF contributed to study design, performed data collection and management, and approved the final manuscript. JS contributed to the conceptualization and study design, analysed data and approved the final manuscript. KB contributed to data collection and management, and approved the final manuscript. R<unk> contributed to the conceptualization of the Model Programme and the study as well, contributed to study design, evaluated data analysis and approved the final manuscript. The final manuscript was read and approved by all authors. --- Funding The Model Programme had been implemented in the framework of the Swiss Contribution Programme SH/8/1. The project was supported by a grant from Switzerland through the Swiss Contribution. K. Kósa, J. Sándor, K. B<unk>ró, and R. <unk>dány were supported during the writing of the manuscript by the GINOP-2.3.2-15-2016-00005 project financed by the European Union under the European Social Fund and European Regional Development Fund. The funders have had no influence on study design, data collection and analyses, interpretation of results, writing of the manuscript or in the decision to submit it for publication. --- Ethics approval and consent to participate The Model Programme had been implemented in the framework of the Swiss Contribution Programme SH/8/1 that specified the indicators for evaluation. Indicators monitored in primary health care are specified in the Hungarian Health Care Act of 1997 (1997. évi CLIV. törvény az egészségügyr<unk>l). The Programme was created on the basis of the Framework Agreement between Switzerland and Hungary (declared by 348/2007. (XII. 20.)) and signed on 20 December 2007. Ethical approval for data collection for research purposes in the Programme was issued by the Scientific and Research Ethics Committee of the Medical Research Council of Hungary (ETT-TUKEB) (16676-3/2016/EKU (0361-16). The protocol of the health status assessment was approved by the Scientific and Research Ethics Committee of the Medical Research Council of Hungary (ETT-TUKEB) (52119-2/2015 EKU). All subjects signed an informed consent form, and the anonymity of personal information was guaranteed during data analysis. --- Consent for publication Not applicable. --- Competing interests The authors declare that they have no competing interests. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: A Model Programme of primary care group practices was implemented in Hungary between 2013 and 2017where virtually all GPs had worked in single practicesaiming to increase preventive service uptake and reduce inequalities based on a bilateral agreement between the Swiss and Hungarian governments. Group practices employed a wide variety of health professionals as well as support workers called health mediators. Employment of the latter was based on two decades of European experience of health mediators who specifically facilitate access to and use of health services in Roma minority groups. Health mediators had been recruited from local communities, received training on the job, and were tasked to increase uptake of new preventive services provided by the group practices by personal contacts in the local minority populace. The paper describes the contribution of the work of health mediators to the uptake of two new services provided by group practices. Methods: Quantitative analysis of depersonalized administrative data mandatorily reported to the Management of the Programme during 43 months of operation was carried out on the employment of health mediators and their contribution to the uptake of two new preventive services (health status assessment and community health promoting programmes). Results: 80% of all clients registered with the GPs participated at health status assessment by invitation that was 1.3-1.7 times higher than participation at the most successful national screening programmes in the past 15 years. Both the number of mediator work minutes per client and participation rate at health status assessment, as well as total work time of mediators and participants at community health events showed high correlation. Twice as many Roma minority patients were motivated for service use by health mediators compared to all patients. The very high participation rate reflects the wide impact of health mediators who probably reached not only Roma minority, but vulnerable population groups in general.The future of general practices lays in multidisciplinary teams in which health mediators recruited from the serviced communities can be valuable members, especially in deprived areas.

INTRODUCTION Disease has classically been associated with disadvantage, with lower socioeconomic and occupational status being associated with chronic disease. 1 Infectious disease has also been disproportionately associated with disadvantage, along with access to health services (expressed by Hart's inverse care law that those who are most in need of healthcare have the lowest access to it 2 ). These patterns of disadvantage extend to targets of discrimination, including women and racial and ethnic minorities. 3 More recently, structural inequalities such as income distribution have been implicated in poor health outcomes and access. 4 Discrimination is largely based on obvious stigmatising characteristics such as race or appearance, 5 and little attention has been paid to less visible or invisible stigmatising statuses, such as homosexuality and homosexual behaviour. As Ross 6 found, there is no close link between experienced discriminatory acts and expectation of discrimination, because those homosexual men who anticipate the worst outcomes remain the most hidden and are thus less likely to experience discrimination. Poor mental health has been hypothesised 7 to result from internalisation of negative attitudes and assumptions about homosexual people, commonly described as 'homophobia'. Internalised homonegativity (IH), internalisation of negative attitudes and assumptions about homosexual people by homosexuals themselves lead to feelings of guilt, inferiority and lack of self-worth. Meyer 7 summarises the evidence and places IH into the conceptual framework of minority stress, in which prejudice, stigma and discrimination create a hostile and stressful social environment that may cause mental health problems. Recent data indicate that IH is directly associated with mental health outcomes, 8 with avoidance of HIV testing and identification as bisexual or straight in men who have sex with men (MSM) (but not HIV-risk behaviours when sexual identification and race/ethnicity were controlled for), 9 and indirectly through the mediators of discomfort with discussing HIV status and condom use. 10 Ryan et al 11 found that family rejection of gay and lesbian adolescents was associated with an 8.4 times higher likelihood of attempting suicide, 5.9 times higher likelihood of depression, 3.4 times higher levels of using illegal drugs and a 3.4 times higher risk of engaging in unprotected intercourse compared with teens who reported no or low levels of family rejection. A recent review of US studies 12 concluded that IH has minimal effect on risk behaviour, although that conclusion has been criticised for being based largely on US data and for assuming a linear relationship between the two variables. 13 However, Ross et al 14 have demonstrated that IH predicts sexual risk behaviour in gay men in Uganda, where homosexuality is heavily stigmatised. The data suggest that IH may vary significantly from country to country as a function of levels of discrimination at legal and structural levels as well as individual experience: that is, ambient homophobia is a major determinant of IH. European data from over a decade ago indicate that general population homonegativity is negatively associated with religion and immigration, and positively associated with urbanisation and economic development. 15 16 The European MSM Internet Survey (EMIS) is a European study with a unique opportunity to research IH in a range of political and social settings. EMIS 17 in 38 European countries is the first pan-European survey of MSM by European researchers. It provides an opportunity to examine IH and both its structural predictors and its associations with HIV risk behaviour in a large and geographically diverse sample, with sufficient countries to enable analysis by countrylevel (macro) as well as local-level (meso) and individual-level (micro) variables. A short form of Ross and Rosser's IH scale has recently been developed with culturally stable items 18 19 to make this possible. We report on an analysis of IH levels by country structural variables including legal discrimination and income inequality, and levels of experienced discrimination and, in turn, its associations with HIV testing and preventive sexual behaviour. We hypothesise that IH will be influenced by legal climate regarding LGB (lesbian, gay and bisexual) variables and possibly other structural issues, and that IH will itself be associated with MSM's visibility and social context and HIV-related prevention behaviours including HIV testing and condom use. --- METHODS The methods of the EMIS have been described in detail elsewhere 17. In brief, EMIS is a joint project of academic, governmental and non-governmental partners from 38 countries in Europe (EU and neighbouring countries) to simultaneously run an online survey in 25 different languages during summer 2010. EMIS was designed to inform the planning of prevention interventions for MSM by identifying prevention needs commonly unmet across MSM ( priority aims), as well as subgroups of MSM who have many prevention needs poorly met ( priority target groups). --- Promotion and recruitment The Internet-based self-completion survey study was promoted online and offline through various media for gay, bisexual and other MSM under the slogan 'Be part of something huge!' More than 235 local, national and international lesbian-gay-bisexual transgender websites undertook online recruitment, and many of these organisations also promoted the survey offline via posters, recruitment cards and face-to-face interaction. By visiting a promotion website, men were exposed to banner advertisements inviting them to participate in the study. PlanetRomeo, Manhunt, Qruiser, Qguys and Gaydar sent individual messages to their users inviting them to complete the survey. By clicking on the banner or the study link, potential respondents were directed to a survey 'landing page' and asked to choose 1 of the 25 languages for completion. Once they had chosen a language, the first page described the study, including its purpose, its voluntary and anonymous nature, and gave contact information on how to contact the researchers. Men were eligible to participate if they were living in Europe, at or over the age of homosexual consent in the country they lived in, sexually attracted to men and/or having sex with men. They also needed to indicate that they understood the nature and purpose of the study and consented to take part. The survey was physically located on servers owned and administered by the company Demographix.com, whose staff was responsible for testing it across different computer platforms/operating systems and with different web browsers, and managing the technical aspects including back-ups and data delivery to the researchers. The data were protected with a 128-bit SSL encryption server not accessible to the public. We neither collected the IP address of participants' computer nor installed any cookies on it. The survey was available online between 4 June 2010 and 31 August 2010. EMIS could be completed in any of the 25 languages and the questions were identical in each language version. It took 21 min to complete the survey on average. No study recompense was offered. All study procedures were approved by the Research Ethics Committee of the University of Portsmouth, UK (REC 08/09:21). --- Data preparation and analysis At the close of fieldwork, there were 184 469 cases in the consolidated file. Partially completed surveys (moving away from the web page before pressing'submit') were not captured. When downloaded three cases were lost because of data corruption, and 2427 cases were removed from the dataset because they gave no country of residence or a country of residence outside the study area. A further 544 cases were removed on the basis that they were women, or men with no homosexual desire or experience, gave no age or were under 13 or over 89 years old. This left a total of 181 495 cases meeting the qualifying criteria. Data from 291 cases from European countries and states that did not reach 100 qualifying cases were removed as were 6995 cases with more than one inconsistency, defined as answers to two questions which could not both be valid. Inconsistent data could be submitted by moving backwards and forwards in the survey and changing previously given answers. It could also be submitted simply by supplying inconsistent answers across one or more questions. Given the length and complexity of the survey, we allowed respondents to have one inconsistency in their given answers; but we excluded all cases with more than one inconsistency, suggestive of random answers being given, which is to be expected in a low-threshold questionnaire like EMIS. This left an analytical sample of 174 209 MSM from 38 European countries. 17 MEASURES AND ANALYSES Measures IH was scored according to the criteria described by Ross et al 18 and Smolenski et al 19 as an additive scale of 7 items, ranging from 0 to 6 (Cronbach's <unk>=0.70). Education was measured using the six levels of the International Standardised Classification of Educational Degrees (ISCED), corresponding to the educational system of each country. 'Outness'-defined as the proportion of people one is out to as gay or bisexual (family, friends, work or study colleagues)-was measured on a 5-point Likert scale from 'all or almost all' to 'none'. The proportion of male friends who are also attracted to men was measured on a 6-point Likert scale from 'almost all of them' to 'almost none of them' and 'I don't have any male friends'. Having ever had an HIV test was a binary (yes/no). We compared this binary variable with the extended variable of how long the respondent had had an HIV test (24 h, 7 days, 1 month, 6 months, 12 months, 5 years, more than 5 years ago) and found that the two measures correlated 0.88. We thus decided to use the binary measure in analyses, particularly given the nonlinear nature of the time measure. Perceived control over sexual risk-taking was measured on a 5-point Likert scale anchored by'strongly agree' and'strongly disagree' with the statement "The sex I have is always as safe as I want it to be." The relative frequency of condom use with non-steady partners was measured as "In the last 12 months, how often were condoms used when you had anal intercourse with non-steady male partners?" (6-point Likert scale 'all or almost all' to 'none'). The number of non-steady anal intercourse partners was asked as 'How many non-steady partners did you have anal intercourse with in the last 12 months?' (number). Sexual diversity of practice: EMIS asked for the recency of a range of sexual practices with non-steady partners. Although this set was originally designed to reflect certain transmission routes for HIV and sexually transmitted infections, it can serve as a proxy for diversity of sexual practices, even if sexual practices with steady partners are not included. Sexual diversity was the sum of the nine different types of sexual behaviours with non-steady partners reported in the past 6 months (mutual masturbation, oral insertive, oral receptive, anal insertive, anal receptive, rimming, being rimmed and insertive and receptive 'fist-fucking'. These were re-coded as binaries to reflect engaging in them or not in the past 6 months and summed to form a measure of the relative diversity of the respondents' sexual repertoires (Cronbach's <unk>=0.77). Size of settlement was measured on a 5-point scale from a 'a very big city or town (a million or more people)', to 'a village/the countryside (less than 10 000 people)'. Legal climate was measured by scoring the presence of the six legislative measures of LGB status (legal discrimination) listed by Wikipedia, 20 with a high score of six for the presence of all legislative protections (homosexual acts are legal, recognition of same-sex relationships, same-sex marriage, same-sex adoption, gays serve in the military, LGB antidiscrimination laws). The Gini coefficient of income inequality in a country was extracted from the UN Human Development Report coefficient for all countries (2008) with the exceptions of Cyprus, Malta, Luxembourg and Serbia, which were derived from an identical calculation method using the CIA Gini score (2008). The Gini coefficient ranges from 0 (all people in a country have equal income) to 1.0 (one person has all the income). It serves as a general economic measure of distribution of wealth. --- Analyses Analyses, unless otherwise indicated, were carried out using SPSS V. 19 and V.20 on the 38-country dataset (table 1). Cluster analysis of the 38 countries was carried out using SPSS V.19: a two-step procedure using Euclidian distances was employed, entering three variables: country LGB legal discrimination scores, proportion of MSM reporting verbal abuse and proportion of MSM reporting physical violence because someone knew or presumed they were attracted to men. Three country clusters emerged including 9, 17 and 12 countries, respectively, with good fit (Silhouette coefficient of cohesion and separation=0.65; figure 1). Bivariate analyses were carried out using <unk> 2 with Cramer's <unk> as a measure of effect size for ordinal data, and t test (using Cohen's D as a measure of effect size) (table 2). Effect size statistic for univariable analysis of variance (ANOVA) for interval or ratio level data was <unk> 2. Because of the large size of the subsample with valid IH scores (N=144 177), almost all analyses were highly statistically significant, so p levels were not reported. Means of IH were graphed against selected structural and behavioural variables in figure 2 2). IH was closely related to the three discrimination-based clusters, with the highest IH in the 'conservative' countries and the lowest IH in the 'liberal' cluster of countries. Using the measures of LGB legal climate (the strongest predictor), the Gini coefficient and the size of place of residence all predicted IH (table 3, figures 2 and 3). IH, in turn, was associated with outness and age. The HIV-associated behaviour most closely associated with IH were (1) ever being tested for HIV (mediated by outness) and (2) a perception of having no control over sexual risk-taking. Sexual diversity had a smaller impact. HIV testing was predicted, following outness as the major component, by the proportion of gay friends, highest educational attainment and age. A second linear regression limited to the demographics, HIV testing, 'outness' and perceived control over sexual risk-taking onto IH indicated that those most closely associated with IH were lack of control over sexual risk-taking and not being tested for HIV. Bivariate relationships between IH and country cluster, Gini coefficient septiles and LGB legal climate (figure 2) show a relationship between these variables and IH. Sexual diversity illustrates (figure 2) an increase in sexual diversity with decreasing IH and a more limited range of sexual activity with higher IH. Table 2 illustrates bivariate associations with strong effect sizes predicting ever having had an HIV test, specifically being 'out' and the proportion of MSM friends, followed by education and the size of settlement. ANOVA of IH score by the relative frequency of condom use with non-steady partners, with covariates, the number of non-steady partners that they had unprotected anal intercourse within the past year and perceived control over sexual risk-taking, demonstrated a small-to-moderate effect size (<unk> 2 =0.03). An accelerating relationship between IH and the relative frequency of condom use (covariates' perceived control over sexual risk and the number of casual partners in the past year) with non-steady male partners was apparent (figure 2). The strong relationship of perceived control over sexual risk-taking (figure 2) with condom use led to an additional analysis of the predictors of this perceived control (table 4). The highest stepwise predictors were the number of non-steady partners one had anal intercourse within the past 12 months, IH, Gini coefficient and LGB legal climate. --- DISCUSSION While these data represent 144 177 MSM in 38 countries, their collection via the Internet constitutes a non-random sample. Previous Swedish and UK data comparing Internet sexual data collection with questionnaire data from a large national random sample from a sexuality study in the same country 21 22 23 indicate that general Internet samples tend to be somewhat younger, more likely to be single, better educated and urban. Internet samples of MSM may be more bisexually identified, worse educated and rural compared with traditional gay venue questionnaire-based samples in Sweden. The Swedish data indicate the known biases for Internet samples by comparing them with a national 'gold standard' study. The EMIS data are likely to be biased towards those who are better educated and Internet-literate and probably more familiar with the gay subculture. This sampling bias will be more substantial in Eastern European countries where household Internet access is less common and a smaller proportion of the whole MSM population participated in EMIS. On the other hand, the Law of Large Numbers would suggest that for the larger country samples, range 3). While the effect size of a <unk>0.1 is minor, generally, the greater the distance of a structural variable from an individual, the more difficult it is to achieve a large effect size. Paths from social climate variables to IH and behaviour are measurable and quantifiable, given that this study has the power to detect theoretically important differences from the 'background radiation' of structural and social climate factors at a country level. Although the study appears to be overpowered at the individual data level, at the country variable level, the n is only 38. Figure 2 graphically indicates the relationships between IH and both structural and behavioural variables. --- HIV testing and IH Higher IH is also associated with decreased levels of HIV test-taking at any time, as mediated by 'outness' (but not directly), and denying that their sexual conduct is always as safe as the respondents want it to be ( perceived control over sexual risk-taking). Thus, higher IH appears to both discourage HIV testing (which may be considered a marker of acknowledgement of stigmatised status) and also to reduce the ability to have safe sex despite the intention to do so. Diversity of sexual behaviour (figure 2) also appears to be restricted by IH, with a lower sexual repertoire, possibly because of the interactions limited by time, locale, anonymity or peer modeling. --- Condom use and IH Condom use for anal intercourse (controlling for the number of non-steady sexual partners for anal sex, and the perceived control over sexual risk-taking) revealed a relationship of lower IH with increasing condom use (figure 2). This suggests that IH is associated with risky sexual behaviour generally and condom use specifically. This unexpected implication of the degree of control over sexual risk-taking bore further investigation. --- Paths between structural variables, IH and HIV risks These data indicate relationships between the levels of legal and experienced discrimination and IH and, in turn, between IH and both outness and gay peer contact and HIV risk behaviours. HIV-risk behaviours associated with high IH include lower levels of HIV testing (not knowing HIV status and/or receiving treatment being a risk), lower levels of perceived control over safe sex and lower levels of condom use for anal sex when this safe sex volition and the partner numbers are controlled for. The relationship of the Gini coefficient and IH is interesting, and here, the Gini coefficient probably represents a level of attitude to equality beyond the purely economic (figure 2) and is consistent with the association found between sexuality-related attitudes and human development indices 15 16 24 25 Impact of homophobic climate and IH and potential intervention levels These data suggest, however, that IH may have pervasive effects on sexual behaviour related to HIV prevention through the association of a homonegative legal and social climate with higher levels of risk and lower levels of preventive behaviours, at both the personal and probably policy levels, although more research is required to elucidate these. Hatzenbuehler et al 26 note that statelevel protective policies modify the effect of sexual orientation on mental health disorders, and it is logical that one of the mechanisms for this may be through the reduction of institutional discrimination, hate crimes, homonegativity and by providing interventions to reduce internalised homonegativity. The public health implications of these data are fourfold. First, it is clear that interventions must incorporate changes in the legal and policy climate (and perhaps religious climate) with regard to MSM. Second, such climate changes need to be communicated to professionals (including healthcare workers) who are at the government or professional interface with MSM clients, possibly with appropriate in-service training. Third, structural interventions should translate into the provision of specific services or outreach to MSM. Fourth, anti-IH measures for MSM might include individual-level opportunities on the Internet anonymously targeting stigmatised sexual minorities. These could be based on short, easily read informational modules, reduction of pathological stereotypes through pride-enhancing modules, role-model stories and approaches to respond to the negative 'voice within' or distancing strategies. The success in reaching some 180 000 MSM via the Internet in three months, Europe-wide, in 38 countries and 25 languages, in this study indicates the potential of this medium and for region-wide rather than countrywide approaches. While effect sizes for some structural components are small, Europe has a population of over 802 million and if we assume conservatively that 2% of men aged 15-65 are MSM, the potential public impact on 5.35 million MSM is considerable. Legislative and legal climate changes and human rights protections are not expensive in monetary terms compared with classic medical screening and treatment. Staff and public servant training or re-training costs, however, may be necessary. As far as we are aware, no study in Europe has previously looked at HIV-preventive behaviour and homonegativity. From these pan-European data, homonegative stigma can be clearly linked through IH to increased HIV risk in MSM at both the population and personal levels. Discrimination is bad law, bad policy, bad psychology and bad medicine: reducing homonegative policy and interactions at the legal and social levels will have a positive impact on the health as well as human rights of MSM populations. --- Author affiliations 1 School of Public Health, University of Texas, Houston, Texas, USA 2 Faculty of Health and Society, Malmö University, Sweden 3 Norwegian Knowledge Center for the Health Services, Oslo, Norway 4 London School of Hygiene & Tropical Medicine, London, UK 5 University College Maastricht, Maastricht University, Maastricht, The Netherlands 6 Regional Center for Health Promotion, Verona, Italy Competing interests None. --- Ethics approval University of Portsmouth IRB. Provenance and peer review Not commissioned; externally peer reviewed. Data sharing statement No additional data are available.

et al. Internalised homonegativity predicts HIV-associated risk behavior in European men who have sex with men in a 38-country cross-sectional study: some public health implications of homophobia.

Overview Early life is a critical period in the programming of the hypothalamic-pituitary-adrenal (HPA) axis, which is essential for maintaining homeostatic balance in the face of stressors (Lupien, McEwen, Gunnar, & Heim, 2009). The social environment plays a crucial role in HPA axis development during this period of heightened plasticity (Lupien et al., 2009). In infancy and childhood, the social environment typically comprises a primary caregiver who is the key external regulator of HPA axis activity. "Social buffering" refers to the ability of social factors (i.e., the primary caregiver) to regulate or buffer (i.e., dampen) the physiological response to a stressor. When a caregiver's ability to serve as a social buffer is compromised, child HPA axis development is affected (Hostinar, Sullivan, & Gunnar, 2014). Research about social buffering may explicate effects of the early caregiving environment on typical and atypical development of the HPA axis; however, much of the work to date has focused on the effect of social support on cortisol reactivity to an acute stressor (Hostinar et al., 2014). This work has greatly increased our understanding of the mechanistic specificity by which caregiver support buffers against stressful early experiences but is also limited in its ability to model the full range of adverse experiences to which children are exposed. In addition to highlighting the effects of acute stressors, research has also shown that the production of basal cortisol is dampened by social support (Hostinar et al., 2014). This review expands the current social buffering framework to incorporate evidence of the mechanistic role of social support in diurnal cortisol patterning in environments of chronic stress. Investigations of stress regulation and the effects of social buffering across time enable us to characterize diurnal cortisol patterning in the home environment in relation to ongoing parenting behaviors. This review focuses on the influence of caregiver support, and lack thereof, on diurnal cortisol patterning for young children in foster care, many of whom experience early and ongoing adversity (see Figure 1). The goals of this review are to (a) review the literature about diurnal cortisol patterning in foster children, with an emphasis on the common experience of neglect (i.e., lack of a social buffer); (b) review evidence that caregiver-based interventions that target responsive parenting strategies affect HPA axis functioning in foster children; and (c) synthesize our findings using the social buffering framework to inform future targeted interventions. --- Caregiver quality and HPA axis patterning In addition to providing a social safety net for maltreated children whose birth parents are unable to provide a safe environment, from a scientific perspective, foster care also represents a valuable "natural experiment" for examining the effects of limited social buffering on children's HPA axis development. Although foster children experience a range of chronic stress, neglect is the hallmark experience leading to out-of-home placement: In a 2013 report, neglect was the most commonly reported type of maltreatment (79.5%; US DHHS, 2015). "Neglect" refers to a caregiver's failure to provide in terms of a child's physical, medical, emotional, and/or supervisory needs (US DHHS, 2015). Within the social-buffering framework, neglect can be conceptualized as the lack of, or absence of, a reliable social buffer during a developmental period when external regulation by caregivers is critical, with consequences for the typical development of the HPA axis and the health and behavioral outcomes influenced by HPA axis activity. A sizable literature demonstrates the impact of early-life neglect on HPA axis functioning in foster children. Though this literature clearly suggests that experiences of neglect affect the HPA axis, the specific nature of this effect varies. The most commonly documented effect of neglect on HPA axis functioning is a blunting of diurnal cortisol (sometimes referred to as hypocortisolism; Badanes, Watamura, & Hankin, 2011;Gunnar & Vazquez, 2001). When the association between neglect and hypocortisolism was first observed, findings were inconsistent with the prevailing consensus in the field that experiences of stress lead to elevated cortisol (Gunnar & Vazquez, 2001). It is important to note that hypocortisolism does not confer the same benefits as blunted cortisol reactivity following successful social buffering. In fact, a peak in diurnal cortisol in the morning has numerous benefits, while a blunted response has been associated with deleterious effects (for full review, see Gunnar & Quevedo, 2007). The consistent finding of hypocortisolism across populations whose early adverse experience largely comprised neglect, such as previously institutionalized children and foster children, challenged the field to consider negative effects on HPA axis functioning specific to the absence of responsive caregiving in addition to effects associated with experiences of abuse (i.e., omission as opposed to commission; Bruce, Gunnar, Pears, & Fisher, 2013;McLaughlin et al., 2015). Research by our group and others has demonstrated a relationship between neglect and hypocortisolism in young foster children. In a sample of preschool-age foster children, our group found the severity of experienced physical neglect to be associated with lower morning cortisol levels (Bruce, Fisher, Pears, & Levine, 2009). In addition, when compared with a low-income community control sample, these foster children were significantly more likely to show low morning cortisol levels (approximately 30% of foster vs. 10% of control children; Bruce et al., 2009). Dozier and colleagues (2006) found foster infants and toddlers, approximately 86% of whom were in care because of neglect, to be more likely to exhibit atypically low or high cortisol patterning at any point across the day than were community controls. Notably, foster children had significantly lower morning cortisol specifically, again pointing to the consistency of the neglect-hypocortisolism relationship. Much of the initial research on cortisol in foster children has been cross-sectional in nature. More recent longitudinal analyses have shown that stability of HPA axis functioning over time may be an additional indicator of early stress effects in foster children. Evidence suggests that diurnal cortisol levels are not only lower but also more variable over time in foster children than in controls (Laurent, Gilliam, Bruce, & Fisher, 2014). A key theoretical link for the extension of the social buffer model from acute stress response contexts to chronic stress response exists in the relationship between acute cortisol reactivity and diurnal cortisol functioning (see Figure 1). Few human studies have directly examined the link between diurnal and acute HPA activity. Recent work from our group that is particularly relevant to our working extension of the social buffering model demonstrates that diurnal cortisol levels measured in a group of children with varying levels of risk (i.e., CWS involvement vs. low income) at regular intervals during the preschool years predicted cortisol response to a laboratory stressor during middle childhood (Laurent, Gilliam, Wright, & Fisher, 2015). Though this study does not specifically address the influence of particular early caregiving environments on the relationship between diurnal cortisol and subsequent response for a stressor, it does serve as a critical first step in the demonstration of this relationship upon which future studies can build to test the moderating effects of early caregiving over time. The exact mechanism through which the social environment affects the HPA axis is challenging to determine because dysfunction could exist at multiple entry points of the system. In humans, measurement of HPA function is typically restricted to cortisol, the end product of HPA activity, due to the level of invasiveness inherent in measuring other products of the system. Animal models have been critical in providing insight into different candidate mechanisms. The HPA axis operates by means of a hormonal cascade: corticotropin-releasing hormone (CRH) is released from the hypothalamus, stimulating the pituitary to release adrenocorticotropin hormone (ACTH), which in turn stimulates the release of cortisol from the adrenal gland. Regulation of the HPA axis is governed by both a direct and indirect negative feedback mechanism; under typical conditions, increased cortisol concentrations directly inhibit the release of CRH from the hypothalamus and ACTH from the anterior pituitary. However, the input into the hypothalamus is multidetermined and includes input from somatosensory, limbic, and higher cortical regions; therefore, glucocorticoid concentrations can indirectly inhibit HPA activity by binding to these glucocorticoid receptor (GR) dense brain structures (Strüber, Strüber, & Roth, 2014). Strüber et al. (2014) propose two pathways through which early life stress can differentially alter cortisol functioning based on quality and quantity of maternal care, suggesting early life stress accompanied by high maternal care results in increased GR expression, whereas early life stress in the absence of high maternal care (e.g., neglect, poor quality care) can result in chronic elevated glucocorticoid concentrations, which in turn leads to a downregulation of GR expression (for full review, see Strüber et al., 2014). Though the measurement of receptor densities requires a level of invasiveness not feasible in humans, the evidence from animal model reviewed above suggests that the differential density of GR expression in brain regions impacting the HPA is a likely candidate mechanism for social environment effects on cortisol functioning. --- Caregiver-based interventions on HPA axis patterning Caregiver-based interventions to improve outcomes for young foster children may function by affecting the quality of the social buffer in the child's life. Strikingly, the interventions reviewed here have also been shown to affect HPA axis functioning. Though these interventions were not explicitly designed to target social buffering, these studies offer examples of opportunities to probe the plasticity of systems such as the HPA axis that we know to be affected by social buffering. In addition, though the specific physiological mechanisms by which these interventions affect HPA functioning remain unclear because of the complexity of this system mentioned previously, these intervention studies demonstrate that the HPA axis is indeed malleable and sensitive to consistent, responsive parenting. Evidence for the effectiveness of these interventions that target consistent, responsive parenting is reviewed in the following paragraphs, with an emphasis on the ways in which this evidence maps onto the social buffering framework. Multidimensional Treatment Foster Care for Preschoolers (MTFC-P) is a multicomponent intervention aimed at improving outcomes for preschool-age foster children by increasing caregivers' positive reinforcement and consistent, nonharsh limit setting (for more details, see Fisher, Ellis, & Chamberlain, 1999). From a social buffering perspective, a mechanism of MTFC-P's action is to increase the experience of reliable and consistent caregiving while in a foster placement and across transitions in caregivers. Results of a randomized, controlled trial showed a preventative impact of the intervention on the HPA axis; specifically, the intervention group did not display the same flattening of diurnal cortisol across time as did the comparison group of children in regular foster care, and the intervention group had cortisol trajectories similar to those of a low-income community control of nonmaltreated children. These findings suggest that MTFC-P protected against this flattening of diurnal cortisol associated with neglect and chronic stress exposure (Fisher, Stoolmiller, Gunnar, & Burraston, 2007). Further investigation of the potential mechanisms underlying this prevention effect revealed the critical influence of caregiver stress, an important correlate of caregiver responsiveness. In particular, foster parents in the MTFC-P condition experienced a sustained decrease in stress associated with child behavior, while the foster parents receiving services as usual experienced an increase during the year-long study period (Fisher & Stoolmiller, 2008). Critically, in families receiving services as usual, higher levels of foster parent stress associated with child problem behavior predicted more blunted cortisol levels in foster children, a relationship not present in the MTFC-P group. From a social buffering perspective, these findings suggest that supportive interventions, such as MTFC-P, may affect HPA axis functioning by bolstering the resources available for foster parents to help them deal with challenging behavior. MTFC-P has also been shown to mitigate the negative effects of a placement change, which can be conceptualized as a shift in the social buffer, on HPA axis functioning by focusing services on providing a consistent caregiving environment in foster care and in permanent placements (Fisher, Van Ryzin, & Gunnar, 2011). Others have focused on foster care interventions in infancy, and as such, have taken an attachment-based approach. A focus on attachment is particularly logical early in life given the critical role of the child-caregiver relationship during this time. The Attachment and Biobehavioral Catch-up (ABC) intervention aims to increase sensitive and responsive parenting following child distress, an aim in line with improving the caregiver's capacity to serve as an effective buffer for the child. Evidence suggests that the ABC intervention is effective at normalizing HPA axis functioning in children with a history of Child Protective Services involvement because of neglect (Bernard, Dozier, Bick, & Gordon, 2015), and that these effects are maintained even 3 years following the intervention (Bernard, Hostinar, & Dozier, 2015). --- Summary and future directions Overall, these studies provide evidence that experiences of chronic adversity across childhood (i.e., neglect and caregiver instability/stress associated with placement in foster care) and diurnal cortisol patterning can be integrated into the current social buffering framework. These studies are a critical starting point for the field's exploration of how interventions designed to influence the quality of caregiving can alter HPA axis functioning. The interventions took a large-scale, multifaceted approach to provide initial evidence for which systems are malleable and which correlates of social support can in fact disrupt (and/or improve) future cortisol trajectories. These data suggest that although the specific mechanisms of interaction between daily parent-child relational quality and daily HPA axis patterning are unknown, interventions designed to improve consistent, responsive, and supportive care demonstrate pliability in this system and are a means by which to normalize HPA axis functioning. And, these findings provide further evidence that the preschool years are a particularly plastic developmental period for social support's influence on HPA axis calibration. Beyond main intervention effects, it is important to acknowledge individual variability in outcomes. For example, it is likely that the level of social buffering required to influence HPA axis patterning exists on a continuum that varies for different individuals. Our review suggests that the absence of a social buffer does not permanently exclude the system from appropriately responding, although it may render the system more vulnerable to downregulation when under stress. Moreover, minimizing parental stress specifically in response to child behaviors and not to overall stress and increasing the stability of the caregiver may be meaningful next steps to probe how the caregiver can buffer HPA axis patterning. However, the large-scale design of these interventions also renders us unable to draw causal inferences about specific mechanisms of change. Future interventions should seek to test these theories by targeting specific components of social buffering to probe the mechanisms of change involved in interventions. One example of this targeted intervention approach is a promising new intervention developed in our lab called Filming Interactions to Nurture Development (FIND). FIND directly targets and aims to increase behaviors that enable caregivers to be more reliable buffers to children. FIND uses the concept of "serve and return" (Schindler, Fisher, & Shonkoff, in press) in the context of strength-based video coaching to demonstrate to caregivers the ways in which they are already effectively responding to their child in developmentally appropriate ways. The intervention includes several elements that are particularly relevant to social buffering, including showing the caregiver instances during which they provided external regulation when the child is distressed. Investigations of FIND's effectiveness and underlying mechanisms are underway. It is essential to note the larger system in which the HPA axis exists and functions as well as methodologies that index chronic stress (i.e., hair cortisol). The HPA axis serves a critical regulatory role in the function and activation of several systems, including brain regions sensitive to levels of glucocorticoids (i.e., amygdala, hippocampus, and prefrontal cortex), immune health, oxidative stress, and the gut-brain axis (Bellavance & Rivest, 2014;Dinan & Cryan, 2012;Lupien et al., 2009). Promising evidence from animal studies and emerging human studies literature suggests that these associations may be multiple entry points into the system for intervention, and that a social buffering framework is a useful means to direct future intervention research and efforts. Future work should continue to seek greater clarity about the mechanisms that link acute and chronic stress and the relative role of social buffering to update and refine our working conceptual extension model. --- Figure 1. Working conceptual model of social buffer theory extension An acute stressor evokes an increase in cortisol reactivity. An effective social buffer can moderate (i.e., dampen) this cortisol reactivity. However, the effectiveness of the social buffer during an acute stressor is dependent on the quality of the parent-child relationship over time. Parents who provide consistent, supportive care are more effective social buffers for their children than those who are not. Over time, these processes (depicted in gray) shape overall functioning of the HPA axis, as indicated by diurnal cortisol, and are moderated by chronic stress, which can take many forms including but not limited to neglect, maltreatment, and socioeconomic adversity. Additionally, we recognize other pathways that likely affect this model and warrant further investigation. The quality of the social buffer may directly impact diurnal cortisol, and the nature of the acute stressor (i.e., lack of food or shelter vs. physical abuse) likely interacts with the quality of the social buffer and social buffering capacity. --- Disclosure statement Dr. Fisher is the developer of the MTFC-P intervention described herein, and a co-owner of Treatment Foster Care Consultants, Inc., which provides consultation on the implementation of MTFC-P.

There is growing evidence that social support can buffer the physiological stress response, specifically cortisol reactivity. We use a developmental framework to review the importance of social buffering in early childhood, a period of heightened plasticity for programming of the hypothalamic-pituitary-adrenal (HPA) axis. The social environment, in which parents play the largest role in early life, is a critical agent in the developmental trajectory of the HPA axis. A prevailing model of social buffering primarily focuses on the role of social support in the context of acute stressors and cortisol response. This review expands this model to provide evidence of the mechanism of social buffering, or lack thereof, across periods of chronic stress by applying the social buffer model to children involved in the child welfare system. We also highlight current interventions that capitalize on the mechanism of social buffering to modify HPA axis functioning across childhood. Last, we synthesize our findings using the social buffering framework to inform future targeted interventions.

Introduction Over the past decade, clinical advances have improved cancer survival, yet morbidity and mortality disparities persist [1][2][3]. These disparities are most pronounced among populations that have historically received inadequate health care and health care services, such as low-income and racially and ethnically minoritized populations. Across the United States, oncology clinicians deliver care for these populations who experience cancer disparities due to structural racism, suboptimal access to healthcare (e.g., due to lack of health insurance, underinsured, geographical isolation), and other causes [3,4]. While studies have addressed the unmet needs among clinicians in the provision of cancer screening [5][6][7] and studies examining the multidisciplinary care for patients outside the United States [8], limited data exists regarding the barriers and facilitators that oncology clinicians within the United States experience in delivering cancer care among populations most at risk for cancer disparities. The American Society of Clinical Oncology (ASCO) strives to support oncology clinicians and other professionals caring for people with cancer. ASCO is committed to addressing cancer disparities and has launched a multipronged approach to achieve cancer health equity [3,9]. Through ASCO's Health Equity Committee (HEC), the Society has advanced several initiatives to eliminate cancer health disparities, including the establishment in 2019 of the "Supporting Providers Serving the Underserved (SUS)" Task Force. The SUS Task Force, composed of oncologists and other cancer professionals with health services and cancer health equity expertise, was created with the goal of developing recommendations to engage, support, and learn from oncology clinicians who deliver care predominantly for populations most at risk for cancer disparities. The SUS Task Force chose to focus on low-income and racially and ethnically minoritized populations given the prevalence of cancer disparities among these groups. As a first step, the SUS Task Force conducted this qualitative study given the limited data available to explore barriers and facilitators to equitable cancer care delivery from the perspective of clinicians. The qualitative study would then inform a quantitative survey that would be distributed as a second step among oncology clinicians nationally to generate data on a national level that could inform recommendations and solutions. --- Materials and Methods The SPUS Task Force utilized the definition of "underserved population" as described by the Health Resource and Services Administration (HRSA), which defines this group as a population of individuals, specifically low-income and racially and ethnically minoritized populations, who have historically received inadequate health care and health care services [10]. The SUS Task Force, in collaboration with prior and current Health Equity Committee (HEC) members, generated a list of 55 medical oncologists across the United States who, based on their knowledge and expertise, spent greater than 25% of their time treating or having had experience treating predominantly low-income and racial and ethnic minority populations with cancer in their daily clinical practice. The list included clinicians who applied to ASCO's grant programs that focused on serving underserved patients. A total of 12-15 participants were anticipated to reach thematic saturation. However, as this study was conducted during the height of the COVID-19 pandemic and given the time limitations of clinicians targeted for study participation, the Task Force sent email invitations inviting 24 clinicians to participate in 30-60 min virtual, semi-structured interviews. These 24 clinicians were purposefully selected from the generated lists of 50 clinicians to participate in interviews based on geographical location. Participants provided verbal informed consent upon initiation of the interview and completed a survey prior to the interview to define their demographic characteristics and to describe the demographic and clinical characteristics of the patient population for whom they deliver care. The interviews were conducted by video conferencing via a web-based platform between 1 March 2021 and 30 April 2021 by investigators (JM and KS) using a semi-structured interview guide developed by members of the Task Force and created in discussion with members of the HEC (Supplementary Materials). The interview questions were developed using the Socio-Ecological Framework [11]. Interview questions explored clinician experiences delivering cancer care for low-income and racial and ethnic minority populations and approaches that could support their care delivery. Two non-clinical members of the team experienced in qualitative methodology, interviewing techniques, and public health (JK and KS) conducted all interviews to limit the impact of clinician-researcher assumptions and experiences on participants' responses. There were no prior relationships or interactions between any of the research team members and the participants. Interviews were recorded, transcribed, stripped of potential identifiers, and imported into qualitative data management software (NVIvo version 12). Descriptive statistics were calculated for demographic and clinical practice characteristics, which included frequencies for categorical variables. The analysis was performed using thematic analysis as described by Braun and Clark [12], given the limited research on the topic. The flexibility of this approach permitted an understanding of the experiences of participants in the broader context of their roles in delivering cancer care for underserved populations while remaining focused on the data and recognizing practical and realistic limitations [12]. The perspectives of the authors were ontological (themes shaped by participants' experiences and reflecting the realities of both participants and researchers) and methodology (shaped by experiences in data collection and analysis). Three investigators with experience in qualitative methodology and backgrounds in public health, anthropology, oncology, psychology, ethics, social work, and health services research (MP, LH, and FH) read samples of the transcribed text to get a broad understanding of the data content. The investigators used a mixed inductive-deductive iterative approach in which the investigators identified constructs driven directly by the data and applied a socioecological framework to the data to generate codes for key points and a codebook. Full transcripts were coded independently by three experienced qualitative coders (as part of Principal Investigator MP's health services research laboratory) who consecutively coded full transcripts, discussed discrepancies, and modified the codebook with the lead investigator (MP). A Cohen's kappa was calculated to measure coder consistency using all quotes from major code categories, with scores that ranged from 93 to 97% suggesting excellent consistency [12][13][14]. Thematic analysis of 1582 unique quotations was conducted, and codes were sorted into broader themes to ensure that data with themes were consistent and that themes were distinct and fine-tuned until thematic saturation was reached, defined as a point when no new additional information was found in the data [12][13][14][15]. Discussions were held between coders and the research team throughout the analysis phase to reflect on how personal experiences were used to attach meaning to themes and if assigned meanings resonated across team members from different clinical and research backgrounds. The research team held reflexive team discussions to ensure meanings assigned to themes were consistently agreed upon by all team members, as described by Barry et al. [16]. This process allowed the research team to identify personal paradigms regarding research involving underserved populations, express an orientation to qualitative research, negotiate a research methodology aligned with the aims of the study, and work to reduce bias in interpretation. Triangulation was applied to compare data gathered from discussions with experts held with other HEC members prior to the study's start with those gathered from the semi-structured interview participants [17]. Analytical memos, field notes, a codebook, coding rules, and meeting notes were kept as an audit trail for dependability and confirmability [18]. The study was reviewed and approved by the Stanford University Institutional Review Board. --- Results Of the 24 eligible participants invited to participate, 12 participated; 10 (2.4%) did not respond to the initial invitation, and 2 (8.3%) agreed to participate but did not have available time to participate in the interviews. Table 1 shows the demographic characteristics of the participants. There were an equal number of male (n = 6; 50%) and female (n = 6; 50%) respondents. Participants self-identified as African American or Black (n = 4, 33%); Asian American Native Hawaiian and other Pacific Islanders (n = 2, 17%); or White (n = 5; 42%); 1 (8%) participant preferred to not answer. Most participants had completed their terminal degree more than 20 years ago (n = 5; 42%) and spent >25% of their clinical effort delivering care for low-income and racial and ethnic minority populations (n = 9; 74%). Most (n = 8; 67%) lived in the community where they delivered care and were employed by a hospital or health-system-owned practice, group, or department (n = 5; 42%) with 10 or less full-time-equivalent oncology clinicians and subspecialists (n = 4; 33%). Most (n = 10; 83%) reported that their patient population was reflective of the broader community in which they practiced and reported that most of their patients were insured by public insurance, either Medicare or Medicare Advantage (35%) or Medicaid (30%), with no insurance or other forms of payment (such as charity care) representing 18% of their patient population. Most participants reported that Non-Hispanic White patients represented 32% of their population, followed by Black or African American (23%), Hispanic or Latino (19%), and Asian American Native Hawaiian and other Pacific Islanders (15%). Thematic analysis revealed three major themes regarding barriers (Table 2) and two major themes (Table 3) regarding facilitators as it pertains to cancer care delivery for low-income and racial and ethnic minority populations with cancer. At least we are aware of the fact that we are short-staffed. We need to hire a dedicated social worker. If we can find the funds, it would be great. So we have to survive with whatever we have. We are working to improve it. We speak closely to the administration, but we got to wait until we see, honestly. --- Table 2. Cont. Because our nurse navigator or social worker are very well-versed in these areas, I depend or rely on them to assist with what resources within the community are available. I have to say that it's still hard. You need administrative buy-in for this as well. But what we have been able to demonstrate to [executive leadership] is that there is actual value to be had, return on whatever upfront investment they made. And we always assure [executive leadership] that we will be good stewards of our resources. We wouldn't ask for stuff without demonstrating upfront the value that that upfront investment was going to make. So that has allowed us to bring in people who are able to do things like understand who needs what service, what support we can deliver beyond the direct cancer-related care and decision-making. Challenge that I deal with every day is fragmentation of care. And so to try and get all these pieces coordinated in a timely fashion, real challenge. So we spend a lot of time and staff effort for that, which the administration here doesn't understand. We don't even have a dedicated social worker here. The patient navigator, she does the social work. Whenever we are faced with transportation challenges or meals on wheels and all that, we ask her if the patient is eligible for those programs. She goes above and beyond to help those patients. So we don't need temples. We need marketplaces. We need places where people can go in and transact. I go in, I have a need, I can be sure that I'm going to get it because, Oh, I am welcome, and the services are there. And yes, the goods are high-quality. Why do I know that? Because the people who run the marketplace are truly held accountable for making sure that the goods they provide for me to purchase are high quality goods. Right now, that's not the way we have [our clinics] set up. Basically, it all boils down to having enough resources to make these things to make these changes that are unique to this population. [Other places] are able to assign additional resources to make the hospital function better. How does it feel to have that luxury? We are asked to keep showing value. For example, in the past, a guy doesn't have insurance, so can't come in. We have been able to negotiate that away with the healthcare system, where there is this idea of accepting the loss lead, if you will, that allows people to come in understanding that [the health system and clinic] is efficient enough to be able to swallow whatever minor up-front expenses there are. But it is a challenge [for them] to see the big picture, but that big picture allows you to understand that your revenue is not based on this single encounter that looks like a loss. I would bet that what happens in other practices like this is that significant corners are cut just because there's no resources to support providing navigation services, and they don't have the nonprofit organization to provide those services; the quality is going to suffer tremendously in this patient population when they don't have access to the navigation services. The main carrot that you'd have to figure out is how do you make providing cancer care in these communities more lucrative so that larger organizations do want to branch out and provide care in those communities and fill the gap that now is being served by most single doctor shops and probably not filled that well. So all of these [approaches for improving care] we have done in such a rigorous way that I have been able to go back to my senior administrators and show them data that has encouraged them to invest more resources into these programs as we're building them. Because of course, you don't want to get administrators in the business of driving their healthcare systems bankrupt, but you also don't want to be in the business of excluding people because of who they are or what they are. And so that has created a real opportunity to look at care delivery programs and how we can optimize them to do both: to be both financially solvent while also opening access to those who truly need it. It's dependent on us, though, to have to make that argument. The inadequacy of things we have around here.. I don't internalize that [the leadership] want to go and kick the dog. But we are always looking for funding for our needs to do this work. I want to say we are able to overcome the disparities because [they] are in a position to invest in the resources that are needed to help our patients. In other words, the challenge is that patients, they may not have the healthcare literacy to follow through the complicated intersections of an oncology patient; we don't have enough navigators to help them. We don't have enough social workers to be assigned to the patient. This is tough for me personally. --- Theme 2: Evidence-based care delivery is challenging due to patients' unmet complicated social and economic needs. I've learned that in my time here, that it's meeting the needs of a patient more so than their cancer. People who don't have access to resources will avoid medical care just because they know it's going to entail something; you know, if they're going to have to pay money or get people to help them get somewhere. So, it's almost as if, so I know that people may delay medical here until they just can't avoid it, and that, I think, is pretty common. For the past several decades, we have made assumptions that we have stereotype patients that are "noncompliant". Not really taking the time to understand what their obstacles might be. Things that we take for granted that we really would never understand. It takes time to get to know your patients [and their needs.] Table 2. Cont. It's a mixed bag. One of the barriers, I think, with the health literacy is a lot of our uninsured patients, their first language.. They don't speak much English at all, so that's been a complicating factor... there is significant issues with health literacy. There are many services patients actually don't have access to... many primary needs, like transportation; patients don't have appropriate phones, they didn't have appropriate Wi-Fi internet service, they did not have email set up to begin with. People who may not have had resources to begin with, and then, when the pandemic hit, being left behind in this big innovative technology shift because they don't have the basics. [My patient] told me that she missed one appointment, and she was fined because the missing appointment. I think it was $66, $65, something like that. She missed the appointment because they couldn't have transportation. So there is a pull-and-push dynamic. So for these different reasons, the proportion of patients who are, I would say, live in poverty, are of certain racial groups, or live away from the Med Center are the ones who come to us because we always are accommodating them. Language is a big barrier. Transportation has been a big barrier, but not at my current institution. They actually have a really good setup where we'll arrange Ubers for any patient that needs transportation issues. So that's been really helpful at this hospital that I didn't have at other hospitals. So I've even been able to get them in a lot easier that way, but yeah, job issues. I've had patients who just refuse to stop working and would miss several appointments because they prioritize their jobs. They needed money over coming in for treatment. That's been a barrier sometimes. Theme 3: Burnout is prevalent among clinical providers due to lack of resources, time, and effort spent advocating for resources and the social and emotional challenges patients face in these settings. The community groups are providing transportation because of the concerns about safety, and family members may have their own concerns. So, the system is stressed and that means the people are stressed, and the doctors are stressed, too. My population has, as it does for cancer in general, skews older, and so patients are reliant on others to be able to get them back and forth for appointments. There is a real health literacy issue. Patients don't want to see the doctor, and so I see patients when they tend to be very far advanced at the time of diagnosis. They don't come in for screening and when they have a cancer, they don't come in until they absolutely have to. So when I was starting as the only full-time oncologist here in this region, I was really worried. How do I be on call 24 h a day every day? And so yeah, lots of challenges in this very rural, very isolated place. It's frustrating when the administration won't listen to me but they'll listen to outside consultants. So yeah, the same things that I was saying and making the case for, the consultants were like, "Yeah, you need someone to help coordinate these new patients. You really need a position; some of the other aspects of running a cancer center". But you had to get someone else to come in and say it. So yeah, story of my life. This contributes to my personal struggles here in a big way. I'm very passionate... and I don't give up, which means I will push and push until I can't. Admittedly, it breaks my heart, but we end up turning away patients that don't have insurance. We just can't financially treat them. The margins are too small to have uninsured patients and write off. We just can't do it; a small center can't do it. There's just not good ways around that, so we directed the patient to a larger center. It breaks my heart. But yeah, there are real issues and I can think of a case recently; man was essentially living in a trailer, didn't have running water, didn't have heat, didn't have any family support. We don't have good options in the community to treat patients like that..living in a situation where we couldn't safely treat him..and so we transitioned to more of a palliative plan of care, and that's just the reality. It's a hard reality for me to see this day in and day out. You know, I think I struggled with that significantly. And also with the fact that, socially, some [patients] are not so supported at home. And sometimes I feel like there's very little I can do to change that. I mean, I can give them the chemo. I can watch them for toxicities, but I really can't take care of so many of the stressors they have at home to make them feel better through their journey. I see a lot of metastatic end-stage patients that with different spectrums, and it's discouraging. The professional aspects related to taking care of this population have been quite a challenge. It's often navigating a patchwork of care that is really challenging for even a medically literate person with means to navigate. This is what has really bothered me the most. When you lose that depth of commitment from the team, especially when trying to do disparities work that's important, but in terms of burnout, I am struggling with figuring out a program that can be put into place. I think being able to get away is important. But it's not always possible. You run into danger when you have just one oncologist or one advanced practice provider. In the last two weeks, I've received the equivalent of four months of referrals, and we are overwhelmed. Patients that are very sick lots of nonmedical needs and medical needs, and my schedule's full, and it's shocking. It feels like it doesn't end. We struggled through COVID, and yeah, the pandemic's getting better, and now we're getting slammed when we're exhausted. So yeah, it just doesn't end. --- Table 2. Cont. More than often, most of us will put 125% effort in doing it. I wonder how much of that can also play into burnout. When people don't feel appreciated, they tend to burn out or they get bored doing the same thing. All the frustrations come out of that fact that we need more resources. So you add all those levels, and you realize that your life is meaningless now. I come here, work hard, go home, work hard. Wake up. Work hard. Come back to work the same story, and I am not doing what I have signed up for, so from my perspective, I think addressing all these three issues becomes extremely important; at least two of them, I would say: the medical records and the mission. We are going to lose our people. There's definitely frustration when you want to see a patient, especially in the hospital, and you'd want to start treatment, but you just can't, and you know that's delaying their care... that can be a huge area of frustration. Working painfully over decades what I came around to recognize was I lost my children's childhood entirely. So that can make for burnout. I struggled for a few years here. Yeah, professionally very rewarding. We've been able to do this and that, but it was tough. The work is time-consuming. It is demanding. With all the rewards that come from being able to serve, we can get blinded to the opportunity cost at the personal level. And families do suffer. The families of healthcare workers suffer. You want to treat that patient and either the health literacy or cultural issues or whatnot, there's a whole variety of reasons why the patient cannot get that treatment. You wonder what else you could have done or what not. I think that takes up a lot of time and can be tiring. So, the cancer center is only a part of the main hospital, right. It's not the predominant part, so, you know, burnout affects the whole hospital, right. It just doesn't affect the cancer physicians. Certainly, there are situations where I feel frustrated by maybe what's happened with the patients before they come into our system or getting certain kind of services that are not allowed by insurance. Here, you know if you order a test, you don't know whether it's going to be done/not done. Are you going to struggle getting that patient a referral for a gastroenterologist? Because no one takes that uninsured charity patient. Now it's unfortunately a lot at the forefront because you know you will not be able to take care of the patient the same way. You're going to struggle through other aspects of care. Table 3. Qualitative Themes Associated with Facilitators of Care Delivery as Described by Clinicians. --- Theme 1: Local connections with community partners and foundations The community organizations not only do they translate and interpret, what they do is, if I say, "Okay, you need to see a GI and surgeon in the next week because this is really important," I message the interpreter, and in addition to the whole explaining to the patient, so that she follows up on it on the backend from the community center. In my time here, I've learned where to send patients, where to do different things, and who to send them to. [Community partners] bring in navigators they funded for several years. One community organization has nurse navigators trained to provide [infusion] services at [community-based] locations for things that are limited and treatments that we feel are safe to give closer to home. We also have a liaison in the community who works with the hospital and our clinic to set up opportunities when people would travel from afar to stay at local hotels. We are able to get a substantial amount of gift cards that help with not just groceries but bill pay because certain agencies can do one aspect but maybe can't meet the other. So we have gotten creative about how we can help pay for a utility bill or a person's groceries. The [community organizations and community members] go out of their way to help get patients here, and we have comfort funds to help patients get care who have difficulty getting care. Our communities have resources to help cancer patients that help to educate our patients. So, I had a patient that I saw who needed palliative treatment. And she said, 'Well, I don't have a car, and nobody in my family has a car.' So, we looked in the community, and there was someone in the community who could drive her. And then the next day we get a call like, 'We can't. The road ends before our house.' So, the person in the freestanding center says, 'I know someone who works in the Department of Transportation,' and she called them up, and they had the road cleared. So, they actually made a road within a two-day period so somebody can drive and bring this patient in for treatment. It's pretty amazing. So, we have our challenges, but we have help as well. I have an intimate, I would say, interaction with leaders in the community. Also, I was introduced to the ins and outs and what happens to folks who don't have the means and how we can support them. We distribute a lot of those grants that we get to local communities to support transportation, sometimes meal vouchers, for those who need to travel and stay away from their homes so they cannot go back and eat that same day. We have these workarounds. Screening colonoscopies is definitely a struggle. We've been getting a few physicians who've been kind enough to say, 'Listen, I would do these without a question. I will not charge them, and I will do them in my private office.' We have multiple connections out there in the community to help empower and engage the lay people. --- Theme 2: It's a calling, not a job This for me is not just a choice, it's a duty; I have to do it. The work I do is definitely more self-motivation. I have always been interested in health disparities, and so I knew early on this is what I was going to do. The self-interest makes you wake up early in the morning and stay up late at night doing stuff to where the stuff you do is aligned with a greater purpose, mission that you subscribe to. That's it for me. This job was a combination of serendipity and recognized opportunity. There are many of us out there that want to help. We just want to be planted in the right direction. We want to be a part of change. We want to be a part of implementation. We want to be able to take the conversation to the next level, we really do. My overall goal is to provide underserved patients with a comprehensive team with differing skillsets to help them navigate and overcome the challenges that they face. I educate myself all the time with regards to no shows and this and that, that dig deeper.. dig a little deeper to better understand where our patients are coming from. I think we have everything we need: native intelligence, interest. And then we just have to apply ourselves diligently to whatever challenge we're grappling with. I left another organization for the opportunity to care for a more diverse patient population. I'm the best chance [the patients who are undeserved] got, and I could try to pass this off to someone else but who? I can't necessarily get them into a primary care provider, or I could get them to a primary care provider but someone who's not able to manage a super complicated situation like this. It's on me. My goal was that patients wouldn't sacrifice anything by being treated here, that they would get the standard of care here that they would get anywhere else. I can confidently say that that's true, but through a lot of trial and error. People need to understand that without job, without housing, without transportation, without healthcare insurance, without health literacy, without all these components, now what you call it collectively the social determinants of health, without tackling all of them at once, we will not be able to provide our patients with the appropriate care that they deserve. Trying to change that culture that wants to ignore all these elements, thinking that it's your responsibility. No. It is our responsibility. --- Theme 3: Experiential Training Was Key to Delivering Care for Underserved Populations The informal training was during my residency. My residency gave me a good foundation going into my oncology fellowship to think maybe how I should think about questions and people differently. We didn't really get any instruction on [how to deliver this care]. Then certainly, in practice, out at meetings and stuff, it's nothing that I focused on. I imagine there probably are some lectures and access to [formal lectures], but it's nothing that I sought out. I learned by doing. It taught me, at least, right off the bat that you have to handle these patients differently. I did my fellowship at an institution where I saw few underrepresented groups. It wasn't until I started practicing that I learned how to do this. When you complete your fellowship and then you transition to a community-based program, there is this idealistic approach that you will have all the resources available. But the demands of the patient population and your time constraints and what you need to do to make sure people get care... well I don't think that that is absolutely clear [in fellowship or other formal training]. Early on in my residency, I learned that if the patient was sitting in front of me, I would actually pick up the phone and schedule the scans and the blood work and set it all up before they left. It was almost like... because there was no real secretary to help with all of that, right? And so, I think early on, I think I learned that there's a population that, unless you set up very clearly all the expectations and line up things, the chances are they may not be able to follow through. --- Barriers 4.1. Theme 1: Lack of Executive Leadership Recognition of Resources Required Participants noted the significant amount of time and resources needed to deliver evidence-based cancer care and the continual need to justify necessary resources to executive leaders. For example, one explained, "..the administration [needs to] understand that we need a staff person just to coordinate records. [It] doesn't make sense to them because they're used to a surgery clinic or a primary care clinic where a patient comes in, sees the doctor, goes on about their day". Another noted, "I don't think we are getting enough support from the outside, so we have to survive with whatever we have". One stated, "it all boils down to having enough resources to make these things to make these changes that are unique to this population". Within the primary theme of lack of administrative support and resources, two additional subthemes emerged. --- Subtheme 1: Reliance on Short-Staffed Team Members All participants described dependence on under-resourced and understaffed teams to address health-related social needs and noted the significant need for dedicated staff to ensure evidence-based care delivery. One stated, "We have quite a busy practice. We are only two oncologists here. We don't have enough navigators to help [patients]. We don't have enough social workers to be assigned to patients". Another noted, "We don't even have a dedicated social worker here. The patient navigator, she does the social work. Whenever we are faced with transportation challenges or meals on wheels and all that, we ask her if the patient is eligible". Another stated, "From the moment a patient is referred to the time they start treatment... to coordinate [care] is really hard. I want to hire a navigator, but we just haven't been able to. We've had an opening for over a year and haven't been able to fill it". One stated, "Interestingly, less-resourced clinics, like ours, do a phenomenal job, considering how difficult it is to generate resources for people who may not be able to pay. In other words, anybody can come here and if they don't have resources, they're not turned away. [They can be turned away] at a private, more-resourced hospital, but it doesn't happen here. That is the challenge. Then, all the frustrations come out of that fact, that we need more resources and [the leaders] are not investing". --- Subtheme 2: Constant Need to Prove Value All participants noted unmet needs in advocating for additional resources and support staff to ensure evidence-based cancer care delivery. One stated, "So me talking to the administration, they don't get it. They don't understand what it takes. What I constantly get told is we don't need the staff to do that, which I hate to think is the barrier". Another stated, "So people want to restrict the lens of what they look at to a convenient sample when, in reality, every life matters. So, the fact that somebody didn't come into your center of excellence doesn't mean that their life is not valuable to someone, doesn't mean that you shouldn't be counting them when you're trying to figure out how well are you doing and where the opportunity is to do better. In fact, it is looking at those places that gives you the greatest opportunity to do better and have greater impact than if we just wanted to keep buffing and shining the glittering part of our healthcare system that we're most proud of". Another stated, "We need [additional navigators and social workers], but we don't have them, and will the hospital invest in it, why? You have to keep asking for certain things.. And you get X amount. So, I feel like it's a big issue". One stated, "The urban poor have not traditionally been the focus on my healthcare system, and so we have to explain and show time and time again why and what is needed. It's exhausting". --- Theme 2: Care Delivery Inhibited by Unmet Complicated Social and Economic Needs Participants noted that the delivery of evidence-based care was adversely impacted by challenging patient social and economic needs. Many discussed how the healthcare system currently does not address these needs or only partially addresses them. One stated, "The [health] system is not set up to do this... I can't get care to my patients if they do not have a way to get to my clinic or they do not have a way to eat or they do not have a place to live. This is not something that our clinic is set up to do". All participants noted that health-related social needs impacted patient wellbeing and frequently inhibited evidence-based care delivery. Many noted that patients lacked transportation to the clinic or had extensive commutes. Almost all participants, however, had identified solutions to overcome transportation-related issues. One stated, "Transportation is a concern, but we have ways

Overcoming cancer health disparities requires an understanding of the etiologies that drive these persistent disparities. To date, little knowledge exists regarding barriers and facilitators experienced by clinicians in their delivery of cancer care for populations most at risk for cancer disparities, such as low-income and racially and ethnically minoritized populations. The aim of this study was to assess the perspectives of clinicians across the United States, with a particular emphasis on understanding modifiable barriers to ensuring evidence-based cancer care delivery for populations most at risk for cancer disparities. Findings revealed the impact of clinical infrastructure, resources, and support to deliver cancer care equitably, social and economic challenges that often inhibit evidence-based care delivery, and the crucial importance of relationships with the community and other clinicians in the community, as well as attention to clinician wellness. These findings reveal areas of unmet need ripe for solutions to achieve cancer health equity.

"The urban poor have not traditionally been the focus on my healthcare system, and so we have to explain and show time and time again why and what is needed. It's exhausting". --- Theme 2: Care Delivery Inhibited by Unmet Complicated Social and Economic Needs Participants noted that the delivery of evidence-based care was adversely impacted by challenging patient social and economic needs. Many discussed how the healthcare system currently does not address these needs or only partially addresses them. One stated, "The [health] system is not set up to do this... I can't get care to my patients if they do not have a way to get to my clinic or they do not have a way to eat or they do not have a place to live. This is not something that our clinic is set up to do". All participants noted that health-related social needs impacted patient wellbeing and frequently inhibited evidence-based care delivery. Many noted that patients lacked transportation to the clinic or had extensive commutes. Almost all participants, however, had identified solutions to overcome transportation-related issues. One stated, "Transportation is a concern, but we have ways to get people the rides they need". All participants noted the prevalence of other health-related social needs such as food insecurity, housing instability, and limited health literacy. One participant stated, "There were patients that would use their rideshare to get to the clinic because that was the one thing that was covered, but they were hungry". Another stated that "patients would forego treatment with a response of 'I don't have money for groceries.'" Housing instability was also noted as particularly concerning and impacting care delivery. One participant stated, "The social circumstances are something that we cannot help. If someone has no place to stay and they stay in a shelter, their capacity to come back and forth is not possible. Somebody who is living in a shelter, then there's nobody at the other end to take care of the poor in a safe manner. These are some of the things that are beyond individual's control and definitely beyond the control of the hospital". Issues involving health literacy were frequently mentioned as barriers to care. "I'm using words like 'doo-doo hole' to explain anal canal to a patient because that's the only word that they would really understand. The challenge is identifying the patients who require that terminology and require that discussion". Another stated, "So my thoughts are that there may be challenges with limited health literacy not just in our non-Englishspeaking population but throughout our population". In response to challenges regarding health-related social needs, participants indicated searching for solutions to ensure continuity of care for their patients. Participants described how they created transportation solutions, such as ride shares, clinic vans, and gift cards for gas money, given the prevalence of this unmet need among their patient population. One stated, "I have patients who live in a place where they don't have personal transportation means. They have to depend on a Medicare/Medicaid van to bring them to get treatment or to be seen, and what have you". Another participant addressed how the clinic provides monetary support for both transportation and food. "We give a lot of gas vouchers to our patients to give credits for that. We do a lot of gas vouchers, meals, meal vouchers as well. This is not something that happens rarely. I mean, every day we have patients who are going to get one or the other or both". Some described how they responded to these unmet social needs by changing their clinical workflows. One stated, "What we've done, at least in my practice, is that we don't let them walk out until they have their appointment and until they have their imaging scan scheduled. We wrap up all of it in the office. It's resource intensive for my medical assistants and the navigators, but that's the only way we can ensure that continuity of care". --- Theme 3: Burnout Prevalent Due to Lack of Resources and Time Spent Advocating and Proving Value All participants noted that a lack of resources to deliver equitable, evidence-based cancer care coupled with increasingly burdensome administrative processes negatively impacted their wellbeing. Many noted that these issues contributed greatly to "burnout". One stated, "So, I would say, my sources of frustration are more systemic. I mean, it affects all of our patients and all of our doctors. Preauthorization, documentation. I'm sure a lot of this was meant [to help]. I don't know about preauthorization, but some of the documentation was meant to make things more efficient but if I made a little graph of my practice of how much time I spend with patients and how much time I spend doing paperwork, I'm sure those lines are becoming inverse". One stated, "Now I understand the price that we often pay as professionals. We feel rewarded by the opportunity to serve others. But oftentimes, we steal time from families and our loved ones. There are only 24 h through any day, and I was alone in my work of filling out paperwork and more paperwork just to get the care to the people who need it". Within the primary theme of burnout, two additional subthemes emerged. 4.6. Subtheme 1: Wellbeing of Clinicians Associated with Delivery of Evidence-Based Care Many participants noted that burnout was associated with their inability as professionals to overcome their patients' social and economic needs-needs that directly inhibited the delivery of evidence-based care. One stated, "So the disparity frustration I've really struggled with like, how am I going to get the patient to see a gastroenterologist, a pulmonologist, who will take them happily without supplementing and pleading and requesting them get squeezed in to do it because they have no insurance". Another stated, "[Our patients] have a significant comorbid health load, and they have significant challenges about what they can't meet with transportation or not having a working phone, being homeless, things that we see all the time. I think that really impacts the way providers try to then deliver the care and then manage their own time and affects us a lot". Another stated, "If you don't have social workers, navigators, good systems in place, then it's very easy to work your clinic load then round at a hospital and not finish your day until 10:00 or 11:00 at night". Another stated, "There are only a few of us. So, we would work through lunch, and it's just not sustainable; you have to use time and manage your time in ways that is still going to protect your ability to not faint, your ability to be appropriate with patients; if you don't have anything left, you don't have anything to give to your patients". --- Subtheme 2: Burnout from Worry about Other Team Members Participants noted their concern about the wellbeing of their team members, including nurses, social workers, and ancillary staff members. One stated, "Yeah, burnout is real. I've remained cognizant of that, but it affects everyone in the care team. So it's not just the oncologist, it's everyone in the care team. So I think I recognize it in nurses. I think I recognize it in clinical staff, support staff. When you're dealing with very sick, high-acuity patients with a lot of needs, burnout problem is real". One stated, "We don't go into this thinking that we go on to fix problems right away. So, we get used to that; psychologically, we have to have some amount of resilience in order to survive. Even as there is progress, it is still a challenging discipline that requires resilience on the part of the physician or the patient care providers. It can be a lot to think about how nurses and others are experiencing the brunt". --- Facilitators While barriers to care delivery were stressed throughout the interviews, participants also noted facilitators. --- Theme 1: Local Connections with Community Partners and Foundations Participants emphasized the importance of forming collaborative relationships with community members and community organizations to facilitate cancer care delivery. One stated, "Without the [community organizations], we would just be in dire, dire straits because so many patients wouldn't be able to get the care they need". Another described how collaboration with community organizations helped them to address health literacy and language barriers. "We went to the Korean center that has Korean-speaking people who go out in the community, and they help translate. They're sort of our translators". Participants discussed how their interpersonal relationships in the community positively impacted their approach to cancer care delivery for low-income and racially and ethnically minoritized populations. One stated, "I guess in community practice that comes along based on your interactions with [other community members]. Like, at least with my gastroenterologist, we've talked so often before about these patients have no real insurance; how do we tackle them? And then we discussed it with the disease management team and came up with the pathway of accepting them". Another stated, "Someone might need something like lymphedema therapy, and they're not in the network with the closest location, but they need to get to another part of town and they can't because of distance because of the transportation. Well, the [community organization] provided with a grant for funding the hospital and also working with the clinic to provide rideshare to provide medical Uber rides, and that has helped a number of individuals get to appointments or be able to get the care they need". Another stated, "It's about who you know and how you know them. You know we go out and talk to people in the town and other advocacy groups and say, 'here's the problem,' and then we can come up with the solutions together". One stated, "We told [the surgeons] about those patients who are waiting for elective surgery related to their cancer diagnosis and how they were at risk for bad outcomes because they were already delayed, and they put them on an emergency list, and they're operating on them". All participants noted that their passion and ties to the community facilitated the delivery of evidence-based care for populations most at risk for cancer disparities. One stated, "If you don't help the underserved population.. well, you see the difference every day". Another stated, "I was born here. I was actually born in this hospital, the hospital where I practice. My great grandfather was sheriff here. So, I have deep ties here, and I don't know; it's really tough to recruit, but I grew up in this rural setting so I always knew I wanted to come back". Another noted, "So I finished my fellowship and started with this practice, and I asked when I was interviewing about where there was a place I could focus on [underserved populations] both in terms of patient population, demographics. and research". One stated, "It's more than just a job to all of us. This is what we are supposed to do". --- Theme 3: Experiential Training Participants noted that they received little to no formal training in working with populations most at risk for cancer disparities. All participants agreed that the care they delivered was based on practical experience. For example, one participant described, "Unless someone is here, all the knowledge you get outside is all book knowledge. It doesn't apply to your circumstances. The learning takes place here on [the] ground in the surroundings in this environment". Another participant stated, "The challenges are incredibly difficult, not insurmountable, when you complete your fellowship, and even if you've spent time in an academic center, and you then transition to a community-based program, there is an idealistic approach. But if you read through the published information, it doesn't help you". Another participant reflected on how their first-hand experiences, such as witnessing racism, influenced how they approach their clinical practice. "I did not have any formal training about that, honestly. I wasn't aware of the magnitude and the history of the United States. I knew about the racism and racial discrimination, and so on and so forth, but I did not realize to what resources are available, and the demands on the patient, and the degree this has also extended to our time now and continues in every single day on certain groups of the population, and your time constraints. I don't think that is absolutely clear [when you move to an under-resourced setting], specifically on our Black community. But I totally understand this problem and in completely different terms and different ways". --- Discussion In this qualitative research study, oncology clinicians described their experiences in delivering care for populations most at risk for cancer disparities, specifically low-income and racially and ethnically minoritized populations. Study findings uncovered key barriers that contribute to ongoing cancer care delivery disparities and clinician burnout. The study also revealed facilitators that can help overcome these barriers and contribute to improving evidence-based care equitably and clinician wellbeing. Clinicians not only delivered care with a dearth of resources but also increasingly devoted time and effort to advocating for resources to facilitate evidence-based cancer care delivery for underserved populations. In 2017, the American College of Physicians position statement recommended that administrators and executives analyze and mitigate or eliminate the adverse effects of administrative tasks on clinicians [19]. Our research findings, four years after these recommendations were made, further support this need, especially in clinical settings where policies and practices may place undue burdens on understaffed and under-resourced teams. For example, a critical evaluation of the care teams' workforce composition, such as the efficient and most effective use of social workers and navigators, can identify if the care team is appropriately staffed to address healthrelated social needs. The correct composition of care team members, with each team member working to the top of their licensure, not only improves workplace efficiency but also more effectively delivers care for populations with complex social and economic challenges. Studies show the value of such team compositions, specifically in the consistent delivery of high-quality care and cost savings via the elimination of time spent by clinicians on activities that may be more effectively and efficiently delivered by other members of the team [20][21][22]. Our findings reveal that the burden of team staffing evaluations was often organic and shouldered by the clinicians who were delivering the care. Such evaluations should not be the sole responsibility of the physicians delivering the care alone but also be heavily influenced and led by the administrative leaders responsible for resource allocation. Our results overwhelmingly highlight the importance of organizational commitment at the executive level, including visible long-term actions and investment dedicated to improving the delivery of care for underserved populations. While most participants had organically created process and practice workflows to ensure that they and their teams could deliver evidence-based cancer care for low-income and racial and ethnic minority populations, participants consistently noted a lack of executive organizational commitment to such improvement efforts. Consistent with prior work documenting the impact of conflicting policies and practices on moral distress among clinicians [23,24], our study revealed that organizational barriers, including a lack of devoted resources and investment, often resulted in conflict with professional commitment to ensuring equity in care delivery for underserved populations. It is well known that such conflict and moral distress can greatly contribute to burnout and impact the wellbeing of cancer care clinical team members [23,25,26]. Clinicians in our study reported that their under-resourced teams, in addition to continually having to advocate for resource allocation, diverted their time and focus from more clinically important activities, which was linked with stress and burnout for them as well as added stress due to their concern for the wellbeing of their team members. Interpersonal and professional relationships with other clinicians in the community and community-based organizations were consistently identified as a major facilitator for evidence-based cancer care delivery among low-income and racial and ethnic minority populations. This organic network of clinicians and community-based organizations served as an "underground railroad" network with a shared vision that collectively mobilized resources and directed them to ensure the delivery of evidence-based cancer care. Such networks of community and local engagement are recognized by the Institute of Medicine and others as the key component of addressing cancer disparities at the local level [27][28][29]. As noted in organizational behavior literature, community capacity is crucial to achieving change and involves building both political and social capital within and outside of the community [30]. As demonstrated in other studies evaluating facilitators for health promotion, skills in community capacity building and collective efficacy [31], or shared values and norms for the common good, were identified in this study as a key component of care delivery in resource-limited settings and were skills attained only through experiential learning, as participants highlighted. A more formalized process to collectively solve prob-lems relating to promoting health equity should be developed and led by the organization as a whole. Such organizational shared vision, leadership, voice, and power can expand the reach and scale of organically derived networks and is necessary to prioritize resources and infrastructure necessary to deliver equitable evidence-based cancer care at the clinic, system, and community levels. Burnout was identified as an ongoing, critical issue due to the many barriers revealed in this study. Administrative tasks, the constant need to prove the value of the resources for ensuring care delivery for underserved populations, and under-resourced teams contributed to ongoing wellbeing concerns among clinicians. While passion for care delivery among underserved populations was noted as a facilitator for ensuring equitable care delivery, such professional fulfillment also contributed to burnout, especially when systemic resource limitations or the complex social and emotional needs of patients inhibited evidence-based care delivery. Concern for other members of the team and their wellbeing also contributed heavily to the mental load and wellbeing of clinicians. Modifiable barriers identified in this study are ripe for potential solutions that ASCO and other organizations may implement. These solutions, depicted in Table 4 and derived from key barriers identified in this first phase of the SPUS Task Force, will be evaluated in a national survey distributed to oncology clinicians across the United States to determine which solutions should be considered and prioritized by the national society. Such findings will help to move from description to action in the support of oncology clinicians caring for populations most at risk for cancer disparities. Our study had several limitations. First, our method of identifying participants relied on a convenience sample. Due to the lack of available data regarding clinicians who predominantly deliver cancer care for low-income and racial and ethnic minority populations, the current study utilized non-random, purposive sampling (based on the geographical location of participants), relying on lists generated by members of the Task Force and the HEOC to identify potential participants for the study. As such, the sampling strategy may bias the results. Second, the study was conducted in 2020, during the early phase of the COVID-19 pandemic, when many healthcare settings and community support organizations experienced unprecedented challenges in delivering care and support for patients. The response rate to our requests for interviews was around 50%, with the primary reason for non-participation given as "time limitations". It is unclear, if these time limitations were due to limitations of routine practice or were due to additional burdens attributable to the COVID-19 pandemic. Either way, it is possible that our study sample was biased towards clinicians within more resourced settings or who have personal characteristics that make them more likely to respond. Third, no participants self-identified their ethnicity as Hispanic or Latino/a/x or a race other than the ones we listed. Due to the small proportion of oncology clinicians nationally with these self-identified characteristics, it is unknown whether experiences may differ. Finally, our study was conducted only among US-based clinicians and in English, limiting findings. Strengths of our study, despite these limitations, were that we had an inclusive, diverse sample of respondents with varied experiences delivering cancer care for low-income and racial and ethnic minority populations, including perspectives of care delivery for unregistered migrants, refugees, and similar groups across the United States. Additionally, we achieved thematic saturation across our interviews, allowing for confidence in the representation of our sample and the generalizability of our findings. --- Conclusions The current study provides key insights into the barriers and facilitators faced by clinical oncologists delivering cancer care for low income and racial and ethnic minority populations. Clinicians identified modifiable barriers, most notably their difficulty in convincing executives and administrators to understand the complexities of care for underserved patients and the need for additional resources. While clinicians are adept at working with what they have and building workflows to ensure the best possible care for their patients, the numerous day-to-day challenges contribute to moral distress and burnout. Collective efficacy and community capacity across multiple levels are important and necessary facilitators for care delivery in under-resourced settings. This work lays the foundation for understanding how and why clinicians work with underserved populations and provides insight for future actions to support clinicians and improve cancer care for these populations. Informed Consent Statement: Informed consent was obtained from all study participants involved in the study. --- Data Availability Statement: Data Availability Statement: All data supporting the findings of this study are available within the article, Supplementary Material files, and from the corresponding author upon reasonable request. --- Conflicts of Interest: The authors declare no conflict of interest.

Overcoming cancer health disparities requires an understanding of the etiologies that drive these persistent disparities. To date, little knowledge exists regarding barriers and facilitators experienced by clinicians in their delivery of cancer care for populations most at risk for cancer disparities, such as low-income and racially and ethnically minoritized populations. The aim of this study was to assess the perspectives of clinicians across the United States, with a particular emphasis on understanding modifiable barriers to ensuring evidence-based cancer care delivery for populations most at risk for cancer disparities. Findings revealed the impact of clinical infrastructure, resources, and support to deliver cancer care equitably, social and economic challenges that often inhibit evidence-based care delivery, and the crucial importance of relationships with the community and other clinicians in the community, as well as attention to clinician wellness. These findings reveal areas of unmet need ripe for solutions to achieve cancer health equity.

September 11 and the war in Iraq (Stevenson 2011: 243) interest in cosmopolitanism as a transformative perspective has not declined (Rovisco and Nowicka 2011: 1). In fact, cosmopolitanism is increasingly regarded as a key analytical tool for studying a variety of outlooks and ethico-political practices, especially since Beck and Sznaider (2006) proposed going beyond the limitations of methodological nationalism by adopting more complex and global perspectives. Alejandro González I<unk>árritu's third feature film, Babel (2006), is global in terms of production and release. An international co-production between France, Mexico and the United States, it was first shown at the Cannes Film Festival and later at a number of other international festivals. It features seven languages -English, French, Spanish, Berber, Arabic, Japanese and sign language-and four settings -the US, Mexico, Japan and Morocco-in which, as in other multiprotagonist films, a number of characters are connected by a global thread. Babel may be seen as both a post-9/11 movie and as a filmic counterpart to Calhoun's insights in that it adopts a cosmopolitan stance for understanding the complexities of the global network society. The film brings to the fore the fact that the war against terrorism, promoted by one nation-state, affects remote places, where innocent people are criminalized as scapegoats for the sake of the protection of more privileged citizens. Hence, it problematizes non-cosmopolitan forms of domination and subordination in the current network society. Some critics have analyzed Babel in relation to Manuel Castells's concept of the "network society", that is, one "whose social structure is made of networks powered by microelectronics-based information and communication technologies" -by social structure, he understands "the organizational arrangements of humans in relations of production, consumption, reproduction, experience, and power expressed in meaningful communication coded by culture" (2004: 3). For example, Paul Kerr explores the film's social relations of production, drawing on Castells's assumption that the work process tends to be globally integrated while labor is locally fragmented. For Kerr, "the film package, Babel, was constructed by globalized, casualized labor, and assembled by international agencies and companies which circulate capital -in much the same way as the circulating object, the Winchester, changes hands in the film" (2010: 39-40). In another example, Celestino Deleyto and Maria del Mar Azcona view Babel as a powerful fictionalization of the network society, characterized by the two emergent social forms of time and space: 'timeless time' and the'space of flows'. The film's narration is simultaneously governed by 'traditional time', that is, linear, irreversible and measurable chronology, and 'timeless time', a new conceptualization of temporality that gives us immediate access to events happening in any part of the global network society. Similarly, the'space of places', characterized by geographical location and physical proximity, is superseded by the intangible space of flows -of information, capital, work and technology-across continents (2010: 51, 66-67). In this article, I propose an analysis of Babel's critique of unequal power relations in the global network society, in particular, after the 9/11 attacks. The film makes extensive use of formal strategies that encourage the viewer to reflect about noncosmopolitan positions of domination and subordination in the network-society of the information era. I will draw on Castells's belief that, although a network-based society is ideally a non-centered form of social structure, it is also a source of dramatic reorganization of power relationships. Babel narrates four stories that take place in three different continents over a fiveday span. The film scrambles the chronology of these storylines by means of a regular pattern that divides the film into twenty-four sections. The opening section, set in the Moroccan Atlas, introduces a goatherd, Abdullah (Mustapha Rachidi), and his two teenage sons, Ahmed (Said Tarchani) and Yussef (Boubker Ali El Caid). We witness the transaction of a rifle between Abdullah and his neighbor, Hassan (Abdelkader Bara), who was once given the gun as an act of gratitude. Hassan gives the rifle to his two teenage sons, who decide to test it, first on some rocks and then on a moving bus. They realize they have hit the bus, and run away, at which moment the second section starts, taking us to San Diego, where an undocumented Mexican nanny, called Amelia (Adriana Barraza), is in charge of two siblings, Mike (Nathan Gamble) and Debbie (Elle Fanning). The events of this storyline shuttle from San Diego to a Mexican village close to Tijuana on the other side of the US-Mexico border. The third fragment, also set in Morocco, revolves around a married couple from the United States, Richard (Brad Pitt) and Susan (Cate Blanchett), who happen to be Mike and Debbie's parents. While on a tourist bus, Susan is shot, and the incident is misinterpreted as an act of terror. The fourth section takes spectators to Tokyo and revolves around Chieko (Rinko Kikuchi), a deaf-mute teenage girl. Apparently unrelated to the other stories, we later find out that Chieko's father, Yasujiro (Kôji Yakusho), gave the same rifle that has now wounded Susan to his Moroccan hunting guide as a present at some point in the past. The fifth section returns to Morocco, where Ahmed and Yussef are running home after having unintentionally shot Susan. This order (Morocco-San Diego/Tijuana, Morocco, Tokyo) is almost strictly followed until the end of the film, resulting in a complex, fragmented and multi-focal film (see Deleyto and Azcona 2010: 50-56). By means of its multi-protagonist narrative, Babel depicts the global reach of the accidental shooting of a United States citizen in Morocco in order to exemplify the devastating effects of the sovereign prerogatives of certain countries, specifically the US, in a global world increasingly driven by fear of terror after 9/11. As Ezra and Rowden argue, "the rhetorical deployment and exploitation of September 11 and the subsequent Iraq war by political officials and some elements of the American media created a popularizing discourse that turned all US citizens into potential victims and all foreigners into potential victimizers " (2006: 11). In Babel, the shooting incident is rapidly labeled as a terrorist act by the United States although, as the Moroccan media claim, the country has been free from terrorist activity for many years. Yet, the connection between the global security network and the international media system is highly effective in quickly spreading the misconception, putting extreme pressure on the Moroccan authorities. Fearing the damage that the international crisis could cause to the economy of the country, they abuse their own subjects, ignoring human rights. As the film shows, diverse forms of domination are ascribed to certain nation-states and their privileged subjects, and treatment by law enforcement authorities under international pressure is uneven. In contrast with the abusive practices endured by the family of goatherds in Morocco and the Mexican undocumented immigrant, the United States and Japanese citizens are treated respectfully. Notably, the suggestion of terrorist links by the US Government is enough to shut down the air space in the area, so that Susan can be evacuated. As Ulrich Beck states, since the global and the local are "mutually implicating principles", global processes "transform the quality of the social and the political inside nation-state societies " (2002: 17). The unbalanced logic of power governing present-day network society is a key theme in Babel, and it runs parallel to the film's uneven network structure. In her analysis of multi-protagonist films, Azcona draws on Deleuze and Guattari's "acentered structure" of "the rhizome" as the structure that best represents socalled mosaic films or network narratives like Babel. As she explains, because these movies lack a narrative center, they potentially disrupt the hierarchical organization reflected in conventional movies -which tend to privilege one character over the others (2010: 21). In the past decade, a number of multi-protagonist films -such as Traffic (2000), Crash (2004) and Syriana (2005)-speculated about the emergent network society, which was being theorized in the social sciences (Wellman 1999;Castells, 2009;Van Dijk, 2006). In these films, the network morphology and the proliferation of points of view it offers tend to counter the inequalities they expound. While Babel roughly conforms to this logic, its presumably even structure disguises unequal power relations at its basis. As the narration unfolds, the apparent structural and narrative balance is put to the test and, gradually, the tourists' (Richard and Susan's) storyline takes shape as the central and dominant one, mirroring the film's central concern with power dynamics. Therefore, against the apparently decentralized narrative conventions of ensemble movies, the shooting incident becomes the focal event in Babel, and the storyline that deals with it, together with its protagonists, are given more prominence. In other words, although the four-part structure of the film, as expected, appears to be balanced due to scrupulous compliance with established turns and an even distribution of screen time, the structure turns out to be onesided for a number of reasons that I aim to explain next. Apart from the higher visibility of the two internationally acclaimed stars playing the US tourists in Morocco (Brad Pitt and Cate Blanchett), at a structural level, one key feature that makes the tourists' storyline superior is its high interconnectedness with the others. Richard and Susan's narrative strand has a higher "networking power". Castells coins this term to refer to "the power of the actors and organizations included in the networks that constitute the core of the global network society over human collectives or individuals who are not included in these global networks" (2009: 42). James S. Coleman refers to a similar kind of power as "social capital": in the network society, the relations between the nodes increase the social capital, facilitating productive activity and "making possible the achievement of certain ends that in its absence would not be possible" (1988: S98). For Mark Buchanan, although the network society is in principle a noncentered structure, it is actually organized through central hubs that dominate network activity. In this sense, he argues that the Internet and the World Wide Web have few elements with a huge number of links, and that this is symptomatic of greater power (2002: 36). As John Urry exemplifies, a small number of nodes, such as Microsoft, Google, Yahoo, CNN, BBC have a very large number of links, and, therefore, control the web. The web is, for him, an "aristocratic network", where those who are connected become ever more connected, their power being increased over time (2007: 215). In Babel, as in the network-based society it represents, switches are relevant sources of power, and Richard and Susan's storyline proves more powerful because it is the only one related to all the other stories. Susan's shooting is causally connected with the events of the Moroccan (Ahmed and Yussef's) narrative strand; it triggers the action of Amelia's narrative, and also infiltrates the Moroccan and Japanese media. None of the other narrative strands is bound to all the rest. For example, Chieko's plotline is faintly linked to the Moroccan story through the transaction of the rifle, and infiltrated by Richard and Susan's strand through the media but it does not interact with Amelia's. In turn, the Moroccan line of action is tied only by coincidence to Richard and Susan's and to Chieko's storylines but it is not directly bound to the Mexican one. Finally, this narrative thread is not connected with the Japanese or with the Moroccan plotline, being attached solely to Richard and Susan's. Hence, the centrality of this storyline is grounded on its higher connectivity, exerting potential influence on the other narrative strands. Positions of structural dominance and subordination are further qualified by the nature and direction of the connections in the network narrative. As in many multi-protagonist films, in Babel, the four storylines are initially presented as independent, and they become gradually interconnected. Yet, the links between narrative strands are not always of the same nature nor do they all have the same relevance. For example, while the Japanese plotline is bound to the two stories that take place in Morocco, its attachment to them is very weak. They are barely hooked up by the slender causal link provided by what David Bordwell calls a "circulating object" (2006: 97): the rifle that is given as a sign of gratitude is sold to Abdullah and later accidentally wounds Susan. It is an "attenuated link" (99), to use Bordwell's terminology, since the interconnectedness is governed by chaos theory, in the shape of the so-called 'butterfly effect'. This impression of chance is reinforced when some of the characters involved in the shooting fleetingly appear in the Japanese news. Although this plotline shares key themes with the others, it mostly serves to establish parallels and contrasts between characters, and causal connections are minimized. Thus, the interconnectedness of the Tokyo line of action is fairly inferred as coincidental and ephemeral. In the lines that follow, I will explore how the level of connection of the tourists' storyline is not only characterized by a larger number of links but also by more consistent narrative ties than those of the three others. Firstly, in a film in which most people are remotely bound by little more than chance, the attachment between Richard and Susan and the characters in Amelia's storyline stands out as the most solid: they live in the same house, Amelia is entrusted with their employers' home and children, and Richard and Susan are Mike and Debbie's parents. Secondly, while many narrative links between the plotlines are based on sheer accident, these two stories are interconnected by causal coherence through a number of telephone conversations between Richard and Amelia, in which he compels her to stay at home with his children the same day her son is getting married. The repetition of one of the telephone calls from different points of view invites the spectators to interpret these conversations as pivotal, to weigh up the scenes from both points of view, and to trace the connections and build structural hypotheses about the nature of the relationship between the two narrative strands. While multi-protagonist films usually depend on chance, and not on traditional causality, to interconnect their narrative-strands (Bordwell 2006: 93, 100;Azcona 2010: 37), I will try to show here that Babel deviates slightly from this logic: the rifle may get the ball rolling but Amelia's fate is governed not just by random chance but by a direct order from her employer. This is also indicative of normalized positions of domination and subordination both in the employers' house and, globally, in other parts of the world. The fact that one of Richard's telephone calls to Amelia is shown twice and from two different points of view urges us to think about and examine the characters' motives and needs. This speculation is symptomatic of the fact that they have choices and a certain amount of control over the events. It points to principles of causality, together with a certain share of responsibility on the part of the characters for the consequences of their actions. Is Richard's request that Amelia stay at home with Mike and Debbie on her son's wedding day inevitable? Is his inflexible attitude towards her a major cause bringing about the near death of Amelia and the children on the border? What options are available to Amelia after Richard's request? Is her decision to go on a one-day trip with the children irresponsible or imprudent? For Azcona, although network narratives seem to be just built out of ephemeral or attenuated links, there often seems to be a place for reflection on the "power of the individual" for the responsibilities behind apparently randomly caused acts (2010: 35). This is also the case in Babel. Susan's shooting in Morocco is clearly marked as an accident, even if it is one that brings light to the contradictions of a global situation in which children are sent off to work with weapons. Amelia's decision to take Mike and Debbie to her son's wedding in Mexico may be motivated by random events (Susan's accident, in turn caused by the transaction of the rifle, and the fact that Susan's sister cannot make it to San Diego in time) but is a conscious decision on her part (we see her trying to leave the children with a friend before deciding to travel to Mexico with them). Since the telephone call is repeated in both storylines from different perspectives, we are encouraged to think that what happened later on the border could have been avoided if the characters had acted differently and, therefore, some kind of responsibility is ascribed to them. Moreover, the fact that the first phone call opens Amelia's plotline and closes Richard and Susan's indicates chronological contiguity between the two narrative strands, further supporting the hypothesis of a cause and effect logic, and departing from the randomness of the rifle, the circulating object that characterizes some connections in Babel. Ultimately, the cause-effect nexus between the two narratives suggests structural subordination of the Mexican storyline to Richard and Susan's. Since the events of this plotline influence to a great extent what happens in Amelia's, the latter can be considered to be an appendage or a prolongation of the former. The repetition of the phone call plays a key role in the articulation of the overarching theme of power relations, hinting that non-cosmopolitan dynamics embedded in everyday dialogue underlie the characters' behavior, bringing about disastrous events. Wanting to know who is responsible for the tragic events on the border, spectators are prompted to wonder how the characters could have acted differently, given that, in the second version of the call, we have more knowledge about their motivations and the effects of their choices and actions. The calls in Amelia's plotline point to her subordinate status and degree of vulnerability. Our first impression is grounded on Richard's authoritative tone, in contrast with the nanny's subservient attitude, and on the fact that the employee is not allowed to defend her position, since Richard hangs up on her. In this scene, we tend to identify with Amelia: although we barely know her, she has been presented as a loving and thoughtful person when we see her tenderly playing with the children who are enjoying their time with her. As Deleyto and Azcona argue, this portrayal of the Mexican nanny as nurturing and protective is enhanced by the warm colors and lighting of the house, including Amelia's red T-shirt (2010: 101). As the phone call unfolds, we only see her image, and feel, therefore, closer to her than to the anonymous, authoritative voice at the other end of the line. While the movie criticizes Richard's unwillingness to genuinely engage with the nanny in the first version of the first call, it maneuvers for the opposite effect in the second -at the end of Richard and Susan's story-when we get his point of view. His responsibility is minimized this time because we know that he has been living an extreme situation trying to save Susan's life, and that he has been pushed to his limits. In this case, we get a close-up of his smiling but tearful face as he speaks to his son; and we can hear a more humanized voice, this time unmediated by technology. We are, thus, invited to sympathize with his role of suffering victim. Yet, even if the movie tends to exonerate Richard in the second version, the many dimensions of the calls in their different versions crucially point to the lack of cosmopolitan interconnectedness in the network society. The film does not give us answers about who is responsible for the events that follow the interaction but asks us to reflect on it. It is hinted that Richard has at least some responsibility, even though something akin to despair has influenced his inflexible and authoritative behavior towards Amelia. Richard's domination over his employee not only parallels the superiority of his storyline over Amelia's but also the overarching theme of unequal power dynamics in the global network society portrayed through the film. In this sense, the oneway communication between Richard and Amelia mirrors her later attempts to be listened to by other characters like the Border Patrol agent and the deportation officer. In other acts of communication in Babel, the emphasis on the contemptuous attitude of law-enforcing agents towards a number of distant people -namely Amelia, Santiago and the Moroccan protagonists-hints that there are global grounds for the unfortunate experiences of some characters (beyond those merely attributable to individual wrong choices). For example, the viewer is invited to suture together Amelia's unidirectional conversations and the US State Department's unilateral management of the shooting. Likewise, the abuse of the Moroccan characters is symptomatic of human rights violations against Islamic suspects pursued in relation to the 9/11 events when, after having declared solidarity with the US people in fighting terrorism, the protection of human rights in Morocco was actually weakened. Clearly, the expanded executive powers in a global conflict become the unifying global force binding the suffering of distant others. Babel critiques the fact that the principle of sovereignty should take precedence over the prevention of human suffering. The film calls for a cosmopolitan consciousness that, as Nick Stevenson would say, reminds global humanity of the ways in which we are morally interconnected with one another, while seeking to struggle for a form of politics that aims to accommodate difference (2011: 249). In other words, the management of the shooting incident in the film serves to exemplify how the United States' economic, political and military supremacy facilitated a nation-centered response to the 9/11 attacks, instead of promoting transnational politics based on a more cosmopolitan understanding of the other. In Babel, one-sided preoccupation with security causes unfounded fears of terrorism and brings chaos to the networked, interdependent nation-states all over. For example, Amelia's ill treatment on the border is articulated as a consequence of the war on terror resulting in more severe immigration laws, as part of the global chain of narrative cause and effect. Thus, it can be argued that the United States' one-sided management of the shooting is at the basis of all the plotlines because it is, to a very large extent, deemed responsible for their resolution. Recreating these dynamics, instead of the centerless narratives of other multi-protagonist films, the structure of Babel places Richard and Susan's predicament at the center of the network narrative. The domination and pervasiveness of the tourists' plotline is further highlighted by the two infiltrations of the shooting event in the Tokyo storyline, functioning as the nexus between the two stories. The first intrusion takes place while Chieko is tediously channel surfing, and a news item about the shooting filters into a random succession of local programs. The scene has a key structural function based on certain expectations and responses rooted in the network movies' generic conventions. For the viewers that are familiar with these films, the intersection of characters from different storylines seems inevitable just because we have been following them from the start: the more the narration focuses on their separate lives, the more we expect relevant encounters among them. Then, when people start to meet, even by chance, we feel what Bordwell calls a "satisfying omniscience" (2006: 99). Since the first intrusion in the TV news linking the tourists' and the Tokyo storylines is the first indication that this narrative strand is attached to the multi-protagonist structure, it makes us experience the rewarding omniscience that Bordwell describes. This reward validates Richard and Susan's plotline as central for its ability to provide the evidence that helps us make sense of the film's structure and plot. A similar intrusion is repeated later in the Tokyo plotline, further reinforcing the role of Richard and Susan's narrative as pervasive and a determining factor in the film. While the Japanese police officer is reading the hand-written note that Chieko has given him in the previous fragment of the Tokyo storyline, we see Susan's face on the Japanese international news and we learn about the outcome of Richard and Susan's story. The contrast between the secrecy of Chieko's note, which to our frustration we are never able to read, and Richard and Susan's conspicuously happy ending reinforces the hypothesis that, in Babel's unbalanced structure, the US protagonists are more worthy of attention than any other characters in the movie. Yet, the officer's uninterested look at the television set, together with Chieko's look of boredom while channel surfing earlier in the narrative, problematizes the omnipresence of the shooting incident and the prominence given to its protagonists. The blatant infiltrations of the shooting incident in the Japanese news is a representation of what Castells calls "the power of the switch", or the ability of dominant actors and institutions to connect different networks, promoting their cooperation by sharing common goals (2009: 45, 51). In Babel, the pervasiveness of the shooting -which points to post-9/11 paranoia about security and the subsequent military operations-and its prominence in the media is a representation of this logic of power. The power of the switch between the security network and the comprehensive media networks resides in its apparent capacity to suppress time -as evoked by the random succession of programs in Chieko's channel surfing scene-and in the capability of the shooting event to obliterate space. The subordination of the Moroccan characters that inhabit the periphery of the network society is visually enhanced when the shooting permeates through the Japanese news across a rapid, random succession of local programs. The Moroccan characters' faces do not fit in and become defamiliarized inside Chieko's bedroom, replete as it is with her personal belongings and pictures. The news beams Yussef and Abdullah's close-ups to this digital tapestry on the Japanese TV, visually representing the space of flows, "made of places connected by electronically powered communication networks through which flows of information [...] circulate and interact" (Castells 2009: 34). Notably, Yussef and Abdullah's local selves are only global when they become secondary actors in the news of Susan's shooting. Their identities are unimportant until they are linked to the presumed terrorist attack of an 'American' tourist. The film emphasizes their insignificance even more when, later in a different sequence of the Tokyo story, the TV news narrates a 'happy ending' for the US citizen's narrative strand, while ignoring the fates of Yussef and Abdullah. Yet, since the movie has previously invited us to engage with the disgrace of the humble Moroccan characters, it also calls our attention to the subordination and the oblivion affecting the lives of the people who are peripheral to the network society. The way in which the film constructs temporality also contributes to Susan and Richard's prominence. As has already been mentioned, Deleyto and Azcona see the film's narration as simultaneously governed by two types of temporality: traditional time and timeless time. They argue that the film's temporal arrangement mirrors timeless time when chronological time is suppressed, and sequencing is cancelled and reversed. Timeless time is exemplified by the links between fragments, evoking a sense of immediacy between events happening in remote places (2010: 50-52). I want to show that timeless time is also marked by the recurrent events connecting narrative strands -Richard's telephone calls to Amelia, and the double intrusion of their narrative into the Tokyo plotline-and that this also contributes to making Richard and Susan's strand more pivotal. As has been mentioned earlier, the repetition of scenes is a generic clue that marks them as highly informative, encouraging us to think about interweaving meanings or building structural hypothesis about the network organization of the film. Furthermore, the repeated scenes in Babel give us information about the actual temporal order between the different storylines. For example, the telephone calls between Richard and Amelia reveal in retrospect that the Mexican story takes place five days later, chronologically starting after Susan has been evacuated from Tazarine and gone into hospital, and not simultaneously, as the actual arrangement of segments seems to indicate. Similarly, in the Tokyo storyline, the TV news makes us realize that this has been brought forward with respect to the action set in Morocco. After the segment in which we see Abdullah and Yussef's faces on Chieko's television, the Moroccan segment that follows shows the kids hiding the rifle under a rock in the Atlas Mountains (in Yussef and Abdulah's plotline the Moroccan goatherd and his sons have not been captured by the Moroccan police yet). In Babel's arrangement, the events on the TV news in Tokyo are broadcast prior to their actual happening in the Moroccan strand. Hence, the recurrent, binding scenes prove to be focal in that they raise awareness about the temporal structure of the film, based on timeless time: they disrupt previous hypotheses about the film's temporal construction. Since those pivotal scenes revolve around the shooting, they further reinforce the preeminence of this event, and the centrality and omnipresence of Richard and Susan's storyline. Therefore, Richard and Susan's relevance and domination in the network society run parallel to the structural supremacy of their storyline and its evocation of timeless time because in the film, as in the network society, timeless time indicates power. For Castells, timeless time is the result of the annihilation of time in the networks of the space of flows, and dominant functions are organized in networks that belong to a space of flows, which ties them up around the world (2009: 20, 34). In Babel, the dominant position of Richard, Susan and the US State Department in the aftermath of the shooting is symptomatic of timeless time, especially when the introduction of dialogue is not synchronized with the matching shots. This is most obvious when telecommunications are involved, for instance, when the voice broadcasting the news is heard before correlative images appear, overlapping with the previous shot. In another example, before Susan is evacuated, we can hear a dialogue from a telephone conversation between Richard and somebody from the US State Department. The metallic voice abruptly overlaps with an extreme close-up of Richard and Susan as they come to terms with their past in the intimate space of Anwar's (Mohamed Akhzam) humble home. This shot is followed by a series of quick shots frantically portraying Susan's evacuation in a helicopter while the voice at the end of the line announces that a helicopter should arrive soon and Richard desperately asks: "how long?" He saturates his time to the limit, struggling to stop the biological time that governs Susan's pulse and imposes a temporal limit on her life. In another sequence in Tazarine, the chronological sequence is suppressed when Richard's authoritative voice, demanding from a Moroccan Police Officer an ambulance "now", is previous to the men's actual meeting, while the officer is still seen approaching Anwar's house. By stressing Richard and Susan's privileged status as white, wealthy, United States citizens in comparison to other characters and the centrality of their narrative strand, Babel speculates about and criticizes unequal relations of power across nations and the distinct consequences across borders. The interest that Richard and Susan arouse is underlined by the use of visual and acoustic contrasts and parallelisms. In the closing section of the film, we see the barren Atlas Mountains in Morocco, where the frightened, unarmed, teenage Ahmed has been shot dead by the Moroccan Police. Yussef stares at his brother's body while it is being removed, a long shot zooming-in to a close-up on Yussef emphasizing that he is virtually the only witness of the killing, that Ahmed has died anonymously and unnoticed. In a brief flashback, Yussef nostalgically recalls happy moments playing with his brother in the wind. The music becomes softer, slows down and almost stops, evoking the void left by Ahmed. The motif of the wind that allows the boys' innocent play in the flashback visually connects with the next shot, narrating Susan's evacuation from Tazarine, the wind now produced by a Red Cross helicopter. Hence, the sad meanings associated with the natural wind of Ahmed and Yussef's game contrast with the meanings of hope but also inequality associated with the powerful artificial wind produced by a helicopter which will save Susan's life. The inescapable strong wind spread by the Red Cross global network blows mercilessly over the anonymous Tazarine bodies, humbly bending down as they try to protect themselves from it. The quick shots render endless examples of similar gestures of people covering their eyes with their hands in order to protect them from the unyielding, penetrating dust, its mighty pervasiveness stressed by the incisive repetition of a few staves from the soundtrack. This sequence dramatizes the overwhelming of the local by the global. At the same time, the music repeats itself in a spiral of increasingly higher volume, indicating that this is the climax of the film. The media emphasize Susan's nationality suggesting that she will be saved because so many efforts have been devoted to this end, given that she is a privileged citizen from the core of the global security network. At the opposite end of the scale, Moroccan Abdullah, his sons and his neighbors suffer ill treatment by the Moroccan Police, who are trying to keep under control what has become an international crisis. The wind motif linking the two narratives and the use of framing, together with editing and music, underline the contrast between the unequal outcomes of the two tragedies and the relatively uneven interest that they arouse. These scenes echo others describing Amelia's subordinate position, when she is treated as mere disposable labor by Richard, and later humiliated by the Border Patrol Agent and the deportation officer. The ruthlessness of the Sonoran Desert, metonymically associated with the immensity and severity of the Atlas Mountains in the Moroccan storyline evoke transnational forms of violence in the network society. In Babel, the global reach of the shooting and its consequences problematize the non-cosmopolitan, sovereign control of security in order to fight a complex matter concerning the world at large. The film also suggests that the United States finds legitimation in defining security as the supreme value due its dominant position in the global network society, and to the interpretation of the September 2001 events as an attack on 'America'. It is made clear that the sovereignty of the nation-state and the security of its citizens take precedence over the suffering of strangers. Babel denounces that, in the wake of 9/11 attacks, the nation-state has become a transnational source of legitimate violence in the global network society. As discussed, the management and global reach of the war on terrorism is at the basis of the four storylines. We can argue that what actually connects the four narrative strands is a critique of the sovereignty of one nation-state in matters of security, which is gradually shaped as the cause of subsequent global effects. Hence, the viewer is compelled

changed the course of contemporary history in more than one sense, including the revitalization of the so-called "war on terrorism". For Craig Calhoun, 9/11 is the date on which the "non-cosmopolitan side of globalization struck back " (2002: 871). Given that most of the terrorists were Arabs who had studied in the West, the attacks unveiled a dark side to globalization in which flows of people, money, weapons and drugs suddenly became a challenge to security and state sovereignty (871). As Calhoun puts it, "the terrorist acts were framed as an attack on America rather than an attack on humanity" (870). The events precipitated a new state-centered politics based on the idea of a 'just war', which made military war on terrorism 'inevitable' and justified, crushing the cosmopolitan concerns that, in the view of many, had started to flourish by the late 1990s (870). The coming down of the Berlin Wall in 1989 had questioned the development of militarism, and undermined the opposition between East and West in favor of a more cosmopolitan focus on human interconnectedness that challenged the hegemony of the nation-state. As Gerard Delanty emphasizes, the awareness of global interdependence in the post-Cold War made cosmopolitanism all the more urgent (2009: 98). Crucially, Vertovec and Cohen (2002) explored the multiple facets of cosmopolitanism as a descriptive social category and as a normative concept focusing on the struggles for global justice and the construction of a new world order. Today, although such ideas seem to be wearing thin afterCrash (Paul Haggis, 2004) Syriana (Stephen Gaghan, 2005) Traffic (Steven Soderbergh, 2000)

mensity and severity of the Atlas Mountains in the Moroccan storyline evoke transnational forms of violence in the network society. In Babel, the global reach of the shooting and its consequences problematize the non-cosmopolitan, sovereign control of security in order to fight a complex matter concerning the world at large. The film also suggests that the United States finds legitimation in defining security as the supreme value due its dominant position in the global network society, and to the interpretation of the September 2001 events as an attack on 'America'. It is made clear that the sovereignty of the nation-state and the security of its citizens take precedence over the suffering of strangers. Babel denounces that, in the wake of 9/11 attacks, the nation-state has become a transnational source of legitimate violence in the global network society. As discussed, the management and global reach of the war on terrorism is at the basis of the four storylines. We can argue that what actually connects the four narrative strands is a critique of the sovereignty of one nation-state in matters of security, which is gradually shaped as the cause of subsequent global effects. Hence, the viewer is compelled to interweave, and establish causal connections between, the events following Susan's shooting and its pervasiveness throughout the other storylines. For example, Amelia's tragic border-crossing story is presented as a consequence of the shooting, together with Richard's one-sided and inflexible behavior, echoing the US authorities' management of Susan's wrongly labeled terrorist attack, and the way they actually handled the 9/11 events. This logic is reminiscent of the practices of control and exclusion set up on the US-Mexico border after the 9/11 attacks. These exclusionary protocols go against the notion of "cosmopolitan hospitality", an ethical attitude already proposed by Immanuel Kant in the 18th century as "the right of a stranger not to be treated with hostility when he arrives on someone else's territory" (1991: 105). In Babel, the viewer is compelled to think about the intertwining structural and narrative layers and the events that lead to Amelia's deportation, together with Ahmed's death and the likely terrible ending for his family in Morocco. Multiple connections lead us to deduce that the unfortunate consequences that some characters suffer are not explained as an apparent converging-fates strategy. While some scenes are riddled with coincidence, misled by the circulating-object device of the rifle and the conventionalized role of chance, the functions of the links between storylines, repetitions highlighting varying points of view, and thematic connections emphasize a causal logic grounded on unbalanced global power dynamics. The uneven structure of Babel mirrors positions of domination and subordination enabling one nation-state to define the goals and values in the global war on terror. In conclusion, this analysis deconstructs Babel's deviant structure as a parallel artifact of the film's thematic critique of state-centered politics in the current network society. The multi-protagonist structure metaphorically embodies the power system of the emergent network society, where dominant functions are organized in networks. In consonance with the unequal network society that Babel represents, the centrality and superiority of Richard and Susan's narrative strand resides in its capacity to influence the other three due to its higher interconnectedness -roughly based on the number and nature of its links-and to its networking power, granting a good prospect for programming priorities and values. Babel supports the view that global forms of conflict, including the war on terrorism, pose profound challenges, given the potentiality of the global network society to become the site of conflict. The film defends the cosmopolitan need to interrogate the nation-state as the basic unit of political activity. Instead, Babel promotes more balanced, transnational political practices grounded on a cosmopolitan understanding of humanity at large.

changed the course of contemporary history in more than one sense, including the revitalization of the so-called "war on terrorism". For Craig Calhoun, 9/11 is the date on which the "non-cosmopolitan side of globalization struck back " (2002: 871). Given that most of the terrorists were Arabs who had studied in the West, the attacks unveiled a dark side to globalization in which flows of people, money, weapons and drugs suddenly became a challenge to security and state sovereignty (871). As Calhoun puts it, "the terrorist acts were framed as an attack on America rather than an attack on humanity" (870). The events precipitated a new state-centered politics based on the idea of a 'just war', which made military war on terrorism 'inevitable' and justified, crushing the cosmopolitan concerns that, in the view of many, had started to flourish by the late 1990s (870). The coming down of the Berlin Wall in 1989 had questioned the development of militarism, and undermined the opposition between East and West in favor of a more cosmopolitan focus on human interconnectedness that challenged the hegemony of the nation-state. As Gerard Delanty emphasizes, the awareness of global interdependence in the post-Cold War made cosmopolitanism all the more urgent (2009: 98). Crucially, Vertovec and Cohen (2002) explored the multiple facets of cosmopolitanism as a descriptive social category and as a normative concept focusing on the struggles for global justice and the construction of a new world order. Today, although such ideas seem to be wearing thin afterCrash (Paul Haggis, 2004) Syriana (Stephen Gaghan, 2005) Traffic (Steven Soderbergh, 2000)

Introduction COVID-19 sparked a global public health catastrophe and a series of additional concerns, including an economic downturn, unemployment, and mental instability. The pandemic has affected individuals from all around the world, causing anxiety, stress, worry, dread, repugnance, and poignancy, in addition to the illness [1]. The epidemic has posed an unprecedented challenge to the restaurant industry. Community lockdowns, social distancing, stay-at-home ordering, travel and mobility restrictions, and other strategies to flatten the COVID-19 curve have resulted in the temporary closure of many hospitality businesses and a significant decrease in demand for businesses that were allowed to continue to operate [2]. the in-restaurant dining: the décor of the venue, the setting of the table, the music, the scents, the lighting, the exquisite cocktails, and the show made by the chefs are all defining elements of the luxury dining experience. Customers that visit fine-dining restaurants do not do so only to satisfy their hunger, but to indulge in the fantasy of a narrative that became compromised during lockdown, when all venues were closed to the public. Secondly, plating is a key element in the fine-dining experience. This became an issue during lockdown, because of the chef's inability to maintain the authenticity of the aesthetic of a restaurant by transferring all the different artistic elements from a plate into a delivery cardboard box. Although the consumer sentiment has been mostly analysed during the pandemic [9], to the best of our knowledge very few studies have investigated consumer sentiment from online reviews on the fine dining. This study aims to fill this research gap by using text-mining approaches to explore the evolution in consumer sentiments and to analyse the change in the overall sentiment from the fine-dining industry in both periods. Moreover, sentiment analysis is a technical tool used to track brand and product sentiment in consumer feedback and better understand customer demands and has become a crucial tool for monitoring and understanding client sentiment. However, to the best of our knowledge, no empirical research to date has considered a comparative analysis of consumer emotional responses in the Romanian fine-dining industry in the pre-pandemic and during the pandemic. Furthermore, no sentiment polarity has been explored in either of the two periods. The paper is organised as follows. The section of theoretical development emphasises the most relevant opinions regarding the customer sentiment over time, while the third section is dedicated to the presentation of the data and methodology. The section of empirical results highlights the most relevant findings, and the paper ends with the main implications, limitations, and future directions of the research. --- Theoretical Development 2.1. Impact of the COVID-19 Pandemic on Global Restaurant Dining The pandemic has had a significant impact on all businesses at a global level, with the hospitality industry being one of the most badly hit by the pandemic [10]. Travel restrictions, lockdowns, and stay-at-home orders restricted people's movement by stringent enforcement, causing major disruptions to the industry [11,12]. It has also been estimated that 13.4 million jobs from the restaurant industry could be affected by the ongoing global pandemic [13]. The food service industry is highly vulnerable during epidemic crises because of its high reliance on physical human interaction [4] (p. 3813). In March 2020, in the early stages of the pandemic, the global table-booking website, OpenTable, launched the State of the Industry website to showcase how COVID-19 has affected restaurants worldwide [14]. The study has been ongoing-the data, which are being continuously collected, are based on a sample of over 20,000 restaurants on the OpenTable network and illustrates the number of seated diners from online, telephone, and walk-in reservations or, in other words, the number of customers that dined in restaurants during the pandemic (2020 and 2021) in comparison with the pre-pandemic days in 2019 [14]. The restaurants provide OpenTable with information on their inventory, which has enabled the company to create a detailed and accurate year-over-year comparison, comparing the same days of the week in 2020 and 2021 "to the same day of the week in 2019 (not the same date)" [15]. The data published by the company clearly indicate a significant drop in restaurant bookings at a global level from mid-March 2020, compared to mid-March 2019: 47% on 15 March 2020, to 83% on 17 March 2020, to a dramatic drop of over 99% in only a few days on 22 March [15]. The percentage of over 99% has been recorded since the beginning of May (3 May 2020) and has not gone below 90% since 20 May [6]. Starting from 21 May, the data show a minor improvement. Between 21 May and 5 June 2020, OpenTable reports a growth of nearly 10% (OpenTable [14]). The percentage improved to below 50% in the 3rd semester of 2020 (49% on 31 July 2020), down to an encouraging 26% on 1 September [14]. Over the last trimester of 2020 and the first trimester of 2021, the booking percentages fluctuate dramatically from as low as 10% on 6 September 2020 to as high as 66% on 1 February 2021 [14]. Looking at the state and country division of the data, it becomes obvious how the pandemic has evolved differently in different corners of the world, depending on the number of infected individuals at different times: for example, on 6 September 2020, when the global restaurant bookings was down by an average of 10% compared to 2019, the percentage in British Columbia was as low as 3%, compared to Hawaii at a complete polar opposite, being as high as 95% [14]. Analysing the changes in the relationships between sustainability and the hospitality industry following the onset of the COVID-19 crisis, Jones and Comfort [10] revealed that that the crisis offered a vision of a more sustainable future with the emphasis being more on environmental and social issues, rather than economic gains, but also on collective, rather than individual, approaches to consumption. However, this vision may pose a major challenge for the industry and for many of its traditional customers. Foroudi, Tabaghdehi, and Marvi [16] investigated customer perceptions of the shock of the coronavirus pandemic, focusing on the influence that it has had on their emotions, as well as how all these emotions could impact the future desire to visit restaurants. The study revealed that trust is the foundation on which the hospitality industry rests, and that the transformation of the restaurant business needs the enhancement of localization strategies, practices, and performance. Based on empirical evidence, Gupta and Sahu [17] proved the positive role of innovative training programmes in the hotel industry in India to support guests and employees during the COVID-19 pandemic, boosting consumer confidence and enhancing their intentions to return. Mason, Narcum, and Mason [18] offer an innovative perspective, from an empirical point of view, on shifts in consumer decision-making behaviors, collecting data for US consumers during pre-pandemic and post-pandemic times and concluding that this sanitary crisis has dramatically altered the consumer needs, shopping behavior, as well as the post-purchase satisfaction level. There is a public hesitance towards eating out post-COVID-19 and the foodservice providers need to redesign their strategies in order to encourage and attract customers [19]. Consumer behaviour will continue to reshape the restaurant industry. In the case of restaurants, the speedy adoption of new digital ordering systems, delivery, and drive-thru innovations will continue to be vital after the pandemic and will require several changes by restaurant operators. However, despite all the industry efforts, returning to pre-pandemic levels in dining-in trends is unlikely. Restaurant formats will look different after the pandemic. With changing consumer behaviour regarding digital ordering, as well as the drive-thru and delivery cultures, it is not surprising that several restaurant chains have introduced new restaurant formats [20]. The future of the hospitality and tourism industries in the COVID-19 era is currently uncertain; therefore, substantial research is required to evaluate how the industry might recover and survive the "new normal" of the COVID-19 world [21]. --- Impact of the COVID-19 Pandemic on Local Restaurant Dining in Romania Social distancing rules have affected economies by reducing the quantity of labour. The most affected work sectors are arts and leisure, hotels, and restaurants, followed by agriculture and business services, activities in which workers rarely use a computer [2]. As previously argued, the food service industry is one of the most susceptible to epidemic disease, because of its high reliance on "human interaction and gatherings" [4] (p. 3813). Prior to the COVID-19 crisis, the hotel and restaurant industry in Romania was blooming; it had a total of 40,000 entirely Romanian-owned companies, with a total turnover of EUR 5 billion, and an estimated 400,000 employees, representing a total of 10% of the total employees from the private sector (around 190,000 direct employees and an estimated 210,000 employees working in related industries, such as product suppliers, manufacturers, and service providers) [22]. It is crucial to acknowledge not only those directly employed by the restaurant industry but also those working in related industries in order to understand how the crisis caused by the COVID-19 pandemic has not only led to the collapse of restaurants but has also had a crucial impact on a series of other stakeholders in the food service industry, such as the state, the customers, the suppliers, the producers, and the banks [23] (p. 816). According to statistics released by the Romanian Hotel and Restaurant Association (HORA), Romania was also severely affected by the pandemic. In 2020, the total turnover of the hotel and restaurant industry registered a total decrease of an astonishing 70% compared to the previous year, with over 40% of local businesses being forced to shut down [22]. Out of these, only 10% are anticipated to reopen after the pandemic, with 30% (approximately 10,000 restaurants) remaining permanently closed due to bankruptcy [24]. Forced by the circumstances, the businesses from the food industry sector have adapted their business model in order to survive the new world order [23]. --- Consumer Perceptions in the Context of the COVID-19 Global Shock Unavoidably, under the influence of a global crisis caused by an infectious disease, the individual's behaviour will adapt to the new context-the world during and after the pandemic, which will give birth to a series of changes in the preferences, attitudes, and cognitive choice-making processes of a population that is increasingly reliant on online ordering and home isolation [4]. The previous section of the Literature Review focused on the impact, from an industry perspective, that the COVID-19 epidemic crisis has had on restaurant demand. At this point, the emphasis is going to shift towards consumers and the key role that the COVID-19 crisis has played in reshaping their consumer sentiments. The question is how does the shock of the pandemic influence customer beliefs, and how could those beliefs have an impact on their anticipated emotions-both positive and negative-and affect their future willingness with regard to dining in restaurants [16]? People's beliefs inform their behavioural intentions [25]. COVID-19 has significantly impacted people's emotions, meaning that it has had undeniable consequences for individual happiness and achievement (Johnson et al. [26]). In their decision-making processes, people's choices are often driven by the anticipation of their feelings regarding the upcoming results [27]. Human emotions have been divided into two distinctive categories, located at polar opposites: positive anticipated emotions, referring to an individual's successful attempt at achieving a goal, and negative anticipated emotions, referring to an individual's inability to achieve their target [28]. During the pandemic, there has been a decline in positive emotion and a considerable rise in negative emotions, such as anxiety and depression: studies reveal that consumers have become worried about their personal health, as well as the health of their families and loved ones, and concerned about whether they will continue to be able to provide for their basic needs, as well as their loss of freedom [29]. These common concerns have manifested themselves in different ways as consumers, influenced by internal (psychological) and external factors, have gradually adopted new consumption patterns. People are no longer interested in holidays and various other leisure activities, including in-restaurant dining, shifting their entire focus onto protecting themselves and their loved ones from illness [30]. The prolonged rise in negative emotion is not only damaging to societies and economies, but also has a damaging impact on an individual's immune system [31]. Jim Samuel et al. [32] addressed an issue of public sentiment, which resulted in increased dread and negative emotion, while Yin et al. [33] proposed a framework for analysing the topic and sentiment dynamics caused by COVID-19 from a large number of Twitter postings. A hybrid technique to finding sentiments on ordinary tweets with polarity calculations was developed in a machine-learning-based sentiment analysis [34]. The polarity score was calculated using three sentiment analysers. Using Twitter data, Ahmed et al. [1] proved that both the users' involvement and their sentiments vary after a particular time. Tardin [35] evaluated the impact of COVID-19 on the Brazilian food service industry, using topic modelling based on online reviews and identified in the pandemic period four of the most relevant topics describing the customer relationship with restaurants: 'delivery', 'employees', 'experience', and 'waiter service'. Using the sentiment analysis, the average value of sentiment of the total sample was 1506, highlighting that the overall sentiment of the consumers towards the restaurants is positive. For the pandemic period, the average is lower than the previous periods, and it becomes clear that the sentiment toward the restaurants reduced in the pandemic period. This aspect has been seen as a reflex of the lack of experience of the restaurants with delivery systems or that the restaurants are no longer capable of delivering the same value to the consumer. The health belief model (HBM) is a theoretical framework used by scholars to "explain and predict health behaviours in public health research" [4]. The model describes how the preventive behaviour of individuals towards illness "can be explained by their risk perceptions and health beliefs" or, in other words, how individuals will act to protect themselves from illness [29] (p. 568). The HBM is positively influenced by three factorsthe perceived susceptibility (people will take measures to protect themselves from a disease if they believe themselves to be vulnerable to a specific condition), the perceived severity (people become wary of a disease "if they believe it would have serious consequences" upon their wellbeing), and the perceived benefits (people become cautious and take preventive measures, under the belief that these would reduce the vulnerability or severity of the disease)-and one negative factor-the "perceived barriers or costs" (factors that prevent individuals from taking the health measures that will protect them) [4] (pp. 3812-3813), [29] (p. 568). The positive factors are visible in the context of consumer behaviour in relation to restaurant dining during the COVID-19 pandemic. Perceived susceptibility captures the individual's perceived risk of becoming infected with the SARS-CoV-2 virus; perceived severity captures the individual's perception of the severity of the COVID-19 infection; and the perceived benefits refer to people's awareness that avoiding restaurant dining will reduce their risk of infection [4] (p. 3813). All of these internal cues to action have also been shaped by external cues, such as risk communication via different mass media channels, health marketing campaigns, and the restrictions enforced by public authorities-full lockdown being one of the most drastic measures taken during the COVID-19 pandemic, aimed at encouraging social-distancing and limiting human interaction and, by doing so, minimising the spread of such a highly contagious virus [4,24]. Ernst and Young have created the "EY Future Consumer Index", a study based on sentimental analysis of individuals in five key markets that showcases how the COVID-19 pandemic has re-shaped consumer behaviours, creating new consumer segments [26]. The study identifies four major consumer groups that have emerged during the pandemic: save and stockpile (35% of consumers participating in the study): those that are not so much concerned about the present, but worried about the well-being of their families and the long-terms effects that the pandemic will have on their lives; cut deep (27%): those individuals, who have been most affected by the pandemic, leading them to be the most pessimistic about the future and causing them to reduce their financial spending across all categories; stay calm, carry on (26%): people that have not been directly affected by the pandemic and, as a result, have not changed their spending habits; last, but not least, hibernate and spend (11%): consumers that, in spite of being most worried about the pandemic, have been in the best position to deal with it, having the financial power to spend more across all different categories [36]. The E and Y [37] report revealed that consumers worldwide are significantly more concerned than they were before the pandemic, arguing that we are witnessing the birth of the "Anxious Consumer." When looking at Chinese consumers, the study pointed out that they were able to return to their "normal life", but in order to facilitate the adaption to this new reality, companies needed to significantly accelerate digital investment in operations and experiences that helped make the consumers feel safe. Based on Yelp online reviews from January-June 2020, Luo and Xu [8] found out that customers elicited a higher level of positive feeling for service in March compared to the previous two months; a potential reason might be that the customers tended to take a restaurant's precarious position into account before evaluating the service quality of a restaurant. Despite the fact that restaurants reopened in May with capacity limitations and social-distancing guidelines, patrons were found to have a generally good view of their eating experiences. In June, sentiments regarding food and place hit an all-time high, indicating that the restaurants were doing an excellent job of maintaining consistency in their performance criteria. One potential recommendation to boost the customers' good feelings about 'location' might be if the restaurateurs provided ample parking to accommodate the customers. Furthermore, having outside seating may make it easier for consumers to locate the restaurant and eat outside in the fresh air. Although the pandemic has had the greatest impact on full-service restaurants, altering societal attitudes and commercial environments will permanently transform all the different restaurant categories. Rather than the establishment of a new business model, the coronavirus may have the most dramatic influence on existing operators' services. Hospitality businesses are expected to make substantial changes to the way they operate in the COVID-19 business environment in order to ensure employee and customer health and safety and enhance the customers' willingness to patronize their businesses [38]. The phantom kitchens would be the food-truck counterpart of COVID-19 (i.e., a new sector pushed by a catastrophe). Hundreds of ghost kitchens have emerged in the last year, either as part of established restaurant chains or as stand-alone businesses, estimated to reach an almost USD 1 trillion commercial potential by 2030 [39]. Operators from the limited, as well as the full-service, spectrum are looking forward to a return to dine-in business. Even models that place a large focus on in-person encounters claim that customer demand for off-premises services will persist long beyond COVID-19. In the restaurant sector, different types of restaurants will soon have less to do with operations and more to do with establishing a specific feeling and brand identity. It has now become mandatory to break down what dining really is, paying particular attention to a guest's needs and their willingness to have whatever their heart desires, whenever they desire it. Many of the safety precautions implemented as a result of the pandemic will continue whenever businesses resume, and the fine-dining sector will maintain, and potentially extend, its focus on outdoor eating [39]. Fine-dining establishments that are still closed, have limited visitor counts, or do not provide delivery/to-go options are under the most strain. Many of these concepts are still seeing 30% to 40% same-store sales drops and will likely only see a modest rebound over the next few months as macro forces begin to replace COVID-19 worries, and customers continue to be wary of dine-in experiences [20]. According to the preliminary findings of longitudinal research undertaken by the editorial team of the Journal of Hospitality Marketing & Management, a big percentage of people (more than 50%) are unwilling to eat in a restaurant right away [40]. The pandemic is far from over, and if the previous year has taught us anything, it is that no amount of preparation can prevent what is yet to come. It is too early to predict how the foodservice industry will develop, how restaurants will define themselves and differentiate themselves from competitors, and what consumer expectations will be. Whatever the case may be, the sector will rebound with a diverse range of cuisines, styles, service methods, and hospitality. The power to adapt is what will maintain the sector's long-term survival. From an industry perspective, restaurants, which are highly reliant on face-to-face human interaction, have been forced to reconsider their marketing strategies and adapt to the "new normal", engaging their customers in virtual communities through social media platforms [41]. KAIAMO, known as the ultimate fine-dining culinary experience, renowned for luxury and sophistication, as well as the artistic plating and exquisite taste, was one of the first restaurants in Bucharest to respond to the changes in consumer behaviour and adapt their business model to the new context, shifting from luxurious in-restaurant dining to dishes that are suitable for take-away and delivery. Soon after the instauration of lockdown in Romania, the restaurant used their Instagram page to announce that the brand is now "Reborn. Refined. Redefined" [42]. Soon after the announcement, the new aesthetic of the brand became visible on Instagram. KAIAMO launched Kaiamo Kanteen, which is, as the word "canteen" suggestively indicates, a less artistic, more realistic version of the original pre-pandemic restaurant, serving dishes from a new menu that focuses less on presentation and more on serving bigger portions at more affordable prices. In other words, there has been a notable shift in their strategy, from unaffordable luxury to "deliverable" dishes, from artistic presentation to comfort food, which is much needed during difficult times. The new culinary aesthetic, as well as the focus on the virtual engagement of customers through online communities, has been visible on the Kaiamo Kanteen Instagram page from the outset of the pandemic: the photos clearly showcase comfort food, instead of artistic dishes, and customer reviews are posted to emphasize the restaurant's ability to empathize with the people during challenging times. Moreover, the restaurant also partnered with the delivery companies Tazz by Emag, Food Panda, and Glovo to offer home-delivery services, which is something that the "pre-pandemic" KAIAMO did not provide to their clients [42]. The Artist, another renowned fine-dining restaurant analysed in the present paper, is another example of a business that has adapted to the new customer behaviour. As a result of the lockdown, the restaurant launched a new menu, specifically designed for home delivery. Moreover, a special home-delivery Christmas menu has also been designed, aiming to bring the luxury experience specific to their restaurant into the safe environment of their customers' homes [43]. The concept was launched under the name of "The Artist @ Home" and, just as in the case of KAIAMO, it has been promoted on the restaurant's Instagram page, once again proving that businesses have understood the new customer behaviour and adapted their "pre-pandemic" business model to serve the "pandemic" needs [43]. KAIAMO and The Artist are proof that luxury restaurants have found ways to manage the crisis caused by the pandemic by adopting their business models to the new rules and regulations, as well as to the more hesitant and worried behaviour of their customers. [8,44] Given that the dark cloud of the pandemic is still ongoing, the recommendation based on the findings of this research is that all fine-dining restaurants should embrace adequate mechanisms in order to survive in the foreseeable future-adaptation of menus for homedelivery and engagement of customers through online communities being two of the recommended survival tools. --- Materials and Methods In order to explore how the customer sentiment evolved over the period, the analysis was based on the online reviews collected for 5 fine-dining restaurants from Bucharest, using the following criteria: the top 100 Restaurants in Bucharest and those ranked $$$$ on TripAdvisor, namely The Artist, Relais & Chateaux Le Bistrot Francais, Casa di David, Kaiamo, and L'Atelier. The sample of reviews extracted from TripAdvisor through webscrapping is formed from a total of 1106 reviews, the earliest review being from 19 May 2010, while the latest dates from 8 March 2021. The timeframes are considered to be the periods from 2010-2019 (for pre-COVID-19 pandemic) and 2020-the present (during the pandemic). The research will take into account a comparative analysis of consumer sentiment for these two periods. The sample of reviews extracted from TripAdvisor through web-scrapping consists of a total of 1106 reviews for five restaurants, the earliest review being from 19 May 2010, while the latest dates from 8 March 2021. The evolution of the total number of reviews per month highlighted a continuous increase in the number of reviews from 2010 until the end of 2018, with, at the end of 2018, a sharp decline for the last period (Figure A1 from Appendix A). The distribution of reviews per restaurant revealed that Casa di David is the oldest luxury restaurant while Kaiamo is the newest restaurant, with reviews starting from 2018 (Table A1 from Appendix A). Analysing the frequency of reviews per restaurant, we can mention that most of the reviews for the whole period were acquired by "The Artist", with almost 400 reviews, followed by "Relais & Chateaux Le Bistrot Francais" and "Casa di David"; at the opposite side we encountered "Kaiamo", a relatively new restaurant. Comparing the consistency of reviews before the pandemic (until 2019) and during the pandemic (2020-March 2021), the research reveals a sharp decline in client reviews as a result of the strict measures imposed by the pandemic-measures that also included closing down restaurants. When customers leave feedback-whether it is to complain or leave a flattering review-there is always an underlying emotion. Having access to the right data at the right time can be a game-changer when it comes to making decisions that boost customer satisfaction and loyalty. Online reviews are still as crucial as ever during COVID-19. During the COVID-19 crisis, timely online reviews might assist potential customers in getting the most up-to-date information about how a restaurant is running. A single poor review might discourage potential customers, making it much more difficult for a food business to survive the COVID-19 crisis. Online reviews have become one of today's most powerful marketing tools, influencing customer behaviour with an astonishing 91% of 18-34-year-old consumers claiming that they trust online reviews as much as personal recommendations [45,46]. Consumer online reviews play a critical role (Del Chiappa et al. [47]; Liu and Park [48]) and have important consequences from a managerial point of view. Customers become "objective voices" (Vermeulen and Seegers [49]), with more than 75% of consumers taking peer reviews into account when planning a holiday (Xie et al. [50]; D'Acunto, Tuan and Dalli [51]). A study by D'Acunto et al. [51], based on a sample of TripAdvisor reviews, analyzed the usefulness of the reviews. From the customers' perspective, online reviews empower them to express their opinion, providing social proof to other potential clients [45]. From an industry perspective, the customers' online reviews can have either a positive or a negative impact on their business. On the one hand, positive reviews have the power to increase customer trust and improve customer experience, whilst on the other hand, negative reviews have the power to reduce the customer base, with 94% of consumers agreeing that a negative review has convinced them to avoid using a specific business [45]. Published literature clearly reveals a division when it comes to the reliability of online customer reviews. On the one hand, the role of customer reviews as rich material is acknowledged, complementing or substituting for existing information sources, whilst on the other hand, they can express the subjective opinion of a limited sample of customers [45]. Online reviews are an effective word-of-mouth marketing strategy in the digital age, providing outside perspectives on products and services. While positive reviews can drive revenue and build a trustworthy reputation, negative reviews, or the absence of reviews, can do the opposite. Understanding the importance of reviews, as well as how to leverage them to boost the business can be a critical way to get ahead in the competitive e-commerce marketplace and be positioned miles ahead of the competition. Therefore, even if there are pros and cons, we decided to use the online reviews from TripAdvisor as the basis for this research paper. A study conducted in partnership with Ipsos MORI polled over 23,000 TripAdvisor users from 12 markets, "across hotel, restaurant and attraction reviews, revealing that more than four out of five (85%) participants report that the reviews they read on TripAdvisor accurately reflect their experience, and 86% agree that TripAdvisor makes them feel more confident in their booking decisions [52]. Moreover, the results of Chua and Banerjee (2013) supported the same conclusion [53]. The research methodology combines elements from two relevant mining analysescontent and sentiment analyses-in order to explore customer sentiment in both the pre-pandemic and the pandemic periods, highlighting which words of emotional and opinion content are important to customers, analysing how consumer emotions evolved in these two periods in relation to restaurant demand in the Romanian fine-dining industry. On the one hand, content analysis is a technique that extracts worthwhile information and represents a solution for the unstructured data. It represents a method of systematic research designed to analyse and infer text, exploring the meanings of different words, themes, or concepts based on word-cloud analysis. Word clouds are visualizations that display words and word frequency to gain an understanding of what consumers specifically like or dislike about a location. On the other hand, the sentiment analysis revealed the emotional tone behind the words used to understand the attitudes, opinions, and emotions and extracted insights from the social data. When human readers approach a text, the understanding of the emotional intent of words is used to infer whether a section of text is positive or negative. The text is considered as a combination of individual words and the sentiment content of the whole text as the sum of the sentiment content of the individual words. Certain emotions are strongly related to specific words. In the sentiment analysis, the proportions of the words that have positive connotations or negative connotations, or are neutral, present interest, and an exploration is made of how many words in a text are also in a predefined list of words associated with a certain sentiment. Within the paper, we have applied sentiment analysis using unsupervised learning, in which the content is characterized by given words or dictionaries; we used the AFINN lexicon dictionary in R software, developed by Finn <unk>rup Nielsen, for a list of words which consists of 2477 coded words and word scores ranging from -5 (very negative) to +5 (very positive). As preliminary steps, AFINN preprocessed the text by removing the punctuation and converted all the words to lower case before analyzing it. One of the drawbacks of using the raw AFINN score is that the longer texts may yield higher values simply because they contain more words. To avoid that issue, we divided the score by the number of words in the text. In order to explore how customer sentiments evolved during the pandemic and to explore potential changes over time, the first step implied the computation of an average sentiment score for both periods, pre-pandemic and pandemic, using the formula: Average sentiment score = Sum(positive) -Sum(negative) Total Words Count (1) Each review was given a sentiment score based on how positive or negative the review was. The final sentiment score ranged between -5 to 5, with the assigning of the following sentiment categories: neutral for a score ranging from -1 to 1, positive for a score ranging from 1 to 2.5, very positive for 2.5 to 5, negative for a score ranging from -1 to -2.5, and very negative from -2.5 to -5. In the second step, based on the values of the average sentiment score, we applied the Welch two-sample t-test in order to highlight reliable change in the average sentiment in both periods, as well as a two-sample, difference-ofproportions t-test to highlight significant differences in terms of positivity and negativity between both periods. The sentiment analysis can experience some difficulty in understanding a few intricacies of human language-polarity, sarcasm, emojis, comparative sentences, or double negatives-referring to many in-between terms, such as "not so bad" or "kind of good", which imply average emotion (mid-polarity), and usually, the sentiment analysis fails to pick up on these emotions. Moreover, sentiment analysis is not able to detect any sarcasm in the comments, it being difficult for the tool to detect the real context behind the sentence, or a double negative, which turns the sentence into a positive. All analyses were developed using libraries tm, tidyte xt, quanteda, tidyverse, corpus, textmineR, tidyr, Rweka, wordcloud2, igraph, ggraph, widyr, stats, ldatuning, stm, readr, readtext, reshape2, ape, and dendextend in R statistical software [54]. --- Empirical Results --- Empirical Results of Content Analysis Customers frequently leave reviews to describe their experiences, which can be key indicators of what specific problems a business is having. Being able to comprehend thousands of reviews through customer sentiment analysis can help identify patterns and behaviors to help improve a restaurant's performance. In order to identify the ways in which customers describe their fine-dining restaurant experience in general as well as in the pre-pandemic and during the pandemic (RQ1), the word-cloud analysis revealed that the most common words characterizing all finedining restaurants from our sample are: "food", "restaurant", "service", "Bucharest", "experience", "menu", "wine", and "excellent" (Figure 1). The sentiment analysis can experience some difficulty in understanding a few intricacies of human language-polarity, sarcasm, emojis, comparative sentences, or double negatives-referring to many in-between terms, such as "not so bad" or "kind of good", which imply average emotion (mid-polarity), and usually, the sentiment analysis fails to pick up on these emotions. Moreover, sentiment analysis is not able to detect any sarcasm in the comments, it being difficult for the tool to detect the real context behind the sentence, or a double negative, which turns the sentence into a positive. All analyses were developed using libraries tm, tidyte xt, quanteda, tidyverse, corpus, textmineR, tidyr, Rweka, wordcloud2, igraph, ggraph, widyr, stats, ldatuning, stm, readr, readtext, reshape2, ape, and dendextend in R statistical software [54]. --- Empirical Results --- Empirical Results of Content Analysis Customers frequently leave reviews to describe their experiences, which can be key indicators of what specific problems a business is having. Being able to comprehend thousands of reviews through customer sentiment analysis can help identify patterns and behaviors to help improve a restaurant's performance. In order to identify the

This research paper aims to analyse how consumer emotions have evolved during the pandemic period in comparison with the pre-pandemic period in relation to restaurant demand in the Romanian fine-dining industry and uses valuable information based on social-media sentiment analysis and content analysis. Focusing on theories of consumer behaviour, the study aims to emphasize how, under the influence of an epidemic crisis caused by an infectious disease, individual behaviour adapts to the "new normal", embracing a series of changes in the preferences, attitudes, and cognitive choice-making processes. The article takes into account a comparative analysis of the consumer emotions between the pre-COVID-19 pandemic period (2010-2019) and the pandemic period (2020-present), based on the online reviews provided by customers for five fine-dining restaurants from Bucharest, the capital city of Romania: The Artist, Relais & Chateaux Le Bistrot Francais, Casa di David, Kaiamo, and L'Atelier. The research was based on two mining analysescontent analysis and sentiment analysis-and explored the emotional intent of words, with the data being collected from TripAdvisor through web-scrapping. The empirical results defined the fine-dining experience during the pandemic as being associated with the quality of the dishes and also with the quality of the service. The overall consumer sentiment in the direction of the restaurants analyzed is positive. The sentiment research found that throughout the epidemic, the consumers' attitudes about restaurants deteriorated. In this sense, consumers seem to be less satisfied with the restaurants' services than before the pandemic. This is another thing that the restaurants had difficulties in when adapting their operations for the pandemic.

so bad" or "kind of good", which imply average emotion (mid-polarity), and usually, the sentiment analysis fails to pick up on these emotions. Moreover, sentiment analysis is not able to detect any sarcasm in the comments, it being difficult for the tool to detect the real context behind the sentence, or a double negative, which turns the sentence into a positive. All analyses were developed using libraries tm, tidyte xt, quanteda, tidyverse, corpus, textmineR, tidyr, Rweka, wordcloud2, igraph, ggraph, widyr, stats, ldatuning, stm, readr, readtext, reshape2, ape, and dendextend in R statistical software [54]. --- Empirical Results --- Empirical Results of Content Analysis Customers frequently leave reviews to describe their experiences, which can be key indicators of what specific problems a business is having. Being able to comprehend thousands of reviews through customer sentiment analysis can help identify patterns and behaviors to help improve a restaurant's performance. In order to identify the ways in which customers describe their fine-dining restaurant experience in general as well as in the pre-pandemic and during the pandemic (RQ1), the word-cloud analysis revealed that the most common words characterizing all fine-dining restaurants from our sample are: "food", "restaurant", "service", "Bucharest", "experience", "menu", "wine", and "excellent" (Figure 1). Analysing the distribution of the most positive and negative words before and during the pandemic, the following can be highlighted (Figure 2): Analysing the distribution of the most positive and negative words before and during the pandemic, the following can be highlighted (Figure 2): --- • the words characterizing the fine-dining restaurants pre-pandemic have been "food", "service", "restaurant", "Bucharest", "experience", "menu", and "wine"; • during the pandemic, it can be observed that even if the restaurants registered a sharp decline in the total number of reviews because of the restrictions and lockdown, the words characterizing the fine-dining restaurants remain almost the same, namely "food", "restaurant", "service", "menu", "dishes", "experience", "Bucharest", "tasting", "staff", or "chef" Therefore, the fine-dining experience in a pandemic is more likely to be associated with the quality of the dishes and also with the quality of service. --- The Empirical Results of Sentiment Analysis Sentiment analysis provides a way to understand the attitudes and opinions expressed in texts. Analysing the whole sample of reviews, it is possible to highlight that the most common negative word is "expensive", followed by "disappointed", "dessert", "bad", "pricey", and "cold" (Figure 3). --- The Empirical Results of Sentiment Analysis Sentiment analysis provides a way to understand the attitudes and opin pressed in texts. Analysing the whole sample of reviews, it is possible to highlight most common negative word is "expensive", followed by "disappointed", "d "bad", "pricey", and "cold" (Figure 3). At the opposite pole, there are the most common positive words for the ent ple, namely "excellent", "nice", "amazing", "recommend", "delicious", "friendly fect", "beautiful", "wonderful", or "fine" (Figure 3). In order to respond to RQ2, a comparative analysis of the word clouds bef during the pandemic in terms of the most common positive and negative words that (Figure 4): --- • if before the pandemic, the most common negative word was "expensive", f by "disappointed", "dessert", "bad"; during the pandemic these were repl "bad", accompanied by "dessert", "rude", and "steep"; • in terms of the most common positive words, before the pandemic these w cellent", "nice", "amazing", and they remained the same during the pa ("nice", "amazing", "wonderful", "excellent"). At the opposite pole, there are the most common positive words for the entire sample, namely "excellent", "nice", "amazing", "recommend", "delicious", "friendly", "perfect", "beautiful", "wonderful", or "fine" (Figure 3). In order to respond to RQ2, a comparative analysis of the word clouds before and during the pandemic in terms of the most common positive and negative words reveals that (Figure 4): --- • if before the pandemic, the most common negative word was "expensive", followed by "disappointed", "dessert", "bad"; during the pandemic these were replaced by "bad", accompanied by "dessert", "rude", and "steep"; • in terms of the most common positive words, before the pandemic these were "excellent", "nice", "amazing", and they remained the same during the pandemic ("nice", "amazing", "wonderful", "excellent"). It is also worth investigating how customer emotions evolved in terms of the most common words contributing to the positive and negative sentiments in the pre-pandemic period vs. the pandemic period (Figure 5) in order to respond to RQ3. By doing so, one can clearly observe that prior to the pandemic the most common positive words were "nice", "excellent", and "amazing" and that they preserve their meaning even during the pandemic. If before the pandemic we identified words such as "expensive", "disappointed", "dessert" as the most commonly used negative words, during the pandemic we found words such as "bad", "terrible", and "steep". From the positive word clouds, it can be observed that the fine-dining experience continues to be defined by the same attributes, with the customers acknowledging that the experience is nice, excellent, and amazing, as those are the largest words in the cloud. However, knowledge gained from the negative word cloud may be more insightful. Before the pandemic it can be observed that some of the larger words include "expensive", "disappointed" and "dessert". If people are using the word "dessert" in their reviews, it may indicate that many customers experienced some issues with the dessert, being disappointed by the quality or finding it too expensive. During the pandemic one can observe an accentuation of the negative feeling, emphasized through the usage of words such as "bad", "terrible", "steep", also revealing issues related to prices. Furthermore, in order to respond to RQ4, a customer sentiment score was built based on how positive or negative the reviews were. In order to explore whether the sentiments of the customers evolved over time, the distribution of the average sentiment score for both periods, as well as the evolution of the sentiment score over time, together with the empirical results of the Welch two-sample t-test were explored. Thus, the modal sentiment (the most frequent sentiment), even if characterized by a small decrease, was slightly positive from an average of 2.03 in the pre-pandemic to almost 1.49 during the pandemic, revealing a decrease in the degree of positivity (Figure A2 from Appendix A). An analysis of the evolution of the customers' sentiments pre-pandemic, as well as during the pandemic, reveals the same decrease in the positivity (Figure 6). The results are in line with the conclusions of Tardin [35], highlighting that the overall sentiment of the consumers towards the restaurants is positive and that the sentiment towards the restaurants reduced in the pandemic period, especially with regard to the lack of experience of the restaurants with delivery systems or to the restaurants not being capable of delivering the same value to the consumer. The empirical results of the Welch two-sample t-test (Table 1) reinforced the same conclusion of a true statistical difference in the average customer sentiment across both periods as the probability of the Welch t-test was smaller than the 5% significance level. demic, reveals the same decrease in the positivity (Figure 6). The results are in line with the conclusions of Tardin [35], highlighting that the overall sentiment of the consumers towards the restaurants is positive and that the sentiment towards the restaurants reduced in the pandemic period, especially with regard to the lack of experience of the restaurants with delivery systems or to the restaurants not being capable of delivering the same value to the consumer. The empirical results of the Welch two-sample t-test (Table 1) reinforced the same conclusion of a true statistical difference in the average customer sentiment across both periods as the probability of the Welch t-test was smaller than the 5% significance level. Whether this difference is due to more positivity towards the pre-pandemic or more negativity towards the pandemic can be determined using the two-sample difference-ofproportions t-tests. (Table 2) Whether this difference is due to more positivity towards the pre-pandemic or more negativity towards the pandemic can be determined using the two-sample difference-ofproportions t-tests. (Table 2). The empirical results of the t-tests for both the differences in positivity and in negativity revealed that statistically significant differences can be observed for both the positivity and the negativity degree, the probabilities (p-values) being smaller than a 1% significant level. Additionally, the difference in the ratios is for positivity 0.157, meaning that the prepandemic period has a 0.157 higher ratio of positive sentiments compared to the pandemic period. For the negativity degree, the pre-pandemic negative ratio is 0.087 higher than for the pandemic; this difference is also statistically highly significant. Therefore, we can conclude that the level of positivity slightly decreased during the pandemic, and so did the level of negativity. A potential explanation for this finding can rely on the fact that positive and negative emotions occur concurrently in a consumption experience [55] as the occurrence of blended emotional experiences has been demonstrated in the consuming of food [56]. Cacioppo et al. [57], Penz and Hogg [58], and Pang et al. [59] have recently proposed that humans can experience various emotions at the same time but of opposing valence. Customers react to extravagant consumption with a combination of emotions, both good and negative, according to Ramanatham and Williams [60]. Manthiou, Hickman, and Klaus [55] (p. 102218) explained very well the co-occurrence of positive and negative emotions, stating that "a couple has dinner at a fine-dining restaurant; although the atmosphere and decoration are splendid, they have to wait long for their main course. The wine is excellent, but the meat is overdone. The salad ingredients are fresh, but the salad dressing is inadequate. The dessert is delicious and delivered on time, but the two servers are not equally reliable and helpful. Will these customers depart feeling satisfied, dissatisfied, or both"? The ambiance was the attribute with the largest improvement in guest sentiment, in conjunction with the change to the growth of off-premises sales as dine-in activity fell alongside the increase in COVID-19 cases. Traditionally, off-premises food sentiment has always been lower than dine-in. All these could be reasons for the decrease in the positivity between both periods. During the pandemic, safety remains the customers' primary concern and meeting or exceeding expectations for safety can contribute to positive feedback online. --- Conclusions and Implications COVID-19 has been the restaurant industry's greatest challenge to date, as well as a severe public-health issue. Never before have such a large number of restaurants been forced to close, some of which will never reopen. Consumer demand will most certainly not grow instantly when restrictions are eased, according to early signs from China and other nations where the epidemic appears to be under control. Restaurants that prepare ahead and are ready to adapt and develop their business model in accordance with the "new normal" will be in a better position to restore pre-crisis sales levels. The study examined online reviews on Romanian fine-dining restaurants, discovering the underlying aspects related to the consumer experiences in these restaurants through content analysis and examining how the customer sentiment evolved over time, based on the results of sentiment analysis. The crisis caused by the COVID-19 outbreak is still an ongoing issue and the foreseeable future is uncertain. Whilst the data clearly indicate that the global economy is entering a recovery stage, scholars and industry experts claim that at the moment the "global trends" and "consumer trends" are nothing but predictions as no one can know for sure how industries, as well as consumers, will re-adapt to "normality" after witnessing the huge pandemic shock. As expected, during the pandemic there has been a sharp decline in client reviews, as a result of the strict measures imposed by the pandemic-measures that also included closing down restaurants. However, Casa di David has been the most constant in receiving reviews, while The Artist has been the restaurant with the highest number of reviews. In describing the customers' fine-dining restaurant experience, it can be mentioned that this experience is mostly related to words such as "food", "restaurant", "service", "Bucharest", "experience", "menu", "wine", and "excellent". If we explore this experience in a comparative way, before and during the pandemic, it can be highlighted that the fine-dining experience before the pandemic is characterised by "food", "service", "restaurant", "Bucharest", "experience", "menu", and "wine", while in the pandemic the words characterizing the fine-dining restaurants remain almost the same, namely "food", "restaurant", "service", "menu", "dishes", "experience", "Bucharest", "tasting", "staff", or "chef". Therefore, the fine-dining experience in a pandemic is more likely to be associated with the quality of the dishes as well as with the quality of service. In terms of negative sentiments, the fine-dining experience could be characterised before the pandemic by "expensive", "disappointed", "dessert", "bad", while during the pandemic these have been replaced by "bad", accompanied by "dessert", "rude", and "steep". In terms of positivity, the most common positive words in the pre-pandemic were "excellent", "nice", and "amazing", and they remained almost the same during the pandemic ("nice", "amazing", "wonderful", and "excellent"). The overall consumer sentiment in the direction of the restaurants analyzed is positive. The COVID-19 pandemic severely affected the Romanian restaurant business, as shown by the radical decrease in the number of reviews. Even while the restaurants continued to operate through delivery systems, it appears that this was insufficient to maintain their customers. The sentiment research found that throughout the epidemic the consumers' attitudes about restaurants deteriorated. In this sense, the consumers seem to be less satisfied with the restaurant's services than before the pandemic. This is another thing that the restaurants had difficulties in when adapting their operations for the pandemic (Tardin [35]). Managers in the food service business will benefit from the findings of this study. The results of the content and sentiment analysis clearly highlighted the most relevant aspects that consumers express interest in with regard to restaurants, showing paths for managers to facilitate and improve their restaurant's operations. The most common element mentioned by the customers is related to the price, which is associated with the lowest associated customer sentiment. Nonetheless, the COVID-19 context changed the socioeconomic environment in which restaurants operate. Consumers are evaluating the delivery aspect, and managers should keep up with this new demand. This research is one of the few types of research utilizing text mining together with sentiment analysis to examine consumer sentiments about the Romanian fine-dining industry, especially with regard to understanding the impacts of the COVID-19 pandemic on this sector. Despite this paper's contributions, one key question remains open: is there going to be a fundamental change in customer behavior towards the food service business after the epidemic ends and everything returns to 'normal'? --- Limitations and Future Directions of Research A first limitation of this research is the small number of reviews of the pandemic period. Collecting more data from this period can significantly change the results. The review website selection is also a limitation of this paper. The selection of only TripAdvisor could induce a platform bias. Therefore, future research can investigate the impact of COVID-19 on other platforms as well. Another limitation was the number of restaurants chosen. Last, but not least, the present article focused on discussing consumer sentiment changes during the pandemic (2020-present), in comparison to the pre-pandemic period (2010-2019). However, with the world economy now entering a recovery stage, consumer behaviour will be subject to periodic change, depending on the progress of the pandemic as well as the ability of individuals to adapt to the ever-changing "new normality". As a result, year-by-year analysis is recommended for the next five years in order to gain a deeper understanding of what the post-pandemic consumer looks like. Future directions of research include developing the analysis further to include restaurants of different price points, located in different cities in Romania, in order to gain a deeper contextual understanding of the current consumer sentiment in relation to the restaurant industry at a national level. --- Data Availability Statement: The main source for the data supporting the reported results can be found on the TripAdvisor website for each restaurant considered within the analysis. --- Funding: This research received no external funding. Institutional Review Board Statement: Ethical review and approval were waived for this study, due to REASON of using online reviews (secondary data). Informed Consent Statement: Individuals' consent was waived due to REASON of using online reviews (secondary data). --- Conflicts of Interest: The authors declare no conflict of interest.

This research paper aims to analyse how consumer emotions have evolved during the pandemic period in comparison with the pre-pandemic period in relation to restaurant demand in the Romanian fine-dining industry and uses valuable information based on social-media sentiment analysis and content analysis. Focusing on theories of consumer behaviour, the study aims to emphasize how, under the influence of an epidemic crisis caused by an infectious disease, individual behaviour adapts to the "new normal", embracing a series of changes in the preferences, attitudes, and cognitive choice-making processes. The article takes into account a comparative analysis of the consumer emotions between the pre-COVID-19 pandemic period (2010-2019) and the pandemic period (2020-present), based on the online reviews provided by customers for five fine-dining restaurants from Bucharest, the capital city of Romania: The Artist, Relais & Chateaux Le Bistrot Francais, Casa di David, Kaiamo, and L'Atelier. The research was based on two mining analysescontent analysis and sentiment analysis-and explored the emotional intent of words, with the data being collected from TripAdvisor through web-scrapping. The empirical results defined the fine-dining experience during the pandemic as being associated with the quality of the dishes and also with the quality of the service. The overall consumer sentiment in the direction of the restaurants analyzed is positive. The sentiment research found that throughout the epidemic, the consumers' attitudes about restaurants deteriorated. In this sense, consumers seem to be less satisfied with the restaurants' services than before the pandemic. This is another thing that the restaurants had difficulties in when adapting their operations for the pandemic.

Background Internationally, there is a growing call to include informal carer costs and benefits in economic evaluations to evaluate the broader impacts of health and social care services [1,2]. In this context, informal carers provide care beyond normal expectations within a pre-existing relationship (family member, relation, friend, neighbour) such as assistance with personal care, household activities or practical support, and generally do not receive payment for the care they provide [3,4]. Over the last 15 years, three carer-specific preferencebased instruments have been developed to measure outcomes for economic evaluations, the Adult Social Care Outcomes Toolkit for Carers (ASCOT-Carer) [5], Care-related Quality of Life instrument (CarerQol) [6] and Carer Experience Scale (CES) [7]. Whilst the number of economic evaluations including carer effects has grown in recent years, few have been conducted in the Australian setting [2,[8][9][10]. In Australia, there are about 2.8 million informal carers with over a third acting as the primary carer, i.e. the person who provides the majority of care [11]. In 2020, informal carers provided an estimated 2.2 billion hours of care, on average 786 h per year or 15 h per week (35.2 h for primary carers) [11,12]. If services were purchased from formal care providers, the replacement costs would be $77.9 billion, almost 40% of the total spending on health in Australia in the same year [11,13]. Given societies implicit reliance on carers' willingness to fulfil this role and the economic consequences if this situation should adversely change, it is crucial that carer costs and effects are considered in economic evaluations [11]. In turn, this would also better inform healthcare decision-makers on actual societal costs, increasing the chance that welfare optimising decisions are made. Two recent studies have investigated the relative construct and discriminative validity, test-retest reliability and responsiveness of the European-developed ASCOT-Carer, CarerQol and CES in a survey of Australian carers [14,15]. Studies in England have also compared ASCOT-Carer, CES, CarerQol and EuroQol-5 Dimension-5 level (EQ5D-5L) [14,16]. These studies indicate that the instruments tap into different constructs of carer-related QoL and caring experiences, reflecting the original purpose of the instruments and suggests the ASCOT-Carer, CES and CarerQol cannot be used interchangeably [16,17]. The ASCOT-Carer was developed to measure social care-related QoL and support of carers in the setting of policy and formal support interventions [5,18]. Whereas the CarerQol measures the impact of informal care on carers' QoL, combining the burden of caring and valuation of their well-being (happiness) in the context of an evaluation in health care [6]. The CES captures the caring experience rather than carer's QoL per se [19]. Validation has been investigated for the constructs in each of the instruments: the ASCOT-Carer with carers in England [5]; the CES with carers of older people in England [19] and the CarerQol with carers in eight European countries [6] [20]. Each of the instruments has preference weights, allowing the calculation of a summary score which reflects carers' preferences for difference aspects of carer-related QoL [7,[21][22][23]. An Australian population was only included in one of the CarerQoL validation studies [22], with the general adult population using hypothetical carer scenarios. Content validation of instruments assesses relevance, comprehensiveness, and comprehensibility of the questions and the overall instrument; ensuring interpretation is as intended, all aspects important to the specific population are included and that the instrument's constructs/domains measured as proposed [24]. Cross-cultural adaptations of instruments are important to capture differences in linguistics, colloquialisms, context and culture, even if translation is not required [25]. Qualitative research is ideally placed to validate the face value and content of instruments by exploring these social and cultural variables that may differ between informal carers in different countries [26]. Given the differences in health and social care support between countries [27][28][29], it is important to evaluate the applicability of these instruments in an Australian setting [25]. Consequently, the aims of this analysis were to identify aspects of carer QoL important to Australian informal carers and explore how well the ASCOT-Carer, CarerQol and CES constructs capture these aspects in the Australian context. --- Methods --- Study design An online questionnaire was administered to a sample of informal carers in Australia between June and September 2018. Participants were recruited through Carers Victoria, a state-wide not-for-profit organisation supporting carers to improve their wellbeing, health, resilience and capacity [30]. This analysis was part of a larger study that investigated the psychometric properties of the carer-related preferencebased instruments [15] and exploratory factor analysis [17]. --- Setting and participants Adults (<unk> 18 years), Australian residents who self-identified as primary, informal carers and able to read the English written study questionnaire were invited to complete a web-based questionnaire. An email invitation was sent to all Carers Victoria registered informal carers who had previously consented to contact for research purposes. The online questionnaire link was also advertised in the Voice: Carers Victoria ebulletin which is distributed to all informal carers registered with the organisation and in researcher's social media posts. Informed consent was collected before starting the questionnaire and a $10 gift voucher was offered to all participants as an acknowledgement of their contributions. --- Instruments --- Adult Social Care Outcomes Toolkit for Carers (ASCOT-Carer) There are seven domains in the ASCOT-Carer, a preferencebased instrument of carers' social care-related quality of life including; control over daily life, occupation (doing things you value and enjoy), social participation and involvement, personal safety, self-care, time and space to be yourself and feeling supported and encouraged [5,21]. The content of the ASCOT-Carer was developed from a literature review, focus groups and interviews with carers and care managers [31], and semi-structured interviews with carers [18,32]. --- Care-related Quality of Life (CarerQol) CarerQol contains two sections; the CarerQol-Visual Analogue Scale (VAS), which measures wellbeing and the CarerQol-7D which measures subjective burden [6,33]. There are seven dimensions in the latter; fulfilment, support, relational problems, mental health problems, problems combining daily activities with care, financial problems and physical health problems. The content of the CarerQol was developed from a survey of carers in the Netherlands and a review of eight popular burden measures [6,23]. --- Carer Experience Scale (CES) There are six dimensions in the CES, a preference-based instrument of caring experiences; activities outside caring, support from family and friends (social support), assistance from organizations and the government (institutional support), fulfilment from caring, control over the caring and getting on with the care recipient [7]. The content of the CES was developed from semi-structured interviews with carers in the UK and a meta-ethnography of qualitative studies on caring. --- Questionnaire The questionnaire was developed online using Qualtrics®. It was piloted to refine wording and comprehension of the activities with a convenience sample of Deakin University Health Economics and Faculty of Health staff members and informal carers (n = 21). Study participants had the option to complete the questionnaire over multiple sessions and all questions were voluntary. Figure 1 shows the sequence of instruments, randomisations and the question wording. Firstly, contextual questions about personal characteristics, caring situation and characteristics of the care recipient were asked. Followed by the randomised ASCOT-Carer, CarerQol and CES instruments to minimize potential ordering effects [15]. To achieve the aims of this paper, the first thirty participants were asked two separate free text questions about the greatest positive and greatest negative effect on their quality of life as a carer in the past week. To minimise survey burden, all participants were randomised to one additional activity: (i) instrument plus free Fig. 1 Questionnaire Flow. *Positive/Negative questions: Please describe the things that have had the greatest positive effect on your quality of life as a carer in the past week?; Please describe the things that have had the greatest negative effect on your quality of life as a carer in the past week? (Response free text). ** [Instrument] followed by question: Thinking about the statements included in this completed questionnaire, please describe any other things that affect your quality of life as a carer that were NOT mentioned. (response free text). *** [Instrument] and following each question: How important is this to your quality of life as a carer? (Response 5-point Likert scale, Very Important to Not Important at all). <unk>CRA = Caregiver Reaction Assessment; Social Isolation = Three item UCLA Loneliness Scale; AQoL-8D = Assessment of Quality of Life-8 Dimensions text question on whether any aspects of carer QoL were not mentioned in the instrument (referred to in this paper as '[Instrument] missing domains'); (ii) instrument plus a 5-point Likert scale of the importance of each domain within the instrument s (referred to in this paper as '[Instrument] domain importance'); or (iii) complete the instrument only. Finally, all respondents were invited to rank the top five (out of the 14 total domains across the three instruments) most important aspects of caring from most to least relevant (referred to in this paper as 'ranking activity'). Other instruments were also completed by all participants in the final part of the questionnaire (Fig. 1, last column), for use in the broader project [15,17]. --- Data analysis Analyses were conducted in STATA Statistical Software: Release 17 [34], QSR NVivo software<unk> (version 11) [35] and Microsoft Excel [36]. Instruments were scored using preference-based weighting of the respective instruments. To ensure consistency, the UK value sets were used for all three instruments. Results were reported in a previous paper [15]. --- Population Descriptive statistics were generated for the demographics, caring situation and care recipient characteristics. --- Relevance and comprehensiveness --- Relevance Responses to the Positive/Negative questions and [Instrument] missing domains questions, were imported into QSR NVivo software version 11 [35] for analysis to identify specific factors influencing respondents' care-related QoL. A four-stage content analysis procedure guided the coding of the open-ended responses: decontextualization, recontextualization, categorisation, and compilation [37]. Conventional content analysis was used in the development of the coding framework and themes were inductive, data driven, and with researchers avoiding using preconceived categories [37,38]. Responses for each instrument and question were coded separately. Coder one (JB) spent time noting any preliminary ideas, codes and themes before building categories and with these, a coding structure. Where responses contained more than one theme/sub-theme they were coded into each. Coder two (AU) reviewed the coding structure and relevant-free text responses. The two coders discussed differing views on codes and discrepancies were settled by authors LE and NM before finalising the coding structure. Proportions of the [Instrument] domain importance question rated as unimportant, neutral or important to respondents' CrQoL were calculated and compared with the Chi-Square test. In addition, for the ranking activity, descriptive analyses were used to determine number of times domains were ranked number one and also the number of times chosen in the top five ranks. --- Comprehensiveness The coding structure of the analysis for questions [Instrument] missing domains was reviewed by authors JB & LE to identify any key aspects of carer QoL that participants identified as missing from the instruments. --- Results --- Population Online Appendix 1 shows the sociodemographic characteristics and caring situation of the informal carers and care recipient characteristics of the total questionnaire sample and for each of the subgroups that received and completed the additional questions that contributed to this analysis. The total sample size was 500 participants with a mean age of 52 and the mean age of the care recipient was 45. Majority were female (79%) and had completed undergraduate and postgraduate education (46%). Just over half of the participants were employed (51%), while the other half were retired or engaging in housework duties including caring (46%). Just over half of participants were sole carers and a quarter providing care to multiple recipients. Most participants shared a household with the care recipient (81%) and had been caring for > 24 months (74%). Relationships with care recipients included children (32%), parents (32%), partners (25%) and other family members or friends (10%) with their most common medical condition being chronic diseases or disabilities (44%) followed by mental health problems (33%). --- Relevance The open-ended responses (N = 115) resulted in 244 units of data and identified eight themes: Behaviour-Mood of care recipient; Caring Responsibilities; Finances; Health; Own Life; Perceptions of carers; Relationship with care recipient; and Support. Although the qualitative positive/negative and Instrument [missing domain] question responses were coded separately, strong similarities between the coding frameworks were very apparent with the same themes present for each question, differences only in sub-themes. Each question appeared to elicit responses that indicated what matters to carers, so the results have been presented together in Table 1 which summarises the themes and sub-themes presented in Online Appendix 2. There was a wide diversity in the openended responses. The number of responses ranged N = 24-33 for each open-ended question (the two Positive/Negative questions and three [Instrument] missing domain questions). Likert responses to the importance of each domain were categorised into: Not Important (Not Important; Slightly Important), Neutral and Important (Important; Very Important) and aggregated within each instrument. The total instrument importance (Table 2) shows that carers in our sample judged all three instruments as important (<unk> 80%) and <unk> 7% not important. There was no statistically significant difference between the ratings across the three instruments (Chi-squared 3.489, degrees of freedom 4, p = 0.479). Similarly, the importance of individual domains within each instrument indicated that all the domains for each instrument were important aspects of CrQoL (Fig. 2). Domains most frequently considered important were the CES domain of getting on with the care recipient (n = 54, 93%) and activities outside of caring (n = 50, 86%) and CarerQol's Mental Health Problems (n = 62, 89%). Domains deemed least important were for ASCOT Carer's personal safety (n = 11, 11%), and control over daily life (n = 9, 9%) and CarerQol's financial problems (n = 7, 10%). Table 3 summarises the instrument domain ranking activity. The top five ranked domains were the same using either analysis method (number of times ranked as position one or number of times ranked in the top 5), only the third and fourth positions (mental health and self-care) were reversed. --- Comprehensiveness After completing the instrument, very few participants reported that the carer-related instruments covered all aspects affecting care-related QoL (ASCOT-Carer (n = 2), CarerQol (n = 3) and CES (n = 2). Content analysis of this question for each instrument (Online Appendix 2) shows that all themes in the coding framework were identified as missing by our sample of carers. This is a particularly interesting result, as many of the themes are constructs measured by the instruments. --- Discussion This analysis identified aspects of carer QoL important to Australian informal carers and explored how applicable the constructs of ASCOT-Carer, CarerQol and CES were to this population. Behaviour-mood of care recipient, Caring Responsibilities, Finances, Health, Own Life, Perceptions of carers, Relationship with care recipient and Support were identified as aspects of caring that affect carer QoL in Australian. Comparing domains most importance in our sample with previous studies that developed preference weights and tariffs for the instruments showed mixed results. Occupation and control over daily life for the ASCOT-Carer where the most preferred among English carers [21], whereas our Australian sample found self-care and time and space to be yourself to be the most important. Suggesting that, with further investigation, Australian preference weights for the ASCOT-Carer may be in need of development. In our sample the importance of domains in the CarerQol (most important, mental health; least important, combining care and other activities) and CES (most important, getting on with care recipient and activities outside of caring; least important, control over caring) were in line with instrument tariffs [7]. CarerQol tariffs, developed for Australia, indicated mental health and combining caregiving with other activities as the most and least preferred [22,23]. Whilst in a sample from United Kingdom, the CES found activities outside of caring and getting on with the care recipient as most preferred and control over caring the least [7]. Almost all participants reported aspects of carer QoL not captured by the carer-related instruments. Many of these aspects that were perceived as not covered by the instruments, could have been included in the domains. This was also the case in a study looking at patient, self-reported, QoL aspects not captured by EQ-5D-5L [39]. As caring experiences are subjective and responsibilities and challenges vary greatly between carers, they may have felt the domain did not completely encompass their experience of carer QoL. This reflects how some aspects of QoL, which can be important to individuals, cannot necessarily be translated into a question for a QoL instrument. Particularly when required to be applicable to a broad range of carers (e.g. caring for partner, child, parent), align with the construct of the instrument, and also fit with other considerations (e.g., timeframe). Alternatively, respondents may have interpreted the questions differently or focused only on certain portions of the question (e.g. heading, examples or explanatory text). A content analysis of the end-of-life patientreported outcome measure showed that interpretation of questions is related to individual circumstances, where in financial matters varying themes of money, investments, funeral arrangements and wills emerged [40]. This could also explain why similar domains across instruments were treated differently. The domain of support is present in all three instruments, however, respondents reported different types of support were missing in each instrument (i.e. formal and informal support for the carer and/or care recipient). Comparable results were also found in more detailed studies of the exploratory factor analysis using this same dataset, where only a moderate correlation was for my son so that I can attend a meeting' (Carer_10) 'There is no support for carers that I know of. In the early days after my husband's stroke there was some support, but everything is capped at a certain amount of visits. And after that you don't meet the criteria anymore so you're on your own. I have not had a proper break from my caring duties for more than 2 years. I care for my husband 24/7. No one works those hours in paid employment. I love my husband and wouldn't have it any other way but to have support and know help is only a phone call away would be great.' (Carer_108, ASCOT-Carer) 'When we look like caregivers who look like they have all the bases covered, you receive less support' (Carer_778, CarerQol) 'this program has now been defunded so I won't have the opportunity to meet with the other carers monthly/bimonthly and the support worker has lost his job so I will no longer have him as a support/source of assistance with my caring responsibilities' (Carer_2) 'Frustration with Government Departments and Caring Organisations who because they are underfunded, over committed or don't/can't do their jobs effectively. I am constantly told they can or will help then simply do not carry out their promises and or don't follow up as promised' (Carer_179, CES) *Government financial assistance coded to 'Support' theme 1 3 found between CarerQol and CES support items and also for relational problems [17]. Content and/or face validation of the three instruments has not been performed with Australian carers, so detailed information of how each instrument's questions are interpreted and understood by this population is not known. The broader project performed a content comparison of the three instruments showing they each perform well in measuring their relevant domains with Australian carers [15,17]. However, the qualitative component of this study suggests that some differences in question interpretation may exist. The majority of sub-themes (over 50%) related specifically to the care recipient. However, only two domains include aspects of carer QoL that are influenced by the care recipient (CarerQol's relational problems and CES's getting on with the care recipient). Consideration was given to a similar domain in the development of the ASCOT-Carer, however, it was omitted because it did not fit with the construct of the instrument (social care-related QoL/impact of care services on carer QoL) [18]. This relationship between the care recipient and carer has previously been proposed as an advantage of the CES in capturing broader aspects of caring [7,15,16]. Given the possible interdependence of care recipient and carer QoL [41][42][43], instruments capturing both could be included in economic evaluations of carer and patient interventions to fully capture the effects of an intervention [44]. However, consideration also needs to be given to the type of evaluation being performed, the perspective taken and the possibility of double counting which could overestimate the benefits of an intervention [2,45]. Keeping these factors and participant burden in mind, an appropriate combination of instruments may be used in measuring carer QoL in studies focused on informal carers. Some of the missing aspects of carer QoL in the instruments, as reported by our sample, are intentionally not covered by the instruments as they each have been developed with different intentions and measure different constructs of CrQoL. The CarerQol was developed and intended to measure the impact/burden of caregiving on QoL and so, appropriately, does not include any themes specifically about the care recipient [6]. Similarly, the CES missing themes of finance and health are reasonably missing as the instrument's purpose is to measure the experience of caregiving. Health problems are not directly measured by CES, as qualitative research indicated that this was linked to other attributes included in the instrument [19]. And the ASCOT-Carer does not measure finance and health, as the instrument was developed as a measure of social care and support services on carer QoL. Financial hardship due to caring and health were considered in the early development of ASCOT-Carer [31], however, it was excluded as it was outside the scope of the instrument's purpose. Although health was not considered as a separate domain in the ASCOT-Carer it is captured by the lowest QoL (high-level needs) response option for each item and indicates that the carer has high-level needs that, if unmet over time, put the carer at risk of poor physical and/ or mental health. --- Strengths and limitations Content analyses have the potential to be influenced by researchers' experiences and preconceptions. Coding framework along with transcripts were reviewed by a second researcher independently and collaboratively discussed, reducing the impact of coder bias. Quotes and sub-themes were classified to themes based on consensus and the coding framework is presented to demonstrate how the data were categorised so that other researchers can consider how their interpretation aligns with the researchers' views. The recruitment of study participants and completion of the questionnaires occurred towards the end of the roll-out of a new government support system, National Disability Insurance Scheme (NDIS), which replaced the existing system of 1 3 disability support. The NDIS caused changes to administrative processes in receiving financial and formal support and may have been particularly front of mind for carers having to navigate this new system. Further, specific issues may have arisen directly due to changes in the systems. The cohort included a greater number of sole carers (55 vs 33%), a greater proportion of female carers (79% vs 57%) and a higher percentage of carers providing more than 30 h of care per week (55% vs 45%) compared to the Australian population of primary carers [11,12]. Income and employment, relationship to care recipients and sharing household with care recipient were similar to the Australian population of carers [12]. Uniquely, participants included carers of people with multiple health conditions. The study included informal carers in the Australian setting only and therefore results may have limited generalisability to other settings. Study participants were self-selected via newsletter advertising and, to reduce burden, randomly allocated to one additional activity described in this paper (excluding the ranking activity that was completed by all). This resulted in a different sub-group completing each (Positive/Negative question, the three [Instrument] missing domain questions, the three [Instrument] domain importance questions). Subgroup characteristic differences (Online Appendix 1) include the Positive/Negative question participants containing only females, being less employed, less likely to be married and more likely to have been caring for > 24 months and > 4 h of care per week and the CES missing domain participants being less educated and caring for more recipients with mental health problems. The open-ended components of this study were embedded in a larger quantitative study [15,17], so there was no opportunity to apply qualitative techniques such as faceto-face interviews or focus groups to explore responses in more depth, clarify the views' expressed or to measure comprehensibility of the instrument questions. Cognitive interviewing, in checking respondent's understanding, mentally processing and response to materials would help with understanding these differences. Cognitive interviewing evidence is present for the ASCOT-Carer in England [18,31,32] and during its translation into German [46,47], Japanese [48] and Finnish [49], as well as for the CES in England [19]. Australian evidence would provide a much greater understanding of the comprehensiveness and comprehensibility of the three instruments in this setting. As questionnaires were completed anonymously, researchers did not have an avenue to discuss findings with participants and receive feedback on the themes and analysis. Importance of domains in this study, were assessed by a sample of informal carers. Whereas carers also participated in the development of preference weights for the CES using a best-worse scale (BWS) valuation exercise [7], the ASCOT-Carer and CarerQol used the general population imagining a hypothetical state of being an informal carer using BWS exercise and a discrete choice experiment respectively [21,22]. The inconsistencies with our sample may be due to the differences in sample (i.e., carers or general population) and also between stated different methodology using preferences (hypothetical situation) or revealed preferences (actual or current situation) or due to different analyses. Three different approaches were used to investigate how well the instruments capture aspects of CrQoL important to Australian informal carers, strengthening conclusions concerning coverage. Some potential cultural/ethnic difference appear in the free text responses. However, with a very small number of participants born outside of Australia and/or speaking a language other than English (Online Appendix 1) these differences were not explored in this paper. A real opportunity exists for future research in this area of an Australian population. --- Conclusions Open-ended responses and quantitative data collected from a sample of Australian informal carers, suggest there are multiple aspects of caring that impact carer QoL. Consideration should be given to measurement of care recipient health and well-being and spillover effects affecting carer QoL, with thought to the risk of double counting. The ASCOT-Carer, CarerQol and CES appear to be relevant for an Australian informal carer population and include most of the aspects of quality of life important to them. The interpretation of questions may differ in Australian informal carers which requires confirmation with a content and/or face validity assessment. The findings support previous research that the selection of an instrument should take into account the aim, purpose and constructs of the instrument. --- Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1007/ s11136-023-03459-1. Author contributions All authors except JB contributed to the study conception and design. Material preparation, data collection and analysis were performed by JB, LE and NM. Double coding was performed by AU. The first draft of the manuscript was written by JB with input from LE and NM. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. --- Declarations Competing interests The authors have no relevant financial or nonfinancial interests to disclose. --- Ethical approval The questionnaire and methodology for this study was approved by the Deakin University Faculty of Health, Human Ethics Advisory Group, Burwood, Australia (reference number HEAG-H 91_2018) and through the Carers Victoria's Carer Participation in Research process. Consent to participate Informed consent was obtained from all participants before commencing the questionnaire. Consent to publish Patients read and accepted, informed consent regarding reporting of research findings in a de-identified format. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Purpose Identify aspects of quality of life (QoL) important to Australian informal carers and explore how well the Adult Social Care Outcomes Toolkit for Carers, Care-related Quality of Life instrument and Carer Experience Scale capture these aspects in the Australian context. Methods Online questionnaires were completed by Australian informal carers. Socio-demographics, open-ended questions: positive/negative aspects of caring and QoL aspects missing from the instruments, and ranking of the instrument domains was used to explore the content of the instruments. Instruments were scored using preference-weighted value sets (reported in another paper). Content analysis was used to analyse the open-ended responses. Chi-squared test looked at differences in domain importance. Descriptive analyses summarised all other information. Results Eight themes were identified: Behaviour-mood of the care recipient, Caring responsibilities, Finances, Health, Own life, Perception of carers, Relationship with care recipient and Support. Many aspects of carer QoL mentioned as missing in the instruments appeared covered by the domains, of which all were reported as important. The highest ranked domain was relationship with the care recipient. The influence of the care recipient specific support, behaviour/mood and health on carer QoL appear absent in all instruments. Conclusion The content of the three instruments appears relevant in an Australian setting. The influence of care recipient's health and well-being on carer QoL should be considered, along with spillover effects. A content and/or face validity analysis is required to confirm differences in item interpretation in Australian informal carers.Informal care • Outcome measurement • Carer-related quality of life • Preference-based measures * Jessica Bucholc

Introduction In recent years, the development of technology and networks has been rapid, and this has a direct impact on the evolution of society. Increasing exposure to new digital technologies has brought about significant changes in many social sectors, and the concept of digitization lies at the heart of all (Castells, 2009). Digitization and development of information and communication technologies are restructuring various aspects of social life such as work, business administration, communication, entertainment, education, and the economy (Parviainen et al., 2022). Even in the field of education, a lot has changed, as learning can be provided without the physical presence of the teachers and/or students. The evolution of society through the development of new technologies and globalization processes is pushing the economy to undergo radical changes. The digital dimension of society and economy has been a huge advantage in COVID-19 lockdown, as without digital technologies, the economic hardships would have been even greater. The concept of digitization is dominating and, in combination with the development of ICTs, changing the way we work, communicate, entertain ourselves, and the way of life in general. In March 2010, the European Union (EU) set forth the Europe 2020 Strategy for Smart, Sustainable and Inclusive Growth in order to lift Europe out of the economic crisis and prepare its Member States' economies for the challenges of the next decade (Communication COM(2010) 2020; Bánhidi, Dobos & Nemeslaki, 2020). To achieve this, specific actions were proposed. One of them is the Digital Agenda for Europe (DAE), which aims to define the key role that ICT will play in Europe's development. The DAE aims to strengthen European economy by providing sustainable economic and social benefits from a digital single market (DSM) (Communication COM/2010/0245 final). To that end, a communication and a recommendation from the Commission were released in order to propose ways to update and improve the EU networks bringing the basic broadband to all Europeans by 2013 (Communication COM(2010) 472 final) and very fast internet by 2020 (Recommendation 2010/572/EU). Part of the DAE's targets were e-government and the re-use of public sector information, which are addressed in an action plan. It was set to exploit the ICTs to support the transition to a new generation of open, flexible, and collaborative e-government services and re-use of public data at local, regional, national, and European levels (Communication COM (2010) 743 final; Directive 2013/37/EU). The DSM is a prerequisite for EU's transformation towards the digital revolution of the 21st century (Troiti<unk>o, 2022). It is one of the EU policy priorities and provides businesses, particularly entrepreneurs, with new opportunities to scale up across Europe. It can expand markets and create opportunities for new start-ups, fostering better services at better prices and creating new sources of employment where advanced digitally skilled employees will be valuable (Communication COM/2015/0192 final). To strengthen human capital, a Communication was released in 2016, focusing on improving the quality and relevant skills formation by enhancing both basic and higher, more complex digital skills of citizens. More opportunities will be available when everyone is equipped with a broad range of skills (Communication COM/2016/0381 final). As digital competence has become key for citizens to participate in today's social, economic and civic life, the Digital Education Action Plan, adopted in 2020, will reset education and training for the digital age (Communication COM/2020/624 final). In 2021, the EU presented its vision for a successful digital transformation of Europe by 2030, putting forward four cardinal points: digitally skilled citizens and highly skilled digital professionals, secure, performant and sustainable digital infrastructures, digital transformation of businesses and digitalization of public services (Communication COM/2021/118 final). Digital competitiveness in our new digital era is challenging and each EU member is required to invest in digital economy and society. The European Commission has gradually developed several indicators to assess the digital performance of its Member States. One of the most popular indices is the Digital Economy and Society Index (DESI). The significant innovation of the DESI is that it represents an overall index that tracks the evolution of EU Member States' (Giannone & Santaniello, 2019). The DESI is a composite index consisting of five dimensions corresponding to five EU principal policy areas: Connectivity, Human Capital, Use of Internet (European Commission, 2020a). It summarizes relevant indicators related to the EU digital performance and monitors the development of its Member States in the field of digital competitiveness. Since 2014, the DESI has been a key analytical tool for the annual measurement of EU countries' progress towards the digital economy and society (OECD, 2018). As the DESI is the indicator that the EU uses to monitor the digital progress of its EU Member States in five dimensions, it undoubtedly plays a significant role in the decision-making process regarding the digital transformation of the EU. Many researchers used the DESI to study the digital progress of an individual country compared with the rest of EU members or with the average digital performance in the EU (Stoica & Bogoslov, 2017;Nagy, 2017;Moroz, 2017;<unk>esnausk<unk>, 2019;Russo, 2020), pointing out digital policies that should be formed to enhance the digital progress at the country level. Moreover, the DESI has been also studied at the EU level (Bogoslov & Stoica, 2019;Raki<unk>evi<unk>, Raki<unk>evi<unk> & Poledica, 2019;Bánhidi, Dobos & Nemeslaki, 2020;Borowiecki et al., 2021;Sevgi, 2021;Liu, 2022;Andrei et al., 2023) and the effect of several socio-economic factors on the DESI is also examined (Scupola, 2019;Stavytskyy, Kharlamova & Stoica, 2019;Marino & Pariso, 2021;Andrei et al., 2023). The current work extends the previous literature in the field by studying the evolution in time of each DESI dimension during the whole period of 2014-2019, using Repeated Measures ANOVA to highlight statistically significant inter-year differences on each dimension. To that end, the five dimensions of the DESI are studied over a six-year period (2014-2019), using the corresponding DESI reports (DESI 2015 -DESI 2020). In the current work, an attempt is made to evaluate the performance of the EU countries on the digital economy and society with respect to the implemented EU digital policies. Moreover, the digital convergence among EU Member States in terms of similarity of their performance in the five dimensions of the DESI, by grouping them according to their optimal number of clusters, is examined. Another novelty of our study is that the optimal number of clusters is determined on the basis <unk>f 30 indicators (Charrad et al., 2014). In addition, the countries' interchanges are also a point of interest. Finally, the impact of some important socioeconomic factors on the DESI was also studied using linear mixed effect models, allowing the incorporation of fixed and random effects into the models since the existing literature is rather limited (Scupola, 2019;Stavytskyy, Kharlamova & Stoica, 2019;Marino & Pariso, 2021;Andrei et al., 2023). In the current study, more factors were in the European Union: <unk> Socioeconomic Perspective TalTech Journal of European Studies Tallinn University of Technology (ISSN 2674-4619), Vol. 13,No. 2 (38) selected to cover a wider area in the social and economic field, to represent the whole of society in a more comprehensive manner. More specifically, the effect of the Gross Domestic Product (GDP) per capita, the average number of weekly working hours, the unemployment rate, the consumption of renewable sources, the lack of corruption, the public expenditures on education and the percentage of gross domestic product spent on basic and applied research and experimental development on the overall DESI was studied. The results reveal the strengths and weaknesses in the countries' digital transformation that are of a great importance for policymakers at the national and European level to plan long-term policies. The rest of the paper is organized as follows. The relevant literature on the DESI and its structure are presented in Section 2 and Section 3, respectively. Methodology and methods are discussed in Section 4. The main results are presented in Section 5 and the discussion and implications are presented in Section 6. Recommendations are provided in Section 7 and some concluding remarks in Section 8. --- Literature review The DESI has gained a lot of research attention in recent years as a global index capable of evaluating states' digital performances related to society and economy, and it has caught the attention of policymakers. The rest of this section presents the most recent related literature. Bogoslov and Stoica (2019) studied the digital evolution of EU Member States during 2014-2019. They also examined the position of the EU at a global level, comparing the digital progress of the EU with non-EU countries. Raki<unk>evi<unk> et al. (2019) studied the overall index in the DESI 2018 report, using logical clustering to measure the proximity among countries forming five (5) groups. Their aim was to identify directions of digital policies to enhance each Member States' digital competitiveness, based on their proximity with the rest of the EU countries. Bánhidi et al. (2020) studied the correlations and partial correlations of the five dimensions in the DESI 2018 report. Moreover, they also grouped EU Member States in five (5) clusters, using hierarchical clustering. According to these results, adequate knowledge of the partial correlation between the dimensions of the DESI will help policymakers in decision-making process to improve competitiveness at local, regional, and European level. Sevgi (2021) used data from the DESI 2020 report and, by applying the k-means algorithm, grouped the EU Member States into four (4) groups, according to their performance in the five dimensions of the DESI. They also examined whether there were similarities between these groups and the classification of the welfare state regimes by Esping-Andersen (1990) in South Europe (Kammer, Niehues & Peichl, 2012), Central and Eastern Europe (Lauzadyte-Tutliene, Balezentis & Goculenko, 2018). According to the welfare of EU Member States, there are six types of welfare regimes: the Social Democratic, Conservative, Liberal, Southern, Central, and Eastern Europe. Each type of welfare state is characterized by the political situation (eligible government or not), social situation (labor market, gender differences in the labor market, demographic situation, population education, living conditions), and financial situation (real GDP) of a country, also government programs provided to citizens. According to their results, comparing the DESI clustering and the first welfare regime, countries applying the social democratic welfare regime (Finland, Sweden, Denmark, the Netherlands, and Norway) are better prepared for the digitization of the economy and future digital competition compared to other EU countries, as they have higher values in all dimensions of the DESI index. They are followed by the countries of liberal welfare regimes (United Kingdom and Ireland), with a small difference from the first category, and by the countries of conservative welfare regimes (Germany, Luxembourg, Belgium, Austria, and France). Finally, the rest of the countries (Italy, Spain, Greece, Portugal, Croatia, Poland, Slovakia, Slovenia, Bulgaria, Estonia, Latvia, Lithuania, and Romania) belong in the last three categories and have the lowest values in all dimensions of the DESI and are more likely to face problems in the future in terms of technological transformation and digital competition with other EU countries. Borowiecki et al. (2021) present the development of digital economy and society in the light of the digital convergence of EU markets using data from DESI 2015 and 2020 reports. They applied dynamic time series models to determine the dynamics of changes in DESI values and classified countries in terms of the similarity of DESI value, forming five clusters. Their results highlight the strengths as well as weaknesses in the Member States' digitization efforts to shape proper domestic governance and strengthen competitiveness in various areas of the economy. Liu (2022) intended to analyze the digital policy performance of EU countries based on the DESI and determined the similarities and differences between their digital performance using data from DESI 2017 report. A graphical analysis co-plot technique was used to categorize EU Member States based on the five DESI variables, creating four groups. Based on this classification, policymakers will design appropriate policies to boost digitization and economic development at national level. Scupola (2019) tried to investigate factors contributing to the digital transformation regarding digital public administration in Denmark. Using the longitudinal case study methodology, they concluded that low unemployment and well-educated population, revenues from exports and public administration which is supported by central government's policies, and political initiatives seem to be the key factors of success. Stavytskyy, Kharlamova and Stoica (2019) studied the effect of per capita expenditures based on purchasing power parity and unemployment on each DESI dimension. A panel regression model was constructed for 28 EU countries, using DESI reports from 2014-2018. Their analysis confirmed that a more prosperous society leads to more advanced digital services. They concluded that there is a certain positive effect of economic development and a negative effect of unemployment on all DESI dimensions apart from Integration of Digital Technology. Marino and Pariso (2021) studied the effect of four socio-economic factors on the DESI and on its dimensions in 2016-2018, performing a statistical regression analysis. They concluded that Social Progress, Corruption Perception, Global Innovation, and Doing Business indices positively affect DESI and most of its dimensions. Andrei et al. (2023) investigated the convergence of digitalization in the EU in terms of the DESI in each Member State during 2015-2020, taking into account indicators representing each country's economy. In their study, econometric models were used to estimate the <unk> and <unk> convergence of the DESI. They also examined the effect of the gross value added and the education index during the convergence process. They concluded that, during the studied period, digital convergence was observed and that gross value added had a positive effect on the DESI, while education appeared not to be statistically significant. The adding value of the current paper is the study of the digital performance of European countries and the evolution of digital competitiveness over a six-year period, from 2014 to 2019, using Repeated Measures ANOVA to highlight statistically significant inter-year differences on them. Moreover, EU Member States are grouped into the optimal number of classes based on 30 indicators applying the k-means clustering method. The stability of the grouping per year is also studied. Finally, several factors, such as income, the average number of weekly working hours, unemployment rate, consumption of renewable sources, lack of corruption, public expenditures on education, and percentage of gross domestic product spent on basic and applied research and experimental development that may affect the DESI were studied. To that end, linear mixed effect models were utilized, which allow for fixed and random effects-an effective tool when there is dependence in the data. --- The structure of the DESI The DESI has a three-level structure, with five dimensions. Each dimension consists of subdimensions and these, in turn, of indicators. In total, the 5 dimensions consist of 12 sub-dimensions and these in total of 37 indicators. The Connectivity dimension examines both the demand and supply of fixed and mobile broadband. As to fixed broadband, it assesses the overall and ultrafast broadband (at least 100 Mbps), the availability of fast broadband access (next generation access (NGA), providing at least 30 Mbps), and the availability of very high-capacity fixed networks (VHCNs) while taking into account retail prices. Mobile broadband includes 4G coverage, mobile broadband adoption (3G and 4G) and 5G connectivity readiness indicator. Digital connectivity is considered a social right in the EU (European Commission, 2020b). The Human Capital concerns the digital skills of EU citizens, ranging from basic skills that enable people to participate in activities involving the use of digital devices and the internet consuming digital goods and services, to advanced skills such as those of ICT specialists (ICT service managers, professionals, and technicians) and ICT graduates that empower the workforce to develop new digital goods and services (European Commission, 2020d). The Use of Internet Services dimension measures the percentage of people who use internet services, and which of the available services they use. Activities include online content consumption (e.g., entertainment, such as music, movies, TV, or games, downloading multimedia-rich information, or engaging in online social interaction) using modern communication activities (e.g., participating in video calls) and activities transactions such as online shopping and banking (European Commission, 2020f). The Integration of Digital Technology dimension measures the digitization of businesses and e-commerce. Digital technologies allow businesses to be more competitive, improving their services and products and expanding their markets. To calculate the five dimensions of the DESI as well as the total DESI, the initial 37 indicators are standardized using the min-max method. The DESI is a weighted sum of the five aforementioned dimensions of the initial 37 indicators. Some dimensions and individual indicators are more important than others with respect to the priorities of EU digital policy and, therefore, have a greater effect on the calculation of the final DESI (European Commission, 2020a). The weights used in dimension level are Connectivity: 25%, Human Capital: 25%, Use of Internet Services: 15%, Integration of Digital Technology: 20%, and Digital Public Services: 15% (European Commission, 2020a). --- Methodology and methods The data used in the current study has been collected from the annual DESI 2015 -DESI 2020 EU reports. The annual DESI reports cover the period from 1 February of the previous year to 31 January of the following one. To determine the socioeconomic factors affecting the DESI for the period 2014-2019, data from both the European Statistical Office (Eurostat) and the World Open Data Bank (Word Data Bank) were used. Real Gross Domestic Product per capita (GDP) were retrieved from the World Open Data Bank and the consumption of renewable sources (RNRGconsumption), the lack of corruption (CORRlack), the percentage of GDP spent on basic and applied research and experimental development (GERD), the average weekly working hours (WORKINGhours), the percentage of GDP for preprimary to tertiary education (EDUexpend), and the unemployment rate (UNEMPLOYMENT) were retrieved from Eurostat. The data refer to 28 countries that belonged to the EU in 2019, observed over a period of six (6) years. The statistical analysis was conducted using R (version 4.2.0) and all the tests were implemented at 5% significance level. Initially, the main descriptive statistics of each of the five dimensions of the DESI are computed for the period 2014-2019. To study the evolution of the five dimensions of the DESI, first, the normality assumption of the data was tested by the Shapiro-Wilk normality test (Shapiro & Wilk, 1965). When the normality assumption was met, as in our case, the ANOVA test in a repeated measures design (Repeated Measures ANOVA) was utilized (Salkind, 2010). When the null hypothesis of Repeated Measures ANOVA was rejected, a post-hoc analysis was performed to explore differences between all possible group pairs using Tukey's test (Tukey, 1977). To group the EU Member States with respect to their performance in the five dimensions of the DESI, the k-means clustering method was implemented, as it is simple, fast, and efficient. Moreover, k-means leads to tighter clusters than the hierarchical clustering and it is easy to interpret the clustering results. As a measure of distance, the Euclidean distance was used. To determine the optimal number of clusters, NbClust package in R was used which "provides 30 indicators for determining the number of clusters and proposes to user the best clustering scheme from the different results obtained by varying all combinations of number of clusters, distance measures, and clustering methods" (Charrad et al., 2014). The performance in the five DESI dimensions was compared between the two groups. Finally, the effect of some main economic and social variables on the DESI was also examined. To that end, linear mixed effect models (LMM) were constructed, which are parametric linear models for repeated-measures data (West, Welch & Galecki, 2014). LMMs are an extension of simple linear models containing fixed and random effects in some analysis and are used when there is dependence in the data, such as repeated measurements on each subject over time, as in our case. The R's lme4 package (Bates et al., 2015) was used to construct an LMM with the DESI as a dependent variable and the aforementioned socioeconomic factors as fixed effects. A model with fixed effect of years and a randomized intercept per country was initially constructed. The fitted model would be expressed as where The two-dimensional fixed-effect vector <unk> consists of the mean intercept, <unk> 1, and the common slope <unk> 2. The one-dimensional random-effects vectors, b i, i = 1.....28 describe a shift in the intercept for each country. Due to there being a common slope, these shifts are preserved for all values of year. A model with random effects for both the intercept and the slope was also constructed (Model 2). The above two models were compared using the function anova in R (Chambers & Hastie, 1992), and Model 2 was selected as optimal based on the Akaike criterion (Burnham & Anderson, 2004). In this model, independent variables were added (Model 3). The basic assumptions of the model, such as independence of data points, absence of measurement errors, independence of random effects from explanatory variables, linearity, and homoscedasticity of residuals (Fig. 1), are satisfied. The normality of the residuals and random effects has not been tested since it has been shown that mixed models are very resistant to the violation of the normality assumption of the residuals and the random effects (Schielzeth et al., 2020). The normality test of random effects is very difficult to implement as it requires additional assumptions for the model without which determining their distribution is impossible (Alonso, Litière & Laenen, 2010). --- Results --- Evolution of the DESI <unk>he main descriptive statistics of each of the five dimensions of the DESI are presented in Table 1, where a steady upward trend through the dimensions of Connectivity, Use of Internet Services, Integration of Digital Technology, and Digital Public Services is observed over time. It is noticed that the median and average performance of the above dimensions are steadily increasing every year. Moreover, it is evident that the median of the Human Capital dimension remains stable in the first years and then an increase from 2018 is observed. Regarding the standard deviation, it remains stable in the Connectivity dimension, while decreasing steadily in the Digital Public Services dimension. For the rest of the dimensions, the standard deviation is stable at the beginning, whereas towards the end of the studied period, an increase is observed. Since the five dimensions of the DESI fulfil the normality assumption, Repeated Measures ANOVA was performed for each dimension and a statistically significant difference in the average performance is observed in all dimensions of the DESI during the studied period. The Tukey's test, which was implemented next, determined between which years these statistically significant differences were observed. As mentioned above, Connectivity evolves very quickly over time, which was also confirmed by the corresponding Tukey tests (Table 2), as in all DESI reports from 2015 to 2020, there is a statistically significant difference to the average performance of EU Member States in this dimension, from year to year. No statistically significant difference in the average performance of EU Member States was observed in the Human Capital dimension between the DESI 2015, 2016, and 2017 reports. A statistically significant difference appeared for the first time between the DESI 2017 -DESI 2018 reports, and between the DESI 2019 -DESI 2020. A delay in the appearance of a statistically significant difference between the average performance of the Use of Internet Services dimension in the EU Member States was also observed, as it appeared from the DESI 2018 report onwards. A statistically significant difference was observed in the average level of the Integration of Digital Technology dimension almost every second year. An exception is the year 2015, where there is a statistically significant difference from the very next year. A statistically significant difference was observed in the average performance of EU Member States in the Digital Public Services dimension in all DESI reports from 2017 to 2018, while it was first observed in the DESI reports from 2015 to 2016. --- Clustering The results of the NbClust package suggest that the optimal number of clusters is two (2) for each year. Figure 2 shows, by year, the map of Europe where the countries with high performance in the DESI are highlighted in darker grey (cluster1) and the countries with low performance in light grey (cluster2). Countries such as Belgium, Denmark, Estonia, Finland, Ireland, Bulgaria, Cyprus, Greece, Czechia, Croatia, Hungary, Italy, Luxemburg, Poland, Romania, Slovenia, and Slovakia remain consistently at a low level of performance. Countries such as Spain, Lithuania, Portugal, and Austria maintain high levels of performance for long periods of time, whereas in 2020, they transpire with a low performance level. Finally, there are countries such as France, Latvia, and Germany that do not maintain a steady course over the years and alternate between low-and high-performance group of the DESI. Additionally, a statistically significant difference is observed in the mean performance on four of the five dimensions of the DESI between the two groups each year. The boxplots of the course of the five DESI dimensions in the two groups are depicted in Figure 3. The Connectivity dimension in both clusters follows the same course with very small differences in the median values. The Human Capital seems to have no development in the countries of the second cluster during the studied period, contrary to the countries of the first cluster which show a significant development, especially in the last years. The Use of Internet Services seems to have a faster development in the countries of the first cluster which is also the case of the Integration of Digital Technology. However, the difference that exists between the two clusters regarding the evolution of these dimensions is clear. For the Digital Public Services dimension, the countries of the second cluster are quite behind compared to the countries of the first cluster. --- Impact of the socioeconomic factors on the DESI Using LMMs, two models were formed, one with fixed effect of years and a randomized intercept per country and the other with random effects for both the intercept and the slope. The two models were compared using AIC and Model 2 was selected as the optimal (Model 1: AIC 674.84, Model 2: AIC 669.12). In this model, independent variables were added (Model 3). In model 3, the variables, RNRGconsumption (p-value = 0.58), UNEMPLOYMENT (p-value = 0.76) and CORRlack (p-value = 0.7) are not statistically significant and were sequentially removed from the model that leads to the final one (Model 4). It is observed that all factors have a statistically significant impact on the DESI in Model 4 (Table 3). The average DESI increases by 1.72 per year for constant values of the variables GDP per capita, average working hours, education expenditure, and R&D expenditure, and by 1.99 for each unit of GDP per capita growth for constant values for the rest of parameters. It also increases by 0.81 for each unit of increase in the percentage of education expenditure and by 1.99 for each unit of increase in the percentage of R&D expenditure, while it decreases by 1.57 for each unit of increase of the average working hours provided that the values of the remaining explanatory variables of the model remain constant. --- Discussion and implications Considering the evolution in the five DESI dimensions for EU-28, the digital evolution of EU Member States is monitored under EU digital policies, established since 2010 by the DAE (Communication COM/2010/0245 final) and Europe Strategy 2020 (Communication COM(2010) 2020). Our findings show a clear increasing trend of the five dimensions of the DESI over time. However, some dimensions are evolving faster than others. The Connectivity dimension is the basis that will give the necessary space for the other dimensions to unfold, as it concerns the technological infrastructure but also the supply and demand of technological benefits. It evolves very quickly over time. EU digital policies seem to fulfil their goal in terms of infrastructure, bringing the basic broadband to all Europeans (Communication COM(2010) 472 final; Recommendation 2010/572/EU). Although the technological infrastructure appears to be present, the Human Capital and the Use of Internet Services dimensions, related to digital skills and using internet services, respectively, are delayed. It is likely that the population may need time to get acquainted with the new technology, to understand how they could use it and get familiar with internet services such as digital communications and transactions, this is also the case for Borowiecki et al. (2021) and Grigorescu et al. (2021). The Integration of Digital Technology dimension shows signs of improvement every second year. It is commonly recognized that a reasonable timeframe is required for the e-commerce and the digital transformation of businesses to take place. To formulate and implement a digital transformation strategy is a time-consuming task (Chanias, Myers & Hess, 2019). Finally, due to the priorities set by each country, whether it focuses on the modernization of its national portals or on an open data policy, the development trajectory of the Digital Public Services dimension is disrupted. The organizational and the technical support in e-government systems is rather challenging (Al-Sebie & Irani, 2005). Furthermore, the EU Member States were grouped in two clusters based on their performance in the five dimensions of the DESI, contrary to Raki<unk>evi<unk>, Raki<unk>evi<unk> and Poledica (2019), Bánhidi, Dobos and Nemeslaki (2020), Borowiecki et al. (2021) and Sevgi (2021) and Liu (2022), where they formed five and four clusters, respectively. A cluster consisting of countries with high performance (cluster1) and a cluster of countries with low performance (cluster 2) in all dimensions of the DESI were formed. Considering that the North-European countries are more digitally developed compared to the rest of the EU countries, these results are not surprising (Castells & Himanen, 2002;Sevgi, 2021;Andrei et al., 2023). Over the six-year period of our study, some countries alternate from one group to another. This is due to the different pace at which each country is evolving in relation to the rest. This is also confirmed by other researchers (Kyriakidou, Michalakelis & Sphicopoulos, 2011;Borowiecki et al., 2021). In addition, our results indicate that the grouping of EU-28 is not affected by the Connectivity dimension even though it evolves quickly and steadily over the studied period. It is revealed that the Connectivity dimension in both clusters follows the same trajectory. However, all the other dimensions are evolving faster in high performance countries than in low performance. Countries in cluster 1 have higher values in all dimensions of the DESI, as they are better prepared for the digitization of the economy and future digital competition compared to countries in cluster 2, as Sevgi (2021) has also indicated. Finally, it is observed that the division into two clusters also matches the economic profile of the countries. According to the proposed model, the DESI is indeed significantly positively affected by GDP, as Scupola (2019), Stavytskyy, Kharlamova andStoica (2019), andAndrei et al. (2023) have also concluded. The most economically developed countries seem to invest more in the digital transformation of both the economy and society (Castells & Himanen, 2002;Sevgi, 2021;Andrei et al., 2023). Moreover, the DESI is positively affected by educational expenditures, as Scupola (2019) also suggests in contrast to Marino and Pariso (2021). Countries which invest more in education from pre-primary to tertiary level seem to have better digital performance, which is also the case for countries which spend a greater percentage of GDP on basic and applied research and experimental development. A decrease in the DESI values is observed as the average number of usual weekly working hours in main job increase. While the working hours are on the rise, population have less time to get acquainted with the new technology either for training or to be involved in other leisure activities. Regarding the other studied factors-the unemployment rate, the consumption of renewable sources and the lack of corruption-it is found that they do not have a statistically significant impact on the DESI in the current study. --- Recommendations As the DESI is the core index, used by the EU to access the digital progress of its Member States, policymakers may rely on it to form appropriate digital policies at international, regional, and national level to strengthen the countries' digital transformation. As indicated, the Connectivity dimension evolves very rapidly over the sixyear period, while both Human Capital and Use of Internet Services keep following the evolution of the Connectivity dimension, even with a two-year delay. The Integration of Digital Technology and Digital Public Services dimensions do not show a rapid evolution. More effective strategies should be implemented to strengthen the more slowly evolving dimensions. Moreover, policymakers at global and domestic level should focus on the countries in cluster 2 to encourage their digital transformation. Long-term policies and supportive initiatives need to be implemented to foster digital transition, considering each country's characteristics. Our research has identified factors that affect the DESI. This knowledge is used by policymakers to develop appropriate policies to promote digital transformation. Policies towards the enhancement of citizens' basic and advanced digital skills need to be implemented to close the digital division across EU Member States (Janssen, Charalabidis & Zuiderwijk, 2012;Grigorescu et al., 2021). The related Communication released in 2016 does not seem to fulfil its target (Communication COM/2016/0381 final). Policymakers have to carefully design digitally based strategies in order to provide appropriately focused learning programs at basic and advanced digital skills. In addition to faceto-face education, distance education could be improved. Also, the gender and the age of citizens should be taken into account so that educational training will be inclusive and accessible to everyone. The Digital Education Action Plan was adopted in 2020 to enhance Human Capital and Use of Internet Services as these dimensions fell behind during the DESI evolution (Communication COM/2020/624 final). In the new digital era, acquiring the relevant knowledge will make citizens active participants, who are able to use internet services such as e-banking, video calling, online shopping, or social interaction. In addition, policies should be designed to foster business and public services digitalization. The digitization of business processes for products and services has helped companies to be more flexible in competitiveness. However, there is still a need to enhance the use of digital technologies such as digital platforms (Apple's iOS and Google's Android platform) and infrastructures (cloud computing, data analytics, online communities, social media, 3D printing, digital marketspaces) in business digital transformation (Nambisan, 2017). Digital Public Services is also a dimension left behind on the evolution path. Strategies for digital governance, such as transparency and openness of government processes to the public, should be developed. DSM also prompts the need for international cooperation with other governments to better serve citizens and businesses across borders (OECD, 2014). In addition, attention must be given to open data policies regarding reusable and available data in national data portals (Janssen, Charalabidis & Zu

The rapid development of information and communication technologies (ICT) in recent years has brought about signifi cant changes in many social sectors such as communication, economy, entertainment, and others. To defi ne the key role that ICT plays in its development course, the European Union (EU) has developed a composite indicator, the Digital Economy and Society Index (DESI), to assess the digital policy performance of its Member States. In the current work, an attempt is made to evaluate the performance of the EU countries on the digital economy and society with respect to implemented EU digital policies by studying the fi ve dimensions of the DESI for the years 2014-2019, using the corresponding DESI reports (DESI 2015 -DESI 2020). Moreover, the digital convergence among EU Member States, in terms of similarity of their performance in the fi ve dimensions of the DESI by grouping them according to the optimal number of clusters, is also examined. Since the optimal number of clusters is two, EU Member States are classifi ed in two groups, one of high and one of low performance in the fi ve dimensions of the DESI. The evolution of each member country and the possible transitions from one group to another during the years 2014-2019 is also a point of interest. The grouping of EU Member States into the two clusters showed that socioeconomic factors may aff ect the overall DESI. Linear mixed eff ect models confi rm the positive eff ect of Gross Domestic Product per capita, the public expenditure for education

COM/2020/624 final). In the new digital era, acquiring the relevant knowledge will make citizens active participants, who are able to use internet services such as e-banking, video calling, online shopping, or social interaction. In addition, policies should be designed to foster business and public services digitalization. The digitization of business processes for products and services has helped companies to be more flexible in competitiveness. However, there is still a need to enhance the use of digital technologies such as digital platforms (Apple's iOS and Google's Android platform) and infrastructures (cloud computing, data analytics, online communities, social media, 3D printing, digital marketspaces) in business digital transformation (Nambisan, 2017). Digital Public Services is also a dimension left behind on the evolution path. Strategies for digital governance, such as transparency and openness of government processes to the public, should be developed. DSM also prompts the need for international cooperation with other governments to better serve citizens and businesses across borders (OECD, 2014). In addition, attention must be given to open data policies regarding reusable and available data in national data portals (Janssen, Charalabidis & Zuiderwijk, 2012). --- Conclusion The evolution of the society through the development of new technologies and globalization processes is pushing the economy into radical changes. Digital competitiveness in the new digital era is challenging and investing in the digital economy and society is required of every EU member. Since 2010, the EU has put in place strategies, action plans, communications, and recommendations related to digital advances, with the aim to support EU Member States on their path in digitalization and digital convergence. The EU introduced the DESI in 2014 as a key analytical tool for the annual measuring of EU countries' digital progress. The DESI is used to evaluate and shape Europe's digital policy towards the digital convergence of all EU countries, as it is presenting the digital performance of each Member State. In the current work, the evolution of the five dimensions of the DESI for the period 2014--2019, using the respective reports DESI 2015 to DESI 2020 was studied. Our study was conducted in the light of how policymakers at national or global level may be influenced by the DESI in their decisionmaking processes. The results revealed the current state of EU Member States' digital performance and highlighted the strengths and weaknesses in the digital progress. Also, the EU Member States were grouped based on their performance in the five dimensions of the DESI and factors that affect the overall DESI were examined. Two groups were created, one including countries with high and one with low performance on the DESI. Finally, socioeconomic factors such as GDP, EDUexpend, and GERD have a positive effect on the DESI, while the WORKINGhours have a negative effect. Our research may help policymakers worldwide to develop innovative strategies in facing digital economy challenges. For a more detailed study of the DESI, other socioeconomic factors could be incorporated in the proposed model and more complex models with interactions could be studied. As the digital economy is a global concern, a review of digital performance of the EU on a global level could be also studied. A study of the International Index of Digital Economy and Society (I-DESI) would help in that direction. Modern society and economy are changing as ICT has developed rapidly in recent years. When taking into consideration the new needs for work, leisure, communication, education, and online shopping, arising from the COVID-19 pandemic, it is commonly recognized how urgent it is for the EU to monitor, understand, and follow the digital evolution. --- Sonia Malefaki is an associate professor at the Department of Mechanical Engineering & Aeronautics of the University of Patras. Her research interests are mainly in computational statistics, simulation methods, Monte Carlo and Markov chain Monte Carlo methods. She also works on Bayesian statistics, Markov and semi-Markov processes, reliability, maintenance longitudinal data, and clustering techniques. She is the author of 37 papers in international journals and has edited volumes and more than 15 papers in international conference proceedings with referees.

The rapid development of information and communication technologies (ICT) in recent years has brought about signifi cant changes in many social sectors such as communication, economy, entertainment, and others. To defi ne the key role that ICT plays in its development course, the European Union (EU) has developed a composite indicator, the Digital Economy and Society Index (DESI), to assess the digital policy performance of its Member States. In the current work, an attempt is made to evaluate the performance of the EU countries on the digital economy and society with respect to implemented EU digital policies by studying the fi ve dimensions of the DESI for the years 2014-2019, using the corresponding DESI reports (DESI 2015 -DESI 2020). Moreover, the digital convergence among EU Member States, in terms of similarity of their performance in the fi ve dimensions of the DESI by grouping them according to the optimal number of clusters, is also examined. Since the optimal number of clusters is two, EU Member States are classifi ed in two groups, one of high and one of low performance in the fi ve dimensions of the DESI. The evolution of each member country and the possible transitions from one group to another during the years 2014-2019 is also a point of interest. The grouping of EU Member States into the two clusters showed that socioeconomic factors may aff ect the overall DESI. Linear mixed eff ect models confi rm the positive eff ect of Gross Domestic Product per capita, the public expenditure for education

The ROI Conundrum This does not actually mean such investments have a poor return; it's just that the principal returns often do not accrue to the investor. Health organization investments typically have positive spillover effects in areas beyond immediate health savings, such as economic improvement, better school graduation rates, and better long-term community health. But this form of value-added is not captured as revenue or savings by the health sector investor-a classic "wrong pockets" problem. In the same way, improvements in housing conditions financed by a housing authority, or a reduction in violence-caused injuries from new policing programs, can generate large spillover health care savings, yet the financial value of those health benefits do not accrue to the budget bottom line of housing authorities or police departments. The result in each case is usually suboptimal levels of investment in the general well-being of a community. The key to achieving an optimal level of health system funding of social services, in which public funding is supplemented with health sector funding aimed at improving health, is to organize these investments through multisector partnerships in a community, with each partner strategically investing in community needs that generate spillover benefits to all partners. If designed and implemented well, this partnership approach encourages investments by all community partners to create a strong social return on investment (SROI), with benefits and savings shared by each partner as it advances its goal. Of course, reaching that result requires the investing partners to agree on their shares. That can easily lead to difficult negotiations and heightened distrust. Nichols and Taylor are among those testing procedures in communities to create a positive climate, in their case through an innovative bidding process run by a neutral broker and designed to align each partner's investment commitment with the benefits it will receive from joint action. --- Government Encouragement There are several ways to encourage appropriate health sector investments in social services. --- Taking a Back Seat in Partnerships Effective partnerships require trust, understanding, and the effective use of complementary skills. Indeed, a review of health-human services partnerships suggests that the quality of the relationship is key. This is not easy to achieve between health and social service organizations when communities often doubt the motives and commitment of large health institutions that, in turn, can be skeptical of the expertise of community organizations. Health systems need to adapt to this reality for their investments to be most beneficial. In 2017, Kaiser Permanente considered supporting a public health and economic development project in Baltimore, Maryland; however, the organization was not then well known in the mid-Atlantic area. So it first partnered with Bon Secours hospital, a very small but highly respected institution in the community; essentially, Bon Secours "credentialled" Kaiser within the community. CommonSpirit Health has been a pioneer in developing trusted partnerships with social service organizations in communities. CommonSpirit recognizes that there is a critically important difference between functioning as a catalyst for action and being in the driver's seat. It has partnered with the Pathways Community HUB Institute (PCHI) in 6 communities. In the PCHI model, a neutral hub operated by a local entity links a network of community organizations, health systems, and community health workers to help coordinate care and address health-related social needs. CommonSpirit, along with competing health systems in each area, funds a "community bank" that helps fund the hub's operating costs and covers otherwise nonreimbursable service costs. It is an intriguing example of how a community partnership can function and how competing health systems can collaborate in a "co-opetition" model for their common benefit. 3 We have come to appreciate that achieving healthier communities requires a larger focus on social factors contributing to ill health. For that to happen, we need to design better SROI techniques and ways in which community savings are distributed. Health systems need to deploy investment resources in ways that prevent illness as well as treat it. Their levels and type of investment should also reflect the broad economic value of prevention and better health. And they must appreciate the importance of sharing control of decision-making over the use of their own investments. That is not an easy equation to get right. But we have been making steady progress and should be encouraging health systems to explore more partnerships. --- JAMA Health --- ARTICLE INFORMATION

In a recent JAMA Health Forum article, Glied and D'Aunno ask whether health sector investment in social services is "a bridge too far" and raise concerns about differing priorities, skills, finances, and functions of the 2 sectors that question whether this is a good match for collaboration. 1 As health care organizations ponder their role, if any, in supporting social services, there is growing evidence of health benefits from certain investments in social services (including housing and nutrition). But it is hard to make a traditional return-on-investment (ROI) case for many instances of health systems funding social service programs. For instance, to the extent that this "upstream" social funding helps improve household and community health, it means reduced revenues to hospitals and fee-for-service medical practices-so, not good for business. Managed care organizations do have a stronger business incentive to address their enrollees' healthinfluencing social conditions. Still, although there is evidence that certain investments in housing, nutrition, etc do yield health care savings, the direct ROI to the health sector is often questionable. 2

Introduction Research over the past two decades have established the difficulties of recruiting ethnic minorities to medical and behavioral research studies (Hilton et al., 2010;Hoel et al., 2009;Hussain-Gambles, Atkin, & Leese, 2004;Shavers-Hornadaya, Lynch, LF, & Torner, 1997;Wang, Fridinger, Sheedy, & Khoury, 2001). Researchers are beginning to understand that such populations often have structural, cultural and language barriers that hinder their participation (Ford et al., 2008;Giuliano et al., 2000). Furthermore, reasons for non-participation are rooted in cultural and religious beliefs (Kwan et al., 2014). Participation rates for Pacific Islanders (PIs) (i.e. people of Polynesian, Melanesian and Micronesian descent) are much lower than their African American, Hispanic and Asian counterparts. Even as a combined group, Asian Americans and Pacific Islanders account for less than 10% of the respondents in most research studies (Chlebowski et al., 2005;Ford, et al., 2008;Giuliano, et al., 2000;Murthy, Krumholz, & Gross, 2004). This paper reports on pilot data regarding the reasons that deter PI participation in biospecimen-related research. Epidemiological data have shown that PIs suffer disproportionately high rates of health disparities compared with most other racial and ethnic groups (Dachs, Currie, & McKenzie, 2008;OMH, 2012). Cancer, heart disease, stroke and diabetes are among the major causes of morbidity and mortality among PIs. Despite these large disparities, research aimed at addressing the underlying causes of disease and illness have had limited success at recruiting and retaining participants. An extensive review examining the barriers to recruiting underrepresented populations to cancer clinical trials found that low participation rates for ethnic minorities are related to lack of awareness about the nature of the studies, lack of culturally relevant materials, and mistrust in researchers (Ford, et al., 2008). Some participants are reluctant to participate because of self-perceived harm of clinical trials and loss of control as it relates to treatment. One study cited found that lack of trust in researchers and the institutions involved were also major barriers for minorities, due to incidents of abuse and unethical conduct in the past (Santos, 2008). (Tanjasiri & Tran, 2008;Tanjasiri et al., 2007) explored the knowledge, attitudes and beliefs that PIs have towards biospecimenrelated research. The study collected both quantitative and qualitative data from adult PIs, 21 years of age or older, and asked what they thought about biospecimen (explained as urine, blood, hair, nail and skin samples) collection, research, and banking. Details about the specific conceptual model, methods, instruments and results have been reported in a previous paper that found general support for biospecimen research among PIs (Kwan et al., 2014). The majority of the PIs interviewed did support biospecimen research, and were willing to donate their biospecimen sample if asked. In order to better understand why such support has not translated into actual donations and inform future biospecimen collection-related studies, this paper reports on the findings that relate to PIs' reluctance to participate in studies that involve biospecimen samples. Based on previous efforts completed among minority populations and general discussions with community leaders, the authors believed that PIs are reluctant to participate in biospecimenrelated research because of a lack of awareness about the studies and its potential benefits to individuals and community (Fong, Braun & Chang, 2003;Ford, et al., 2008). It should be noted that the results presented in this paper pertain only to those respondents who responded with hesitation about donating biospecimens for research. --- Methods --- Study Design Utilizing a community-based participatory research (CBPR) approach, PI community leaders from five community-based agencies in Los Angeles, Orange and San Diego counties worked with the academic researchers to conduct this mixed methods study. The community-university team developed the conceptual model, screening and recruitment materials and qualitative and quantitative study instruments. --- Sample Eligible participants were those who selfidentified as being PI, over the age of 21 years and living in the southern California area. Participants were recruited through word-ofmouth, at PI festivals and events and through flyers posted and distributed in community partner organizations. Since this was a pilot study, the sample was a convenience sample of PIs who received services from the community organizations (i.e. health education classes, health insurance information, health screenings and referrals, etc.) or those who knew the health educators in various capacities. In addition, study team members actively recruited cancer survivors because of their unique experiences with biospecimen collection during their cancer diagnosis and/or treatment activities. --- Measures Approved by the Institutional Review Board at Claremont Graduate University, the study instruments consisted of a 46-item paper and pencil questionnaire and a semi-structured oneon-one interview. The paper and pencil questionnaire included demographic, acculturation, health status and cancer history items. Whenever applicable, standardized scales such as the health locus of control scale (Wallston, 1978) and the acculturation scale (AHIMSA) (Unger et al., 2002) were adapted and used in the questionnaire. The interview questions assessed respondents' satisfaction with providing biospecimen samples, willingness to donate, and perceived threats and risks of participating in biospecimen-related studies. For example, we asked "What would stop you, if anything, from personally volunteering to take part in tissue sample research?", "What might be some reasons why a person might not want to participate in genetic research?" and "In your opinion, how might tissue sample research harm your community?" --- Data Collection Data was collected by trained PI health educators at their offices, and all took place in one day. Health educators scheduled one-on-one meetings with each respondent after they were recruited to the study and agreed to participate. Each meeting ranged from approximately 1 to 1.5 hours (i.e. 20-30 minutes to complete the paper and pencil questionnaire which was followed immediately by a 40-60 minute oneon-one interview). Respondents were offered breaks in between the questionnaire and interview portions, although most did not need it. Due to the nature of the data collection method, respondents were not scheduled together and all data collection was done individually. --- Data Analysis Quantitative data analysis was conducted in STATA version 10.0. Basic univariate data analyses were conducted to find the mean age and gender, ethnic and educational level distributions. Frequencies are presented in Table 1. ATLAS.ti was used to code the transcripts into various themes such as reservations, satisfaction, willingness and trust. Each time a respondent provided an answer, it was coded under a particular theme and sub-theme, if appropriate (e.g. "reservation: fear" or "reservation: lack of knowledge"). It was not uncommon for a certain theme to be expressed by a respondent more than once and thus certain themes and subthemes appeared more frequently in an interview transcript than others (i.e. a respondent may have expressed that he/she had "reservations about providing samples because of fear" multiple times throughout the interview). The percentage of coded responses for a particular concern about biospecimen donation is defined as the number of responses coded for the reason divided by the total number of coded responses for the entire sample (i.e. 1601 coded responses). --- Results As shown in Table 1, a total of 60 PIs were interviewed for this study, half of whom were over the age of 65 years. Twenty-one out of the 60 respondents were male. The largest PI ethnic groups in the sample were Native Hawaiians, Samoans and Chamorros. A majority of the respondents had completed high school and post-high school education. Forty one (68.3%) respondents shared that they had some reservations or reluctance about providing their biospecimens for research. A total 53.3% of the respondents aged 65 years or older had reservations about donating biospecimen samples, compared to 83.3% of respondents younger than age 65 who expressed their reservation. A similar proportion of males and females expressed some form of reservation. A majority of the Chamorro and Tongan respondents reported having reservations about participating in biospecimen studies (83.3% and 77.8%, respectively). Education level was not related to whether or not a respondent had reservations. A majority of the female respondents expressed "fear" (76.5%) as their main reason for having reservations about donating their biospecimen sample followed by "lack of knowledge" (56.3%) and "God or spirituality" (72.7%). Male respondents shared similar reasons except that "fear" was cited less often than "lack of knowledge" among males (23.5% and 43.8%, respectively). When analyzed by age, respondents over the age of 65 years indicated that "fear" was their main reason for having reservations (35.3%) while those younger than 65 years cited "lack of knowledge" as their main reason (81.3%). "...a lot of our Polynesian society doesn't want to know. They actually feel that as long as they don't know, it is not happening. Cause there's a lot of people still living in the back, you know. You know, back in the olden days when you don't talk about it, you don't see it, then it is not happening. I think, they have to be educated. But educated in the basics, so they can understand what is being done and what is being said them." -Female, Native Hawaiian --- God, spirituality (11) "...the whole idea with genetic research is um, something that I need to think about more, cause I consider myself a spiritual person and I believe God is the only one that can, should create life and um, so it, it might bother me more than normal" -Female, Native Hawaiian "Yup, yup. And I understand they need to do some of these things better, but sometimes they get a little crazy and you know, to me, overstep the bounds between man versus God, our creator" -Female, Native Hawaiian "...I'd rather leave that up to God" -Female, Marshallese Cloning, scrutiny of genes (5) "Um, more Frankenstein kind of stuff. You know, what the heck are they going to do with me. Are they going to slice my genes and goat and see if we are more alike than not. Genetic research kinda scares me. Spiritually, I don't. I back away from it." -Male, Native Hawaiian "yeah I just don't want people using my genetics to make monkeys," -Female, Tongan --- Discussion To our knowledge, this pilot study is the first of its kind to report on the reservations that PIs living in the continental U.S. have towards biospecimen research. We found that PIs have authentic reservations that hinder their participation such as fear of pain and not wanting to find out about potential health problems. These findings are consistent with previous research studies conducted among African American and Hispanic communities (Corbie-Smith, Thomas, Williams, & Moody-Ayers, 1999;Giuliano, et al., 2000;Shavers-Hornadaya, et al., 1997). Through our interviews and informal discussions with PIs, we learned that some PIs would rather not know about their health problems. These individuals think that if they do not know about their health problems then the problems do not exist (i.e. it is better to not know). Some respondents expressed that even if they learned about their health problems, they would not change their views because of the belief that disease and death are beyond their control. As mentioned earlier, the major finding obtained from the overall pilot study was general support for biospecimen research (Kwan, et al., 2014). Many PIs indicated that they support research that will benefit others in their community and are willing to provide biospecimen samples for that specific purpose if asked to do so. However, a small but important proportion of respondents expressed reluctance to participate personally in research. "Reservations" as a combined theme made up less than 6% of the total coded responses. This means that among all the responses that were coded (i.e. 1,601 total coded responses), reservations only comprised 6% of those responses. It is imperative that researchers better understand respondents' fears and misgivings if we are to increase PI and other minority involvement in research in the future. --- Limitations Several limitations existed in this study that should be noted. First, because this was a pilot study the sample size was small (N=60), and thus respondent fears may not reflect the beliefs of the larger PI population. Further, we did not employ random sampling and thus our findings might not reflect the knowledge, attitudes and beliefs of PIs living in other parts of the U.S. or around the world. For instance, a majority of our respondents were U.S. born, and thus their relatively higher levels of western acculturation may be unique to PIs in the region covered by our study. Second, close to two-thirds of our sample was female and over half were cancer survivors. The experiences of cancer survivors are different from that of the general population because they have been more exposed to clinical practices like blood drawing and biopsies. Cancer survivors also have more exposure to health care providers and medical and research staff during their diagnosis and treatment. --- Implications Although drawn from a small sample of PIs in southern California, we hope the findings from this study provide researchers with valuable information that helps in the planning of and recruitment to research studies among PIs and other minorities. Since the participation of minorities in biospecimen research is essential to reductions in health disparities, it is important that all factors, whether they encourage or hinder participation, be reviewed, carefully considered and properly addressed. If we can address the fear and reluctance that potential respondents have about our studies -whether through improve recruitment methods, more outreach and education about our studies or through a more user-friendly informed consent form -participation, enrollment and retention of minorities may be improved.

Background and Significance: Despite high rates of chronic diseases like cancer, diabetes and cardiovascular disease, Pacific Islanders (PIs) are underrepresented in clinical and genetic studies designed to identify the physiological causes of poor health outcomes. There are limited genetic data and biospecimen samples from PIs under study. This paper described why PIs have reservations about donating their biospecimen samples for research. Methods: Data were drawn from a pilot study designed to assess the knowledge, attitudes and beliefs surrounding biospecimen research among PIs in southern California. Utilizing a community-based participatory research approach, community and academic partners collected quantitative and qualitative data from a total of 60 PI adults with a mean age of 61 years (SD 13 years). Results: "Fear", "God or Spirituality" and "Lack of Information or Knowledge" were the most cited reasons for not participating in biospecimen research. Respondents younger than age 65 years expressed more concerns about donating their biospecimen samples than those older than age 65 years (p<0.012). No significant gender differences were found (p=0.84). Conclusion: Our results emphasize the need to conduct relevant and appropriate biospecimen education among minority communities in order to address misconceptions and build support to increase PI and other minority participation in biospecimen-related studies.

INTRODUCTION Since the first case, reported in December 2019 in China, the severe acute respiratory syndrome coronavirus-2 (SARS-CoV2; COVID-19) pandemic has caused unprecedented global challenges due to rapid interpersonal transmission. This virus causes symptoms ranging from mild, such as sore throat and fever, to severe pneumonia resulting in death (1). Due to a higher transmission rate than other coronaviruses (reproduction ratio: 2.44-4.18) and a high proportion of asymptomatic infectious people (2), the global pandemic has grown significantly, causing nearly 271.4 million cases with 5.3 million deaths (as of 16th December 2021) according to the World Health Organization (3). In Korea, since the first case of COVID-19 in a person who visited China was identified on January 20, 2020, multiple clustered outbreaks associated with religious followings, call centers, and courier services led to a surge in the number of disease occurrences; this was followed by enhanced strict counteractive measures, including social distancing, that were enforced by health authorities, which reduced the weekly average number of cases to single digits (4). However, due to increased outdoor activities, large-scale gatherings during the holiday season and seasonality, the number of newly infected cases grew dramatically to more than 1,000 cases daily, mostly driven by a substantial increase in infections in the capital region, where 25.92 million people live within 11,851.26 km2, one of the most densely populated areas in the world. Historically, disadvantaged people have been highly vulnerable to emerging infectious diseases, especially when they become a persistent epidemic (5). In recent studies on COVID-19, historic evidence showed that socioeconomically vulnerable individuals were more likely to have higher incidence and case-fatality rates of COVID-19 (6,7). This indicates that underlying socioeconomic gradients are strongly associated with the distribution of incidence and fatality rates of COVID-19, due to variations in personal hygiene, access to testing and treatment, compliance level with social distancing policy, and the ability to work remotely (8). In recent studies regarding COVID-19 in the United States, low-income individuals were less able to reduce their mobility or maintain social distancing, indicating that economic activity is highly associated with behavioral responses to social distancing policy (9,10). In addition to individual socioeconomic vulnerability, area-level socioeconomic disadvantages have consistently been associated with COVID-19 incidence. Area-level socioeconomic status (SES) tends to depend on territory-based communities that characterize human society because of a shared socioeconomic basis, commonality in available services, living culture, and lifestyle (11). Area-level socioeconomic measures have been identified in various ways and typically measured using an aggregate variable (e.g., median household income) or a composite measure (e.g., deprivation index). Each measure represents a unique contribution to the socioeconomic association. Specifically, associations with COVID-19 were consistently observed for median household income (12,13) and minor ethnicity (1,14,15) but findings for deprivation index (16,17) and unemployment rate (13,14) were inconsistent, indicating that area-level SES measures have different values across time and place and that how they are measured is important (18). Individuals from lower SES areas are more likely to be infected for various reasons; however, in most studies, the primary cause was the lack of mobility reduction resulting in the inability to maintain social distancing. However, the mediating role of mobility was advocated in other studies to explain area-level socioeconomic inequalities in COVID-19 infection based on the high correlation between area-level SES and mobility reduction (10,16,19,20). Despite wide acceptance of the explanation, studies in which the underlying relationship was investigated using both measures are scarce. Thus, firm empirical evidence is lacking on whether the effect of area-level SES on COVID-19 incidence depends on the level of mobility. This concept may be particularly relevant in countries like Korea, where socioeconomic inequalities in COVID-19 incidence may not be straightforward because affluent areas are also a central business place. As noted below, Korea had been undergone a relatively lower level of COVID-19 incidence compared to other countries (21). Nevertheless, a better understanding of regional disparity in COVID-19 incidence is a huge challenge because it is essential to monitor the pattern of spread into subsegment of the population, let alone the incidence from the entire population. Thus, we investigated the socioeconomic inequalities in COVID-19 incidence at the level of a primary administrative unit of local government in Korea, using a diverse range of socioeconomic indicators including a mobility measure. In this study, we investigated (1) whether area-level socioeconomic measures are associated with COVID-19 incidence at the municipality level; (2) whether the associations' differences in the association between socioeconomic inequalities and COVID-19 incidence in two different epidemic phases with disparate social distancing enforcement; and (4) whether socioeconomic inequalities in COVID-19 infection are mainly due to mobility differences. --- MATERIALS AND METHODS --- Study Base Overall, Korea experienced favorable outcomes of COVID-19 compared with other countries in terms of incidence and mortality through the pandemic and the study period (21,22). To evaluate the effects of socioeconomic inequalities on COVID-19 incidence at different epidemic levels, the epidemic period was divided into two phases based on the daily number of cases and the accompanying social distancing intensity level as shown in Figure 1: low phase (from May 6 to August 14, 2020) in which less than 100 mean daily cases were confirmed with the eased social distancing regulation (level 1) and rebound phase (from August 15 to December 31, 2020) in which more than 100 mean daily cases were reported with stricter distancing imposed (level 2). Because the early phase of the epidemic was induced by a specific religious congregation concentrated in very limited municipalities, the starting time point in this study was March 5, 2020, to ensure the validity of the results (23). The social distancing level enforced by the Korean government was classified into two levels during the study period through the guidelines underwent several changes afterward. For example, under level 2 social distancing, the use of face masks in public became mandatory, social gatherings with more than a certain number of individuals were prohibited and restaurants must be closed after a specific time point but without movement restriction. Lower social distancing regulation (i.e., level 1) began from May 6 to August 14, 2020, and stricter social distancing measure (i.e., level 2) was enacted from August 15, 2020. --- Socioeconomic Status Measures and Covariates The information on COVID-19 incidence as an outcome variable was collected from 229 municipalities and compiled from the KCDC and the local administration's official websites. As listed in Table 1, six area-level (i.e., municipality) socioeconomic factors were used to investigate the effects of inequality on the incidence of COVID-19 in Korea. The indicators were classified into two subcategories, SES, and economic activity, based on the corresponding attributes. SES measures included the following: national insurance contributions as the proxy of area-specific income level; material deprivation index (MDI); nonemployment rate; the proportion of basic livelihood security recipients; financial autonomy of the area. Economic activity included mobility at risk. Data on national insurance contributions in the first quarter of 2020 were obtained from the Korean National Health Insurance Services. MDI for each area was a composite index derived from the sum of standardized Z-scores for eight measures based on data from the national population and housing census conducted by the National Statistical Office of Korea; the proportion of nonemployed males, manual laborers, households under the minimum housing standard, nonsecured housing tenure, nonapartment housing, lower educational level (<unk>middle school), single-parent household and school dropouts between 9 and 24 years of age (24). The higher the MDI score, the more the area is deprived. The nonemployment rate was calculated as the proportion of individuals who were unemployed or out of the labor force (e.g., early retirement, studying, and disability) between 30 and 64 years of age (25), based on data from the National Population and Housing Census in 2015. The proportion of basic livelihood security recipients at the area level in 2019 was retrieved from the Korea Social Security Information Service. Financial autonomy for each area was defined as a ratio of total revenue generation to the total expenditure per municipality as provided by the Korean Statistical Information Service for 2019. To determine the socioeconomic strata of socioeconomic factors, those continuous values of socioeconomic factors were converted into quintiles of their distribution (i.e., each stratum accounted for 20% of the number of municipalities) (26). In addition, the municipality-specific economic activity variable, including the volume of traffic for mobility at risk represented by a Z-score, was added. Mobility at risk was equal to the proportion of the traffic volume of work-related movement utilizing public transportation, which was calculated by multiplying the volume of public transportation and the volume of works-related traffics (e.g., commuting to work and field trips). This variable was obtained from a transportation survey conducted by the Korean Transport Institute in 2018. Finally, three covariates, namely, municipality-specific median age, population density, and the number of healthcare workers per 1,000 inhabitants, were used to adjust for the demographic composition and the local health care capacity of the areas in our analysis. The variables were derived from the data obtained from the Korean Statistical Information Service for 2020. The data in our study were extracted from open sources, which are aggregated by administrative subdivisions. Therefore, do not contain any information that is indicative of information about personal or household level. The Institutional Review Board (IRB) of Korea University granted an exemption for this study (IRB exemption number: KUIRB-2020-0297-01). --- Statistical Analyses Several steps of the analytical process were applied to examine socioeconomic inequalities in COVID-19 incidence. Due to the nature of spatial data, spatial autocorrelation on the SES variables and three covariates were examined using Global Moran's I test before investigating the association between SES measures and COVID-19 incidence. The statistical significance of the Global Moran's I was estimated with 999 simulations. Following identification of the presence of spatial autocorrelation in socioeconomic indicators, the association between socioeconomic measures and COVID-19 incidence was estimated as an incidence rate ratio (IRR) using a spatial negative binomial model with marten correlation function for spatial correlation term (Model 1). To account for potential confounding factors, adjustment was initially made for three covariates (i.e., median age, population density, and healthcare workforce at the area-level) (Model 2). In addition, we conducted a regression with a further adjustment for economic activity to evaluate the mediating effect of mobility on the association between area-level SES and COVID-19 incidence (Model 3). We built a Bayesian generalized linear model to estimate the posterior marginal distribution of IRR of each SES measure. Because the observed incidence rate by the municipality, used as the outcome of interest, was overdispersed, it was modeled as a negative binomial random variable with overdispersed variance instead of Poisson regression. In addition, the Besag, York, and Mollié (BYM) model was used to account for spatial autocorrelation of residuals by adding a spatial random effect using intrinsic conditional autoregressive (iCAR) function and extra residual term for spatially independent variation that was independent, identical, and normally distributed as follows: Y i <unk>NB (<unk> i,r i ),Y i : Number of COVID -19 cases by municipality i <unk> i = r i r i +<unk> i,E (Y i ) =<unk> i log (<unk> i ) =<unk>+log population i +<unk> 1 <unk>SES i,k + N j=2 <unk> j <unk>covariate i +u i +<unk> i u 1:229 <unk>ICAR W,<unk> 2 u <unk> <unk>N 0,<unk> 2 <unk> u <unk>N0, I-C -1 <unk>M,C=<unk> <unk>W,M=I<unk> <unk> 2 u where u i is the conditional autocorrelation regression term, the covariance matrix of the parameters calculated based on the neighboring regions, <unk> i is the nonspatial structured term, u is the spatial correlated random effect calculated by averaging neighboring random effects, I is the identity matrix, and W is the spatial weights matrix constructed by an inverse distance function with the exponents followed by row-standardized such that each row sums to 1 for interpretation of the parameters (27). The neighboring region at each municipality was defined as the administrative division located within the geographical distance that was not spatially correlated in a variogram generated using a Bayesian generalized linear model without the spatial correlation term. The spatial correlation parameter denoted as <unk> was set to 1. The models were run with three chains with different starting values in which sampling values in the MCMC process with a burn-in of 4,000 iterations and a thinning rate of 10, and 50,000 iterations were used for each posterior distribution of parameters for SES and covariates. Convergence of the chains was assessed by visual inspection of the posterior distributions and computation of the Gelman-Rubin statistic. The Deviance Information Criterion (DIC) was used to measure and compare the goodness of fit for the model. The prior distribution for each parameter and hyperparameter is described in the Supplementary Material. R2WinBUGS R software package version 2.1 (28) with WinBUGS software version 1.4.3 was used to carry out given statistical approaches (29). The map presented in this study was created by Esri ArcGIS software version 10.8.1 using the South Korea map which is publicly available (30). All analyses were separately performed for two different phases of the COVID-19 pandemic; the low and rebound phases. We repeated a similar analysis to estimate the relative index of inequality (RII) as a supplementary measure of inequalities in the COVID-19 incidence rate at the area level. RII is a commonly used measure of health inequalities that summarizes the distribution of a health outcome measure against an SES as a relative difference of the least and most deprived subgroups (31). RII in this study corresponds to the relative risk of the incidence for COVID-19 in the lowest and the highest socioeconomic strata and, therefore, is directed by changes in two strata (Supplementary Material). The RII was also estimated using a spatial negative binomial model with marten correlation function for spatial correlation term, 95% CI was estimated by bootstrap. RII estimation was made as follows. Y i <unk>NB (<unk> i,r i ), Y i : Number of COVID -19 cases by municipality i <unk> i = r i r i +<unk> i,E (Y i ) =<unk> i log (<unk> i ) =<unk>+log population i +<unk> 1 <unk>SES i,k + N j=2 <unk> j <unk>covariate i +u i +<unk> i u 1:229 <unk>ICAR W,<unk> 2 u <unk> <unk>N 0,<unk> 2 <unk> u <unk>N(0, (I-C) -1 <unk>M), C=<unk> <unk>W,M=I<unk> 2 u where x i denotes the mid-point of municipality i in socioeconomic class k with number 1 assigned to the highest class of SES, as opposed to the lowest strata. The mid-point was derived for each SES class. In addition, SES variables are likely to be mutually correlated. Thus, Spearman's correlation coefficient between two paired SES variables was estimated to exclude the correlated combinations for subsequent multivariate analyses. --- RESULTS --- Overview of COVID-19 Incidence and Socioeconomic Characteristics The COVID-19 epidemic in Korea showed two distinctive phases in terms of the incidence level over the study period as illustrated in Figure 1 Geographically, a significant difference was observed in the area-level COVID-19 incidence rate as shown in Figure 2. On average, 12.7 cases were reported per area [minimum (min) -maximum (max) = min -max = 0-127 cases] in the low phase and 177.1 cases (min -max = 0-1,653 cases) were reported in the rebound phase. The majority of COVID-19 cases were reported in the Seoul metropolitan area (81.8% in the low phase and 72.8% in the rebound phase) where 50.28% of the total Korean population resides within 11,851.26 km2 (11.8% of the land size of Korea). The average nonemployment rate was 13.9 and 4.9% of households received basic livelihood security (Table 2). All variables, in particular, economic activity, showed significant spatial autocorrelation in the Global Moran's I test indicating that the association of those variables with COVID-19 should be measured with consideration of spatial autocorrelation. Generally, socioeconomic measures were significantly correlated with each other (Figure 3) but heterogeneous in direction. For example, national insurance contributions as the proxy of personal income level had a negative correlation with indicators of social exclusion and poverty [e.g., MDI (Spearman coefficient = -0.84], the proportion of basic livelihood security recipients [Spearman coefficient = -0.75), and the nonemployment rate (Spearman coefficient = -0.13)]. Notably, a negative correlation of economic activity (i.e., mobility at risk) was observed with indicators of social exclusion and poverty such as the proportion of basic livelihood security recipients (Spearman coefficient = -0.38) but not with nonemployment rate (Spearman coefficient = 0.69), and a positive correlation of mobility at risk was shown with national insurance contribution (Spearman coefficient = 0.69), indicating economic activity (i.e., mobility at risk) was characteristic of affluent areas. The incidence rate ratio (IRR) was estimated using a Spatial and Bayesian negative binomial model with marten correlation function and BYM for spatial correlation term, 95% confidence interval was estimated by bootstrap, denoted in the parenthesis. <unk> Model 1: unadjusted model. <unk> Model 2: socioeconomic indicators were remained to estimate the associations, adjusting for covariates (human density, median age, and health care workforce). <unk> Model 3: two significant variables in Model 2 were retained to estimate the associations, adjusting for covariates from Model 2+ mobility at risk, separately. In turn, the incidence rate ratio for mobility at risk returned two estimates for each of two corresponding socioeconomic status variables. The incidence rate ratio of mobility at risk in this table was given as an adjustment factor for basic livelihood security recipients variable. <unk> denotes a given value is >1. --- Associations Between Area-Level Socioeconomic Measures and COVID-19 Incidence Table 3 shows the estimation of IRR for the association between area-level SES measures and COVID-19 incidence using a Bayesian negative binomial regression. Overall, two area-level SES measures, nonemployment rate and the proportion of basic livelihood security recipients, were consistently associated with COVID-19 incidence based on unadjusted and adjusted modeling in the low and rebound phases. Specifically, in the low phase, the adjusted IRR corresponding to an increase in 1% of the nonemployment rate and the proportion of basic livelihood security recipients was estimated as 1.20 (95% credible interval (Crl) = 1.13-1.28) and 1.23 (95% Crl = 1.07-1.40), respectively (Model 2). In the rebound phase, the same SES measures presented inconsistence association with COVID-19 incidence. For example, the nonemployment rate showed a significantly negative association with COVID-19 incidence adjusted for only covariates (model 2), but for both covariates and mobility at risk (model 3), while the proportion of basic livelihood security recipients had an only univariate association with COVID-19 incidence (model 1) (Figures 4,5). An economic activity indicator (i.e., area-level mobility at risk) was positively associated with COVID-19 incidence rate in both the low (IRR = 1.67, 95% Crl = 1.26-2.17) and rebound phases (IRR = 1.28, 95% Crl = 1.14-1.44). When assessing the mediation of mobility at risk in the association between two SES measures and COVID-19 incidence, the magnitude of the associations was attenuated but remained significant in the low phase, but no associations were observed in the rebound phase (Model 3). For instance, the area with a higher nonemployment rate had a higher risk of COVID-19 incidence in the low phase (IRR = 1.61, 95% Crl = 1.09-1.25) but independent associations were not observed in the rebound phase (IRR = 1.02, 95% Crl = 0.99-1.06). When assessed with RII, nonemployment rate and the proportion of basic livelihood recipients showed a similar pattern of associations with COVID-19 incidence (Supplementary Table ). --- DISCUSSION In this study, a significantly positive association was found between area-level nonemployment rate and the population share of basic livelihood security recipients and COVID-19 incidence. However, area-level socioeconomic effects were stronger in the low phase when the prevalence of COVID-19 was low, with less strict governmental measures (Figures 4,5). In other words, the strength of the association of those SES measures decreased as the level of COVID-19 incidence rate across the country increased. Similarly, the inequalities in the COVID-19 incidence rate concerning nonemployment and basic livelihood security recipients were significantly high in the low phases. In addition, higher mobility at risk, indicating active economic activity at the area level, increases the risk of COVID-19 incidence in both phases. In this context, when both area-level SES measures and mobility were simultaneously adjusted, SES measures remained significant in the low phase, suggesting they were independent of each other in the low-risk period. However, in the rebound phase, adjustment for economic activity variables showed no association between SES measures and COVID-19 incidence. Overall, partial existence of COVID-19 inequalities in some measures may have occurred as a result of counteraction between risk raising and lowering area-level effects (e.g., poor communities are less mobile). Among five measures of area-level SES, the areas with a higher level of basic livelihood security recipients and nonemployment rate showed a higher risk of COVID-19 incidence; however, an association was not observed for other area-level socioeconomic measures. A partial observation of area-level socioeconomic inequalities in COVID-19 incidence differs from most previous studies from the United States (32), the United Kingdom (20), and Spain (33,34), and is similar to a previous Korean study (35) in which no or a partial association was observed. Inconsistency in inequalities in COVID-19 incidence across measures may possibly be interpreted using the socioeconomic context of Korea. Korean government measures were impartially imposed regarding case-identifying processes, awareness of the process, access to COVID-19 testing, and contact tracing, which may provide a relatively equal chance of being diagnosed. Close supervision by national mandatory conduct systems was uniformly applied regardless of area-level SES. This universal approach is not exclusive to Korea, but the outcome may be proequity in countries with high levels of public support for strong governmental measures. In addition, inequalities may be greater in places where COVID-19 diagnostic testing is often delayed, and choosing which patient to care for first is an issue when the number of patients is overwhelming (36). In contrast, Korea has maintained a lower COVID-19 incidence by enhancing rigorous contact tracing and extensive testing with no discrimination, which may have minimized some forms of socioeconomic disparity across areas. Inconsistent inequalities may be also explained by different conceptualizations of the five area-level SES measures. Basic livelihood security recipients are mostly older adults, and the age composition of this measure better reflects diagnosed COVID-19 cases; the majority (35.1% in the low phase and 31.6% in the rebound phase) were older adults (<unk>60 years of age), according to the Korea Central Disease Control Headquarters (http://ncov.mohw. go.kr/en/). This finding is in agreement with an individual-based Korean study in which higher SES was associated with higher COVID-19 incidence in the older population, and both higher and lower SES were associated with the younger population (37). Similarly, the area-level nonemployment rate largely depends on the proportion of individuals who are not in the labor force. However, the national health insurance premium and financial autonomy address directly the income level of the working population. The deprivation index is a composite measure developed using six variables of material circumstances. Deriving the material deprivation index by assigning the same weight to each individual variable may mask socioeconomic patterns existing in the COVID epidemic (18). In this study, a high level of mobility was associated with an increased risk of COVID-19 incidence, consistent with recent studies from the United States. However, this study results are in contrast to previous studies in terms of which areas are more mobile. In most previous studies, socioeconomically disadvantaged areas were reportedly more likely to have higher mobility (9,10); however, this study results showed that a high level of mobility was characteristic of affluent areas in Korea. This finding is understandable because mobility using public transport is concentrated in densely populated areas in the capital and large cities in Korea and within-city mobility is distributed across places of social gatherings and business meetings. High mobility observed in affluent areas may offer another plausible explanation as to why socioeconomic inequalities differ based on the measure. Collectively, area-level socioeconomic disadvantages concerning COVID-19 incidence were mixed with lower economic activity in poor communities. Notably, when simultaneously adjusted for mobility, SES measures of basic livelihood security recipients and the nonemployment rate remained significant in the low phase but not in the rebound phase. This result indicates that mobility is a major contributing factor to the association between area-level SES and COVID-19 incidence in the rebound phase, but mobility alone does not fully explain the association; other vulnerabilities (e.g., a larger poor older population) are likely to be involved. The area-level socioeconomic effect was stronger in the low phase, when the prevalence of COVID-19 was low, with less strict governmental measures, indicating that the area-level socioeconomic gradient is less likely to affect the variation in COVID-19 occurrence. Hypothetically, the socioeconomic inequalities in COVID-19 incidence were not exacerbated in the rebound phase. A larger inequality in the low phase may be attributed to people in poor communities being less responsive to an initial spread of COVID-19 when government public health measures were not sufficiently implemented nationwide. With progression to a widespread stage (rebound phase), the Korean government launched the testing and contact tracing system as a key part of the control strategy. The relatively effective performance of the strong government measures, with public compliance, applied in a nondiscriminatory manner, irrespective of SES, led to subsequent improvement in regional variations in incidence. The strength of this study includes the use of nationwide incidence data and various socioeconomic measures. In particular, concurrent use of SES measures with mobility measures enabled us to obtain a better-fitted model and identify any existing associations. This study had several limitations. First, the mobility measure was obtained from the previous year and does not reflect the mobility changes induced by the COVID-19 pandemic. However, to some degree, the use of previous mobility data may serve as a proxy indicator in this interpretation because mobility change depending on SES appears minimal in Korea. The only study in which the average mobility patterns were compared during the COVID-19 period in Korea showed no significant change in mobility shaped by socioeconomic differences (38). Second, the findings in this study are limited to area-level interpretation, due to the inherent nature of ecological studies, which could not be directly applied at an individual level. Third, the variables associated with living conditions, such as poor hygiene conditions and overcrowding, could not be included due to data availability, although this would be relevant information regarding the association between SES and COVID-19 incidence. Furthermore, it is noteworthy to investigate the impact of inequalities on the incidence of COVID-19 in countries with a relatively lower number of cases and during the post-vaccination period to understand the direct effect of SES disparity on the infection adjusted for vaccination coverage. In conclusion, COVID-19 does not occur randomly but follows socioeconomic patterns; socioeconomic inequalities in COVID-19 incidence occur concerning the unique context of a society in response to the pandemic. Despite similar contexts, each SES measure represents a specific factor and has a different ability to identify socioeconomic stratification caused by COVID-19. In Korea, where government control measures were effectively applied, with high compliance and with relatively low incidence, SES measures, such as basic livelihood security recipients, reflecting age stratification, may be preferable. Mobility was associated with COVID-19 incidence and partly explains the correlation between area-level SES and COVID-19 incidence during a high incidence period in countries such as Korea, where mobility is characteristic of affluent areas. The results confirm the necessity for emergency policy priorities concerning the older population in disadvantaged areas, including faster vaccination, and underscore a further need for socioeconomic support, including emergency relief funds. --- DATA AVAILABILITY STATEMENT The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author. --- AUTHOR CONTRIBUTIONS D-sY conceived and designed the study. D-sY and MK acquired the data and wrote the original draft. MS and N-KS supported data collection. MH contributed data standardization. D-sY performed analyses. BC, MH, MS, N-KS, and SK edited the subsequent drafts. All authors read and approved the final manuscript. --- SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmed. 2022.840685/full#supplementary-material --- Conflict of Interest: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Objective: Area-level socioeconomic status (SES) is associated with coronavirus disease 2019 (COVID-19) incidence. However, the underlying mechanism of the association is context-specific, and the choice of measure is still important. We aimed to evaluate the socioeconomic gradient regarding COVID-19 incidence in Korea based on several area-level SES measures. Methods: COVID-19 incidence and area-level SES measures across 229 Korean municipalities were derived from various administrative regional data collected between 2015 and 2020. The Bayesian negative binomial model with a spatial autocorrelation term was used to estimate the incidence rate ratio (IRR) and relative index of inequality (RII) of each SES factor, with adjustment for covariates. The magnitude of association was compared between two epidemic phases: a low phase (<100 daily cases, from May 6 to August 14, 2020) and a rebound phase (>100 daily cases, from August 15 to December 31, 2020). Results: Area-level socioeconomic inequalities in COVID-19 incidence between the most disadvantaged region and the least disadvantaged region were observed for nonemployment rates [RII = 1.40, 95% credible interval (Crl) = 1.01-1.95] and basic livelihood security recipients (RII = 2.66, 95% Crl = 1.12-5.97), but were not observed for other measures in the low phase. However, the magnitude of the inequalities of these SES variables diminished in the rebound phase. A higher area-level mobility showed a higher risk of COVID-19 incidence in both the low (IRR = 1.67, 95% Crl = 1.26-2.17) and rebound phases (IRR = 1.28, 95% Crl = 1.14-1.44). When SES and mobility measures were simultaneously adjusted, the association of SES with COVID-19 incidence remained significant but only in the low phase, indicating they were mutually independent in the low phase.The level of basic livelihood benefit recipients and nonemployment rate showed social stratification of COVID-19 incidence in Korea. Explanation of area-level inequalities in COVID-19 incidence may not be derived only from mobility differences in Korea but, instead, from the country's own context.

Introduction AgeTech refers to the use of technologies and services to support aging. The new generation of AgeTech encompasses emerging and advanced technologies in areas such as artificial intelligence (AI), robotics, machine learning, e-health, and mobile technologies to support the health, independence and well-being of older people (Sixsmith, 2021). AgeTech has been explored in terms of supporting older people to remain at home for longer (Verloo et al., 2020), to provide social connectedness (Baez et al., 2019), support wellbeing (Astell et al., 2016) and mental health (Andrews et al., 2019) and connect the older person to their wider community (Fleming et al., 2018). However, the use of technology to provide support for older people to age well in place may also bring with it the potential to increase inequalities in access to health and health outcomes for vulnerable and marginalised people. This is especially the case when the needs and resources of individuals as well as groups are not fully acknowledged, when their characteristics are reduced to oversimplified or stereotypical narratives, and when barriers to equitable access exist. Thus, technology design and its functional requirements necessitate deeper insight into the range of diverse factors that may shape its context of use. Research has indicated a crucial need to fully understand who and in what ways emerging and advanced technology interventions such as AI can have positive benefits for older people, or further exacerbate experiences of marginalisation (Sixsmith, 2006). A critical discussion of ethical design, digital equity, and policy pathways is required if we are to fully understand the positive and negative intended and unintended consequences of AI as an AgeTech solution to drive practical, equitable, and inclusive multilevel solutions to support healthy, active ageing. Regulations and the accompanying policies are without a doubt crucial instruments when it comes to avert harm from vulnerable groups, safeguard their civil rights, and avoid marginalisation. However, apart from the rather passive or reactive view that regulations and policies should act as a protective shield, a more active approach could be taken in regard of technology design and implementation. In order for AI-based AgeTech to realise its full potential, the focus on equity is key. Equity -as it pertains to the changing digital landscape -is defined in this paper as fairness and equality of access and use of AgeTech regardless of the myriad social characteristics which one holds, that combined, may create social disadvantages. As a mechanism to help to ensure equity by this definition, the threshold of having meaningfully addressed AgeTech can be achieved by (Fang, 2018;Fang et al., 2019): (1) assessing the intended and unintended positive and negative consequences of AgeTech; (2) considering who is most likely to benefit from AgeTech; and (3) responding to the challenges of those who experience the most significant barriers to access and use. To help to ensure AI-based AgeTech to truly support an independent lifestyle and improve the quality of life of older adults, these technologies need to be designed in a way that enables agency as well as social participation. Therefore, equitable access has to be a key feature of AI-based AgeTech. The precondition for this is to acknowledge the diversity of older adults and to tailor technologies to their multifaceted needs and resources. Equity should thus be an explicit aim of designing AIbased AgeTech, and not just seen as an extra benefit or minimum requirement. This approach does not only aim at protecting older adults against the possible negative effects of AI-based AgeTech. It actively addresses the social determinants of health as defined by the WHO as non-medical factors that influence health outcomes by shaping the daily life and social situation of a person (WHO, 2010) (such as age, ethnicity, gender, socioeconomic status, and level of education) which may create experiences of marginalisation. We will explore how a particular facet of AI systems 1 3 -complexity management -can potentially create and exacerbate social inequities, and subsequently, make recommendations as to how inequities that stem form AIbased AgeTech can be addressed. Although clinical as well as legal issues may also arise in this context, our specific focus is on the primarily ethical aspects connected to complexity management. Each of the aformentioned social determinants can be a contributing factor for marginalisation, yet in reality, we seldom see these determinants as sole factors to shape inequity. More often, they are mutually dependent. For example, lower education and belonging to a social minority often correlate with lower socioeconomic status, which in turn, can lead to poorer health outcomes in old age. Fang et al., (2019) have identified this as a "wicked" problem, in view of three key principles of intersectionality (Hankivsky, 2014): 1) Single traits such as age, gender, or ethnicity are insufficient when it comes to understanding individual experiences -as reducing individuals to a single trait results in the oversimplification of their lived accounts. 2) Social determinants are not objective or fixed categories, but are seen as fluid and flexible social constructs that vary depending on the contexts of time and place and are shaped social processes, structures, and power relations. 3) Social justice and equity are not merely add-ons, but have to be considered as crucial aspects when it comes to policy-making. --- Complexity, Complexity Management and AI A main ethical issue when it comes to AI-based AgeTech in the light of interconnected social determinants is complexity management. AI-technologies are built to detect patterns in large data sets and derive algorithms for predicting future events or controlling processes. In AgeTech, AI-technologies face the inherent complexity and unpredictability of everyday behaviors, situations, and contexts. For AI systems to operate within the current technological limitations, it is essential that complexity is managed, which often means reducing complex aspects to simple factors. Complexity management thus means to align the data with a given framework for processing it, which often means to standardise, decontextualise, and quantify data. This can become a problem when qualitative and context-rich data is reduced to manageable data formats, thus losing some of its crucial features. This is not merely a technical, but primarily a social and ethical issue, since it affects the way we are dealing with diversity in society and the specific needs and resources of individuals. Complexity in this context however does not only refer to different contexts of use, but also to the variety of user characteristics. Here lies the connection between complexity and vulnerability: the diversity of users regarding the aforementioned social determinants causes complexity, which in turn poses a problem that AI-based systems try to resolve by reducing complexity. Therefore, AI-based AgeTech aimed to support the personalisation of services may inadvertently do the exact opposite and thus posing a risk to vulnerable groups. In order to analyse ethical issues connected to complexity management, we suggest examining the way social determinants are represented in the data that is used, 1 3 how social determinants are taken into consideration when processing data, and how social determinants may shape the use of technologies that operationalise the data. Therefore, we identify three dimensions of ethical issues: The inappropriate representation of social determinants may cause flawed or stereotypical concepts od user characteristics or bias. The context-insensitive procession of data, i.e. an oversimplifying or reductionist approach, may lead to standardisation of user characteristics. Ignoring social determinants in technology use may negatively affect access to these AI-based AgeTech. Complexity management thus has three crucial aspects which we explore in more detail in the following sections and, subsequently, recommend strategies for dealing with the issues at hand: • Bias-Bias in relation to certain social characteristics may be introduced into AI systems at different stages in the development and innovation process, such as in the AI training data that is used or decisions made regarding the commercialisation and marketing of products and services. • Standardization-the problematic aspect here is the underlying assumption that objective parameters can be defined and operationalised to identify and respond to ambiguous behaviors and situations. • Access-Some AI-based technologies are designed in a way which makes them difficult to use by some individuals and groups. --- Complexity Management and Bias Bias is one of the most fiercely debated issues in AI-based technologies (Challen et al., 2019;Mittelstadt & Floridi, 2016;Neven, 2015;Safdar et al., 2020;Wong, 2020). Bias can be a result of complexity management in two ways. Firstly, algorithm-based systems strongly rely on training data that has been used in their development. These systems learn by processing large amounts of this training data, detecting connections and patterns, and inferring general rules. The quality of the training data thus determines the quality of the algorithm-based system. The training data is usually taken from large cohort studies (Jones et al., 2018), in which certain social groups are often notoriously underrepresented. As a consequence, the bias within the training data is transferred into the system, leading to a biased algorithm. Thus, the needs and resources of various social groups are not considered by the system. Secondly, AI-based systems and hardware devices are produced on a certain scale in order to be cost-efficient. In order to make the product suitable for a certain mass of users, complexity management regarding concepts of old age is necessary. Further, by way of complexity management, the risk of so-called age scripts arises, consequently, stereotypical concepts of what it means to be older aged, regarding needs and resources of older adults, are inscribed into the technology. --- 3 Training Data AI-systems learn by inferring algorithms from data sets (Jones et al., 2018). In order to operationalise this, the systems are fed with training data, mostly taken from large cohort studies. The more good quality training data a system is fed with, the more accurate its algorithms become. That also means that the quality of the algorithm, its accuracy and discriminatory power, depends largely on the quality of the training data, and lack thereof can lead to bias. A bias within the training data is often perpetuated in the system's algorithm, as bias and complexity management often go hand in hand. This is largely due to the fact that prior to processing for analysis, the data is decontextualised, meaning as part of data preparation, the data becomes separated from its specific spatial, temporal, or social determinants. Racial bias is a prominent example for the connection between complexity management in the shape of decontextualization and bias. Obermeyer et al. (2019) have demonstrated this connection in a striking way upon analysing an algorithm used within the US health services. The task of this algorithm was to identify people with complex health needs. The algorithm assessed individual health costs and predicts future costs. It followed the basic assumption that individuals where more health costs have been invested in the past, have higher health needs and should therefore have easier access to health services. As a result of challenges that stem from socio-structural, and historical issues, when less health spending are applied in certain groups (i.e., African-American people), the algorithm assigns them a lower risk-score. This example demonstrates that the inherent technological need to reduce complexity in order to process data efficiently may lead to bias and thus conflict with the basic goal of providing enhanced personcentred health care. The result is a harmful feedback loop whereby health disparities manifest themselves in training data that do not represent social realities. Biased algorithms and predictive models are subsequently built based on these training data which can in turn aggravate the structural discrimination already inherent in the healthcare system (Walsh et al., 2020). --- Age Scripts AgeTech relies on specific concepts of age, a narrative that defines characteristics of older adults as well as what age means. These age scripts are written into AgeTech and define the scope and purpose of the technology (Peine et al., 2015). Age scripts can be developed based on various sources such as societal views, design traditions, or individual interpretations (Peine & Neven, 2021). This implies a certain risk for stereotypical or poorly-informed views about older adults, their needs and resources, and their desired way of life becoming the basic design framework of AgeTech. Usually, such ageist stereotypes present older adults as 'problem focused' which assumes older people as a homogeneous group with associated traits such as frail, vulnerable, and in need of help (Ayalon & Tesch-Romer, 2018). The notion of age is fundamentally viewed as a problem to be solved by AgeTech (Rubeis, 2020). Although enabling agency is the intended goal, limitations arise when fundamental concepts and purposes of AgeTech are not determined with the end users (Neven, 2015;Peine et al., 2015). The result is the emergence of a sense of 'benevolent paternalism' that 1 3 predefines the appropriate lifestyle of older adults without including them in the design process (Manzeschke et al., 2016). This is especially the case when digital equity was not considered as part of the intended goal. Focusing on equity may help to diversify training data as well as the basic scripts that are written into the technologies. It is especially important to acknowledge that older adults do not constitute one homogenous group, but a diverse population who are shaped by a complex interplay of social determinants. --- Complexity Management and Standardisation Standardisation means that certain parameters must be predefined for AI based systems to process. For example, a crucial parameter in some systems for fall detection is gait pattern (Piloto et al., 2018). These particular systems measure the typical gait pattern of a person, and whenever the gait pattern changes and/or there is a deviation from the standard, the system interprets this as a sign for an impending fall. The system may then trigger an alarm and inform caregivers. The issue here is not so much the definition of a standard as such since it is defined on the basis of a person's individual health data. What is problematic, is the fact that specific parameters are defined as objective indicators of health and well-being. This suggests that primarily qualitative categories like well-being can be easily quantified by choosing the right parameters. This is an oversimplification that ignores the rich contexts of a person's attitude towards well-being, health, and others, which forces them to adapt to prefixed definitions of well-being. In the fall detection example and as seen in other examples, this may not seem like an issue since it is quite reasonable to define gait pattern as a parameter for assessing the fall risk. Another example is systems that recognise whether shutters are open or closed, and if closed, how long they remain closed during daytime. In this case, systems that draw conclusions based on the mental health status of a person because closed shutters throughout the day may imply a mood change, or even a depressive episode. It becomes clear that in this case, complexity management (closed shutters equals depression) can lead to a standardisation of behaviour. Standards and standardised parameters play a key role in all data-driven technologies. The scope and characteristics of data that are deemed as useful or viable depend on the parameters for measuring them. In the context of digital health, the main parameters are digital biomarkers. Biomarkers are biomedical or behavior-related indicators used to measure, predict, or evaluate health-related outcomes (Guthrie et al., 2019;Sim, 2019). Raw data as detected by sensors are of no use for health care professionals without predefined biomarkers that allow for scaled and contextualised data. Digital biomarkers are therefore crucial for personalised interventions such as AgeTech. In AgeTech, digital biomarkers may be used in sensor technologies, either in smart home sensors or smart wearables, in order to measure and predict changes in symptoms and behavior. Digital biomarkers have to be defined, meaning that a certain trait has to be singled out and deemed as a viable indicator. This may not be problematic in a biomedical context, e.g., determining blood pressure as biomarker for cardiovascu-1 3 lar health. However, even in this context, blood pressure would first and foremost be seen as one factor amongst others with which it interacts and thus constitutes cardiac health. Regarding individual behavior, it is even more difficult to define clear-cut biomarkers. One example is a framework for recognising and regulating emotions in older adults presented by Castillo et al. (2014). The framework consists of sensors that measure physiological signals, facial expression, and voice in order to determine the emotional status of a person. A social robot reacts to these emotions and the color, lighting, and music in the room is adapted in order to enable positive emotions. In this setting, certain facial expressions for example are defined as indicators for a certain emotional state. A certain colour scheme or music is defined as a means of regulating emotions. The underlying assumption is that clear-cut parameters can be defined for distinguishing between positive and negative emotions as well as for regulating the negative ones. This blurs the line between support and conditioning, i.e. the regulation of a person's conduct. Conditioning signifies an action directed at creating a specific behavior that is deemed as appropriate or desirable. For example, some stakeholders may have a personal interest in reducing costs of health care services. Thus, instead of modifying health care services to address the specific needs of individuals, methods of conditioning could be applied in order to shape the individual to fit the most cost-effective service. Taking the aforementioned example, the use of technology for regulating emotion may be understood as a way to achieve a certain behavioural standard which offers cost-effectiveness as opposed to providing personcentred care and tailored services. The same method could be applied to condition people to eat healthy or exercise in order for cost-saving purposes. Conditioning has already been recognised as a risk in AgeTech (Hummel & Braun, 2020;Manzeschke et al., 2016;Mortenson et al., 2015;Petrakaki et al., 2018;Rubeis, 2020). What is seldom discussed is the connection between epistemological and ethical aspects in this context (Morley & Floridi, 2020). According to this view, the problem resides in the underlying assumption that objective parameters can be defined for appropriate or desirable emotions and behavior. There is a certain practical and ethical risk of creating parameters deliberatively designed in order to produce a specific behavior -a type of nudging (Thanler & Sunstein, 2008) that contradicts to the goal of person-centred care. An equity-focused approach could minimise the risks of standardization by taking the diversity of user preferences, needs, and resources into account, thus providing a more balanced and person-centered baseline for technology design. --- Complexity Management and Access Another ethical quandary of complexity management is that it manifests itself by way of optimal AI design with limited consideration for issues relating to access i.e., lack of cultural appropriateness. "Culture" in this context may refer to certain codes, concepts, and behaviors that stem from the ethnic identity of older adults. It may, however, also be understood in terms of the life experience of certain generations. As such, the way in which individuals have been exposed to technology across their lifespan influences their attitudes and behaviour towards AgeTech in old age. Tech-1 3 nology that is designed without consideration for cultural appropriateness can mean that they are more difficult to access resulting in reduced uptake. Aligned with the issue of access is usability, and links to cultural appropriateness. For example, digital literacy, more precise the lack of it, can be an access barrier when it comes to AI-based AgeTech. This is especially the case with technologies that require direct interaction, like human-machine interfaces (HCI) or technologies for self-management and self-monitoring common in mHealth and telehealth (Fang et al., 2018). The lack of experience with digital technologies or affinity towards their use may thus prevent older adults from accessing them (Fang et al., 2019). Empirical evidence shows that digital literacy and health outcomes correlate with education, age, gender, and socioeconomic status (Ang et al., 2021). The need for complexity management in AI-based AgeTech also affects access to AgeTech. One important factor in this regard is diversity within the older population (Fang et al., 2018;Haufe et al., 2019). Older adults vary not only in health status, ethnicity, and socioeconomic status, but also in health literacy and affinity to technology. These factors often interact and create a dynamic that may prevent access to AgeTech. As a result, the most complex outcomes of digital health disparities arise across social intersections, e.g., between age, gender, ethnicity, and socioeconomic status (Fang et al., 2019). Studies show that people over 65 have generally less desire and intent to use ICTs (Fang et al., 2018 b). Especially older adults with less exposure to technology throughout their work life show a lower uptake of digital health technologies (Fang et al., 2019). Other reasons that impact technology use are health-related. Co-morbidities and functional disabilities, e.g., vision impairment or difficulties in memorising passwords, are barriers to digital access in this regard (Chen & Chan, 2013;Fang et al., 2018). This is especially the case when interfaces and systems lack the appropriate level of user-friendliness (Haufe et al., 2019). AgeTech access and uptake is further complicated when gender is added i.e., as another layer of social complexity. In general, women use ICTs less than men, but more often for health purposes than men. Education, socioeconomic status and age are associated in this regard, since younger, more educated women are more likely to use ehealth than older, less educated men with low incomes (Fang et al., 2018). Last, individual beliefs, attitudes, and fears of technology are associated with age and generation and often play a role in preventing older adults from using AgeTech (Chen & Chan, 2013;Fang et al., 2018;Haufe et al., 2019). Older adults often consider themselves as incapable to handle the technologies due to the aforementioned factors and their perceived age or lack of digital competencies. As well, internalised-stigma including negative self-perceptions of being frail and in need of help when using AgeTech hinders technology use (Haufe et al., 2019). Such beliefs and attitudes may be the result of lack of experience with technologies, educational status, with the root of such perceptions shaped by intersectional social factors such as ethnicity, gender, and socioeconomic status. As a consequence, the interplay of all of these technology deterrents creates a situation where individuals who can benefit the most from AgeTech are the same that people who experience the greatest access barriers (Fang et al., 2018a). A key explanation for this may be that the combined effect of social determinants is overlooked 1 3 when designing AgeTech. Thus, a focus on equity already in the design process is crucial. --- Discussion and Recommendations Although complexity management as an explicit concept has not been widely discussed so far, several strategies have been suggested for dealing with its outcomes. We will provide a short overview of these strategies, based on the main currents of the ethics of AI-based health technologies in the research literature. These strategies are often labelled as democratisation of AI-based health technologies and consist of at least three objectives (1) diversifying training data, models, and algorithms, (2) engaging relevant stakeholders as well as communities in design and implementation of AI-based health technologies, and (3) granting access to medical to all social groups and on a global scale. --- Diversifying Training Data, Models, and Algorithms (1) Discrimination and bias in AI-based AgeTech may be a result of ignoring social determinants when selecting training data or designing models and algorithms. This issue is mainly discussed under the header of algorithmic fairness (Wawira Gichoya et al., 2021). One way to address this issue is to question existing proxies and biomarkers and to include social determinants (Walsh et al., 2020) -meaning that the socio-demographic contexts in which data has been collected must be considered. Furthermore, various social determinants should be actively integrated into the data procession process. This means that when selecting data sets for training, there needs to be assurance that the data does not focus exclusively on one social group. The epistemic scope and limits have to be evaluated based on the quality of the data in this respect (McCradden et al., 2020). If an algorithm performs poorly for specific groups, additional data from these groups should be collected and induced into the machine learning process (Walsh et al., 2020). This implies a process of model auditing throughout the design process that focusses on the reliability and validity of models as well as the assessment of confounding errors (McCradden et al., 2020). Also, upon implementation of these technologies, a local evaluation should be conducted in order to investigate hidden stratification effects (McCradden et al., 2020). A key takeaway message is that fairness should not be an afterthought or post-hoc consideration (Wawira Gichoya et al., 2021). Rather, fairness in terms of diversifying training data, models, and algorithms should be operationalised through model reporting guidelines, clinical trial guidelines, and regulatory approaches. This is an issue for AI generally and is not limited to older populations. However, the complex intersectionality of factors contributing to aging, the prevalence of ageist attitudes, and indeed increasing heterogeneity in later life create additional challenges. --- 3 --- Engaging Relevant Stakeholders and Communities (2) Participatory methods for designing AgeTech are widely seen as ways to prevent discrimination and enable equity (López Gómez & Criado, 2021). Engaging relevant stakeholders and communities may be a crucial measure throughout the life cycle of an AI-based AgeTech product. At the research stage, the perspectives of diverse end users may be required for inclusivity training data as well as user-centered technologies (Fohner et al., 2019). By using a community-engaged approach, educational aspects as well as shared decision-making are emphasised, and mutual benefit may be achieved. This approach may also foster trust in technologies, which is a crucial aspect as it pertains to the acceptability of AgeTech (Walsh et al., 2020). Using a community-engaged approach may also help to situate AI-based AgeTech in view of context-specific healthcare infrastructures and communities. Consequently, technologies tailored for the specific requirements of a given context as opposed to a one-size-fits-all-approach can mean that caregivers and care receivers are potentially more likely to use them. (Fohner et al., 2019). However, this approach has some drawbacks. Engaging stakeholders and communities can be a challenging and often very time-consuming approach, which also means more cost-intensive. Furthermore, participatory processes need a moderation in order to integrate and balance different expectations (Merkel & Kucharski, 2019). As with AgeTech itself, there is also no one-size-fits-all-approach regarding stakeholder and community engagement. In order to choose the right approach or method, it has to be clarified why a stakeholderand community-engaging approach should be applied, what individuals or groups as well as future users will be included and at what stage of the lifecycle of the product they will be involved (Merkel & Kucharski, 2019). Another difficulty here is how to best involve older adults who are isolated, hard to reach, and who may be non-tech users in the co-development process. --- Use of Personas for User-Centered Design (2) Developing a 'persona' and'scenario' -a technique often used in technology design to ensure that the tech product is conducive to the end-user. A 'persona' is a description of a fictitious individual based on data or information from real people (Adlin & Pruitt, 2010), while scenarios provide context of the persona, which include stories of personal experience -a setting or situation in detail which a person performs a sequence of actions (possibly involving other people) to produce an outcome. In technology design, developing personas and scenarios can be viewed as a tool for considering how products can be delivered in an ethically, socially-responsible and culturally-sensitive way. The use of personas and scenarios are intended to promote our empathy with the people who we aim to serve (Jackson & Hwang, 2020). While each of us come from different backgrounds, personas can help tap into our innate human tendency to generate detailed and complex models of people and their behaviours, even when those people are fictitious (Jackson & Hwang, 2020). This is demonstrated by the ways in which we naturally try to relate to or develop understandings about fictional characters in stories or films. Using personas in teaching can help tap into this natural human aptitude that we all possess (Jackson & Hwang, 2020). --- 3 --- Ethical and Responsible AgeTech (3) Previous research has also established that factors influencing digital marginalisation are multi-faceted and that an intersectional approach needs to be adopted to fully understand how people can be disadvantaged by the increasing reliance on AgeTech (Fang et al., 2019); and in the context of AI, identify and understand if and in what ways this technology can be created with the intent to ameliorate traditional forms of marginalisation (and for whom). There are various policy pathways considered to help to ensure AgeTech access, equity and other social determinant barriers. Stix's (2021, p. 15) notion of actionable principles for AI policy highlights how "successful actionability in policy" requires going beyond AI-based AgeTech ethics principles as a reference point, and this can be achieved, in part, by referencing transdisciplinary theoretical perspectives from disciplines of gender studies (i.e., intersectionality), urban studies (i.e., sense of place), and health sciences (i.e., digital social determinants of health). Critical analysis of existing AI design and policy by viewing the problem area through diverse theoretical viewpoints can help shed light on how we can better understand and respond to the various ways in which social positioning create distinct, and often multiple, barriers for various subgroups. According, the three AI policy pathways presented by Stix ( 2021): (1) preliminary landscape assessments; (2) multi-stakeholder participation and cross-sectoral feedback; and, (3) mechanisms to support implementation and operationalisability can be supported by integrating equity-driven theoretical models and frameworks such as the Social Justice Framework for Bridging the Digital Divide, Health Equity Impact Assessment (HEIA) tool, and the Intersectional Place Perspective for AgeTech solutions (Fang et al., 2019(Fang et al.,, 2020;;Ontario Ministry of Health and Long-Term Care & Toronto Central LHIN, 2009). HEIA originated from the Health Impact Assessment methodology and has been heavily used worldwide over the past decade as a decision-making tool to facilitate the development of healthy public policy (Ontario Ministry of Health and Long-Term Care & Toronto Central LHIN, 2009). HEIA can be applied to describe the individuals and groups most affected by the progression of AI in healthcare, characterise the ways in which possible inequities can occur to support the responsible design of AI interventions and initiatives to optimise equitable access, positive ageing outcomes whilst identifying factors that may unintentionally exacerbate experiences of vulnerability and disadvantage. The Social Justice Framework for Bridging the Digital Divide framework stems from a realist review and affirms that individuals exist within structures and systems designed by and for persons in more advantageous social positions, which creates modes of differentiation across groups and divisive access to digital resources (Fang et al., 2019). This framework may be particularly useful for recognizing and responding to the multiple layers of access and use inequities that older people might experience (Sixsmith et al., 2019), when introduced to AI interventions. Last, the Intersectional Place Perspective designed to identify individual, social and place-based factors that shape opportunity and oppression, has been used to better understand the combined effects of the digital determinants of health and wellbeing (Fang et al., 2020(Fang et al.,, 2021)). This theoretical model can help to characterise the ways 1 3 in which AI can compound or alleviate inequity, through consideration of socio-cultural and environmental contextual factors that shape lifetime health and wellbeing outcomes of older people -recognising the importance of intersectionality and place. Overall, such equity-driven resources can support AI developers and shape policy pathways by providing a spotlight on how digital social determinants are operationalised in real life scenarios, which can help to ameliorate inequities associated with AI design and rollout how these might be produced and utilised across different contexts, at scale and in an ethical and sustainable way. --- Granting Access (3) As we have seen, access barriers may arise when AI-based AgeTech does not fit the needs and resources of users. One strategy for fostering access is universal design, sometimes referred to as inclusive design or design for all (Ma et al., 2021). The aim is to reduce access barriers by designing technologies that are simple and intuitive and allow flexible use with a certain tolerance for error. This requires a well-informed design process that includes user experiences from various user groups. Another factor preventing certain user groups from access is contextual bias (Weissglass, 2021). Most AI-based AgeTech is designed in high-income countries and adapted to the structures as well as contexts of use within their respective health systems. The resulting technologies might not be fitting for the systemic or institutional conditions in low-income countries. Disclosure and validation policies might be instruments for ensuring that the context of technology design is made transparent (Weissglass, 2021). However, also in this regard, a more inclusive design process that takes diversity on a global scale into account would be needed. Another approach is to make algorithms as well as data-bases consisting of training data accessible on a global scale. This would provide researchers, clinicians, and developers from lower-income countries the opportunity to adapt technologies to their own respective contexts of use. Finally, empowering e-health literacy of older adults may reduce the digital gap, which is one of the main access barriers (Seifert et al., 2019). E-Health literacy could be improved by providing learning tools, e.g., through existing educational services. This would mean to acknowledge that older adults are not inherently technologyadverse and have both the resources and the willingness to learn, which is often ignored due to stereotypes about older adults. --- Conclusion AI-based AgeTech has the potential to support older adults towards living a more active, independent, and healthier life. In facilitating more personalised healthcare services, AgeTech may potentially be used not only for the purpose of dealing with deficits, but rather to improve the quality of life of users. This potential, however, can only be realised in full if equity is seen as major principle for the design, implementation, and use of AI-based AgeTech. Equity is neither an add-on nor can it be an afterthe-fact consideration. Rather, equity should be the guiding principle in developing AI-based AgeTech at the outset and should also guide its implementation as well as 1 3 the practices it enables. This requires a well-informed design process that takes the combined effect of social determinants into account, provides accompanying measures for educating users as well as caregivers, and defines regulatory approaches that address the issues of bias and discrimination. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. --- Authors and Affiliations Giovanni Rubeis 1 • Mei Lan Fang 1,2 • Andrew Sixsmith 1,3 Giovanni Rubeis

AgeTech involves the use of emerging technologies to support the health, wellbeing and independent living of older adults. In this paper we focus on how Ag-eTech based on artificial intelligence (AI) may better support older adults to remain in their own living environment for longer, provide social connectedness, support wellbeing and mental health, and enable social participation. In order to assess and better understand the positive as well as negative outcomes of AI-based AgeTech, a critical analysis of ethical design, digital equity, and policy pathways is required. A crucial question is how AI-based AgeTech may drive practical, equitable, and inclusive multilevel solutions to support healthy, active ageing. In our paper, we aim to show that a focus on equity is key for AI-based AgeTech if it is to realize its full potential. We propose that equity should not just be an extra benefit or minimum requirement, but the explicit aim of designing AI-based health tech. This means that social determinants that affect the use of or access to these technologies have to be addressed. We will explore how complexity management as a crucial element of AI-based AgeTech may potentially create and exacerbate social inequities by marginalising or ignoring social determinants. We identify bias, standardization, and access as main ethical issues in this context and subsequently, make recommendations as to how inequities that stem form AI-based AgeTech can be addressed.

the potential to support older adults towards living a more active, independent, and healthier life. In facilitating more personalised healthcare services, AgeTech may potentially be used not only for the purpose of dealing with deficits, but rather to improve the quality of life of users. This potential, however, can only be realised in full if equity is seen as major principle for the design, implementation, and use of AI-based AgeTech. Equity is neither an add-on nor can it be an afterthe-fact consideration. Rather, equity should be the guiding principle in developing AI-based AgeTech at the outset and should also guide its implementation as well as 1 3 the practices it enables. This requires a well-informed design process that takes the combined effect of social determinants into account, provides accompanying measures for educating users as well as caregivers, and defines regulatory approaches that address the issues of bias and discrimination. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. --- Authors and Affiliations Giovanni Rubeis 1 • Mei Lan Fang 1,2 • Andrew Sixsmith 1,3 Giovanni Rubeis [email protected] Mei Lan Fang [email protected] Andrew Sixsmith [email protected]

AgeTech involves the use of emerging technologies to support the health, wellbeing and independent living of older adults. In this paper we focus on how Ag-eTech based on artificial intelligence (AI) may better support older adults to remain in their own living environment for longer, provide social connectedness, support wellbeing and mental health, and enable social participation. In order to assess and better understand the positive as well as negative outcomes of AI-based AgeTech, a critical analysis of ethical design, digital equity, and policy pathways is required. A crucial question is how AI-based AgeTech may drive practical, equitable, and inclusive multilevel solutions to support healthy, active ageing. In our paper, we aim to show that a focus on equity is key for AI-based AgeTech if it is to realize its full potential. We propose that equity should not just be an extra benefit or minimum requirement, but the explicit aim of designing AI-based health tech. This means that social determinants that affect the use of or access to these technologies have to be addressed. We will explore how complexity management as a crucial element of AI-based AgeTech may potentially create and exacerbate social inequities by marginalising or ignoring social determinants. We identify bias, standardization, and access as main ethical issues in this context and subsequently, make recommendations as to how inequities that stem form AI-based AgeTech can be addressed.

Introduction and background The purpose of this qualitative hermeneutic phenomenological study is to describe the meaning that members of the Parys community ascribe to their experiences of a Dalcrozeinspired workshop. In this project, members of the Research Niche for the Cultural Dynamics of Water (CuDyWat) at North-West University applied Dalcroze-inspired activities to create a sense of social cohesion among a diverse and randomly selected group of residents and officials of Parys. The workshop enabled them to use music and movement to express their emotions and opinions about their communal water service delivery problems. The research question that guided this inquiry was: What meanings do participants ascribe to their experiences of the Dalcroze-inspired workshop? The problem is that water resources in South Africa are running dry. There are growing concerns that South Africa, as among the 30 most arid countries in the world, with an average rainfall of less than 600 mm, may not be able to cope with erratic drought conditions, because of anticipated climate change conditions. The latest version of Department of Water and Sanitation (2021) National Water Resources Strategy (Version 3) accentuates the need for local authorities to apply water conservation demand management strategies. All municipal, industrial, and commercial water consumers need to contribute to reducing water loss as far as possible. The country' s municipal water and sanitation infrastructure systems are notorious for leaks and the wastage of costly potable water resources. The DWS now plans to actively promote water use conservation strategies (Department of Water and Sanitation, 2021;Mchunu, 2022). The town of Parys, which falls under the Ngwathe Local Municipality and operates under the jurisdiction of the Fezile Dabi District Municipality, is frequently subjected to severe water shortages. The town's water purification infrastructure cannot consistently cope with the inferior quality of raw water at its point of intake on the Vaal River. At the same time, there is an increasing demand for more municipal water due to local population growth. The town's overworked water and wastewater infrastructure systems are also subject to deterioration. Since the early 2000s, Parys has experienced extensive periods of water-quality issues and infrastructure collapses. In response to a request by a former premier of the Free State Province, the Sasolburg-based company, Sasol Limited, funded a multi-stakeholder consultation research project by the CuDyWat Research Niche at North-West University. The research team focused on the analysis of the impact that the dire water crisis in Parys had on local residents. Since people use musical experiences to create meaning during difficult times (Bonde, 2011), the research team decided to use a Dalcrozeinspired approach to engage with local stakeholders. The use of Dalcrozeinspired activities in this study is informed by previous research studies (Dutton, 2015;Habron, 2016) that highlight the positive influence of Dalcroze in promoting social transformation within individuals, local communities and the environment. These activities cultivate a desire for self-expression and communication, promoting interpersonal connections and a sense of belonging and well-being (Navarro Wagner, 2016;Sutela et al., 2016). Furthermore, our study draws on the findings from a previous study conducted in the town of Brandfort (Van der Merwe et al., 2019), which demonstrated the significant contribution of a Dalcroze-inspired workshop during a communal water crisis. The findings further revealed that the implementation of the Dalcroze approach effectively fostered a feeling of connectedness and a shared sense of purpose among the participants by making them feel welcome; promoting active engagement; becoming aware of themselves, their space and each other; promoting cooperation and playfulness; experiencing enjoyment; and enabling emotional expression concerning their communal crises. The findings of these studies motivated the researchers to employ the Dalcroze-inspired approach again, this time in a different context, Parys. This was an ideal opportunity to use Dalcroze-inspired activities as a basis for constructive communication and to interpret the perceptions of municipal water users. --- Music, health, and well-being The World Health Organization describes health and well-being as a "state of complete physical, mental and social well-being and not merely the absence of disease or infirmity" (Grad, 2002, p. 981). The positive effects of the arts on health and well-being have been studied extensively. Findings have shown that the arts can affect physical and mental health by preventing poor health, promoting good health, and managing and treating various illnesses (Fancourt and Finn, 2019). The growing body of literature on the transformative value and benefits of music for health and well-being illuminates the effective contribution of music participation to the quality of life of a community (Dillon, 2006;Hallam et al., 2012;Lenette and Procopis, 2016). MacDonald et al. (2012) developed a conceptual framework illustrating that the growing multidisciplinary interest in the relationship between music, health, and well-being could enhance musical activities' social, therapeutic, and communicative benefits. Paton (2011) states that music has a supportive function in enhancing a community's emotional, mental, physical, and social well-being that goes beyond purely musical entertainment or performance. Participatory music-making serves various social functions in the community, such as involving community members in social communication and activities, and building trust and unity among individuals in the group (Garrido et al., 2016). Music does not only play a sociocultural role but also has a significant impact on quality of life as it is "an inherently social act and one which contains enormous potential to bring people together and to facilitate various forms of social action" (Murray and Lamont, 2012, p. 76). Music making promotes the community's wellness by having a calming influence on community members in despair and discomfort; it can be used to inform community members of important health and welfare matters, and music making "can lighten the load of the suffering" (Reigersberg, 2017, p. 134). Small (1998, p. 2) uses the term "musicking" to illustrate that music entails activity and adds that meaning is generated through social and cultural interactions. Stige (2003) views health musicking as an interdisciplinary field that promotes well-being through musical participation. Musical activities in health musicking create meaning in times of hardship and teach strategies to cope with different social problems (Bonde, 2011). A health-musicking perspective supports Dalcroze-inspired activities since the emphasis is on the inter-and intra-personal relationships which develop through musicking to promote health and well-being (Navarro Wagner, 2016). DeNora (2013, p. 1) developed "a grounded theoretical account of how music can be understood to create conditions conducive to wellbeing. " The Dalcroze-inspired workshop for the community members of Parys relates to her theory in so far as the musicking provided momentary respite from distress and afforded participants the opportunity to renew their environments. --- The Dalcroze approach Émile Jaques-Dalcroze (1865-1950) was a Swiss composer, pianist, and music educator interested in training the body to facilitate inner hearing, musical thinking, learning, and understanding. He was appointed as a professor of solfège1 and harmony at the Conservatory of Geneva from 1892 to 1910 (Choksy et al., 1986). After the First World War, Jaques-Dalcroze was concerned with social reform and cohesion by encouraging students to live in harmony with themselves and those around them. He envisaged the musical expression of human emotion and created an approach that included three branches, namely eurhythmics, solfège, and improvisation; of these, eurhythmics proved to be his unique contribution (Jaques-Dalcroze, 1919). The distinctive Dalcrozian approach to musical learning involves rhythmic bodily movements in response to the elements of music. As a result, scholars develop a bodily awareness awakened by active listening and inner hearing (Le Collège de l'institut Jaques-Dalcroze, 2019). In a Dalcroze class, students explores aspects of time, space and energy through locomotor and non-locomotor body movements. Movement is the link between the ear and brain, leading the student to an embodied and deeply internalized understanding of music (Le Collège de l'institut Jaques-Dalcroze, 2019). In the Dalcroze approach, the muscular and nervous systems develop simultaneously with the body's natural rhythms, stimulated by the auditory and visual imaginations of the musical mind. This approach is explained by Juntunen (2004, p. 68) as follows: "Dalcroze Eurhythmics primarily teaches habits of musical action or, more generally, 'a bodily way of being in sound', rather than conceptual or abstract knowledge of music. " Dalcroze envisioned making the whole human being more aware, receptive, and imaginative. This approach has spiritual, holistic, and intra-disciplinary dimensions (Habron and Van der Merwe, 2017) and is usually implemented in a group so that learners respond to one another, learn from each other, become aware of the group, and practice mutual respect (Le Collège de l'institut Jaques-Dalcroze, 2019). In the hands of a caring facilitator, the atmosphere of a Dalcroze class is a safe environment with an active, alert, and reflective approach, and participants learn by "doing" and are encouraged to take risks and develop inter-and intra-personal skills, as well as musical skills (Le Collège de l'institut Jaques-Dalcroze, 2019). In the process, they become more aware of themselves as human beings intersecting in a social ecology where music plays a formidable role in helping them articulate-through movement-how they feel and express what they experience subconsciously. Therefore, Dalcrozeinspired activities were selected as the most appropriate collaborative music-making experience since humans construct meaning from bodily experiences. The research group was particularly interested in understanding the meaning participants ascribed to their experiences of a Dalcroze-inspired workshop. By using music and movement to improve communication between all relevant stakeholders in Parys who wished to engage with the research group on the question of water-related problems, we wanted to find solutions to this critical issue in the community. In the following sections, we first discuss the research procedures, the Dalcroze-inspired workshop, and the types of activities used. Secondly, in the findings section, we interpret the lived experiences of those who participated in the Dalcroze-inspired workshop and their views on local water-delivery problems. Thirdly we discuss the findings in the context of the relevant scholarly literature. --- Procedures The best research approach to answer the research question of this study was hermeneutic phenomenology since we wanted to describe the experiences "together with its meanings" of those involved (Hendriksson and Friesen, 2012, p. 1). With this hermeneutic phenomenological inquiry, we intended to describe the experiences "all participants have in common" (Creswell and Poth, 2018, p. 75) during the Dalcroze-inspired workshop. A one-day workshop was held at Stonehenge on Vaal, a conference venue outside Parys, to ascertain first-hand views and opinions about the water situation from various stakeholders affected by this problem. The research team chose the venue because of its location just outside Parys on the riverbank. Stonehenge has two sizeable halls, where the Dalcroze music and movement session and subsequent discussions were held. The research team met there the day before the workshop to make final preparations for the next day. Invitations to participate in the workshop were randomly distributed to members of the Parys community and to specific key stakeholders identified in previous studies in Parys. The invitations were communicated verbally and via SMS, email, and fax. The verbal invitations were distributed by members of the research team, who went to different areas in Parys to alert stakeholders about the workshop. The municipal management and workers, the Ngwathe Water Forum members, and community members from Tumahole, Schonkenville, and the Parys town area were invited. The invitation indicated that the stakeholders were invited to participate in a music and movement workshop dealing with waterrelated challenges facing their community. It also explained that North-West University's Research Niche for the Cultural Dynamics of Water (CuDyWat) would facilitate the session. Stakeholders were also notified that the transdisciplinary research group was collaborating with Sasol Limited, a partner in a cooperation agreement with the Free State provincial government. It mentioned that the Free State Premier had asked for the assistance of Sasol Limited to investigate the water contamination and shortage challenges in the Ngwathe Municipality. The initial invitations were sent a week before the workshop, up to a day before the event, but did not specify the venue where the workshop would take place. As a precautionary measure to prevent potential disruptions of the event, the venue was only announced 48 h before the workshop. --- Participants Fifty-three stakeholders from Parys and its neighborhoods and eight researchers participated in the event. The transdisciplinary research team included a biochemist, banker, chemist, political scientist, two musicians, environmental engineer, and two historians. Each one was also encouraged to describe the meaning they ascribed to the Dalcroze-inspired workshop and provide their perspectives on solutions to the water problem. Therefore, there were 62 participants in this study. Two musicians organized the Dalcroze-inspired workshop in collaboration with the researchers and rehearsed a possible version of the workshop with the researchers the day before. The Parys workshop commenced with a two-hour Dalcroze-inspired session. The purpose of this workshop was to give an opportunity for the expression of emotions about the water challenges through music and movement and to promote creativity, communication, and problem-solving. The Dalcroze-inspired workshop generated a unique opportunity for the diverse participants representing different stakeholder sectors in the community of Parys to interact with each other. Frontiers in Psychology 04 frontiersin.org Following the Dalcroze-inspired session, the water meeting focused on participants sharing their encounters regarding the water dilemma. Participants were then asked what action they would take to address the water situation if they were the municipal manager of Parys. The positive emotions generated by the Dalcroze approach broadened their minds (Fredrickson, 2013) and facilitated creative problem-solving. Their suggested solutions were included in an unpublished research report conducted for Sasol Limited on behalf of the office of the Free State Premier. Lastly, the participants were asked what meaning they ascribed to the Dalcroze-inspired workshop. Although some members were initially hesitant when invited to participate in a Dalcroze-inspired movement-to-music session, they were fully prepared to share their deepest feelings with us openly and transparently. The group developed a strong sense of togetherness, and participants could exchange their opinions. This article reports all the findings related to the Dalcroze-inspired workshop. --- Ethics statement A pre-defined, ethically approved framework guided the research team's actions. Significant effort ensured an inclusive communication platform between the stakeholders and researchers. Informed consent was obtained from each participant before the workshop to ensure we could use the data for research. Furthermore, pseudonyms were used to ensure participant anonymity and confidentiality. A feedback session with the stakeholders took place six weeks after the workshop to give them an opportunity to give input. --- Data collection The data-collection process continued during the workshop. While the Dalcroze-inspired workshop was in progress, data were captured of the group in action using photos, videos, sound recordings, and active information gathering by means of notes taken by members of the research team. Data were also collected through focus group interviews and open-ended individual interviews. The objective was to use diverse data collection strategies to glean relevant information on the perceptions and opinions of community members. Immediately after the Dalcroze-inspired workshop, when the focus group interviews were conducted, participants were specifically asked: What meaning do you ascribe to your experiences of the Dalcrozeinspired workshop? --- Data analysis All data generated during the workshop were qualitatively interpreted as the data were coded, categorized and thematised. The textual data were organized and coded in one heuristic unit in ATLAS. ti, a computer-assisted qualitative data analysis software program (Friese, 2014). ATLAS.ti is a useful tool to organize, manage and support the process of qualitative data analysis. Using the NCT (noticing, collecting, and thinking about things) method (Friese, 2014), data were analyzed and interpreted by moving back and forth between these three iterative steps. Interesting aspects observed in the data were collected and collated in the process of constant comparison (Butler-Kisber, 2018). Throughout the process, we identified patterns and links in the data. This approach to data analysis is illustrated in, among other things, the network views. The data analysis took place in two phases. First, there was descriptive-level open coding (Salda<unk>a, 2013). Later we entered the next conceptual phase in the data-analysis process, and patterns and links were identified. The themes were subdivided into seven "problem" themes, five "solution" themes, and five "music experience" themes. This article focuses only on the "music experience" themes, namely the meaning participants ascribed to the Dalcroze-inspired workshop and does not elaborate on the water delivery problems and solutions that the stakeholders identified because of the Dalcroze-inspired workshop. The water delivery problems and suggested solutions expressed during the water meetings that followed the Dalcroze-inspired workshop, were included in a report to the Free State Provincial Premier. --- Activities in the Dalcroze-inspired workshop In a situation where conflict resolution is necessary, the appropriate choice of music and activities is crucial. The workshop aimed to encourage communication between community members by creating an opportunity for them to express their emotions through music and movement. Although the researchers made suggestions for music that they believed could inspire movement, the first author guided the choice of music and activities with reference to her knowledge of the community and its context. As this was mainly an intuitive and spiritual process for the music facilitator, she walked to the music in her living room to get a feel for the music and asked herself: "Will this music make things better?" The topics of the music were carefully chosen to facilitate reconciliation, such as Reconciliation Ballet by Rachel Portman, You and Your Crown by Matthew Mole, Baby can I Hold you by Tracy Chapman, Simple Gifts by Yo-Yo Ma and Alison Krauss, and others. Each activity was built sequentially from the previous activity, and each activity had a social intention, such as making eye contact, being aware of other people, leading and following, synchronizing the beat, making physical contact, celebrating together, giving recognition to the other person, communicating to whom you are going to pass the beat, being expressive and creative together, being sensitive to another person etc. For each activity, clear instructions were given, for example: "Walk with someone you have never met before. Change directions together only by being sensitive to the other person (no talking). " Another example of an instruction for an activity was: "Pair up with someone you do not know. Mirror that person's movement. Take turns to mirror and be mirroredtake turns to lead and follow. " These are just a few examples. Our approach relates to Higgins, 2012 third perspective of community music, namely an intentional intervention emphasizing people, participation, diversity and accessibility. We recognized "the value of music to foster intercultural acceptance and understanding" (Higgins, 2012, p. 5). Our choice of musicking was Dalcroze-inspired activities since we have succeeded with this approach in similar community contexts. We acknowledge that other community music approaches that require active engagement in interesting, playful, musical movement activities could have similar findings. In designing the workshop, the first author was also informed in her choices by the seven types of touch promoted by Greenhead and Van der Merwe et al. 10.3389/fpsyg.2023.1176691 Frontiers in Psychology 05 frontiersin.org Habron (2015). She agrees with their statement that "the touch-like nature of sound not only makes contact with the body, inciting physical and emotional movement but also develops awareness of self, others and environment due to the social nature of musical participation in general" (p. 93). They further state that touch activities are suitable to "communicate intentions and feelings to others and receive their responses" (p. 101). Therefore, five of the seven types of touch activities were used as well as one additional category, namely eye contact and imagined contact. Massage and therapeutic touch were not included since sensitive cultural and gender issues were at stake. Recorded music was used since there was no piano available, and the participants related better to more familiar music. The six types of touch/contact included were: 1. Eye contact and imagined contact, greeting with eye contact and throwing an imaginary ball to each other. 2. Direct physical contact with others by clapping each other's hands. 3. Self-touch: body percussion. 4. Touch and play easy percussion instruments: sound shapes, rattles, bamboo claves and rattles. 5. The touch and manipulation of materials: using balls, ropes, scarves, and other equipment. One group event involved using a large elastic band to connect participants. 6. To touch or be connected using an object such as a rope, scarf or elastic band. --- Findings The Dalcroze-inspired approach to the water workshop helped people get to know one another, think in a new way about an old problem, put conflict behind them, and, importantly, work on creative solutions. The findings in this article draw on an unpublished research report. In addition, the findings related to the Dalcroze experiences were expanded on, interpreted, and interrogated in the context of the relevant scholarly literature. --- Suffering It became clear from the people of Ngwathe's stories that they suffered tremendously (Figure 1). This suffering is caused by the health risks posed by dirty water. People who do not have money to buy treated water end up drinking the dirty water and become ill. Old and ill people do not even have water to drink with their medicine. In many cases, there is also no water supply, which leads to vulnerable people having to fetch water for themselves. The water problem jeopardizes children's education. They spend time fetching water when they are supposed to be learning. People also experience embarrassment and frustration with the unfairly and unevenly distributed water. Most participants stated simply and powerfully: "We need water. " --- Emergent themes With the use of the ATLAS.ti qualitative data analysis software, we identified the following five themes from the interviews, focus group interviews, photos, videos, and observations during and after the Dalcroze-inspired workshop (Figure 2): 1. Joyful experiences when participating in Dalcroze activities. 2. Dalcroze activities facilitate social interaction. 3. The Dalcroze approach transforms relationships. 4. Dalcroze-inspired activities foster virtues. 5. Music and movement support well-being. --- FIGURE 1 The suffering that the people of Ngwathe endure. The ATLAS.ti networks present the findings visually to illustrate the relationships between codes, categories, and themes. The first number after a code refers to the groundedness, the number of quotes associated with the category. The second number reflects the density, the number of links to other codes (Friese, 2014). Van der Merwe et al. 10.3389/fpsyg.2023.1176691 Frontiers in Psychology 06 frontiersin.org Theme 1: Joyful experiences and emerging categories. --- Theme 1: Joyful experiences when participating in Dalcroze activities Participants had a joyful experience participating in Dalcrozeinspired activities (Figure 3). They enjoyed it because they found the movement invigorating and exciting. They also enjoyed it because they found the music interesting and the movement activities entertaining. The joy they experienced was expressed as fun, happiness, and a good experience. One of the participants joyfully expressed how the Dalcrozeinspired activities broadened their outlook. The social component of the embodied approach may be the most important reason participants find it enjoyable, which brings us to our next theme, social interaction. --- Theme 2: Dalcroze activities facilitate social interaction When people are actively engaged, Dalcroze-inspired activities enable social interaction (Figure 4) because they enable communication. Participants moved in synchrony and could communicate better because of music' s expressive capabilities. One female participant expressed it beautifully: Yes, and then it leads us nicely into the session, and you know, I think it just oiled the whole meeting. Even the other person said it stopped us from pointing fingers. It helped us to relax and talk from the heart. Five emergent themes on the meaning participants ascribed to their experiences of the Dalcroze-inspired workshop. Van der Merwe et al. 10.3389/fpsyg.2023.1176691 Frontiers in Psychology 07 frontiersin.org Participants explored their roles in the group. They took turns leading and following. This made them realize that they can depend on each other in this workshop and the community when trying to solve the water crises. Their active engagement in the workshop was evidence of their commitment to addressing the water problems of Ngwathe. This social interaction helped relationships to transform. --- Theme 3: The Dalcroze approach transforms relationships One of the participants, a representative of the water forum, had instituted a lawsuit against members of the municipality. These members from the municipality were also present at the workshop. The highlight of the meeting was when this participant said: "Uhm, before I went in there, I had three enemies. We are now friends. It was fantastic!" This transformation process (Figure 5) was possible because the music "opens up the boundaries." Because this was such an interesting new concept for the participants, it opened their minds. Participants got to know each other more quickly. "I've never made friends so quickly; it is like quick you meet someone, then you greet them here, then you greet them at another corner, and your mind recognizes, and they also recognize you." Through the playful interaction, people connected. Music and movement facilitated social integration across ages and cultures, as one participant explained:... you could even see the different race groups... it was almost like...I do not know... There was no hierarchy, there was no color, there was also no gender. Everybody was trying, following their leader, doing whatever is being done. I was in awe.... It really broke down those barriers. Not only does the process help with conflict resolution, but it also helps to build friendships. This transformation process brings out character strengths in people. --- Theme 4: Dalcroze-inspired activities foster virtues The Dalcroze-inspired activities brought out an uninhibited spontaneity and honesty in participants (Figure 6). "I'm happy. I'm myself. I'm willing to engage, and there is nothing for me to hide from you or anything of that sort. " People felt cared for: "I am very happy there are people concerned about the water problem. " The safe space that was created supported people's experience of well-being. The photo and video data showed that participants were helping each other during the Dalcroze-inspired activities. They worked together to create water wave movement improvisations. Furthermore, participants engaged in giving and taking activities. For example, they moved a large elastic band and handed it over to someone else without the elastic snapping. The beat was passed around; scarves were thrown to each other, and balls were rolled to each other. Participants also took turns leading a movement or following someone else's lead. They gave recognition to each other's movements in a mirror activity. Recurring quotes after the workshop included "We can depend on each other" and "We should help each other. " --- Theme 5: Music and movement support well-being Participants felt welcome at this workshop. Through music and movement, they could express their emotions and sense the music in their bodies. This promoted relaxation and was a source of Theme 2: Social interaction and emergent categories. Van der Merwe et al. 10.3389/fpsyg.2023.1176691 Frontiers in Psychology 08 frontiersin.org stress relief. "It was powerful. Music takes the mind far away from stress." The participants described it as a meaningful experience and a gift. The music helped them to be present in the moment (Figure 7). They were filled with awe and wonder and realized that the Dalcroze approach could help them all. "If we do this on a daily basis, it could help us all. It is better than the six o'clock news on TV." Theme 4: Virtues and emergent categories. --- Water narratives The closing activity of the workshop required participants to divide into groups to create a movement narrative about their experiences of the water situation in Parys. They were asked to show in movement how they experienced the water situation in Ngwathe. The participants imitated water sounds with percussion instruments, body percussion, and their voices for these improvisations. Some small groups used dramatic storytelling with spoken word, sound effects, and movement to express their experiences. These narratives were devised spontaneously and highlighted the improvisatory nature of the Dalcroze approach. The groups shared the following messages: • Group 1: How the water purification process at the water plant works and communicated their intention to bring "water to the people. " • Group 2: The water pressure and quality decrease in the lower-lying areas, and although the municipality provides the water, it does not reach the people, especially not the people in the rural areas. • Group three 3: The water comes from rain and boreholes, and we need to be proper stewards of the groundwater and surface water. • Group 4: The impact of leaking pipes. • Group 5: How the water in the river flows, the beauty of the river and their gratitude for the river. • Group 6: There is no clear water, and the available water gives them stomach cramps and diarrhea. They pray for water and praise God when it rains. • Group 7: Tells a story of rain, full reservoirs and an operator that pumps the water to the people, but then the water dries up, and the people are angry at the operator, and they are sad. --- Discussion We now relate the five themes to the relevant scholarly literature and Jaques-Dalcroze's own writings. Theme 1: Joyful experiences Elliott (1995) identifies enjoyment as one of the three essential values of music-making as a human pursuit, while Freeman (2002) found heightened enjoyment to be one of the four ways in which students engage in music when a spiritual approach to music education is followed. It is, therefore, a valuable finding of this study that the participants enjoyed the Dalcroze-inspired workshop. This was also a finding in a previous study that the stakeholders of the Brandfort community enjoyed the Dalcroze-inspired workshop experience (Van der Merwe et al., 2019). Jaques-Dalcroze (1912, p. 31) himself said, "I like joy, for it is life. I preach joy, for it alone gives the power of creating useful and lasting work. " Joy is also important because it opens us up for social interaction (Johnson, 2020). Theme 2: Social integration McGuire (2003, p. 15) states, "Collective embodied practices, such as singing or dancing together, can produce an experiential sense of community and connectedness. " In the Brandfort study (Van der Merwe et al., 2019), the researchers found that the Dalcroze-inspired workshop facilitated interaction, cooperation, and connection between people. The current study takes it one step further. Not only did participants interact, but they also experienced synchrony and improved communication as a result. Furthermore, they learnt that they could depend on each other. Ansdell (2014, p. 71) explains that "Musicking allows us to explore, affirm and celebrate our diverse real and symbolic human relationships" and, according to Small (1999, p. 13), "It is in those relationships that the meaning of the act of musicking lies. " Jaques-Dalcroze (1919, p. xii) knew that his approach could help people to express emotions regarding a communal crisis. In 1919, he said, "Now the War is over, the coming generation will experience this need of forming groups for the expression of common emotion. " Theme 5: Well-being and emergent categories. The four lifeworld existential of Van Manen (1997, p. 39) form the categories of the well-being theme: lived body (corporeality), lived time (temporality), lived human relation (relationality), and lived space (spatiality). These categories also relate to Van der Merwe and Habron's (2015) conceptual model of spirituality in music education. Van der Merwe et al. 10.3389/fpsyg.2023.1176691 Frontiers in Psychology 10 frontiersin.org Theme 3: Transforms relationships Not only can Dalcroze-inspired activities facilitate social integration, but they can also enable the transformation of relationships. Boyce-Tillman (2000, p. 93) states that "Words divide, but sounds unite" and that "When a group of people makes music together, unity is restored." The Brandfort study (Van der Merwe et al., 2019) revealed that the participants reached a phase of heightened awareness of each other by feeling part of the group. We argue that in the current study, this awareness facilitated the transformation of relationships by opening up boundaries between people. Jaques-Dalcroze (1930a, p. v) says: "I am certain of one thing: that the rightly-directed will can convert mean and selfish instincts into generous and altruistic ones, negative resolves into positive." Theme 4: Virtues expressed Jaques-Dalcroze (1930b, p. 93) considers art "the outward projection of love and knowledge of beauty and truth. " For him, "receive and give" is the "golden rule of humanity" (1919, p. 63). In the Brandfort study (Van der Merwe et al., 2019), the participants felt positive about being open to each other's needs and feelings. In the Parys study, openness led to people helping each other. Boyce-Tillman (2007, p. 1418) explains that in a globalizing culture, musical experiences might help us connect to the Other "in ways that are characterized by a combination of respect and empathy. " Virtuous behavior supports our well-being. Theme 5: Support well-being Croom (2015, p. 44) supports the claim that "music practice and participation can positively contribute to one living a flourishing life by positively influencing their emotions, engagement, relationships, meaning, and accomplishment." The Brandfort study (Van der Merwe et al., 2019) showed that the Dalcroze-inspired workshop positively contributes to participants being able to express their emotions regarding the water crisis, active engagement, and connection with each other. In the Parys study, more explanations were found why Dalcroze-inspired activities supported participants' well-being: they felt welcome, relaxed, present, and could escape reality. These aspects contribute to participants' overall well-being. Jaques-Dalcroze (1930c) states that "it cannot be denied that rhythmic movements possess a calming influence upon the nervous system" (p. 159). --- Conclusion Moving from conflict to open communication and problemsolving is challenging for communities in crises. MacLaren (2009) believes it is more effective to overcome emotional tensions when a conflicting situation is approached positively and meaningfully. She emphasizes that shared bodily experiences are useful for making sense of a mutually conflicting experience, encouraging positive individual transformation. Similarly, the Brandfort study showed that embodied experiences promoted social interactions and joyful emotions during a communal crisis. The positive findings of the Brandfort study (Van der Merwe et al., 2019) motivated this study's research undertaking and purpose. Conceptual replication is relevant in qualitative research as its interpretive strengths can lead to numerous discourses on the same phenomenon (Tuval-Mashiach, 2021;Makel et al., 2022). Although the same qualitative approaches were used in both studies, they were applied in different contexts, which supports transferability, adding trustworthiness, dependability, and validity to the research findings (TalkadSukumar and Meto

This qualitative hermeneutic phenomenological study aims to describe the meaning that members of the Parys community ascribed to their experiences of a Dalcroze-inspired workshop. Stakeholders were a number of local residents of Parys, a scenic town situated on the banks of the Vaal River in South Africa's Free State Province. Local residents were asked to share their water-related experiences at a workshop organized by a North-West University research group. It became clear from their stories that they had experienced severe stressful circumstances. They had suffered stress because of the health risks posed by polluted water and the frequent lapses in the town's water supply system. Dalcroze-inspired activities were used to facilitate this meeting. Data were collected through focus group interviews, open-ended individual interviews, photos, videos, and observations. All these data were consolidated in one heuristic unit in ATLAS.ti, a computerassisted qualitative data analysis software program. The codes were organized into categories and themes. Friese's (2014) notice, collect and think (NCT) method for computer-assisted qualitative data analysis was used. From the data analysis, five themes emerged related to the Dalcroze-inspired activities. It included: joyful experiences, which facilitated social interaction that made it possible for personal relationships to be transformed. Virtues arose from this transformation, and participants' experience was that the group engagement supported their well-being. We, therefore, argue that the Dalcroze approach can be used in communities in crisis to facilitate conflict resolution and transform relationships.

"it cannot be denied that rhythmic movements possess a calming influence upon the nervous system" (p. 159). --- Conclusion Moving from conflict to open communication and problemsolving is challenging for communities in crises. MacLaren (2009) believes it is more effective to overcome emotional tensions when a conflicting situation is approached positively and meaningfully. She emphasizes that shared bodily experiences are useful for making sense of a mutually conflicting experience, encouraging positive individual transformation. Similarly, the Brandfort study showed that embodied experiences promoted social interactions and joyful emotions during a communal crisis. The positive findings of the Brandfort study (Van der Merwe et al., 2019) motivated this study's research undertaking and purpose. Conceptual replication is relevant in qualitative research as its interpretive strengths can lead to numerous discourses on the same phenomenon (Tuval-Mashiach, 2021;Makel et al., 2022). Although the same qualitative approaches were used in both studies, they were applied in different contexts, which supports transferability, adding trustworthiness, dependability, and validity to the research findings (TalkadSukumar and Metoyer, 2019). We developed a distinctive and cumulative body of knowledge, increasing understanding and moving toward a theoretical understanding (see Figure 2) of the phenomenon. The unique contribution of this study is that the Dalcroze workshop was designed for a community in crisis to facilitate conflict resolution. This article is the first study to show that Dalcroze-inspired activities can facilitate conflict resolution. The Dalcroze-inspired activities encouraged embodied expression and creativity, enabling the participants to develop a new understanding of their association with conflict. The participants recognized the negative effects of conflict situations and identified the need for conflict resolution peacefully to transform relationships. This study highlighted that joyful, communal, and embodied musical experiences can open participants' minds to connect, communicate, and creatively solve problems. This joy was affiliated joy (Gabriel et al., 2020); in other words, "joy that was shared with others" (Johnson, 2020, p. 7). The level of arousal during the Dalcroze-inspired workshops was high (Van Cappellen, 2020), generating a lot of energy and urging participants to move. Joy motivates us to reach out to others (Van Cappellen, 2020), as the participants repeatedly mentioned: "We should help each other." Joy sparks an interest in fellow human beings and makes us socially responsive (Izard, 2013). The joyful experience from the communal movement activities was the catalyst for their meaningful experiences. Meaning is associated with "being a giver" (Dwyer et al., 2017, p. 200), helper, andcontributor (Baumeister et al., 2013). Li et al. (2019) explain that meaning leads to "self-transcendent outcomes" (p. 408) when serving something greater than oneself. During this Dalcrozeinspired workshop, participants became aware that they were serving something larger than themselves, namely the community. Meaningful experiences are often associated with a struggle (Dwyer et al., 2017), effort, and worrying (Baumeister et al., 2013) because of a concern for others (Vohs et al., 2019) or an issue, for example, a struggle against injustice (Metz, 2009). Ngwathe stakeholders could share and express their struggle through the Dalcroze-inspired workshop, voice injustices and their needs, and explore possible collaborative solutions. Furthermore, a sense of belonging (Martela and Steger, 2016) and taking care of others (Metz, 2009) enabled the community members to experience the workshop as meaningful. Negative experiences and suffering can lead to transformation and even meaning when the individual can make sense of difficult events (Vohs et al., 2019). Moving together with others and creating water narratives helped participants understand their water crises' complexities. The Dalcroze-inspired community engagement presented in this article might be transferable to other contexts. In this hermeneutic phenomenological inquiry, the Dalcroze approach was used for a community in crisis to facilitate conflict resolution and transform relationships. --- Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. --- Ethics statement Ethical review and approval was not required for the study involving human participants in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. --- Author contributions LV facilitated the Dalcroze-inspired workshop, collected and analyzed data, and wrote a large part of the article. DJ wrote the section on Music, health and well-being, and did extensive editing. JT was the leader of the transdisciplinary research team that facilitated the meeting about water, actively collected data in the field and wrote the section on the water, and gave feedback to the various stakeholders. All authors contributed to the article and approved the submitted version. --- Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

This qualitative hermeneutic phenomenological study aims to describe the meaning that members of the Parys community ascribed to their experiences of a Dalcroze-inspired workshop. Stakeholders were a number of local residents of Parys, a scenic town situated on the banks of the Vaal River in South Africa's Free State Province. Local residents were asked to share their water-related experiences at a workshop organized by a North-West University research group. It became clear from their stories that they had experienced severe stressful circumstances. They had suffered stress because of the health risks posed by polluted water and the frequent lapses in the town's water supply system. Dalcroze-inspired activities were used to facilitate this meeting. Data were collected through focus group interviews, open-ended individual interviews, photos, videos, and observations. All these data were consolidated in one heuristic unit in ATLAS.ti, a computerassisted qualitative data analysis software program. The codes were organized into categories and themes. Friese's (2014) notice, collect and think (NCT) method for computer-assisted qualitative data analysis was used. From the data analysis, five themes emerged related to the Dalcroze-inspired activities. It included: joyful experiences, which facilitated social interaction that made it possible for personal relationships to be transformed. Virtues arose from this transformation, and participants' experience was that the group engagement supported their well-being. We, therefore, argue that the Dalcroze approach can be used in communities in crisis to facilitate conflict resolution and transform relationships.

Introduction Like other populations in the Middle East, Qataris have increased susceptibility to some diseases, including: diabetes, cardiovascular disease, obesity, and some neurological disorders. Genetic discourses have predominated in Qatari approaches to combating these conditions. Discussions around genetics also prevail because of the high frequency of consanguineous marriages in the country and concerns about its genetic effects. Rates of genetic disorders and birth defects are relatively high given the small population size. Autosomal recessive disorders are increased due to consanguinity, but also common, multifactorial disorders such as diabetes mellitus type 2, obesity, psychosis, and congenital malformations are seen in excess (Bener and Alali 2006;Bener et al., 2007). There are particular concerns about high rates of sickle-cell aneamia; thalassemia; homosystinuria; and mental retardation. Qatar provides an opportunity to explore issues around discourses of genetic risk where despite the global spread of biomedicine, a culture of transnational research, and a commitment to modernization we witness a continuing dedication to the practice of consanguineous marriage. The paper investigates the way medical technologies and research; particularly those that focus on genetics are the focus of the Qatari state's positioning as a global political and economic presence. The research on which this paper is based contributes to the literature on modernity and globalization by focusing on the Qatari context where rapid social change and globalizing forces have been met by negotiations and, in particular, containment at the individual, group and state level. The paper thus contributes to the literature on the impacts of discourses of genetic risk in consanguineous marriage in diverse global settings (Shaw and Raz 2015). This paper is derived from an exploration of Qatari encounters with globalizing discourses of genetics, particularly as they relate to notions of risk. I draw on my experience of living in and conducting research in Qatar (2010Qatar ( -2016)). During this period I conducted two research projects: one an investigation into miscarriage and the other a general investigation into public understandings of genetics and notions of risk, disability and illness. During the course of my research interlocutors commented on issues around inheritance, genetics, risk, illness and health. In addition, the paper is based on analysis of Qatari medical research and government documents. I situate this paper's findings amongst social science research focusing on how new medical technologies and knowledge, including those involving genetics, are accommodated, contained and resisted within local moral landscapes (Inhorn 2009). The analysis and interpretation of research findings is informed by current anthropological knowledge and analytical approaches to the topic of genetics in the Middle East region. A variety of scholars (i.e. Panter-Brick 1991, Raz 2005, Beaudevin 2013, Parkhurst 2014) explore genetics in the Middle East. Raz (2005) explores what happens when this "newly discovered scientific notion of 'fate' encounters its traditional and religious conception" (xv) amongst the Negev Bedouin community in Israel. Beaudevin's (2013) research on inherited blood disorders in Oman reveals, "the way increased biomedical knowledge and the outcomes of its applications can affect a society" (p.185). Panter-Brick's (1991) research in Saudi Arabia shows the profound impact that knowledge of genetics has on family relationships. Scholarly works such as these have shed light on the way encounters with genetics alter social life and understandings in fundamental ways. The paper explores Qatari negotiations of biomedical genetic knowledge, particularly in light of marriage practices and ideas about reproduction. The paper first presents material on the medical and research platform in Qatar, particularly the central role genomics and genetic medicine hold within the development agenda. In particular, I argue that a focus on genetic discourses and genetic research underpins the Qatari quest for modernity. The paper explores the way in which genetic medicine and research is at the forefront of the Qatari state's mission as a "modern" nation and a world leader in research, education and international politics. This impetus can be at odds, however, with the way that Qataris themselves understand and incorporate genetic knowledge into their lives. Thus, the second part of the paper considers local Qatari knowledge systems and explores the tensions and negotiations in their intersections with modernizing genetic discourses. I show that Qataris meet such discourses with processes of containment, in a balancing of the traditional and the modern. With regard to the dominant discourse around genetic risk and cousin marriage, the increase in family marriage can be seen as one of the main platforms of resistance and negotiation. --- Methods This paper is based on several years of research on/ in Qatar including two ethnographic research projects. The paper primarily draws on fieldwork conducted over 12 months (2012)(2013) during which accounts were collected about notions of health risk and, particularly, genetic risk (for a full description of methods see XXX 2015). Interviews were conducted with 45 families whose children were pupils at a school for special needs. Because of the presence of the medical genetics centre at the same facility, families of students were at the interface of public understandings of genetics. Families had been exposed to genetics through general discussions by staff or more directly exposed through genetic testing. Interviews included collecting basic demographic information about the family, marriage patterns, education and occupation. General questions about notions of health, illness and risk were asked. Questions probed notions of inheritance and genetics, i.e. 'who provides the genetic material to a child' as well as questions to explore how traits (i.e. intelligence, physical traits, and so on) were passed down through generations. Qualitative, flexible, and open-ended methods were used to focus on the perceptions, experiences and views of interlocutors. Professionals involved in the creation of genetic knowledge and those at the interface between the public and genetic discourse were interviewed including: social workers, geneticists, and those involved in medical screening programmes. Additional information about public understandings of genetics is provided by ongoing exploration of documents, including media and government policy documents as well as data derived from 18 months (2013)(2014)(2015) ethnographic research on pregnancy and miscarriage (for description of methods see XX in press). The latter involved interviews with 60 women and 55 secondary participants (family members, health care providers, religious leaders etc), whose discussions about conception and inheritance inform this paper. Analysis of the qualitative data sets generated by the research was through a process of thematic coding and analysis that make use of a grounded theory approach (Strauss and Corbin 1990). Regular reading and re-reading of interviews and field observations identified themes for exploration in subsequent interviews. Analysis for this paper involved identifying themes related to the topics of risk, genetic risk, illness causality and inheritance. As themes emerged, data were re-analysed to develop and refine the findings. Several over-arching themes were identified, providing a framework through which I present and contextualize the findings. --- Qatar: modernity Visitors to Doha often express surprise at its state of the art appearance; its skyline dominated by high-rise modern buildings raised from the ground at an astonishing rate. The city is continuously changing, growing, and evolving: one is overwhelmed by sounds of around the clock construction work and left confused by suddenly altered roads. Photographs of 1980s Doha show a desert with a few low buildings: surrounded by desert sand, the pyramidic Sheraton Hotel stands out as the only building taller than a few stories. Today, the hotel is dwarfed by the modern steel and glass skyscrapers, which surround it. The city itself is a symbol of Qatar's rapid social and economic development, representing the county's ability to change itself and its surrounding environment at a rapid rate fuelled by its high-income economy. Qatar is characterized by the striking modernity of its urban built space and its pace of social change. A small country occupying the Qatar Peninsula on the northeastern coast of the Arabian Peninsula, Qatar's recent social and economic changes have occurred as the result of the discovery of natural gas and oil in 1940. This discovery turned Qatar from a largely Bedouin society that depended on fishing and pearl fishing to the richest country in the world. In 2013 its population was 1.8 million people, with Qatari citizens representing a minority at only 278 000 people. The rest of the population is comprised of migrant workers from all over the world. A British protectorate in the early 20 th century, the country gained independence in 1971. An absolute monarchy headed by Emir Sheikh Tamim bin Hamad Al Thani, it is a conservative country with most Qataris adhering to strict Salafi interpretation of Islam. Sharia law is the main source of legislation in the country. Despite its conservatism, the country has endeavoured to be an influential link between the Arab world and the west and has worked to develop a reputation as a progressive Arab nation in terms of education, research, as well as economically and politically. The home of Al Jazeera Media Network, Qatar positions itself as an influential player in the Arab world and supported several rebel groups during the Arab spring. The country has had a high profile of late due to these issues, its investment in a number of high profile holdings (i.e. in the UK: Harrods, The Shard) but also because it is to host the Fifa 2022 World Cup and has been embroiled in the accusations of corruption that has accompanied its bid. Further damaging Qatar's reputation has been criticism of its treatment of migrant workers, which has gained a great deal of media attention in recent years. Qatar is positioning itself as a world leader in research and medical technology: "In striving toward this vision, Qatar will distinguish itself within the region and world as a cosmopolitan nation that embraces scholarly excellence, innovation, creativity, inclusiveness and merit."(http://www.qnrf.org/en-us/About-Us/Vision-Mission). But what does cosmopolitan mean in this context? The Oxford dictionary suggests a number of meanings: familiar and at ease with many different cultures, including people from many different cultures, exciting and glamorous character associated with a mix of cultures. One can look to a number of recent initiatives to better understand what cosmopolitanism means in this context, such as "Education City," a 14 square kilometer area in Doha, which houses branch campuses from a number of leading American and one (UCL) British university. An initiative of the Qatar Foundation (QF), it is conceived as a forum for exchange, where universities collaborate with businesses and institutions in both the public and private sector. It is seen as a cosmopolitan space, where Qataris, other Gulf Arabs, and students from other nations mix and are taught by foreign (predominantly American and European) professionals. The aim of Qatar's focus on creating world class education for Qataris at Education City predominantly is built upon a notion of education as a "means of creating marketable, international skill; education is focused on connecting Qatar to the outside world" (Fromherz 2012:10). Education City stands at the intersection of two sets of interests: being at one with the globalised world while aspiring to stand apart from it, as Kane (2013) describes in her research into medical education. The Qatari leadership did not recruit from the pool of established regional medical schools, but deliberately chose to import one from the U.S, notes Kane (2013). Although a regional medical college would "offer a professional training more in keeping with core Qatari values and culture, the US programme was perceived to be superior both in terms of its quality of training and its capacity to develop a domestic research platform which could communicate at an international level" (Kane 2013: 101). Diverging from the aims of other pursuants of globalised and globalising education, the Qatari state is: Not aiming to produce neoliberal subjects trained for employment in global knowledge-driven economies. Rather, citizens are being presented with opportunities to equip themselves with specific skills in order to participate in the development of a domestic knowledge-based economy (Kane 2013:101). To this end, a number of measures are activated to ensure the maintenance of Qatari values and to nourish a local set of principles of development. Concerned about their dependence on foreign labour and outside expertise, the Qatari state has introduced a number of governmental initiatives devised to increase the number of Qatari citizens employed in public and private sectors. These initiatives are part of the so-called "Qatarisation" of the nation, a focus of the Qatar National Vision (QNV) 2030. This development plan was launched in 2008 by His Highness Sheikh Hamad bin Khalifa Al Thani to provide a "clear roadmap of Qatar's future" and to "transform Qatar into an advanced society capable of achieving sustainable development" (http://www.gsdp.gov.qa/portal/page/portal/gsdp_en/qatar_national_vision). The structure of Education City is indicative of the general dependence on the outside, which can be seen throughout Qatar economic, social and political arenas: 90 per cent of Qatar's food comes from overseas. This dependence has become a preoccupation of Qatar's political elite (Fromherz 2012:11). Thus, the model is bringing from the outside to create Qatari citizens and improve Qatar, but such imports are contained and controlled. As Douglas (1966) has suggested, when boundaries between categories are unclear, contradictory or broken down, the resulting category confusion is seen as polluting and potentially dangerous, requiring containment strategies. Containment is seen throughout Qatari policy and approaches to outside forces, which focuses on limiting such forces seen polluting or corrupting Islamic and Qatari values: such as the tight restriction on the sale and use of alcohol; restrictions on the behavior and movements of certain migrant workers and, of course, the policy of Qatarisation itself. The state restricts the ideological influence of others. Commentators have suggested that the focus of education in Qatar is on connecting with the outside world rather than inward criticism or reform it is, "not on the issues of governance and society within Qatar itself" (Fromherz 2012:10). The main instrument of containment is through constant emphasis on tradition. Qatar positions itself as innovative, modern and open to learning from other cultures. However, the emphasis is on retaining Qatari customs and authenticity as well and benefitting Qataris specifically. Gulf regimes, such as Qatar, actively seek to create, "invented traditions" in order to maintain and deepen their legitimacy and relevance, with such "heritage engineering" represented by the emphasis on cultural activities such as folkloric customs, national dress, as well as the passion for museums and a renewed attention to suqs, as in Doha's reconstructed and reimagined Suq Waqif (Cooke 2014). In this context the Qatari notion of cosmopolitan seems to mean welcoming in, making use of, and learning from outside influences, but only whilst retaining a strong sense of continuing Qatari identity and culture. --- Qatari Medical Research landscape Qatar has pledged to spend 2.8% of its GDP on research (of 211.82 USD billion estimated) with the majority of these funds being managed by the QF, which establishes the country's research strategy, heavily influencing where funding is channeled. The Vision of QF is to enhance the education of its citizens and training of its workforce and fostering improvements in health, wellbeing, environment of "its own people and those of the region". The QF launched Qatar National Research Fund (QNRF) in 2006 as part of its ongoing commitment to establish a knowledge-based economy. Applications must emphasise how the research will directly benefit Qatar, this being key criterion for evaluation. In 2012 Qatar presented its first National Research Strategy (QNRS), whose mission is: "to build and maintain a competitive and diversified economy; improve the health and social wellbeing of Qatar's population; and support Qatar's distinctive culture and the security of its people" (http://www.qnrf.org/en-us/About-Us/QNRS). Included in the QNRS is the development of "plans to tackle Cancer, Diabetes & develop personalised medicine" and to develop a "dedicated research program focused on the understanding of the genetic causes of diseases and conditions in Qatar." Indeed, considerable state funds have been channeled towards understanding the country's high rates of diabetes, cardiovascular disease, obesity, as well as certain neurological and genetic disorders. In light of increasingly high rates of diabetes, heart disease and obesity, popular health discourse in the region has emphasised the emergent Arab genome as the primary etiological basis of major health conditions (see Parkhurst 2014, this issue). Local health authorities focus on programmes looking at genetic susceptibility for certain health concerns. In recent years, the Qatari government has implemented dozens of public awareness campaigns intended to educate Qataris about healthy eating, exercise, and fitness. It has been reported that "more delicate cultural issues", such as family marriage "are being addressed by 'higher-ups in society." Community leader and sheikhs are engaged by public health drives to communicate such messages around risk (http://www.theatlantic.com/health/archive/2011/11/the-richest-fattest-nation-on-earth-itsnot-the-united-states/248366/). Research, including the mapping of the genome, has been undertaken in order to understand Qatari genetic difference and its impact on health. The basic goal of the QNRF-funded Weill Cornell Medicine-Qatar (WCM-Q) research, as outlined by the lead, Dr. Crystal: "involves trying to understand the structure of the genome of the Qataris and put that in context with the environment to see how we can use that information to help the population in terms of general health, understanding disease, developing appropriate therapies" (QNRF 2015). Such research exemplifies the next phase of genetic research and involves localized research to discover patterns in heritage and genetic susceptibilities to disease. In Qatar this has focused on the local Qatari population in order to map ancestral background and isolate areas of the Qatari genome that could potentially allow prediction and intervention. Rodriguez-Flores et al 2014 outlined three categories of the Qatari local population: Q1 Bedouins, Q2 Persian or South Asian mixture and Q3 African-derived Qataris. This research team plans to build upon this knowledge to study the responses of the different genetic populations to the same environmental stressors. As more studies into the local population take shape, researchers will also be able to isolate more monogenetic diseases prevalent in Qatar, such as thalassemia. Such research aims to understand the differences between the Qatari populations and other populations in order to learn why Qataris are more susceptible to certain disorders. A number of projects have been developed to further Qatar's vision. In 2013 The Qatar Genome project was launched in order to identify diseases and illness "in order to have personalised treatment and medications for such diseases", according to HMC Managing Director Hanan al-Kuwari (Gulf Times 2013). The SIDRA Medical and Research Centre is to be an "ultramodern, all-digital academic medical center which will set new standards in patient care for women and children in Qatar, the Gulf region and internationally" (http://www.sidra.org) and "help Qatar grow as a hub for genetic medicine": [It] will house a first-of-its-kind facility, the High-Throughput Genomics Center, to undertake population studies and genetic sequencing for the MENA region. The facility will provide Qatar and the region access to the latest technology to help advance genetic mapping projects. The facility's early phase of development is the creation of the Arab consensus genome, which will allow for a deeper understanding of genetic variants that contribute to the health of the Arab population.... The initial focus of the facility will cater primarily to high-end applications, including human whole genome sequencing for rare genetic diseases and population studies.... Genome sequencing and population studies are an exciting new field of research, enabling scientists to create truly personalized care based on an individual's own genetic makeup. WCM-Q is conducting research involving the sequencing of 1,000 Qataris representing "0.3 percent of the whole Qatari population" (QNRF 2013). However, this does not include the entirety of the population, but only Qatari nationals. This is particularly exclusive given the way that nationality is conferred: through the male Qatari line. Hence, offspring of Qatari women are not considered nationals if their father is non-Qatari. The genomic revolution has furnished potent resources for the expression of nationhood (Atkins and Glasner 2007). The research projects outlined above resonate with the nation's overall research strategy, which is tied in to the Qatar National Vision (QNV) 2030, which "is based on the guiding principles of Qatar's Permanent Constitution. It is the primary catalyst driving change and growth across Qatar [and] reflects the aspirations of the Qatari people and the resolve of its leadership" (General Secretariat For Development Planning 2008). One must remember that Qataris make up 12% of the country's population and the rest are migrant workers, many of whom have different experiences of health and access to services and opportunities as outlined by such mandates. Such research raises questions about who benefits from investment in genomics. Are migrant workers, for example, able to access genetic interventions? Are they potential research participants in genetic research? Gardner et al (2013) found that 7% of the foreign workers reported that they did not have a QID (residence permit). However, what is particularly relevant for the discussion here is that a much larger percentage (56%) of the workers lacked a government-mandated "health card" (Gardner et al 2013). This document is required for accessing health care in the state's expansive public health system. Thus, over half of the foreign residents are unable to access basic state funded health care, let alone genetic interventions. Genetics provides a powerful idiom for the expression of individual and collective identity. QNRF has funded genome-sequencing projects and is independently pursuing the whole genome sequencing of 300,000 Qatari nationals, representing almost the entire Qatari population (Al Mulla 2014; http://www.gulf-times.com/qatar/178/details/374345/qatarlaunches-genome-project). The nation appears in discussions around genomic patrimony and sovereignity (Rabinow 1999;Benjamin 2009). Genomic researchers and proponents of genome projects in Mexico and India'strategically calibrate' contrasting modes of groupmaking (Benjamin 2009). The Qatari project resonates with Rabinow's (1999) interpretation of the reality of "French DNA" through his account of the undoing of the commercial exchange of DNA from French families to a US biotech company. The French government considered DNA samples to be part of the collective patrimony and therefore not appropriate as items of commerce: Rabinow, thus, contrasts this with the US system where body parts (i.e blood) can be sold. An individual's body and body parts are seen as part of the French patrimony. The economics of biovalue mean that DNA may be regarded as a national resource, as well as a repository of national characteristics (Atkinson & Glasner 2007). The rhetoric of biological inheritance and relatedness is not new and has long provided a vocab of nationality, nobility and purity (Atkison & Glasner 2007). The intersection of national and biological identity has been documented in many contexts and is particularly true of Qatar. Saudi Arabia has initiated the sequencing of about 100,000 Arabs from around the Kingdom (http://rc.kfshrc.edu.sa/sgp/). The ambitious remit of these programmes resonates with the calls to widen programmes of genetic research to global health care arenas outside Western Europe and North America and to include genotype information from'minority populations' and 'other ethnic groups' (see Bustamante et al 2011). The need to include genotype information from broader and more diverse groups is desired mainly so that those most in need will be included in research and subsequent interventions. However, in Qatar the aim is to provide specific information related to Qatari bodies and diseases and, thus, seems less about underserved populations and more about national relevance. Concern has been raised that data generated from such genome projects will be unconnected because of its politically driven nature. As Kuwaiti based geneticist Al-Mulla suggests, genome sequencing "is viewed only as a national endeavor rather than a humanitarian triumph or necessity" and argues "locking genomes... contradicts the goal of genomic research, namely that the benefits be shared globally" (2014:133). It is likely that such research will produce findings that will help other populations, but this may be a side effect rather than the central aim. For example, Crystal, the lead of the WCM-Q genetic mapping project noted that the mapping and analysis of Qatari genomes has provided unexpected insight into a genetic variation that affects other populations all over the world. One study focuses on a variation gene "ApoE", which makes carriers susceptible to having increased levels of unhealthy fats called triglycerides in their blood. Considered rare, the variant was far more common in those with sub-Saharan African extraction; its presence is associated with disorders such as heart disease, type-2 diabetes and stroke (Abou Ziki et al 2014). The research included 1266 New York African-Americans and found that four per cent carried the R145C variant and, thus, Crystal points out, has implications for African derived populations throughout the world: "it was extremely gratifying for us to make a discovery in Qatar that provided insight into the health of people in New York" (QNRF 2015). Whilst the Qatari government has emphasised genetics research and medicine as keys to creating a healthier society, the particularities of the Qatari context have attracted global genetic research initiatives. The generous funding programmes, emphasis on research and development, and possibilities for international collaborations have enabled an impressive research landscape. The genetic singularity of the Qatari population, high rates of consanguinity and the high prevalence of certain conditions both highlights the need and offers the opportunity to study the genetics of recessive single-gene disorders (Marincola & Sheikh 2012). Research into the relationship between consanguinity and the severity of multi-genic complex diseases of adulthood could also be investigated in Qataris better than in most populations worldwide: for example, research into the high prevalence of type II diabetes in Qataris and the connections between socio-economical factors, endogamy and/ or consanguinity could be disentangled (Marincola & Sheikh 2012). Indeed, genetics studies related to diabetes and obesity are currently being conducted at WCM-Q as well as QBRI. --- Genetics and genomics: public understandings and negotiations Genetic discourses have a high profile in Qatar through local media reports, public health initiatives and the research landscape. Seen as the key to improving the health of Qataris they are part of a state political drive to be seen as a harbinger of cutting edge research and knowledge: to be a modern and "cosmopolitan" country. I now explore how this vision of Qatar, particularly with its harnessing of global discourses on genetics, is experienced and interpreted by Qataris themselves. In particular, I explore the discrepancy between scientific implications and lay interpretations of inheritance. Central to this is a process of negotiation and containment in order to re-establish the balance between tradition and modernism that is at the heart of contemporary Qatari social life. Thus, the role of genetics as part of the Qatari state's mission to become "modern" can be at odds with the way that Qataris themselves understand and incorporate genetic knowledge into their lives. Despite years of public dissemination of genomic knowledge in the region, its commonality and widespread acceptance; the rates of diabetes, obesity and heart disease continue to rise. Many have commented about the sudden, severe problems of obesity and diabetes among wealthy citizens in the Gulf over the past decades. The link is made between these condition and modernity: these are medical conditions that were almost completely non-existent before the oil boom: These observers seem to feel that the diabetes, an inability of the body to absorb glucose because of lack of insulin, is analogous to a deeper intellectual and philosophical problem existing in the social fabric of Qatar: the inability to digest and process the meaning and possible negative consequences of modern change (Fromhertz 2012). The suggestion is the high rates of diabetes and obesity are related to an inability to digest change. Gulf Arabs themselves often see these as conditions of modernity and many do not appear willing to consider personal habits as responsible for their wellbeing (Parkhurst 2014;this issue). Lay observations about the rise of such "new" diseases resonate with theoretical observations of risk as an aspect of modernity (Beck 1992, Giddens 1991, Caplan 2000) and with lay observations by, for example, British Pakistanis that risk is greater in modern societies (Shaw and Hurst 2008). Parkhurst observes that Gulf Arabs often cite the presupposed Arab predispositions to Type 2 diabetes as the ultimate issue in discussions of treatment and lifestyle choices: These habits are a form of personal violence embedded in many Emirati lifestyles. This type of violence is internal, and genes, as a category of fate are inducted into indigenous cosmology, are perfect foreign agents. They are embedded in the body, but not part of familial kinship... They are perhaps perceived as sacred when they are categories of fate, or agents of Allah, and they are mundane as agents of disease. They might cause disease, but they are also thought to be at the heart of being Arab (2014:193). Qatar has reacted to modernization uniquely: the "'heroic' in Qatar is associated not with the 'now', a now dominated and built by migrant labour, but with what is 'behind it'-a past reconstructed and reconstituted into nationalized historical moments" (Fromhertz 2012:15). The government has perpetuated neo-traditional myths of culture and authenticity: Qataris have instead maintained tradition or at the very least, constructed newtraditional notions of identity. Shielded from the most ravishing consequences of modernization by enormous wealth and by a dependence on a massive expatriate community, Qataris can maintain a bubble of culture and internalized authenticity. It is no mere fashion that leads all Qatari men to wear their traditional thob at all times in Qatar, moving through Western spaces and even influences while maintaining lineage and family as the primary determinant of destiny (2012:13). --- Consanguinity: containment and tradition Despite the high profile of genetic discourses and particularly public health messages about the risks of family marriage, consanguinity is on the increase in Qatar. The global spread of biomedicine has brought with it a globalizing discourse of risk accompanying the practice. Attitudes of Western cultures and medical professionals toward consanguinity are often negative, ostensibly because of the impact on health (see Shiloh et al. 1995;Bittles and Makov 1988). As Shiloh et al (1995) suggests such notions have "penetrated even in societies where this kind of marriage is common, and have been integrated into more general attitudes and beliefs about consanguinity" (p1301). Such is the case in Qatar where despite being a dominant marriage pattern, the negative discourse of risk has been integrated into public discourse, linked, in part, to the country's continuing emphasis on modernization. As Bittles (2012:1) outlines, 'it has become virtually impossible to persuade members of the general public that inbreeding, and by extension marriage between biological relatives, can be anything other than harmful.' This is despite the fact that there is growing evidence that the deleterious effects of consanguinity are exaggerated (i.e. see Shiloh 1995). Shaw (2009 49) states: Public perceptions of risk are not neutral: a 'climate of disapproval grounds the belief that certain deeds are dangerous' (Douglas 1992:27). Moreover, in the current political climate of Muslim/non-Muslim relations in contemporary Europe, the biological risk of cousin marriage provides'scientific' grounds for disapproval of a marriage practice... for persisting in risky behavior, resisting cultural conformity. By actively drawing in these discourses, there is in Qatar a focus on the dangers of consanguineous marriage that is communicated through public health drives. Mandatory premarital screening programme for all couples intending to marry are one of the main ways that Qatari citizens confront genetics. Across the Middle Eastern region, the rates of consanguinity range in most societies from 20% to 55% of all marital unions (Bittles et al 1991;Bittles 2012). However, unlike in many parts of the world (Lebanon, Saudi Arabia, Kuwait, Jordan, Israeli Arab communities and the Palestinian territories) the practice is on the increase, as it also is in the United Arab Emirates (UAE), Yemen, Iran (Bittles 2011:62). The increasing rates of consanguinity contradict the assumption that modernization will "inevitably reduce the incidence of consanguinity" (Jurdi&Saxena 2003:2; in Qatar see Harkness 2014; Al-Ghanim 2010)). A recent study found that 54% of Qatari marriages are consanguineous with first cousin unions being the most common: 34.8 % of all marriages and 64.4 % of all consanguineous unions (Bener & Hussain 2006). Consanguineous marriage is common not only in the Middle East, but also in South Asia, North Africa and parts of Southern Europe. Shaw (2009), found a similar increasing trend amongst British Pakistanis over the past decade, contrary to expectations. Rozario's (2013) study of British Bangladeshi Muslims found that, despite negative views of cousin marriage amongst health care professionals, respondents saw the practice as a way to retain a "purity of blood". The "assumption that the decision remains in Allah's hands makes it possible for people to go ahead with cousin marriages despite being perhaps at some level aware of the risks" (Rozario 2013:201). Research from a variety of cultural contexts has revealed that lay and public health/media discourses about genetic risk and cousin marriage do not always correspond (Shaw 2009). Genetic risk was seen as irrelevant and thus, a "differently constituted perception of risk" was present (Shaw 2009:3). Scholars have shown that encounters with genetics alter social life and understandings in fundamental ways in the Middle East (i.e. Panter-Brick 1991, Raz 2005, Beaudevin 2013). Parkhurst (2014) shows the robustness of UAE native knowledge system: "Local systems of cosmology, kinship, and fate... all have their own language in which biology must be maneuvered" (p.167). In discussions with my interlocutors, genetics remained entrenched in the language of disease, illness and disability, resonating with Parkhurst's findings: "While there exists a robust local knowledge of the mechanisms of inheritance and kinship, genes as biological entities are not part of, and not associated with this inheritance and kinship.... They are widely known to be carriers of disease, but are not often understood to contain the essence of, or the benign traits of a person" (2014:188). Individual Qataris are still grouped according to lineage: one's qabila: one's extended "tribe" or family, remains the fundamental determinant of an individual Qataris social position and future (Fromherz 2012:7). Wealth has re-inforced certain aspects of history and lineage: families now have the opportunity to live in large family compounds with their own mosque (Fromherz 2012), emphasizing family allegiances. Fromherz' discussion of the spatial arrangements of buildings is taken further in Gardner's (2013) account of Doha itself. An anthropologist who has conducted research in Qatar, Gardner makes use of Dresch's discussion of the integral feature of Gulf societies as the accommodation of "foreign matter" and applies it to Do

Genetic discourses have taken a predominant role in approaches to combating a number of conditions that affect Qataris. This paper is derived from an exploration of Qatari encounters with globalizing discourses of genetics, particularly as they relate to notions of risk. It explores Qataris negotiations of global interactions and influences, including the discourses around genetic risk and cousin marriage. It suggests that family marriage can be seen as one of the main platforms of resistance and a means for modern, cosmopolitan and tradition to be negotiated.

and disability, resonating with Parkhurst's findings: "While there exists a robust local knowledge of the mechanisms of inheritance and kinship, genes as biological entities are not part of, and not associated with this inheritance and kinship.... They are widely known to be carriers of disease, but are not often understood to contain the essence of, or the benign traits of a person" (2014:188). Individual Qataris are still grouped according to lineage: one's qabila: one's extended "tribe" or family, remains the fundamental determinant of an individual Qataris social position and future (Fromherz 2012:7). Wealth has re-inforced certain aspects of history and lineage: families now have the opportunity to live in large family compounds with their own mosque (Fromherz 2012), emphasizing family allegiances. Fromherz' discussion of the spatial arrangements of buildings is taken further in Gardner's (2013) account of Doha itself. An anthropologist who has conducted research in Qatar, Gardner makes use of Dresch's discussion of the integral feature of Gulf societies as the accommodation of "foreign matter" and applies it to Doha's urban spatial discourse. The urban spatial discourse involves spatialization and enclaving of globalization and neoliberal flows: The compartmentalization of foreign matter -people, practices, and ideas-is fundamentally an assertion of cultural sovereignty over the domain punctuated by these exceptional spaces. Put another way, this urban spatial discourse is an assertion of Qatari cultural sovereignty over the cosmopolitan and transnational heterogeneity so essential to the developmental goals of state and citizenry. This urban spatial discourse reflects the relationship between a people and the foreign matter they both host and depend upon. This relationship is woven into the shape of the city (Gardner 2013) There is a similar containment of genetic discourse as it relates to risk. Skepticism reflects a rejection of a dominant discourse of genetic risk that stigmatizes cousin marriages (Shaw and Hurst 2008). Such a process resonates with the containment of foreign influences and bodies in the urban space of Doha and the notion of cosmopolitanism found in Education City. Thus, in the case of Qatar, modernization has provided opportunities for containment of the global flows that accompany it. --- Traditional modernists One of the ways that bodies, spaces, ideas, practices and discourses are contained and altered is through an emphasis on tradition. A constant balancing between the forces of modernization and tradition is felt acutely in Qatar: it pervades most aspects of social life. Qatari identities mix modern (i.e., Western) and cosmopolitan (i.e., global) whilst adhering to local customs and traditions. The distinct amalgam of customary and contemporary is a crucial element of most projects in Qatar, from higher education to museums to sports (Harkness & Khaled 2014). Rajakumar (2014) argues the embodiment of this delicate balancing can be seen in the form of Sheikha Moza, the second wife of the former and mother of the current Emir. The Sheikha has played a key public role and was the driving force behind a large number of initiatives, including the development of Education City. She espouses a "plural, liberal interpretation of Islam, which focuses on the religion's emphasis on education, the role of the individual in society, and the betterment of the world" (Rajakumar 2014: 127). Positioning herself as a "modern traditionalist," the Sheikah's recent speeches have increasingly emphasised the blend of Arab identity, modern pragmatism, and importing the best educational models to make Qatar a hub for innovation in the Arab World (Rakakumar 2014:128). This emphasis certainly resonated with what many of us working in/ on Qatar were experiencing: a re-alignment in such a way to de-emphasise the role of foreign experts, with purposeful alignment between Qatar, Arab identity and an emphasis on Arab intellectuals (Rajakumar 2014:138). This seems to be part of a general re-balancing in favour of tradition and Arab culture in the face of increased development and foreign influences. Historical data demonstrate that social changes in the Middle East are sometimes met with conservative reactions that include the enactment of traditional policies and cultural expectations (Harkness & Khaled 2014). Scholars working in Qatar have shown that tribal "tradition" is often increased not decreased with wealth (see Fromhertz 2012, Gardner 2013, Al-Ghanim 2010), particularly in relation to social roles, marriage practices and adherence to traditional dress. For example, the increased use of the hijab in the wake of escalating rates of female employment, education, and athleticism in the region (Jawad, Al-Sinani, & Benn, 2011). Thus, family marriage may be part of this negotiation of the traditional and modern. The continued and increasing popularity of consanguineous marriage may reflect a desire to moderate modernising trends by preserving traditions and customs. Abbasi-Shavazi et al ( 2008) explain the increased rates of consanguinity in Iran as "cultural maintenance... in the persistence of traditional behaviour despite the forces of modernization" (p.917). In a quickly developing nation such as Qatar, the benefit of the effective transmission of culture could create social stability, which in a period of change could have enormous benefits (Sandridge et al. 2010). The potential advantages attributed to family marriage include: greater autonomy for women, benefits such as familial unity, decreased pressures on the bride in her new home, a stronger marital bond with less risk of divorce, greater compatibility of the bride with her husband's family, property retention and effective transmission of the culture from generation to generation (Sandridge 2010;Barth 1953;Khlat et al.1986;Bittles 1994;Ottenheimer 1996). Interlocutors spoke of the importance of familiarity with customs and traditions when looking for a potential spouse for their children. They hinted at notions of purity and authenticity: familiar customs and practices would allow women to effectively impart them to future generations. Indeed, the importance of this is central to Qatari society, as outlined in the Qatar Development Strategy: The family is the basis of Qatari society... Qatar's strong Arab and Islamic identity pervades all aspects of family life and continues to inform the family structure, but changes brought on by external pressures and internal evolution are changing family dynamics.... Women are central to this positive, evolving nature of the Qatari family. Even as they maintain an adherence to valuable traditions, women are adapting to the impacts of modernization. They exemplify the new opportunities available to all Qataris as a result of the country's rapid economic growth and social transition. (p.17). We have argued that whilst aware of the discourse of risk and close marriage, Qataris negotiated with other forms of risk, such as the potential dangers of marrying a daughter to "strangers" (Kilshaw et al 2015). Thus, marriage in the family mean the woman would be more comfortable, have closer ties to her nascent family and would share cultural traditions in order to effectively impart to children. Our interlocutors were often unwilling to accept the idea of disease as inherited, stating that not all members of the family were affected. Panter-Brick (1991) similarly found that unless family history made an inheritance factor obvious, Saudi participants were unlikely to see the condition as genetic and thought genetic illness should affect all children and should appear soon after birth. Whilst aware of the discourse of consanguinity risk, interlocutors seemed unsure of its reality and used anecdotal evidence to support uncertainty. Qatari interlocutors cited examples where cousin couples had many healthy children or referred to families where offspring had disabilities despite the parents being unrelated (see also Panter-Brick 1991). Wadha, a 33year-old mother of nine explained that people often suggest that her son's disability was due to 'inheritance', but she refers to it as 'fate' One of my sisters is married to my father's sister's son and all of her children are normal. Many cases of marrying within the family in our family and nothing happened. When discussing whether or not they thought their child's disability was genetic many would respond saying: "there are no other cases in the family" or "there is no one in the family like [her]." Amna, who offered such an explanation further expanded, saying it was common practice within her family and did not increase risk: No, I don't think so. I think anything is destined to happen whether they are cousins or not. Aisha, the mother of a daughter with autism, believed her daughter's condition to be caused by her emotional state whilst pregnant: I was upset because of problems between her father and I. I was very upset. We had many daily problems. When I was pregnant, I was very stressed and I was always feeling down. It was [the] probable cause of Maryam's condition. In interviews, respondents often saw the benefits of close marriage as outweighing potential risks (Kilshaw et al 2015). Although familiar with the discourse of genetics, participants were ambivalent about genetic risk and pointed to other possible causes. Panter-Brick's (1991) research found that despite an awareness of genetic risk, participants often accounted for illness and disability through explanation such as: 'evil eye', 'God's will' and illness or upset during pregnancy. In particular, in Qatar a woman's physical and emotional state during pregnancy took importance over other forms of explanation. A woman's emotional state would also affect the baby's temperament. In the UAE Parkhurst (2014) found traits mainly coming from the father, with others are "picked up" or absorbed from the mother in utero. Similarly, in the Qatar the father is seen as providing the dominant template for traits, and the mother as "influencing". Indeed, some suggested that all genetic material comes from fathers. Qatari models of inheritance, including social and biological, which are seen as tracking primarily through men have parallels outside this region. The Alawi of Southeastern Turkey have notions of inheritance that stress patrilineaty based on a concept of the differential transmission of male and female blood / genes (Prager 2015). Shaw and Hurst (2008) identified a "cultural blueprint" amongst British Pakistanis that appears to prioritize inheritance through men. Men are seen as the perpetrators of lineage or identity, with a father's genetic contribution to a child, through the substance of blood, stronger than that of the mother's because semen is concentrated in blood (Shaw 2015). A mother's influence is felt in utero but also through her role in raising children including breastfeeding: women were told not to breastfeed when angry or sad so as not to transmit to the baby. A Qatari pregnant woman is supposed to be calm and relatively tranquil. She is to protect her unborn child from outside risks and also from stress, anxiety and upset (Kilshaw et al, 2016), reflecting notions of maternal influence. Pregnant women, women who had recently miscarried and family members of a disabled child all spoke about the importance of a woman's emotional, psychological and physical state when pregnant. Such discussions appeared to take precedence when people were making sense of problems and linked to cultural systems of blame and risk. Thus, despite the fact that everyday notions of inheritance emphasised the man's contribution, problems in pregnancy or with a child are often attributed to maternal influences. Such findings have been found in other contexts: Shaw and Hurst (2008) found that their interlocutors reported that problems in a pregnancy or a child were blamed on the mother's actions during pregnancy (see also Rozario 2013), particularly in relation to attracting jinn as the result of some bad or careless behaviour (Shaw and Hurst 2008;Shaw 2000:209-212). Ultimately, God determines a person's fate, including in granting health, illness or pregnancy loss. Personal or religious beliefs and local understandings of illness causality may mean that genetic problems are viewed as matters of fate, destiny or God's will, as among Muslims in Saudi Arabia (Panter-Brick 1991). --- Conclusions Genetic discourses have taken a predominant role in approaches to combating a number of conditions that affect Qataris. Qataris have increased susceptibility to some diseases and the state's focus on genetics shows its dedication to improving the lives of its citizens and desire to strengthen the nation. Whilst the incentive behind the development of genomic medicine is to improve the nation's health, this currently rests on a narrow understanding of nationhood. This is particularly interesting in a climate where there are increasing calls to widen the scope of genomics to include a wider diversity of people and to harness such knowledge and information for those most in need. This campaign should be seen in relation to the way genetic discourse is accommodated and understood by in Qatar and with an understanding of local configurations through which genetics has been absorbed as part of the "Qatarisation" process. I have pointed towards the multiplicity of discourses where both the global and the local participate in constituting the meaning of genetics, relevance and scope in the Qatar context. The drive is also part of the nation's positioning itself as a modern, leading and cutting edge entity in regards to research, education, and health care. However, whilst the state is eager to incorporate the modern and continue to develop, there is a desire to balance this with an emphasis on Qatari and Arabic tradition. Thus, there is a constant shifting, balancing, re-positioning and re-inventing as the country and its people negotiate global interactions and influences. This paper has described the unevenness of change in Qatar. There have been dramatic lifestyle changes in some areas but not others. Family marriage is one of the main platforms where one can see the way modern, cosmopolitan and tradition are negotiated. Indeed consanguinity has been shown to be at the heart of tensions and negotiations of modernity and tradition in a variety of global contexts (i.e. Shaw 2015;Prager 2015;Shaw and Raz 2015). This paper contributes to the scant literature on the impacts of genetic discourse on genetic risk in consanguineous marriage in diverse global settings (Shaw and Raz 2015) through a focus on the Qatari context. I have aimed to reveal some of the ironies and contradictions in the focus on genomics as a key area in the quest of modernity because inheritance is simultaneously the key to tradition. Tradition itself is central to the process of containment of outside forces and undesirable change.

Genetic discourses have taken a predominant role in approaches to combating a number of conditions that affect Qataris. This paper is derived from an exploration of Qatari encounters with globalizing discourses of genetics, particularly as they relate to notions of risk. It explores Qataris negotiations of global interactions and influences, including the discourses around genetic risk and cousin marriage. It suggests that family marriage can be seen as one of the main platforms of resistance and a means for modern, cosmopolitan and tradition to be negotiated.

INTRODUCTION Despite advances in medicine, social and economic factors contribute to 50% of a population's health status 1 (Canadian Institute for Advanced Research, Health Canada, 2002), and some estimate that less than 10-15% of mortality is preventable by medical care, with the remainder being attributed to social factors 2. For example, citizens living in the 1% or 5% highest income counties in the USA had better health outcomes compared to average US citizens 3. In contrast, those with lower socio-economic status had a greater prevalence of psychological and chronic health conditions 4. These health disparities arise from historical inequities that result in decreased access to clean environments, housing, quality nutrition, and health care. This in turn predisposes to chronic stress and chronic disease development 5. Patient-centered care is a vital component of health care that improves the physical and psychosocial well-being of patients 6. Shared decision-making, a component of patientcentered care 6, involves assessing the patient's decisionmaking needs, providing individualized support and evaluating patient goals to arrive at a quality decision informed by evidence and patients' values and preferences 7. Shared decision-making is associated with improved patient satisfaction and engagement in care 8. Shared decision-making is facilitated by empathic communication 9 ; it encompasses the cognitive capacity to understand a patient's needs, an affective sensitivity to the patient's feelings, and a behavioral ability to convey this to the patient 10. A 2002 meta-analysis of medical interactions in primary care demonstrated that increased physician empathy was associated with increased patient satisfaction, adherence, comprehension, and perception of a good interpersonal relationship 11. Diabetes is a complex chronic disease that disproportionately affects racialized groups and those of lower socioeconomic status worldwide, with these groups experiencing increased prevalence, lower life expectancy, and increased complications of diabetes 5. For example, in the USA, the risk of type 2 diabetes is 66% higher in Hispanic people and 77% higher in Black people 12 than white people. In Canada, Indigenous people are three to five times more likely to have type 2 diabetes than non-Indigenous people 13. Both low income and education are also associated with increased prevalence of diabetes: individuals with lower income and education are two to four times more likely to develop diabetes than more advantaged individuals 14. In those with diabetes, low socioeconomic status was associated with a two-fold greater risk of all-cause, cardiovascular-and diabetes-related death compared to high-income counterparts 15. A recent population-based study in the UK by Riley et al. (2021) showed that social deprivation is an independent risk factor for developing diabetic foot disease and related complications 16. Health disparities also exist in the receipt of patient-centered care, which may then worsen care gaps. National survey data demonstrate that racialized low-income patients in the USA perceive that they receive less patient-centered care, including less shared decision-making, trust and empathic communication, and as a result are less satisfied with their care 17. Similarly, people living in areas of high deprivation (i.e., low income and education) in Scotland perceive their physicians as less empathic and had less desire for shared decisionmaking 18. A survey study found that physicians viewed Black patients and patients of low and middle socioeconomic status as less intelligent, less rational, and less likely to adhere to medical advice or follow-up, than White and high socioeconomic status patients 19. Additional studies using audiotapes and videotapes of patient-clinician encounters, as well as patient self-report, demonstrate that clinicians exhibit less empathic and participatory communication (characterized by information sharing and patient involvement in discussion) towards racialized and low income patients 20. These implicit biases impact physicians' communication and clinical decision-making 19,20, and may impact patient outcomes. Specific to diabetes care, clinician empathy has been associated with increased patient satisfaction, quality of life, reduced HbA1c, LDL cholesterol, and fewer diabetes complications 21. However, little is known about the relationship between social determinants of health, clinician empathy, and shared decision-making in a diabetes-specific population. Thus, we sought to quantify the relationship between empathic communication, shared decision-making, and sociodemographic factors of income, education, and ethnicity, using validated scales. Our primary objective was to evaluate the relationship between empathic communication and patient education, income, and ethnicity in individuals with diabetes attending primary care clinics in the Greater Toronto Area. The secondary objective was to evaluate the relationship between shared decision-making and patient education, income, and ethnicity in this same population. We hypothesized that patients with lower education, income and from ethnic minorities would experience less empathic communication and shared decision-making compared to those with high education, income, and patients who identified as white. Specifically, empathic communication will be quantified by the Empathic Communication Coding System (ECCS) 10 an observer-rated measure of empathy, which has not been studied in this context before. Shared decision-making will be quantified by with the Decision Support Analysis Tool-10 (DSAT-10) 22 an observer-rated measure of a clinician's ability to engage a patient in shared decision-making. --- METHODS --- Overview and Study Design This is a cross-sectional study and secondary analysis of clinical encounter transcripts from a large randomized controlled trial that evaluated the impact of interprofessional shared decision-making tools for patients with diabetes and other comorbidities, on decisional conflict 23. Sample size was based on 48 available audiotapes from the original study. We reported according to Strengthening the Reporting of Observational Studies (STROBE) guidelines for a cross-sectional study (Supplemental Table 1), with details on the original study and recruitment published elsewhere 24. --- Settings and Participants The previous study was a 10-site cluster randomizedcontrolled trial 25. The trial recruited 53 clinicians from primary care practice groups across the GTA, one of the most multicultural cities in the world, where 51.5% of the city belonged to a visible minority 26. Within each consenting clinician's practice, patients 18 years of age or older, with diabetes and 2 other comorbidities were randomly selected and invited to participate in a study using a web-based goalsetting and shared decision-making aid via telephone, with a total of 213 patients included. Exclusion criteria included those who did not speak English, had documented cognitive deficits, were unable to provide consent, had limited life expectancy (<unk>1 year), or were not available for a follow-up. In order to assess intervention fidelity during the trial (that is, how was MyDiabetesPlan used during the encounter), 48 clinical encounters were audio-recorded then transcribed. This constituted the data source for the current study. --- Study Outcomes The primary outcome was empathic communication, measured using ECCS. The secondary outcome was shared decision-making, measured using DSAT-10. --- Data Sources We used patient-reported sociodemographic information and transcripts of clinical encounters. Patients self-reported their ethnicity, education, and income through an online or mailed survey at the start of the prior trial. In the original study, clinical encounters were audiotaped for qualitative analysis to inform future iterations of the shared decision-making intervention. Of note, the prior study also consisted of patient questionnaires for patient-reported outcomes of decisional conflict, diabetes distress, assessment of care, and quality of life; however, these were not used in the present study. --- Data Collection Tools Assessment of Empathic Communication. We conducted qualitative coding of the clinical encounter transcripts to derive a score for empathic communication, using ECCS. ECCS is an observer-rated measure of clinician empathy that measures empathy by examining clinician empathic responses to patient-created opportunities, with responses subsequently categorized on a scale from 0 (denial) to 6 (shared experience) (Supplemental Table 2). Because it is observer-rated, it eliminates biases associated with self-report. Assessment of Decision Quality. We conducted qualitative coding of the clinical encounter transcripts to derive a score for decision quality using DSAT-10. The DSAT-10 evaluates the clinician's ability to address the status of the decision; the patient's knowledge of the options, benefits, and harms; the patient's values and preferences associated with the decision; assessment of the involvement of others; and the patient's preferred role in the decision-making and the next steps 22. The DSAT-10 scale ranges from 0 to 10, with higher scores indicating more decisional support during the patient-clinician interaction (Supplemental Table 2). Because it is observerrated, it eliminates biases associated with self-report. --- Data Analysis Audio-recordings of the clinical encounters were transcribed verbatim and coded independently using ECCS and DSAT by two team members with expertise in qualitative coding. The first 20 transcripts were double coded until an inter-rater agreement of 75% was attained. Coders were blinded to participant characteristics. We then calculated the weighted average empathy score from the ECCS and the total score from the DSAT-10. The unit of analysis was the transcript, so the average empathy score was calculated by dividing the total score by the number of empathic opportunities per encounter. The DSAT-10 score was reported as a total score out of 10. We a priori selected to evaluate the relationship between sociodemographic factors of education, income, and ethnicity with ECCS and DSAT-10. --- Statistical Methods We used descriptive statistics to describe the characteristics of participants (patients and clinicians) and clinical encounters. For the primary outcome, we used one-way ANOVA to examine the effect of the categorical independent variables (income and education) entered as between-subjects factors, on our continuous dependent variable ECCS. If there was a significant effect of any factor, we conducted exploratory post hoc Tukey's to determine which groups differed from each other 27. For ethnicity (white vs. non-white; binary outcome), we used 2 independent sample t-test. For the secondary outcome, we used the Kruskal-Wallis test to examine the effect of the categorical independent variables (income and education) on our ordinal dependent variable DSAT 28. If there was a significant effect of any factor, we conducted exploratory post hoc Dunn tests to determine which groups differed from each other 27. For ethnicity, we used Mann-Whitney test. We conducted a Benjamini-Hochberg procedure to correct for multiple comparisons, a moderate false discovery rate of 0.15, given the exploratory nature of this study 29. All analyses were done using SPSS Statistics for Windows, Version 20.0 30. --- RESULTS --- Characteristics of Patients, Clinicians, and Clinical Encounters We analyzed a total of 48 clinical encounters, involving 30 unique patients and 23 unique clinicians. Sociodemographic characteristics of patients and clinicians are indicated in Table 1. There were 26 male patients (54%) and 22 female patients (46%). The majority of patients were within the age ranges of 65-74 years old (50%) and retired (64%), with annual income >$60,000 (53%). The sample was primarily white (69%), with income and education being fairly uniform among patient demographics. All patients had type 2 diabetes, with the exception of 1 nonrespondent. Clinicians consisted mostly of family physicians, 61% of which were female, and majority (48%) had more than 16 years of practice (Table 1). Clinical encounters ranged in length from 3 min 55 s to 1 h 30 min, with a mean length of 31 min 26 s (SD 15 min 33 s). Mean ECCS score across all clinical encounters was 3.5 (standard deviation (SD) = 0.8). The most frequent empathic responses were "acknowledgement with pursuit" (29 %) and "confirmation" (30.0%) (Table 2). Mean DSAT score across all clinical encounters was 3.9 (SD = 1.8). The most frequently assessed DSAT components were the stage of decisionmaking (present in 86% of encounters) and intervening to provide the knowledge of options (present in 82% of encounters). The least frequently assessed DSAT components were assessing and intervening regarding the preferred role of the patient (present in 16% and 14% of encounters respectively) (Table 3). --- Relationship Between Patient Sociodemographic Factors and Empathic Communication (Primary Outcome) We found that ECCS was varied by education (p=0.030) and ethnicity (p=0.030), but not income (Table 4). Post hoc analyses of the former revealed that patients with only a college degree received more empathic communication than patients with bachelor's degrees or more, and South Asian patients received less empathic communication than Asian patients (Table 4). --- Relationship Between Patient Sociodemographic Factors and Shared Decision-Making (Secondary Outcome) We found that DSAT varied with ethnicity (p=0.07) but not education or income (Table 4). Post hoc analyses of the former revealed that white patients experienced more shared decisionmaking than those in the "other" category (Table 4). To correct for multiple comparisons, we conducted a Benjamini-Hochberg procedure using a conservative false discovery rate of 0.15. The comparison between DSAT and patient ethnicity had the highest P (=0.07) that was less than its critical Benjamini-Hochberg value (0.08), thus confirming that all preceding comparisons (ECCS and ethnicity, ECCS and education) were significant. The rank table is included in Supplemental File 2. --- DISCUSSION Our study demonstrated that ECCS was varied by education and ethnicity, such that patients with only a college degree received more empathic communication than patients with bachelor degrees or higher, and South Asian patients received less empathic communication than Asian patients. In addition, we found that DSAT varied by ethnicity, such that white patients received more shared decision-making than nonwhite patients/"other." --- Interpretation of Findings/Relevance to Literature These findings are consistent with the existing literature that racialized individuals and those with lower SES perceive less empathy and shared decision-making in healthcare interactions 31. However, we found that patients with only college education received more empathic communication compared to those with a bachelor degree or higher, which reveals a new finding that is contrary to the overarching trend in the literature, despite controlling for income and ethnicity 32. Although several studies have demonstrated that individuals with low income, education and from ethnic minorities experience less empathic communication and shared decision-making, we provide objectively assessed, quantitative evidence of these relationships, in a clinical populationindividuals with type 2 diabetesthat is characterized by ethnic and socioeconomic diversity; these relationships existed despite the study being conducted in a multicultural geographic setting during an era with growing awareness of considerations for equity, diversity, and inclusion. These associations are concrete evidence of systemic bias in healthcare, with roots in medical education. Empathy decreases throughout medical training 33. Further, studies of medical trainees' attitudes towards racialized populations have demonstrated a "pro-white" bias in empathic communication and treatment, in that medical trainees held false beliefs regarding black patients' perception of pain which led to undertreatment 20. This was confirmed in a systematic review by Hall and colleagues, which showed that healthcare providers have implicit biases in terms of negative attitudes towards people of color, which impacted patient-provider interactions, treatment decisions, and patient health outcomes 34. Similarly, patients of lower socioeconomic status perceive less access to care, altered physician-patient interaction (feeling like they are listened to), and differences in management plan (such as reduced diagnostic testing) 35. Taken together, strategies must be implemented at the medical education level to foster empathy and address these biases that are often part of the hidden curriculum. Examples of strategies to enhance empathic communication include assessing provider patient-centered communication at the point-of-care, education among peers, and mentorship by clinicians who score highly on patient-rated scales of patient-centered communication 36. Strategies to address systemic bias in medical education include standardized anti-racism and anti-bias training 37 as well as implementation of a structural competency framework, including improving recruitment, promotion and retention processes of faculty, "stop the line" processes for racism, and the use of a community council to review health equity initiatives and provide feedback on performance 38. --- Strengths and Limitations First, our study was limited by small predominantly white sample; however, this was a hypothesis-generating exploratory study wherein we a priori selected specific outcomes and statistically controlled for multiple comparisons. Second, because our findings were based on audio-recordings of clinical encounters alone, we were unable to assess non-verbal empathic communication, a key component of empathic communication 18. Third, heterogeneity of appointment type (i.e., follow-up vs. initial appointments) may have resulted in differing levels of empathic communication and shared decisionmaking, given that shared decision-making is longitudinal in nature occurring over several appointments 8. We tried to account for this heterogeneity by adjusting for the number of empathic opportunities per encounter; however, we could not adjust the overall DSAT score based on encounter length because of scale properties. Fourth, we did not examine the impact of clinician factors on empathy or shared-decisionmaking, including burnout, workload, gender, and training 32, which may have influenced our results. Study strengths include our use of objective third-party observer-rated scales not previously used in this context that reduce bias associated with patient or clinician self-report. Second, we captured representative clinical encounters with different clinicians, including physicians, nurses, and dietitians, consistent with interprofessional diabetes care. --- NEXT STEPS AND IMPLICATIONS Our research has implications for medical education, clinical practice, and research. That empathy declines with medical training and that racial biases are prevalent in medical trainees call into question the adequacy of medical education in preparing physicians to care for patients with a lens of social justice. Several professional organizations have advocated for education interventions to prepare medical trainees to care for the needs of a culturally diverse population including cultural competency training 39, and incorporating critical reflection and dialogue into curriculum to address biases and assumptions that shape healthcare interactions 40. In terms of clinical implications, our research underscores the importance of clinical interventions such as shared decision-making tools to empower patientsin particular racialized individuals and those of lower income and educational attainmentto become involved in their healthcare. Additional supports such as interprofessional teams and peer coaches 36 should be leveraged to enable patients in vulnerable groups to play an active role in their care. In terms of implications for research, future studies should confirm our findings, and specifically assess the trend of education level and empathy in a larger and more diverse patient population during chance encounters. Triangulation of objective rating of empathic communication (as in our study) with patient self-report of empathy as well as the patient lived experience using qualitative methodology would enhance our understanding. Future studies could also test the impact of patient-directed interventions (such as peer coaches) as well as clinician-directed interventions (such as professional development regarding implicit bias aimed at improving empathic communication or shared decision-making) in specific vulnerable populations. --- Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s11606-021-07230-5. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. --- Author Contribution Publisher's Note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

To assess the relationship between empathic communication, shared decision-making, and patient sociodemographic factors of income, education, and ethnicity in patients with diabetes. RESEARCH DESIGN AND METHODS: This was a crosssectional study from five primary care practices in the Greater Toronto Area, Ontario, Canada, participating in a randomized controlled trial of a diabetes goal setting and shared decision-making plan. Participants included 30 patients with diabetes and 23 clinicians (physicians, nurses, dietitians, and pharmacists), with a sample size of 48 clinical encounters. Clinical encounter audiotapes were coded using the Empathic Communication Coding System (ECCS) and Decision Support Analysis Tool (DSAT-10).The most frequent empathic responses among encounters were "acknowledgement with pursuit" (28.9%) and "confirmation" (30.0%). The most frequently assessed DSAT components were "stage" (86%) and knowledge of options (82.0%). ECCS varied by education (p=0.030) and ethnicity (p=0.03), but not income. Patients with only a college degree received more empathic communication than patients with bachelor's degrees or more, and South Asian patients received less empathic communication than Asian patients. DSAT varied with ethnicity (p=0.07) but not education or income. White patients experienced more shared decision-making than those in the "other" category. CONCLUSIONS: We identified a new relationship between ECCS, education and ethnicity, as well as DSAT and ethnicity. Limitations include sample size, heterogeneity of encounters, and predominant white ethnicity. These associations may be evidence of systemic biases in healthcare, with hidden roots in medical education.

Introduction Discrimination is a systemic arrangement that molds opportunities and assigns value depending on the societal interpretation of physical appearance, which is frequently referred to as "race." (Jones CP, 2023)Certain individuals and communities are unfairly disadvantaged by this system, while others are unfairly benefited by it. As a result, the strength of the entire community is ultimately depleted as a result of the inefficient utilization of human resources. In order to promote talks, it is helpful to define racism as a systemic issue rather than an individual weakness, personal moral shortcoming, or psychiatric disease. This shift away from the contentious notion of categorizing individuals as either racist or not constitutes a significant step toward achieving this goal. It emphasizes that everyone has a stake in destroying this system and replacing it with one that enables all individuals to fulfill and develop their full potentials by recognizing that racism weakens the strength of the entire society. This is done by realizing that racism undermines the strength of the entire society (Miller et al., 2019). For example, my novel Gardener's Tale depicts three layers of racism: institutionalized, personally mediated, and internalized (Jones CP, 2000). It also places a significant emphasis on the necessity of addressing institutionalized or structural racism in order to repair flaws that are present in our society's "garden." Additionally, the relevance of addressing the ways in which racism shapes possibilities and the ways in which it assigns worth is brought to light by this narrative. Even if the gardener were required to enhance the poor, rocky soil to equal the richness of fertile soil, the preference for red blooms over pink blossoms would continue to exist if the issue was not addressed. In the context of our attempts to combat racism, this narrative highlights the significance of addressing both opportunity structures (the unequal access to social goods, services, and opportunities that are based on "race") and value assignments (the ideology of white supremacists) (Jones CP, 2000). The impact of racism on social media is quite substantial and has a variety of complex implications in today's day and age society. The use of social media platforms offers a wide range of opportunities for social contact; but, in certain instances, these platforms have become a medium through which racist attitudes are disseminated, reinforced, and maintained. Social media platforms have the potential to serve as a venue for the propagation of hate speech and racial prejudice. Numerous users of social media sites make advantage of these channels in order to publicly express their racist beliefs, so creating an environment that is both unsafe and destructive for minority groups. During the same time period, this phenomenon has the potential to generate polarization as well as wider social conflicts. The use of social media platforms can be a tool for the perpetuation of racial prejudices and stereotypes (Chetty &Alathur, 2019). Through the transmission of photos, memes, and other content, these platforms have the potential to reinforce limiting beliefs about particular groups, so establishing perceptions of those groups that are shallow and negative. The problem of "hate algorithms" operating on certain platforms might result in the construction of information bubbles, in which people are largely exposed to viewpoints that are congruent with their own ideas. A greater understanding between groups can be hampered as a result of this, which can also exacerbate social division.On the other hand, social media platforms can also serve as a means of resistance and activism against white supremacy. Through the use of these platforms, campaigns, protests, and civil rights movements are frequently coordinated and documented (Matamoros-Fernández & Farkas, 2021). This provides a voice to underrepresented groups and raises awareness about racial concerns.Therefore, the impact of racism on social media generates a number of complex challenges that must be overcome in order to achieve an online environment that is inclusive and equitable. It is necessary for platforms, users, and politicians to work together in order to reduce the negative effects of social media and make the most of its potential as a constructive instrument in the fight against racism and the promotion of mutual understanding among various groups. Social media serves as a powerful arena for public discourse, offering a dynamic space where individuals from diverse backgrounds can voice their opinions and engage in discussions about the cultural implications of films like "The Little Mermaid." Through hashtags, tweets, and posts, users amplify their voices, sparking a digital dialogue that critically assesses the film's representation of race. This essay aims to explore how these discussions unfold, examining the various perspectives that emerge on platforms such as Twitter, Instagram, and TikTok (Chetty &Alathur, 2019). Humans are created with various physical forms. Starting from face shape, body, hair color, skin color and others. We will discuss skin color, human skin color is very diverse, ranging from white, olive, brown and black. On various continents in the world, people have different skin colors. For example, on the European continent, the population there is predominantly white, whereas on the African continent the majority of the population has dark skin. Well, this is one of the differences between each continent throughout the world.Each country has its own beauty standards, such as in Asia the standard of beauty is having white skin, in Japan having crooked teeth, in America having dark skin, in Mindanao the Philippines having sharp teeth, and many more, because of this. there is a lot of racism when we visit other countries.Actually, not in all countries this can happen. However, there have been many incidents such as cases of people with dark skin that have occurred in several countries such as America and China. As happened this year when the film was released with the title The Little Mermaid. This American film has caused a lot of controversy because the main actor has dark skin color.Especially in China and South Korea, their lack of interest in the film The Little Mermaid is because, as is known, Ariel is known to have pure white skin. However, the reality is that the player who plays Ariel has dark skin color. Apart from that, they thought that the film had damaged their imagination about the figure of Ariel, who they knew had white skin. We will discuss this in more depth. In this paper, we can see the impact of racism in The Little Mermaid on society by using Wolfgang Iser's literary theory approach. Iser's theory of reading activities includes the concept of meaning mining and the role of the reader as an active constructor in understanding literary works. By applying this theory, we can analyze how the messages of racism in this film can affect our interpretation and perception of racial and ethnic differences. sIn an era where inclusion and awareness of cultural diversity are increasingly important, movies, as one of its popular forms, have the power to shape people's opinions and attitudes. For this reason, it is important to examine how racist messages in films such as The Little Mermaid are widely received and perceived, especially by younger audiences, who are the main target audience of this animated film. Iser's theory introduces a useful tool to examine how audiences interact with film narratives and how they contribute to constructing the meaning of the film. In the context of racism in The Little Mermaid, Iser's theory can help us understand how audiences interact with elements of racism hidden in the film, both consciously and unconsciously. Through an in-depth analysis of stereotypical representations, the portrayal of certain characters, and the narrative as a whole, we can gain a greater understanding of how these influences can create negative and detrimental impressions of certain groups in society. By considering Iser's theory, we will also look at how individual interpretations play a role in influencing our perceptions and attitudes towards the heightened racial and ethnic issues in the movie. In addition to looking at the consequences of racism in The Little Mermaid on an individual level, we will also look at how screenings of this movie can influence children's mindsets and the long-term impact in creating an inclusive culture or reinforcing harmful stereotypes. Then, by considering Iser's theory, we can see how the role of the audience, especially children, contributes to shaping their own understanding of racial and ethnic issues. Throughout this paper, we will discuss and critically evaluate racism in The Little Mermaid using Wolfgang Iser's theoretical approach. Through in-depth understanding and analysis, we will understand the impact it may have on society. In addition, we can also reflect on how the handling of racism in this movie can provide opportunities for better teaching and understanding of racial and ethnic issues that exist in and around us. By using Iser's theory as a framework, we can dig deeper to gain a broader and more objective perspective on racism in The Little Mermaid and its implications on our society. The film "The Little Mermaid," a beloved classic from the Disney canon, has been a source of fascination and critique from various points of view, particularly in its portrayal of black people. This iconic animated tale, initially released in 1989, has recently come under scrutiny for its casting choices and representation, sparking discussions about diversity and inclusion in the world of entertainment. "The Little Mermaid" follows the journey of Ariel, a curious and adventurous young mermaid who dreams of experiencing life on land. While the film's narrative primarily centers on Ariel's pursuit of love and independence, it has also been examined through a lens that highlights the lack of racial diversity and representation in its characters. One notable point of view centers around the casting of the main character, Ariel. In 2019, Disney announced that Halle Bailey, a talented African-American actress and singer, would be portraying Ariel in an upcoming live-action adaptation of the film. This casting decision was met with both enthusiastic support and backlash, shining a spotlight on the broader conversation surrounding racial representation in the entertainment industry. Critics argue that the rejection of the world community, in this context, reflects a historical pattern of underrepresentation and exclusion of black people in mainstream media, including animated films. The discussion extends beyond "The Little Mermaid" to encompass the larger issue of diversity in Hollywood and the importance of offering more inclusive narratives that reflect the diverse makeup of our global society. In this introduction, we will explore various perspectives on the rejection of the world community's views toward black people within the context of "The Little Mermaid," shedding light on the film's impact and its role in the ongoing conversation about representation in the entertainment industry. --- Discussion Researchers and academics utilize a complicated set of criteria to identify and assess the multifaceted nature of racism, which is a social problem that has been around for a long time and has deep historical roots (Banaji et al., 2021). The fact that racism is a systemic phenomenon, meaning that it goes beyond individual prejudices and encompasses institutional and structural inequities, is one of the most important categories of criteria. One of the most important indicators of systemic racism, according to the arguments of academics, is the fact that racial inequities continue to exist inside a variety of society institutions, such as the criminal justice system, healthcare, and education. For the purpose of comprehending the pervasive influence that racism has on a societal level, it is essential to investigate policies and behaviors that disproportionately affect specific ethnic groups (York, 2018). In the academic discourse on racism, implicit bias is another criterion that is regularly covered and investigated. This idea acknowledges the fact that racial biases can arise unconsciously, so affecting the decision-making process and contributing to the perpetuation of discriminatory behavior (Banaji et al., 2021). The psychological aspects of racism are investigated by researchers, who analyze the ways in which deeply ingrained prejudices and stereotypes can influence the attitudes of individuals and lead to larger-scale institutional disparities. When it comes to understanding the subtle yet significant ways in which racism manifests itself in a variety of aspects of life, it is vital to unravel the nuances of implicit bias (Sophia &Kulaszewicz, 2015). One of the most important criteria for comprehending racism is the historical background. The historical legacies of colonization, slavery, and structural oppression are believed to be necessary in order to comprehend the modern expressions of racism, according to the arguments of academics. In order to establish comprehensive solutions to address and repair systemic racism, it is necessary to investigate the ways in which systemic racism continues to resonate in the power structures and societal dynamics of the current day.One of the criteria that emphasizes the connectivity of different forms of oppression and social identities is called intersectionality. It is the contention of researchers that racism does not exist in a vacuum but rather interacts with other types of discrimination, such as biases based on classification and gender (Matamoros-Fernández & Farkas, 2021). A more nuanced understanding of the various manifestations of racism can be gained by first acknowledging the complicated ways in which individuals experience several layers of oppression. The identification of racism in everyday interactions can be accomplished through the use of microaggressions, which are small but significant criteria. Derogatory messages are communicated through these brief, often inadvertent, verbal or behavioral slights, which lead to an environment that is hostile toward people of different races. When it comes to understanding the varied ways in which racism presents itself on an interpersonal level, academics stress the need of recognizing and resolving microaggressions.Through the appropriation or denigration of cultural practices and symbols that are linked with particular racial or ethnic groups, cultural racism develops as a criterion that investigates how racial biases present themselves in the world. The media and popular culture are responsible for the perpetuation of cultural stereotypes, the reinforcement of racial hierarchies, and the formation of biased perspectives, as this criterion sheds light on (Shiao& Woody, 2021). Certain regions might become racist as a result of structural disadvantages or discriminatory behaviors, and racialized spaces and place-based discrimination are two criteria that bring attention to this phenomenon. The geographical elements of racism and the unequal distribution of resources are brought to light by the research conducted by academics who explore the ways in which individuals may be confronted with exclusive issues as a result of the racist environments in which they reside (York, 2018). Walt Disney Pictures created the animated feature The Little Mermaid. The film is based on the beloved tale by Hans Christian Andersen about Ariel, a mermaid who longs to be a human. The following elements can be taken into account while analyzing this movie. First, narrative the movie has a compelling plot with a distinct beginning, middle, and finish. Ariel's ambition to travel and find her true love are the primary sources of tension in the story. Characterization: This movie has welldeveloped characters. Ariel is characterized as a bold and adventurous mermaid with a strong personality. Ursula and other antagonistic characters possess powerful traits as well. Images and Animation, this movie has strange and magical animation. Beautiful visual elements abound, particularly when illustrating the undersea environment and well-known musical moments like "Under the Sea" and "Part of Your World." Music, this movie has a number of well-known and memorable tunes. Alan Menken's music and Howard Ashman's words combine to provide the audience a fully immersive experience. Tema, this film explores themes of self-identity and searching, selfeducation, and cinning with tulus. King Triton, the head of the duchy, and Ariel's sahabat, Flounder, were also Ariel's neighbors and benefactors. Moral lesson,the Little Mermaid emphasizes the need of accepting oneself and forgiving oneself. Ariel accepts the consequences of her actions and learns to understand what is, in all honesty, important in her life. This is a summary of the Little Mermaid movie analysis. Every film analysis might vary depending on the background information and advice provided by writers. According to Iser literary texts can only produce a response when read. Therefore, it is impossible to describe the (reader's) response without analyzing the reading process.The analysis carried out was on social media, where the search for comments on social media related to the film The Little Mermaid. Below are some of the comments found. --- Data Presentation a. Table This research is based on negative comments for the film on Instagram, where several comments we took as'screenshot' evidence prove that many people still criticize the black main character in this film. Because it's mostly Disney. Princess has white skin according to Brownmiller (2013, 145) indeedconsidered feminine and often complimented. In terms of the pros and cons of the main actors, there are still many people who think straight that each actor or character cannot be seen physically, especially skin color, because basically, the remake of this film will be judged based on the quality of the actor's acting. The following are some comments regarding the Little Mermaid film: 1) 2) This work is licensed under Creative Commons Attribution-ShareAlike 4.0 Internasional. Copyright <unk> 2024, Elvira Eka Jayanti, et al. --- 3) 4) From the data table, several comments were found which were more towards hate comments, but there were also some positive comments. This explains that the little mermaid film is actually a good film, but it goes back to the beginning because the main character has a very different skin color from the Ariel in the cartoon. This makes many assumptions of negative comments. This data was taken from several social media such as Instagram, YouTube and Twitter. --- A. Comments 1 In comment one as we saw in the initial comment he mentioned that stop blaclwashing because he hasn't seen the film. Some people may not have watched the film because after seeing the trailer, the main character turns out to have different skin, so maybe they are too lazy to watch the film. But in the second comment, it is actually the opposite of the first comment, in the second comment it actually leads to a worse comment because there are black putrid words. --- B. Comments 2 The second comment is in the form of an opinion that expresses someone's disappointment with the film The Little Mermaid. He probably had more hopes for the film because the Ariel cartoon was his favorite cartoon. But what can I do just because the main character causes him to have an opinion like that? --- C. Comments 3 Comment 3 in the first comment said that it was not as expected but he still said that the film was still good. The second comment is more insulting or mocking because it says that the main character lives mostly at sea and forgets to wear sunblock so his skin burns. --- D. Comments 4 In comment 4 it is almost the same, in this comment they also express their disappointment with the film which was later released but did not meet their expectations. These comments are a sign that the main problem is about the main character. Even though in fact there is nothing wrong with the main character just because of the color of his skin, this has caused a lot of controversy. --- Conclusion The controversy surrounding the casting of Halle Bailey, a black actress, as Ariel in the live-action adaptation of "The Little Mermaid" has sparked discussions about racism and representation in the media. Some people have rejected the idea of a black actress playing the role of a character who was originally depicted as white. The controversy highlights the ongoing struggle for representation and diversity in the entertainment industry. The purpose of this research is to see how people respond to the film The Little Mermaid in real life with a black main character. The Little Mermaid, a Disney classic, has been the subject of controversy due to the casting of a Black actress, Halle Bailey, as Ariel in the upcoming live-action remake. Some people have refused to accept a Black actress in the role of Ariel, which illustrates how a white-centric perspective requires deeper understanding. The backlash against the movie has led to abysmal box office numbers in China and South Korea, showing the global reach of anti-Blackness. However, the importance of representation in media has been highlighted, and the casting of a Black actress as Ariel is seen as a positive step towards diversity and acceptance. The negative attention on The Little Mermaid has fueled a groundswell of support from other fans who view such anger as founded in racism. Humans are created with various physical forms, including diverse skin colors. Each country has its own beauty standards, leading to racism when visiting other countries. The film "The Little Mermaid" has caused controversy due to the main actor having dark skin color, leading to lack of interest in China and South Korea. This paper aims to analyze the impact of racism in The Little Mermaid on society using Wolfgang Iser's literary theory approach, focusing on how the messages of racism in the film can affect interpretation and perception of racial and ethnic differences. In an era where inclusion and awareness of cultural diversity are increasingly important, it is important to examine how racist messages in films are widely received and perceived, especially by younger audiences. The film "The Little Mermaid" has been a source of fascination and critique, particularly in its portrayal of black people. The casting of Halle Bailey, an African-American actress, as Ariel in an upcoming live-action adaptation of the film, has sparked discussions about diversity and inclusion in the entertainment industry. Critics argue that the rejection of the world community's views toward black people reflects a historical pattern of underrepresentation and exclusion of black people in mainstream media, including animated films. The discussion extends beyond "The Little Mermaid" to encompass the larger issue of diversity in Hollywood and the importance of offering more inclusive narratives that reflect the diverse makeup of our global society.

The controversy surrounding the casting of Halle Bailey, a black actress, as Ariel in the live-action adaptation of "The Little Mermaid" has sparked discussions about racism and representation in the media. Some people have rejected the idea of a black actress playing the role of a character who was initially depicted as white. The controversy highlights the ongoing struggle for representation and diversity in the entertainment industry. This research aims to see how people respond to the film The Little Mermaid in real life with a black main character. The Little Mermaid, a Disney classic, has been controversial due to the casting of a Black actress, Halle Bailey, as Ariel in the upcoming live-action remake. Some people have refused to accept a Black actress like Ariel, which illustrates how a white-centric perspective requires a more profound understanding. The backlash against the movie has led to abysmal box office numbers in China and South Korea, showing the global reach of anti-Blackness. However, the importance of representation in media has been highlighted, and casting a Black actress as Ariel is seen as a positive step towards diversity and acceptance. The negative attention on The Little Mermaid has fueled a groundswell of support from other fans who view such anger as founded in racism.

INTRODUCTION Physical activity levels and impacts Across Europe, less than one-third of people aged 2-18 years old achieve the recommended 60 min of moderate-to-vigorous physical activity per day. 1 Such high levels of physical inactivity are fuelling a worldwide public health problem and negatively impacting children's physical fitness, cardiometabolic health, bone health, cognitive outcomes, mental health and adiposity. 2 3 Physical activity levels reduced further during the COVID-19 pandemic, 4 with recent evidence suggesting the reductions have remained post-lockdown. 5 Children from socially disadvantaged areas and ethnic minorities-who were already the least active-were most negatively impacted by the COVID-19 pandemic, further exacerbating health inequality. [6][7][8][9] Additionally, across all age groups, girls are less active than boys, creating gendered health inequity. 1 Progressing to whole-systems approaches to physical activity To date, most interventions have focused on individual behaviour change, resulting in minimal effects on physical activity behaviours. 10 11 To address this issue, the WHO Global Action on Physical Activity report proposes a systems-based approach, involving cross-government, multisectoral partnerships and community engagement. 12 To enact effective systems-based approaches, it is recommended that programmes work closely with local people to develop solutions --- STRENGTHS AND LIMITATIONS OF THIS STUDY <unk> The study design allows children and families to be at the centre of our understanding of what encourages and discourages them to be active. <unk> By conducting citizen science as part of a reactive process evaluation, improvements to the research and the implementation can be made in real time, centred around those who matter most. <unk> The data collection methods are designed to facilitate a positive participant experience. <unk> The study has a small sample size, which will likely make it more difficult to capture a diverse range of experiences. 2 Frazer M, et al. BMJ Open 2023;13:e069334. doi:10.1136/bmjopen-2022-069334 Open access tailored to intended recipients' context and experience. [13][14][15] Recently, Sport England (the arms-length body of government responsible for growing and developing grassroots sport and getting more people active across England) has invested £100 million across 12 Local Delivery Pilots. 12 The purpose of each pilot is to design and implement a whole-systems physical activity approach, providing a unique opportunity for evaluation. --- Evaluations of whole-systems physical activity interventions There is a growing body of evidence surrounding wholesystems evaluations 14 16 17 that suggests mechanisms underpinning complex and whole-systems interventions are likely more varied and dynamic than singular or less complex interventions. 18 As physical activity interventions become more complex, it is increasingly important to explore what works, where, for whom and in what contexts. 19 This is essential to understand the transferability, replicability and upscaling of interventions and to inform policy change. 20 Additionally, timely and integrated evaluation can inform dynamic systems change through continuous improvement of the intervention design and implementation. 14 Literature reviews of evaluations of system approaches to health recommend several approaches and methods: the embedded researcher approach, qualitative inquiry (process evaluation) through a systems thinking lens, systems mapping, network mapping, ripple effect mapping and dynamic systems modelling. 14 17 21-23 Such methods are applied in evaluations of whole-systems approaches. 14 17 19 24 However, to our knowledge, there are no existing guidelines, protocols or evaluation studies reporting on evaluating whole-systems approaches focused on children and families. Given the increasing adoption of whole-systems approaches to physical activity and/or other health behaviours, there is a need to build further evidence around appropriate, effective and innovative methods for evaluating systems-based interventions, 21 22 including those involving children and families. --- Using citizen science to evaluate whole-systems approaches to physical activity The United Nations asserts that children have the right to contribute to decisions that impact them personally and affect the services they use. 25 Therefore, it is essential that children and their families are integral stakeholders within the evaluation of whole-systems physical activity approaches. Moreover, including local people promotes a more comprehensive and contextual understanding of what works and for whom. 14 At present, previous evaluations of whole-systems physical activity approaches have not placed children and families at the centre. One way to address this is through involving the public through a citizen science research approach, which could improve research quality and lead to system changes. 26 By placing children and families at the heart of research, 27 28 their needs can be better understood, and programmes adapted accordingly. A growing cross-disciplinary body of evidence demonstrates the benefits of taking different citizen science approaches with young people. [29][30][31][32][33][34][35] Citizen science has been used successfully to understand young people's physical activity experiences. 36 However, the potential to understand whole-systems physical activity approaches using citizen science has not been realised. 22 A key principle of citizen science is that citizen scientists should benefit from participating. 37 Research indicates that positive experiences for young people can be achieved by considering power dynamics, relationships and personal growth within citizen science. 29 32 38-40 Furthermore, it is recommended that citizen science projects evaluate participant experience to understand the value of young people's contribution and to improve outcomes. 41 42 The purpose of this paper is to describe a citizen science evaluation approach of the Join Us: Move, Play (JU:MP) whole-systems physical activity intervention with children and families. A secondary purpose is to outline the evaluation of participant experience within the citizen science process. --- METHODS AND ANALYSIS Aims and objectives This paper describes a protocol for a citizen science research study, aiming to understand the mechanisms through which a whole-systems approach to physical activity (JU:MP) influences behaviour change among families and to evaluate participant experience within this citizen science project. The specific objectives are: 1. To understand perspectives and lived experiences around the physical activity of children and families in JU:MP delivery areas. 2. To assess the feasibility, fidelity and acceptability of JU:MP among children and parents/carers. 3. To examine the mechanisms of change that underpin the physical activity behaviour of children and families, when, how and why this happens within JU:MP. 4. To contribute to dynamic systems change through informing programme refinement based on ongoing findings from objectives 1-3. 5. To formatively understand children and families' experience as participants within a contributory and collaborative citizen science approach, to inform continuous study delivery improvements. --- Study context During the pandemic, 73% of Bradford children (9-13 years) were not meeting physical activity guidelines. 4 On average, children of South Asian heritage and females were less active than their white British peers and males, respectively. 4 Within the Bradford district, the number of children overweight or obese is higher than the national average (37.9% vs 34.2%), with higher levels in the most deprived areas. 43 Specifically, within the JU:MP area, average income is significantly below the UK average. 44 Bradford is the youngest city within the UK, with 24% of residents under the age of 16 years. 43 Open 1). Further information can be found in Hall et al. 19 The overarching JU:MP evaluation The JU:MP evaluation employs a mixed-methods approach, which sits within the complementary philosophies of realist and'systems thinking' methodology. 48 The concurrent mixed-methods approach contains two main elements: an effectiveness and a process evaluation (figure 1). The effectiveness evaluation includes: (1) a controlled trial focused on primary-age children, examining effectiveness at the neighbourhood level, and (2) a pre/post-evaluation of children within the Born in Bradford cohort study at age 7-11 years, 49 and again at age 13-15 years, examining effectiveness at the population (North Bradford) level. The primary outcome is children's moderate-vigorous intensity physical activity. 50 The process evaluation includes an examination of the mechanisms and contextual factors influencing the implementation and impact of JU:MP and includes a focus on policy and strategy, overarching work streams, the JU:MP neighbourhood approach, and children and families. A mixed-methods data collection approach includes semistructured interviews, observations, documentary analysis, surveys and participatory evaluation methods (eg, reflections and ripple effect mapping). For further information on the overarching process evaluation, see Hall et al. 19 This paper describes two citizen science evaluation studies that are part of the overarching process evaluation: (1) an interview and focus group study with primary-age children and their families; and (2) a year-long collaborative study with secondary-age children and their families. Patient and public involvement Public involvement through the citizen science methodology is integral to this study, as described in the Methods section of this protocol. Wider public involvement with youth research ambassadors from Born in Bradford 50 shaped the overarching study design, including, for example, incentives, methods and realistic time commitments. Citizen science approach Two interlinked longitudinal citizen science studies will help understand child and family experiences of the JU:MP programme. These will be complemented by an evaluation of the citizen scientists' experiences. Citizen science is a transdisciplinary participatory method. 51 While there is no agreed definition, [52][53][54] for this paper, we define citizen science as the involvement of members of the public who work with professional scientists to advance research. 55 Citizen science projects can be viewed on a continuum. On one end, citizens are 'passive contributors' to activities 56 ; on the other, citizen scientists are fully immersed in a local community, and the research is a joint enterprise to help identify and solve societal issues. 56 Using Shirk et al's 57 typology, study one adopts a contributory model approach, where projects are designed by researchers and members of the public primarily contribute data. Study two adopts a collaborative model approach, where researcher staff create the project and'members of the public contribute data and help to refine project design, analyse data, and/or disseminate findings'. 57 Guidance exists on conducting high-quality citizen science research, 37 including young people in research 58 and co-production in a Bradford context. 59 The following seven principles, adapted from the above documents, will guide the current citizen science approach: 1. Child-friendly involvement, including clear communication and feedback. A longitudinal research design will provide an understanding of families' direct experience with JU:MP and how this evolves. 60 61 The longitudinal nature will facilitate the development of a meaningful relationship between the child, families and the researchers, 28 29 Open access provide time to explore change mechanisms and take seasonal variations in children's physical activity levels and sedentary time into account. 62 --- Study design Herein, we describe the design of the two studies that make up the citizen science evaluation, including contributory citizen science (study one) and collaborative citizen science (study two) (see figure 2). The two studies have been designed in a way that the level of citizen science participation required is age appropriate. [63][64][65] The contributory citizen science study will accommodate the views of younger children and Open access their families through focus groups and interviews. 64 For the collaborative citizen science study (ages 12-14 years), participants choose the study method, which may involve independent data collection in different neighbourhood locations and/or independent use of technology. These methods may not be suitable for primary-age children due to parental concerns of personal access to smartphones. 66 67 Ongoing findings from both studies will be added to the agenda for the six weekly research and implementation meetings, where they will be discussed, and actionable outcomes created to inform the JU:MP programme development and delivery. Actions are added to future meeting agendas and progress is recorded. Ethics approval was granted by the Chair of Humanities, Social and Health Sciences Research Ethics Panel at the University of Bradford for both studies (June 2022). For study two, given the collaborative citizen science approach, ethics approval covers participant on boarding, the study in principle and the first workshop. Ethical amendments will be submitted at (at least) two further points: (1) once the data collection method has been decided on and (2) once the data analysis approach has been finalised (further details in the Data analysis section). Equity of access has been considered in both studies. In addition to the information mentioned in the study design, any reasonable accommodation will be made to include eligible children who wish to take part. This will include following schools' instructions on how to accommodate learning differences, as well as ensuring physical accessibility for children with disabilities. If language is a barrier to participation, suitable accommodations will be made, for example, the provision of study information in the relevant language or a translator. If further specific barriers to accessing the research arise, current best practice will be followed to ensure inclusion. Study one: contributory citizen science with primary-age children and their families Study one uses two data collection methods: focus groups (separate child and parent focus groups) and parent-child dyad interviews. Focus groups allow for a diverse range of responses and provide children with less intimidating environments than interviews. 64 The parent-child interviews-which pair a child with their primary caregiverallow for more detailed and longitudinal exploration of people's experiences and engagement with JU:MP, and physical activity behaviour change mechanisms. Parentchild dyad interviews will aid understanding of how parents shape children's behaviour as well as allowing parents to expand on the child's verbal expression. 68 69 The study draws on a range of concepts relevant to intervention evaluation that the focus group and interview topic guides were informed by 48 --- (see table 1). Study one: sampling, recruitment and incentives Focus group and interview participants will be recruited from the three neighbourhoods included in the JU:MP control trial (see The overarching JU:MP evaluation section) to permit integration of findings across evaluation components. Each family will receive a £20 voucher per interview as a token of appreciation for their time. Focus group recruitment: parents and children will be recruited from the sample who provided consent --- Open access and assent to engage in the JU:MP control trial when in primary years 1-3 (aged 5-8 years). The sample for the focus groups will be recruited when the children are in years 2-4 (aged 6-9 years). Schools will be selected from different areas to ensure a geographical spread of participants. Parent focus groups will typically take place in different schools to child focus groups to enhance the diversity of responses. Children and parent participants will initially be randomly selected but revised based on advice from teaching staff (eg, if the family has since left school, or if the sample lacks diversity in socioeconomic backgrounds, ethnicity or physical activity behaviours). Different participants will be selected at different data collection time points to allow a range of families to share their experiences. Two weeks before the child focus group, parents will receive a letter or email from their child's school to notify them and provide an opportunity to withdraw consent. Verbal consent (parent group) and assent (child focus group) will be recorded at the start of each focus group. Parent-dyad interviews: the eligibility criteria are as follows: (1) those who live in the JU:MP control trial neighbourhoods and (2) including a child in primary years 2, 3 or 4. Participants will be recruited through local community networks and social media. A random sample will be selected from those who express interest in participating and have suitable availability. Informed verbal consent will be obtained from adult participants, and assent from child participants. --- Study one: data collection methods Interview data collection will take place approximately 9, 21 and 27 months after the JU:MP acceleratorphase programme launch (September/October 2022, September/October 2023 and March/April 2024). Focus group data will be collected at the first two time points. Multiple data collection time points enable findings to provide repeated feedback on programme delivery to enhance the likelihood of engagement with and impact of JU:MP. Participants' basic demographics (postcode district/name of neighbourhood; gender; age; ethnic group) will be collected during the interviews and focus groups to describe the sample characteristics relative to the population. Focus groups: six focus groups with children and three focus groups with parents will be held in schools. Two researchers will be present at each focus group: one to facilitate the session and one to take reflexive field notes. The children's focus groups will incorporate the Write, Draw, Show & Tell 70 technique to explore children's understanding of physical activity. This participatory visual method allows children to express their views, thoughts and emotions non-verbally and/or verbally, facilitating inclusivity and engagement. 70 Images depicting various <unk> What has made it easy/hard to engage with JU:MP <unk> What would make it easier to engage with JU:MP Participant fidelity [89][90][91] The extent to which participants understand the concepts and purpose of the intervention and are exposed to and engage with the intervention. <unk> Exposure participants have to different elements of JU:MP <unk> Receipt of knowledge of physical activity and its benefits; and knowledge and understanding of the JU:MP intervention <unk> Responsiveness-the extent to which participants feel the various components of JU:MP are useful <unk> Engagement-the components they report seeing/ hearing about and/or engaging with/or taking part in Acceptability 92 Anticipated or experienced cognitive and emotional responses to the intervention. <unk> Affective attitude (anticipated or experienced thoughts and feelings about JU:MP) <unk> Burden (the anticipated or experienced amount of effort required to engage with JU:MP) <unk> Ethicality (the extent to which JU:MP fits within their value system) <unk> Self-efficacy (participant's confidence that they can perform the behaviour required to participate in JU:MP) <unk> Opportunity costs (anticipated or experienced benefits or values that are given up to engage with JU:MP) Mechanisms of change 93 How does the intervention produce change within the delivery context? Open access components of JU:MP will also facilitate discussions (see table 1 for an overview of focus group discussion topics). Parent-child dyad interviews: nine parent-child dyads (three from each neighbourhood) will be interviewed at each time point. Photo elicitation methods will be used to stimulate discussion. Before each interview, families will be asked to take photos showing their physical activity as part of a normal week and images depicting various components of JU:MP will be shown in interviews. Both sets of images will be used to facilitate discussions around child and family physical activity behaviours. Interview questions will focus on the parent and child's understanding of physical activity, their (change in) physical activity behaviours, why they may or may not have engaged with the JU:MP programme and what impact this has had (table 1). Study one: data analysis Data analysis will be undertaken using a framework approach. 71 Interviews and focus groups will be audiorecorded and transcribed verbatim. Once imported into NVivo (QSR International, Melbourne, Australia), data will be coded into a framework previously developed to analyse wider JU:MP process evaluation data. This will facilitate the integration of data and findings with the broader process evaluation. Further detail as to how the framework was developed and the rationale for this approach can be found in Hall et al. 19 A recurrent crosssectional analysis will be applied for the focus group data to explore differences across the three different focus group cohorts. 51 For interview data, a trajectory analysis will be used to examine change over time within the parent-child dyads participating in the study. --- Study two: collaborative citizen science with secondary school-aged children and families Study two is a collaborative approach-with data collection developed in partnership with the citizen scientists. The longitudinal study will take place over a year (September 2022-September 2023) and focus on the citizen scientists and their families' experiences of physical activity and the JU:MP intervention. --- Study two sampling, recruitment and incentives We aim to recruit six female citizen scientists, aged 12-14 years, and their families. We are targeting females due to females being consistently less active than males 1 and our public involvement work indicating that families, particularly those from South Asian backgrounds, may be more accepting of a single-sex research project. The study will be advertised as female only; when the children and guardians give consent, this will be taken as gender selfidentification as female. Participants will be selected to ensure a diverse sample according to family composition, ethnicity, geographical location and previous interaction with JU:MP. Three families will be selected from one pioneer neighbourhood (60% Pakistani ethnicity) and three from an accelerator neighbourhood (75% white British ethnicity), enabling exploration of families' experiences at different stages of the JU:MP delivery process. The household earnings in both areas are below the national average, 44 and the researchers have taken steps to ensure socioeconomic status should not be a barrier to recruitment. Families will be approached through JU:MP Community Engagement Managers and other partners who work directly with the communities, to facilitate trust. 30 72 Informed consent from the parents and assent/consent from the children will be obtained before data collection commences and at key data collection points as the project evolves. The study two incentive package was developed in partnership with the youth research ambassadors from Born in Bradford prior to the commencement of the study. The incentives (to the total value of £400 per family) given throughout the study include family vouchers, child vouchers, money donated to the child's charity of choice and team-building activities. Due to the small sample size if the citizen scientists withdraw from the study, we will continue to recruit up to halfway through the study. Amended information and consent forms will be used to recruit once the study has taken shape. If citizen scientists are unable to make a workshop, any progress made will be communicated to them by the facilitator at a time convenient to the citizen scientist. If the citizen scientists do not return data at agreed time point, prompts will be sent to encourage submission. --- Study two: data collection methods Over the longitudinal study, the citizen scientists will be expected to collect data every fortnight on personal and family experience of JU:MP and physical activity. Depending on the data collection method selected, the data could be entered securely online or physically brought to the workshop. The citizen scientists will attend training workshops on how to conduct ethical research, plan and carry out data collection and analysis. Families will be asked to contribute to the data collection and, if the citizen scientists see it as appropriate, comment on the data analysis. There will be six group workshops in addition to regular contact between the researcher and citizen scientists across the year (see table 2). The first two workshops will be facilitated by a researcher, and subsequent workshops may be co-led with citizen scientists as the project develops. The first workshop will be held in a location mid-way between the recruitment localities. Citizen scientists will be asked if this is a suitable location for future workshops. Workshops one and two will focus on training and project development. Key decisions will focus on: finalising the research question, selecting the data collection method(s), ethical issues, how the citizen scientists want to transfer data, data analysis and dissemination. The citizen scientist will have the opportunity to use Open access different methods, surveys, interviewing, photovoice and journaling within the workshop to inform their chosen data collection method. In later workshops, if appropriate, families will be brought into the workshops to see the data analysis. The workshops have been designed to be both accessible and engaging for children. 73 Optional study Citizen scientists will have the option to design and conduct a wider research study to understand physical activity experiences among their peers (of all/any gender). There is potential for the citizen scientists to gain breadth of data and unique insights, given evidence that young people can be more open and willing to share experiences with peer researchers than professionals. 74 This optional study will be introduced in workshop three, and if the citizen scientists want to take it forward, they will guide the study's scope, reach and audience. Through this, the citizen scientists will learn further skills on how to plan and deliver a research project from the start, including first-hand experience of the academic ethics process. Study two: data analysis Citizen scientists will be introduced to different analytical approaches appropriate to the data collection method(s) selected, and will be supported to decide on an analysis method based on the options presented and their own ideas, and to undertake analysis of the data. Given the onerous nature of traditional data analysis approaches, such as thematic analysis, 75 76 citizen scientist and their families' analysis is likely to take an adapted approach, such as producing stories or case studies from the data. We anticipate that citizen scientists will be involved in analysing data collected in the period in-between workshops, from workshop three onwards, taking a crosssectional approach. Alongside the collaboratively chosen cross-sectional data analysis, data will also be analysed using a trajectory approach; this approach has been decided in advance by the researchers to take full advantage of the longitudinal data. It is adapted from the idea of rivers of multilingual reading. 77 The citizen scientists will choose which data are to be included and how. At each workshop, selected data will be placed along the 'river' and the families will note links or changes over time. This process of displaying the snapshots from each wave together will allow for identifying critical moments. 78 --- Overarching participant experience evaluation The research team will evaluate the process and impact of the citizen science approaches on participant experience across both studies. 79 The guidelines adopted in this research intended to promote a positive participant experience, outlined in the Patient and public involvement section, 37 58 59 and inform the evaluation. Three data collection methods will be used: a body sort exercise, an individual reflection card and researcher observations. Any adaptations to the citizen science process suggested by participants and/or researchers, and agreed upon by the participants, will be implemented on an ongoing basis. --- Open access The body sort exercise 73 77 80 will take the form of a workshop activity with the citizen scientists. Cards will prompt key participant experience elements (eg, power relations, inclusivity) in an age-appropriate format. Participants will be able to add further cards detailing other aspects of their experience. Participants will place cards representing their experience on the body outline, and engage in discussion around this. Within study one (the Study design section), the body sort exercise will occur at the end of three of the six children's focus groups, and parent-child dyad participants will be asked to attend a focus group to discuss their interview experience, at each time point. Within study two (see the Study one: contributory citizen science with primary-age children and their families section), participant experience will be evaluated at the end of each workshop and will encompass citizen scientists' experience since the last workshop. Citizen scientists will also have the option to complete an individual anonymised reflection card in case they do not want to voice their opinion in front of the group. A secondary researcher will collect participant experience data to mitigate social desirability bias. 81 Researcher observations, which allow an independent record of events and behaviours to be captured in real time, 19 will be conducted to inform evaluation of participant experience. Researchers will observe focus groups (study one) and workshops (study two). Observations will not occur within the parent-child dyad interviews (study one) to avoid creating a power imbalance between the researchers and the participants. The lead researcher will complete a reflexive research journal, focused on participant experience and improving our research to create a better participant experience. 82 --- ETHICS AND DISSEMINATION The current study adopts and advocates for a bespoke ethical approach for collaborative citizen science projects. Training for the participants, alongside a flexible, responsive approach to changes made as the project develops, ensures the projects meet high ethical standards, 82 for example, if tablets are to be used in study two, data collection training will be provided on data management for the citizen scientists. Ownership of the research is an important ethical concept. 30 Where information is to be disseminated, work will be credited to the citizen scientists while ensuring the participants' and their family's views on anonymity are respected. The collaborative citizen scientists are of an age where evidence suggests they can carry out a research project and express their expert knowledge. 29 82-84 However, as researchers under the age of 16 years, their consent alone is not legally adequate, and parent or guardian consent must be provided prior to research participation. Given the status granted to the citizen scientists, it seems incongruous not to allow them to provide their consent to take part in the project. 80 As a result, to meet institutional ethical standards, both the children (citizen scientists) and their families will be required to give informed consent before any child's participation in the research. If any safeguarding concerns are raised as part of the research project, the university's safeguarding procedures will be followed. A key element of citizen science is acknowledging the voluntary contributions to research by citizen scientists, with debates on how this should be achieved. 37 85 Within both studies, incentives are used to compensate participants for their time. In line with European Citizen Science Association's characteristics of citizen science, incentives differ according to the project context and type. 86 One of the key characteristics of citizen science is that it is a voluntary undertaking, and therefore there is a fine line between incentives and payment, which has been carefully considered. 86 There is an emerging call for diversity in citizen science projects, with socioeconomic status being a known barrier to participation. 87 To reduce the participants' economic situation as a participation barrier, travel will be kept to a minimum, with research being conducted locally and any necessary travel costs reimbursed. Within study two, if technology is a barrier to a family participating, JU:MP will provide reasonable equipment (eg, a tablet) and reimburse reasonable data transfer costs if needed. The progress and findings of the study will be communicated to the citizen scientists and families in various ways, for example, animated videos, and citizen scientists in study two will contribute to wider dissemination of study findings. Results will be published in peer-reviewed journals and summaries will be provided to the participants, through schools or directly. Twitter Amanda Seims @dramandaseims --- Contributors This study is part of the evaluation package of the JU:MP wholesystems physical activity project in Bradford. SB and AD-S led the development of the overarching JU:MP research design in partnership with JH, DB and AS. The conception and design of the studies presented in this paper were led by AS (study one) and MF (study two). JH and AD-S provided equal senior author support for both studies, ensuring the alignment with the overarching process evaluation, with feedback from all authors. MF led the writing of the initial manuscript with significant input from AS, JH and AD-S. Subsequent drafts were commented on by all authors, and revisions were made by MF. All authors have approved the submission. --- Funding

Introduction Whole-systems approaches are being adopted to tackle physical inactivity. The mechanisms contributing to changes resulting from whole-systems approaches are not fully understood. The voices of children and families that these approaches are designed for need to be heard to understand what is working, for whom, where and in what context. This paper describes the protocol for the children and families' citizen science evaluation of the Join Us: Move, Play (JU:MP) programme, a whole-systems approach to increasing physical activity in children and young people aged 5-14 years in Bradford, UK.The evaluation aims to understand the lived experiences of children and families' relationship with physical activity and participation in the JU:MP programme. The study takes a collaborative and contributory citizen science approach, including focus groups, parent-child dyad interviews and participatory research. Feedback and data will guide changes within this study and the JU:MP programme. We also aim to examine participant experience of citizen science and the suitability of a citizen science approach to evaluate a whole-systems approach. Data will be analysed using framework approach alongside iterative analysis with and by citizen scientists in the collaborative citizen science study. Ethics and dissemination Ethical approval has been granted by the University of Bradford: study one (E891focus groups as part of the control trial, E982-parentchild dyad interviews) and study two (E992). Results will be published in peer-reviewed journals and summaries will be provided to the participants, through schools or directly. The citizen scientists will provide input to create further dissemination opportunities.

Introduction Persons living with cancer in under-resourced and vulnerable populations are disproportionately affected by their disease and have less access-to-care compared to persons from high-income settings [1]. In small island developing states, the challenges encountered when accessing care are further amplified. The health care systems in these islands are fragile and are constantly threatened by natural disasters. Implementation of innovation therapeutics and diagnostic services are difficult due to inherently small populations and limited economic capacity [2]. --- of 11 Saint Lucia is a small island developing state in the Caribbean. In 2020, there were 449 new cases of cancer and 232 deaths for a Afro-Caribbean population of about 180,000 [3]. There are three public hospitals, one private hospital, and many health centers distributed across the island. Universal health care does not exist in Saint Lucia. However, residents have access to a national social security system, which subsidises some health care costs and private health care insurance policies [4]. Different cancer advocacy groups operate on the island including "Faces of Cancer Saint Lucia". Faces of Cancer Saint Lucia started in 2009 in order to assist patients going through their cancer journey. Faces of Cancer Saint Lucia has a membership of over 215 persons including volunteers from rural communities. All services and social events offered to survivors of the group are free of charge. Services include education, health fairs, emotional and spiritual support, and chemotherapy and post-surgical support [5]. Faces of Cancer Saint Lucia is currently developing activities beyond cancer advocacy towards involvement in patient navigation. Having an in-depth comprehension of patients' experiences during the delivery of cancer care is of growing value as it improves understanding of patient expectations, and therefore informs community interventions to accompany cancer survivors throughout their journey aiming for better survivorship [6]. We have knowledge of cancer care experience from pacific islanders [7] and other populations from small islands and under-resourced populations [8][9][10]. Survivors from these populations have expressed financial barriers to care, difficulties navigating the health system, and travelling long distances for care due to geographic isolation. These survivors rely heavily on assistance and encouragement from family, and members from the community and faith-based organizations as a means of psycho-social support. Patient navigation programs are also highly valued by these survivors, and have been used to improve equity with counterparts from high-resource settings [6,7]. The under-resourced populations from the Asian/pacific region and Africa for which data is available live either in large countries with developing economies or live on small islands governed by a developed country. However, unlike these populations, Caribbean populations are mostly small island developing states. This status often leads cancer survivors to travel overseas for care in the hope of accessing better services [11]. However, this practice could be counterproductive as it exposes survivors to isolation, which can worsen health outcomes [12,13]. Considering the socio-cultural differences, varying levels of access-to-care, and health seeking behaviors between regions (Asian/pacific, Caribbean and Africa), patient perceptions may not be the same in the Caribbean. Data on this in the Caribbean are scarce. A previous study quantitatively assessed the overall care experience [11]. The only qualitative study on perception of cancer health services was from a health care provider perspective of breast and cervical cancer patients [8]. We sought to describe for the first time the views of cancer survivors residing in Saint Lucia on their overall care experience using a qualitative method. --- Materials and Methods The present study is a secondary analysis drawing data from a community-based study referred to as "the DCAP study"(Description of the Cancer Health Services: Diagnosis and Treatment Pathways). The protocol for this study has been fully described elsewhere [11]. --- Patient Recruitment The DCAP is a cohort of cancer survivors between May 2019 and August 2020. Eligible patients were greater than 18 years of age, able to communicate in English or Creole (without cognitive impairment), with an invasive cancer diagnosis (any cancer site, histology, and year of diagnosis), and having accessed health services in Saint Lucia due to cancer. Participation included authorization to access a patient's data from medical records in health care institutions and centres. Sources for subject recruitment were Faces of Cancer Saint Lucia (FOCS), Victoria Hospital, the Oncology centre, and key informants. Patients at health care establishments were recruited during opportunistic cancer navigation assistance by a FOCS representative. Key informants were recruited using purposive sampling. We aimed to constitute a sample that would reflect the cancer survivors in Saint Lucia by sex, cancer site and district of residence. When possible, we recruited key informants during cancer advocacy activities organized by FOCS. Snowball sampling was used during interviews to identify prospective participants [14,15]. We screened data sources for potentially eligible participants and then invited as many patients as possible. Next-of-kin were interviewed where the index patient was deceased, or not physically able to undergo an interview. The DCAP study was granted ethics approval by the ethics committee from the Medical and Dental Council (Saint Lucia, WI). All participants provided written informed consent prior to the study-required interview. --- Data Collection and Questionnaire Eligible participants were interviewed face-to-face by trained field investigators using a standardized questionnaire. The content of this questionnaire has been fully described elsewhere [11]. Participants were asked to have on-hand their test reports and personal clinical documents, to use as memory-aids during interviews. The questionnaire was developed to ascertain sociodemographic variables such as education level, private medical insurance, hot water at home, employment and clinical characteristics, such as cancer stage at diagnosis, and comorbidities. Participants' personal appraisal of their experiences for major events was ascertained throughout the interview. Interviews took place at the Faces of Cancer office or at the participants place of residence. They lasted on average one hour and 24 min (standard deviation: 34 min). This current study was based on three open-ended questions ascertaining information on their overall care experience (clinical and nonclinical aspects). Cancer survivors were asked specifically "Was there anything in particular that made your experience easier?", "Was there anything in particular that made your experience harder?", and "Do you have any suggestions to help improve the experience for other people in similar circumstances?". Probing was not used for these questions during interviews. --- Variables and Definitions Education level refers to the highest level of education that was completed. Private health insurance refers to coverage at the moment of the interview regardless of the person who pays the policy. Hot water at home refers to the availability of hot running water through a heating system in their primary place of residence. Professional status refers to a form of paid employment at the moment of the interview. The variable was divided into two categories: still working and not working. Not working includes unemployment, volunteer work and retirement. Diagnosis abroad was defined as a medical test performed that required physical travel outside of Saint Lucia. Treatment abroad was defined as a therapeutic intervention administered outside of Saint Lucia. --- Data Analysis We extracted the responses for the three questions based on patient overall care experience. Two of our authors independently analyzed and coded the same sample of responses. Interview responses were blinded for the other variables (age, sex, cancer site etc.) to prevent them from influencing the research findings. After saturation was reached, a thematic analysis approach was used to categorize key codes into themes and subthemes. Thematic analysis is a rigorous, yet inductive, set of procedures designed to identify and examine themes from textual data in a way that is transparent and credible. This method draws from a broad range of several theoretical and methodological perspectives, but the aim is ultimately to present stories and experiences voiced by study participants as accurately and comprehensively as possible [16]. Guest et al. described basic steps in undertaking thematic analysis [16]; Familiarization with and organization of transcripts; Identification of possible themes; Review and analysis of themes to identify structures. Coders met at different intervals to discuss emerging themes until a general consensus was achieved. --- Results --- Characteristics of Cancer Survivors Of the 50 cancer survivors from the initial DCAP study, 44 provided responses for this current analysis. Table 1 shows the sociodemographic and clinical characteristics of those participants. The majority of survivors were black, female, and diagnosed with breast cancer. On average, survivors were 53 years at diagnosis and were interviewed about five years after (standard deviation: 5.3). Years of survivorship were heterogeneous. Most participants were interviewed between two and eight years after their diagnosis. A little over a quarter of participants were more recently diagnosed (<unk>2 years). At diagnosis, 62% reported an early-stage cancer and 47% reported a history of medical conditions. The most frequent conditions reported were hypertension (32%) and diabetes (14%). The majority (73%) had finished their initial active treatment at the time of their interview. In terms of socioeconomic variables, at least half of these survivors had a spouse, a professional activity and hot water at home. Only 40% of survivors were covered by private health insurance. Twenty-eight percent had only primary school education. More than half of the survivors had cancer treatment done outside of Saint Lucia. The proportion of survivors travelling for diagnostic tests was slightly higher (66%). --- Thematic Analysis of Patient Experiences We analyzed the responses from the three open-ended questions for which participants provided a response. Based on the responses, saturation was achieved. We noted 39 responses for the question "Was there anything in particular that made your experience easier?", 37 for the question "Was there anything in particular that made your experience harder?", and 41 for the question "Do you have any suggestions to help improve the experience for other people in similar circumstances?". The length of responses were mostly one to three sentences. On one hand, family support was by far the most common code among the responses for the positive aspects of care, and represented about half of the survivors. On the other hand, responses for the negative aspects and suggestions were more heterogeneous. Fifteen themes emerged from the three open-ended questions (Figure 1). Table 2 shows key quotes from survivor responses contributing to the development of themes. Of the 44 interviews conducted, three were with caregivers/next-of-kins. Caregiver codes were similar to those from cancer survivors and did not contribute to any distinct themes (Supplementary Tables S1 andS2). We identified four themes that were common to the three questions: (1) Availability of support groups, (2) Importance of support from family and friends, (3) Access to finances, and (4) Health education and Patient navigation. Hereafter, we provide a detailed analysis of the content leading to the formation of these themes. common to the three questions: (1) Availability of support groups, (2) Importance of support from family and friends, (3) Access to finances, and (4) Health education and Patient navigation. Hereafter, we provide a detailed analysis of the content leading to the formation of these themes. Positive: "Was there anything in particular that made your experience easier?", Negative: "Was there anything in particular that made your experience harder?" and Suggestions: "Do you have any suggestions to help improve the experience for other people in similar circumstances?". HCP: Health care provider. --- Open-Ended Question Patient n° Quotes Was there anything in particular that made your experience easier? 1 Joining Faces of Cancer Saint Lucia. --- 2 The support of family and friends who provided housing spiritual and emotional support. 3 Yes, the almighty, I trusted him to give me the strength to endure. --- 4 Family support (My sister was always here), insurance (Money was not a problem), my employer supported me mentally and financially. 5 Treatment at Tapion hospital was excellent but costly. 6 Family support, natural medications. Positive: "Was there anything in particular that made your experience easier?", Negative: "Was there anything in particular that made your experience harder?" and Suggestions: "Do you have any suggestions to help improve the experience for other people in similar circumstances?". HCP: Health care provider. Table 2. Key quotes from participant responses contributing to the development of themes. --- Open-Ended Question Patient n • Quotes Was there anything in particular that made your experience easier? 1 Joining Faces of Cancer Saint Lucia. --- 2 The support of family and friends who provided housing, spiritual and emotional support. 3 Yes, the almighty, I trusted him to give me the strength to endure. 4 Family support (My sister was always here), insurance (Money was not a problem), my employer supported me mentally and financially. 5 Treatment at Tapion hospital was excellent but costly. 6 Family support, natural medications. Was there anything in particular that made your experience harder? --- Availability of Support Groups A total of 15 categories were identified to create this theme. Survivors expressed how support groups assisted them in getting information and in coping strategies with their illness, "Faces of cancer because of information they gave and hope". Relating to a more difficult experience, many survivors expressed their dissatisfaction with not having the knowledge of existing support groups, "Not having knowledge of who to contact for support". Survivors also gave advice to other cancer survivors relating to cancer groups. Many insisted that it is quite valuable throughout the journey to share your experiences with other survivors. "We need a cancer center to provide support and counselling to patients who are diagnosed with cancer. Increased support to Faces of Cancer to assist patients in care and treatment." --- Importance of Support from Family and Friends Support from family and friends is one of the most frequent themes in this study. Survivors who had an easier journey expressed their gratitude for having family members and close friends supporting them throughout the journey, "The only thing that made my experience easier is the fact that my sister accompanied me at every visit to the doctor and to the hospital". Survivors who travelled overseas for care spoke about the burden associated with being isolated from their family "I could not see my husband and children", "Having to leave my son in Saint Lucia made it a bit hard". --- Access to Finances Many survivors felt overwhelmed with the burden of finding the funds to pay for their treatment, laboratory tests, medication, etc. throughout their journey. Survivors showed their disappointment with the lack of enthusiasm from medical professionals to assist them when they were unable to make payments, "Having no finance to pay for treatment and the doctors would not see you if you have no money, they would rather you die". Some survivors emphasized the need to have medical insurance early enough in the case of being diagnosed with cancer. A common way of paying for treatment as expressed by survivors was by raising funds whether it be in the form of having fundraiser barbeques created by family members, close friends or even members in the community, "... Just if one doesn't have insurance one should start asking for money early so they can do the treatment without missing any treatment". On the other hand, one survivor described their experience as being easier since he/she "did not have to worry about finances". --- Health Education and Patient Navigation Survivors have shown profound interest in having a navigation system within cancer care. Survivors are burdened with having no knowledge on "what's next" after being diagnosed. They feel that inadequate information is provided by health care expertise after their diagnosis; "Lack of team structure to deal with issues together... ". Other survivors also expressed the need for health care professionals with the support from government to raise awareness, to educate the public, and to invest in cancer research. However, one survivor who had knowledge of the disease described having a more difficult experience, "Having the knowledge and being the patient is heart wrenching". --- Discussion This is the first study focusing on cancer care experiences from a patient perspective in the Caribbean. This study underscores the importance of family and social support for a positive cancer experience in Saint Lucia. Although the themes that emerged from our study were widely consistent with those from existing patient experience framework (patient preferences, emotional support, physical comfort, information and communication, continuity and transition, coordination of care, involvement of family and friends, and access-to-care) [17,18], we showed that survivors often have to travel overseas and leave their family in Saint Lucia to access care, and this impacted negatively on other aspects of the cancer experience. This finding is novel, and distinguishes our survivors from those in larger countries. We compared our themes with the few data from other small islands. Themes from the Caribbean islands of Dominica, Grenada, and Saint Vincent and the Grenadines were consistent with ours [8]. Our survivors had a deep appreciation for the support from family and friends, and for support groups like Faces of Cancer Saint Lucia. Interestingly, the indigenous populations from the Torres strait islands of Australia also appear to have similar cancer experiences to our survivors [7]. They both have similar experiences while they access care overseas (e.g., isolation from family, language barrier, and cultural differences). However, compared to these same islands, survivors from Saint Lucia had great enthusiasm for religion and faith, a factor which was uncommon in other small developing islands outside of the Caribbean [7]. Many survivors from our study looked towards higher spiritual powers, praying and having faith as a means of support or coping mechanism. The role of religion and spirituality has also been described in larger LMICs [10]. However, compared to developed countries, we believe that family support and faith appeared to be valued more among our survivors in Saint Lucia and other LMICs [19][20][21]. Access to finance was a recurring theme across the three open-ended questions. Cost of treatment and obligation to fundraise were mostly cited as sources of financial hardship. Financial hardship is well-known to be associated with the cancer experience in both highincome and low-income countries [22,23]. Qualitative studies on the financial hardship in LMICs and small islands are particularly scarce [23]. One study using a semi-structured interview was conducted in Iran [24]. Iranian survivors reported financial difficulties due to interference with their ability to work [24]. This was not observed in our study. We hypothesized a possible relationship between some themes. Survivors expressed concerns with lack of empathy from providers, trust in their expertise, and health education. Lack of empathy may explain the distrust in expertise and motives of providers [25]. Consequently, the patients-provider relationship is suboptimal and diminishes the quality of knowledge transfer and care [26,27]. In addition, many survivors also spoke about emotional distress. Travelling to more developed countries for treatment often resulted in leaving behind family. Knowing the emphasis placed on family support by our participants, there is likely a strong patientburden generated by travelling for care. This may contribute negatively to health outcomes of survivors from Saint Lucia. This link between social support and improved quality of life is well established [9,28,29]. A previous study on breast cancer patients showed that having social support mediates the choice of coping strategies toward positive reframing, which leads to better emotional well-being [28]. Social support is defined as a network of family, friends, neighbors, and community members that is available in times of need to give psychological, physical, and financial help [30]. Our findings add further understanding to what survivors in Saint Lucia view as important during their care. Patient satisfaction appears to be influenced more by interpersonal and provider factors rather than objective system measures. In a previous study, quantitative ratings of overall care experience showed 76% of these cancer survivors from Saint Lucia were satisfied (rated as "good/very good") [11], whereas the responses of our current analysis were heterogeneous, revealing both a strong role of family/friend support and also numerous difficulties. Indeed, our current analysis revealed notable dissatisfaction with the medical professionals' delivery of care, notably when conveying vital information and timely diagnosis. This incongruity between quantitative ratings and qualitative responses raises two potential explanations on the perception of the overall experience of cancer survivors in Saint Lucia. Firstly, we believe that the effect of psychosocial support may outweigh that of suboptimal health services in Saint Lucia. Secondly, health literacy is also a probable factor as we previously described [11]. Survivors may not view long delays or low-quality services as problematic due to lack of knowledge of best practices and standards in cancer care. There may be other factors that contribute to the perceptions of cancer survivors. This information is vital for tailoring interventions. Future work should particularly assess potential associations between the socioeconomic status and patient perceptions of care using a mixed methods approach [31]. This study had several strengths and limitations. Recall bias is likely since we ascertained information on patient experiences several years after their diagnosis. However, the effect of recall bias is unlikely to be alarming. The average delay was only five years, and our major themes corroborate with data from providers treating cancer patients from Saint Lucia [8]. Given the purposive sampling performed for this study, our results may not be an accurate representation of the views of all cancer survivors in Saint Lucia, a common weakness of qualitative studies. Some survivors were also recruited by Faces of Cancer Saint Lucia. Consequently, responses on support groups may have been overrepresented. In addition, most of the sample comprised women. Greater participation from women is a common occurrence in research studies [32]. However, we previously showed that our sample was indeed representative of the most common cancer sites by sex in Saint Lucia [11]. In addition, 44 participants spoke about the factors they perceived as important. This is a substantial sample size for a qualitative investigation, and is also comparable to other studies on this topic [7,9]. Furthermore, our study also adds new information to cancer care from a patient's and caregiver's perspective. The patient perspective gives a more holistic depiction of the difficulties encountered unlike studies from a provider perspective where comments on provider performance may be omitted [8]. In addition, our investigation included both quantitative and qualitative data. Small developing islands like Saint Lucia are not always equipped with the resources needed to adequately treat cancer patients. The qualitative data gathered from this study raises awareness of the importance of capturing patients' perspectives when receiving treatment. With our description of patients' experiences, the local government and patient associations have an opportunity to plan and implement more successful evidence-based patient-centered interventions focusing on reducing the patient-burden associated with social isolation, notably from separation from family. Although these patient experiences are from an island context, we showed that these findings are also relevant considerations for planning of cancer control in under-resourced and uninsured populations in more developed countries. --- Conclusions In light of our study, multi-disciplinary case-conferencing, patient education programs, and patient navigation could contribute to substantial improvement in the care experience and better survivorship among cancer survivors in Saint Lucia. These survivors depend --- Data Availability Statement: The datasets analyzed during the current study are available from the corresponding author on reasonable request. --- highly on support from family for a positive cancer experience. Travelling overseas for more comprehensive care often translated into separation from family and an additional burden on survivors. The above interventions can be implemented immediately awaiting the development of more tertiary cancer services on-island. We believe that support groups such as Faces of Cancer Saint Lucia have an important role in building scientific evidence to strengthen the advocacy for better cancer control in Saint Lucia. Small developing islands of the Caribbean and similar under-resourced populations with limited offerings in cancer health services may also benefit from these new findings. --- Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph19116531/s1, Table S1: Characteristics of cancer patients who were represented by a caregiver during the study interview; Table S2 Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. --- Conflicts of Interest: The authors declare no conflict of interest.

Understanding the views of cancer survivors on their experience is important for informing community-based interventions. We studied, for the first time, the views of cancer survivors residing in Saint Lucia on their overall care experience. We used interview data from a cohort of adult cancer survivors from Saint Lucia between 2019 and 2020. We performed a thematic analysis to derive themes from codes. Forty-four survivors provided responses to at least one of the three questions. The majority of survivors were black, female and diagnosed with breast cancer. Survivors were interviewed on average five years after diagnosis. Four common themes emerged; "Availability of support groups", "Importance of support from family and friends", "Access to finances" and "Health education and patient navigation". Travel overseas for health services was common among survivors. Survivors expressed emotional distress during travel due to isolation from family and local providers. This is typical among island populations and is distinct from existing patient frameworks. Survivors also suggested that networking amongst providers and interventions assisted families of cancer survivors. Although tertiary care services are limited, we showed that survivors deeply value and depend on their inter-personal relationships during care. Interventions aimed at strengthening the inter-personal environment of survivors are warranted.

understandings of how women manage health under the highly challenging conditions of reentry after incarceration. --- Women's Incarceration and Women's Health Jail admissions in the United States total over 11 million per year, with daily population censuses for jails numbering about 2 million (Minton & Zeng, 2015). Women make up approximately 15% of those totals, their rates increasing 18% between 2010 and 2014, whereas men's rates of jail incarceration decreased by 3.2% (Minton & Zeng, 2015). Much has been written about the distinctive pathways that lead women to incarceration, and it is well documented that rising rates are tied to trends in the policing, charging, and sentencing of drug and property crime in underresourced, often racially targeted communities (Alexander, 2010;Wacquant, 2010). Justice-involved women, including both those who are incarcerated and those under criminal justice supervision in the community, often come from and return to underresourced areas, where jobs and social capital to improve socioeconomic status are lacking (Sered & Norton-Hawk, 2014;Western & Pettit, 2010). In addition, incarcerated women struggle with substance abuse disorders, mental illness, and severe personal trauma at higher rates than women in the general population, including experiences of childhood sexual abuse (CSA), intimate partner abuse, and adult rape (Grella, Lovinger, & Warda, 2013;Lynch, Dehart, Belknap, & Green, 2012). In seeking to gain control over the cycle of incarceration, women with a history of incarceration face multiple barriers that affect their health and safety (Lorvick, Comfort, Krebs, & Kral, 2015;Swavola, Riley, & Subramaniam, 2016). --- Trauma and Trauma Narratives of Justice-Involved Women High rates of lifetime trauma from interpersonal violence afflict women with a history of incarceration (Grella et al., 2013). Reports of CSA and adult sexual violence are prevalent in incarcerated populations, with studies finding CSA rates for incarcerated women between 32% and 68%, and sexual violence as high as 86% (Briere, 1996;Cusack, Herring, & Steadman, 2013;DeHart, Lynch, Belknap, Dass-Brailsford, & Green, 2014). Women who experience traumatic life events often adopt coping strategies such as drug and alcohol abuse that lead to revictimization and additional health risks (Cusack et al., 2013). Symptoms from unresolved trauma responses can take a toll on physical health and are associated with dysregulation of stress responses and associated health outcomes, including musculoskeletal pain and digestive, endocrinal, cardiac-respiratory, and immunologic conditions (Pacella, Hruska, & Delahanty, 2013;Weiss, 2007). Trauma has special significance in a study that focuses on how women narrate relationships. In her landmark text Trauma and Recovery, Herman (1992) explained that trauma, a psychic injury or wounding, refers to the profound disruption in a survivor's understanding of self in relation to the world, a loss of perceived "control, connection, and meaning" (p. 33). Caruth (1996) has stressed that the key aspect of trauma is the difficulty with which survivors understand and make meaning of negative experience owing to the peculiar way trauma is recorded in memory, its tendency to be unanchored contextually-not part of the flow of recalled experience but separated and locked out. Survivors often can only incorporate the meaning of traumatic experience into their lives indirectly and belatedly. Janoff-Bulman (1992) has argued that trauma entails such a shattering of a person's basic assumptions about the self in its connection with others that it might require construction of a whole new selfnarrative or schema. The process of recovery from trauma can be an extended and uneven process marked by reexperiencing, avoidance, and dissociation (Bromberg, 2003;Janoff-Bulman, 2004), sequelae that interfere with a survivor's ability to make sense of, form, and manage relationships (Golin et al., 2016;Herman, 1992). As women with significant lifetime trauma might suffer retraumatization through unintended triggering of symptoms within the carceral setting, trauma-informed corrections has been a focus of research in recent decades (Harner & Burgess, 2011;Machtinger, Cuca, Khanna, Rose, & Kimberg, 2015). However, the attention to targeted, trauma-informed support during incarceration rarely extends beyond release for these women. Post incarceration, many receive only minimal social assistance and inadequate mental and physical health care (Belknap, Lynch, & DeHart, 2015;Mallik-Kane & Visher, 2008;Richie, 2001). To change how women are supported in the community after release, providers can begin by hearing what women are already doing, taking stock of the particular ways in which justice-involved women with a history of trauma perceive self and other in managing health and safety risk through informal social support. --- Feminist Standpoint and Relationality Feminist standpoint theory and its emphasis on relationality informed the overall objective of the study: to learn from justice-involved women's own accounts of trauma how they manage social ties for health and safety. Feminist standpoint theory holds that women share certain ways of viewing and understanding the world as a result of being socialized as women in a specific social structure at a specific historical juncture (Ramazanoglu & Holland, 2002). While subject to a good deal of variation and modification due to intersectionality (i.e., the synergistic role played by factors of socioeconomic status, race, and religion in forming women's experience), feminist standpoint theory argues that an emphasis on relationships represents a primary way in which women know and experience the world (Freedberg, 2015). Connectedness and transaction rather than autonomy and detachment thus become key motifs around which feminist epistemology (Haraway, 2014) and feminist approaches to philosophic and scientific inquiry take form (Gilligan, 1982;Harding, 2004;Miller, 1986;D. E. Smith, 1990). The relational aspect of feminist theory further supports a focus on stories about trauma, as trauma is often defined as the rupturing of connection or relationship (Comstock et al., 2008;Herman, 1992). --- Narrative Inquiry for Health Research Story-based or narrative research assumes that narrative is a basic means by which humans make sense of and communicate experience (Bruner, 1990;Polkinghorne, 1988). Often, narrative refers to the temporal ordering of a series of events or conditions by a speaking (or writing) subject (Mishler, 1995), whereas story refers to a more purposeful performance that might involve greater attention to scene-setting and evaluation of meaning than is found in the mere narration of events (Maynard-Moody & Musheno, 2009;Mishler, 1995). Notwithstanding distinctions made between story and narrative, the two are often used interchangeably in the social sciences to refer to segments of talk that form a plotted sequence, though it might not be chronologically ordered in the telling (Riessman, 2008). I use the terms narrative and story interchangeably in this article. Narrative inquiry has been developed for social science research by Clandinin and Connelly (2000), Gubrium and Holstein (2009), Hall (2011), Maynard-Moody and Musheno (2014), and Riessman (2008), who argued that personal narratives provide rich, detailed information about how people infuse experience with meaning and emotional valence, which in turn inform attitudes and behaviors (see also Sandelowski, 1991). The narrative analytical methods used in this study were specifically adapted from Riessman (2008), who demonstrated how narratives convey meaning through thematic, structural, and performative dimensions, as well as through intertextual connections-the way stories told in one context converse with stories told in another, as in the broad social narratives or stories that a culture tells itself about itself. Inquiry based on personal narratives has had special significance for the health sciences (Priest, 2000), where stories and storytelling are employed as a mode of intervention (Charon, 2006; M. J. Smith & Liehr, 2013) and as a method to build knowledge about the subjective experiences and needs of patient populations (Draucker & Martsolf, 2010;Polzer, Mancuso, & Rudman, 2014). For health researchers, storied data can render visible the emotional and perceptual patterns that organize the ways in which people manage health and safety risk in their worlds (Bally et al., 2014;Lee, Fawcett, & DeMarco, 2016). --- Accessing Support Through Social Connection This study focused on justice-involved women's stories about relationships to understand how they perceive self and other in mobilizing social support. According to social network theory, the webs or networks of social relationship that bind persons to one another in a community often serve as a conduit for social support (Berkman, Glass, Brissette, & Seeman, 2000). Social support might take form as emotional, material, instrumental, and informational resources exchanged across specific relationships (Cohen, 2004;Uchino, Bowen, Carlisle, & Birmingham, 2012). In low-resource situations, support from close network connections might be necessary for survival or "getting by" (Briggs, 1998, p. 178;Dom<unk>nguez & Watkins, 2003). For a population, with limited resources, it is important to observe that social support operates according to a principle of reciprocity, the achievement or promise of some form of give and take (Uehara, 1990). The central tendency in the conceptual framework for this study-which includes feminist standpoint theory, narrative inquiry methodology, and social support theory-is relational. It is thus well suited to investigating what trauma stories of women with justice involvement have to say about how women perceive and manage social ties to access the social support they need to protect health and safety. --- Method Recruitment and Case Selection Women were invited to interview for this study from a larger, ongoing interventional and ethnographic study (Ramaswamy et al., 2017b). That parent study had two aims, the first being to assess changes in women's cervical cancer knowledge, attitudes, and self-efficacy to screen and seek treatment before and after a brief education and empowerment program delivered in the jails. The second aim of the parent study involved interviewing and observing a subsample of the women after release to learn more about how women navigate health care and social service barriers (Ramaswamy et al., 2017b). The convenience sample (n = 182) for the parent study included English-speaking women in three county jails in a Midwestern metropolitan area from September 2014 to March 2016 (Ramaswamy et al., 2017a(Ramaswamy et al.,, 2017b)). The story-based, relationship-focused interview study reported here involved a purposeful selection of 10 cases from the parent study sample. Selection was based on the variety and complexity of the women's social networks and experiences of trauma as expressed in the interventional phase of the parent study. In determining which participants to invite, I evaluated women's discussion during the interactive group sessions, inviting participants who reported numerous, shifting, and overlapping roles and relationships in the most detail. Invitations for the story-based relationship interviews were issued on a rolling basis between December 2015 and July 2016. Recruitment of participants for the story interviews ended when I judged that the data were yielding sufficient complexity of form, richness of detail, and diversity of experience to answer the research question (Patton, 2015) about how women narratively represent the management of relationships to protect health and safety in stories of trauma. No one who was invited to participate in the interviews declined, and (unusual for work with this population) none of the participants was lost to follow-up during the 12 months of interviewing. --- Data Collection Procedures Two in-depth, semistructured interviews were conducted with each of the 10 women who agreed to participate. All interviews were audio recorded, transcribed verbatim, and stored in Dedoose (a password-protected online data management system) and on a secure server at the University of Kansas Medical Center. Interviews in both the initial and follow-up series averaged 60 minutes and ranged from 31 to 91 minutes. The average time between each woman's initial and follow-up interview was 5 months, with a minimum of 3 weeks and a maximum of 12 months. In the initial interview with each woman, a life-story trope was employed in which the participant was invited to imagine her life as a book, similar to what McAdams (2008) described, but with each chapter centering on one or more key relationships. The relationships could be with family, friends, intimates, or acquaintances. Participants were encouraged to order their chapters and the stories within them however they liked and were not held to the book format if they seemed resistant to it. After a woman narrated her last chapter, she was asked which (if any) of the chapters she would remove from the story and why, what she would make up and add in as a chapter, and what title she would give her overall story. In the follow-up interview with each woman, I presented a typed outline or a verbal description of the book's chapters and key contents for verification and invited additional storytelling prompted by that frame. All interviews were conducted in person, with the exception of one follow-up, which took place on Skype after the participant moved out of state. Most interviews occurred in the community, at sites that included participants' residences, the researcher's car, coffee shops, a public library, a church-choir changing room, and various McDonald's restaurants. Three interviews were recorded in a private meeting room in the jail. --- Data Analysis Procedures Each transcribed interview was reviewed and cleaned against the original recording before being loaded into Dedoose, where I open-coded transcriptions independently (Salda<unk>a, 2016). Two members of the parent study analytic team, who also read the transcriptions, met with me biweekly throughout data collection to discuss themes based on field notes, coding, and analytical memos (Emerson, Fretz, & Shaw, 2011). Over the course of data collection and analysis-a yearlong, iterative process of reading, coding, memoing, and discussion of my interview data and field notes with colleagues-I discerned the presence and main attributes of a framework of strategizing and fatalizing talk in the management of relationships for social support. The unit of analysis in this study was two-part and included the overall life stories constructed by the women in interviews and field encounters and the embedded trauma stories. The larger life stories were constructed as described above from the chapters narrated by the women. Embedded trauma stories were extracted for analysis based on a trauma event series that had a discernible abstract or beginning (often instigated by an interview prompt) and an ending, in which, after recounting the shifts in action or condition, a participant gave some form of resolution and provided a "coda" (Labov & Waletzky, 1967, p. 39). The coda or ending was usually a sentence or two that brought the account into the present perspective. Embedded stories of trauma could extend a few lines or a few pages; some were peppered through with probing questions, whereas others issued forth as an unbroken flow of narration. The four trauma stories analyzed in this report were selected on the basis of their detail, complexity, and to a lesser extent their completeness of form. I selected stories that illuminated the characteristics of the two poles of strategizing and fatalizing to better define those modes of talk, though what was in fact more common was a combination of the two, an example of which I also present. For each case, I contextualize the trauma narrative(s) with detail from the participant's life story and include my interpretive analysis to clarify the reasoning by which I drew conclusions from the findings. --- Ethical Issues All participants gave informed, written consent to interview as part of the enrollment process in the parent study. Prior to each interview, I further reviewed the purpose, benefits, and possible risks of the substudy and obtained assent to continue participation and to audio record. To protect participants' autonomy, I reemphasized in each encounter that the women were under no obligation to continue with the parent study, the interview study, or any particular line of discussion within an interview. I stressed that the project would probably not benefit the participants in any direct way but I also let them know that their contributions might help care providers and program designers improve conditions for women in like positions by designing better services. I warned women that our discussions could cause unwanted memories or emotions, and I reminded them that the parent study team could give referrals for mental health with experienced providers affiliated with the project if requested. During informed consent, the women were apprised of our obligation to report child abuse or endangerment, and I remained alert to any such disclosures during all encounters. Participants in this study were initially encountered during an incarceration, and most continued to be subject to criminal justice supervision, several returning to jail during the course of the study. As such, they were deemed vulnerable and merited special protections as human subjects. Amplifying their potential for harm and exploitation was the continuing involvement of many of the women in drug use, drug sales, and prostitution-illegal activities that participants discussed freely in the interviews. A National Institutes of Health Certificate of Confidentiality was obtained to protect recordings and other materials in this research from seizure or discovery by law enforcement. Data were stored in passwordprotected computers and on a secure university server. Initials and pseudonyms were used to identify the women in notes and transcriptions. All the names that appear in this report are pseudonyms, most of them chosen by the women themselves. Women were remunerated US$10 for each interview. The amounts were applied to debit cards held by the women as part of their ongoing participation in the larger investigation. The substudy was approved through the parent study by the institutional review board at the sponsoring university medical center. The programs administrator at the jail where three of the interviews took place additionally approved the audio-recording of interviews. --- Findings The objective of this study was to achieve depth and detail of data in specifically selected cases rather than representativeness or probabilistic generalizability (Miles & Huberman, 1994;Patton, 2015). Even so, demographic information in a qualitative study can provide a snapshot of a group of cases and give some idea where diversity of experience was achieved or lacking in case selection (Sandelowski, 2001). The supplemental file provides a summary of characteristics of the 10 participants in the interview group and some indication of trauma exposure as revealed during interviews. Of particular note, various combinations of interpersonal violence, drug abuse, sex exchange, and insecure housing put these women at elevated risk of infectious and chronic disease and injury (Kelly, Cheng, Spencer-Carver, & Ramaswamy, 2014). Four women reported exchanging sex for food, shelter, money, or drugs either regularly or intermittently during the 12 months, and two others had done so in the past. All the women reported a history of substance abuse, and, in either their initial or follow-up interviews, all but two of the women acknowledged continuing drug use during the period of the study-including the two incarcerated women, who described using while in jail. One of the most pressing risks related to health and safety in the women's day-to-day lives and the one to which women most often attributed shifts in relationships was the threat of injury or death due to violence. Most women reported multiple interpersonal traumas. Six women described being raped as adults, often multiple times. Seven were beaten by boyfriends, husbands, or johns severely enough to require hospitalization. Three described abductions by abusive partners that involved being held captive for a period of time. Two reported having been hospitalized with gunshot wounds. Three of the women described the death or life-threatening injury of an infant in the home. Indeed, the women's stories at times gave an impression of constant crisis, yet within the narration of trauma was also evidence that the women labored to shape meaning in ways intended to facilitate support or minimize harm. --- Strategizing Talk and Fatalizing Talk in Narratives of Trauma The initial interviewing and analysis of data in this study was guided by the broad question, "What do trauma narratives of recently incarcerated women reveal about how they perceive and manage social ties to access support for health and safety?" The trauma stories displayed in the list titled "Trauma Narratives" (Supplemental file) and explicated below represented comparatively extreme moments of health and safety risk in the women's lives. Janoff-Bulman (1992) has argued that research based on extreme events might provide particular insight into "basic aspects of human thought and behavior" (p. 3). The trauma narratives analyzed herein indicated that justice-involved women in volatile environments with compromised access to resources perceived relationships of support and their own capacity to engage them along a continuum extending from greater (strategizing) to lesser (fatalizing) ascriptions of agency. --- Strategizing talk In strategizing talk, women emphasized agency and an implied self-efficacy, sharing in some form the goals, plans, and actions that made up a process of mobilizing social support. Natalie-Natalie's criminal justice history included a 4-year stay in prison and more than 10 short-term jail stays, mostly for drug-related offenses. Chapters in Natalie's life story described relationships with her mother, the fathers of her children, two same-sex partners in prison, and the man she married during the year of this study. Key events in Natalie's life story included rape by an acquaintance at age 14, the birth of her children, the loss of her parental rights, and the sudden death of her mother. Natalie identified herself as an alcoholic and reported abusing methamphetamine and prescription opioids. Natalie's overall life story was shaped by her almost constant maneuvering to secure sanctuary, both physical and emotional, for herself and formerly for her children. The embedded trauma stories within Natalie's life story exemplified the use of strategizing talk. In her first chapter, Natalie described leaving her mother's home at age 15 to live with a teenage boyfriend, Darren, aged 14, in his mother's house. Natalie spent nearly 7 years and had two daughters with this partner in a relationship characterized by severe physical and emotional violence. The extended, disjointed account of her first opposite-sex relationship included allusions to being punched, kicked downstairs, and raped. However, despite the recitation of chronic, severe abuse, in the story labeled "I Finally Did It," Natalie first narrated a predicament over which she perceived she had little control and then, despite that lack of control, followed with strategizing talk: variously establishing intent, assessing options, making plans, and taking action. Structurally, Natalie set two plots in motion in "I Finally Did It," an elaborately planned scheme that involved a secret job and savings but required time to bring to fruition, and a second, the fight, enacted on the fly, its results put to immediate use. The movement of the narrative from one to the other highlighted Natalie's tactical aptitude as she pivoted abruptly from describing the first approach (i.e., getting a secret job, saving money) to seize on and develop the circumstances surrounding the second (i.e., staging a disclosure). In the second gambit, she transformed the "something stupid" of a minor fight with Darren's mother into a narrative of imminent danger. Strategizing was made even more apparent in the final lines of "I Finally Did It," when Natalie pointed to her reasonable assumption that, even though her mother had not responded to previous requests for sanctuary, appealing to fear of HIV and the safety of her granddaughters would prove more successful. Natalie's strategizing talk took form in "I Finally Did It" through her presentation of self, which focused on the depiction of a woman isolated and yet continually plotting to rebuild connections. Her emphasis was on strategizing-with little help and much hindrance from others-to manage relationships in a way that would secure her own and her children's safety. Similar to other trauma stories that featured strategizing talk, in Natalie's second trauma story, "[M]aybe Now I Could do a Little Bit More," the presentation of a self in isolation, lacking either formal or informal social supports, was parlayed paradoxically into a hardbought agency. The entire first passage of the second narrative was dominated by expressions of helplessness and bewilderment, exemplified by her anguished wondering: "Where's my support group?... What the hell is going on?" In Natalie's storytelling, alienation did not lead her to submit to circumstance but to reassert agency: though utterly alone, she was "working on this"; once the children are in the care of family members, she would be able to "do a little bit more" to achieve time with her kids; and, in narrating what occurred after the sexual molestation of her daughters by their cousin, she underscoredwith a single, important "[b]ecause"-that it was her own rage that impelled her brother finally to step-up. Although not always successful and certainly filled with heartache, frustration, and sometimes recklessness, Natalie's narrative constructions of her own efforts to achieve safety highlighted strategizing talk, wherein plans were laid and goals for safety were pursued by an individual who evaluated motivations to manipulate relational ties and assert agency over events. Natalie chose as the title for her overall story "Don't Throw Me No Pity-Party," sounding a kind of credo against being defined as powerless by others. Jennifer-Jennifer, a white woman in her late 30s, also emphasized strategizing talk in her approach to managing support in a trauma narrative. Jennifer's strategizing talk underscored differently from Natalie's the ways a woman with criminal justice involvement might find herself having to navigate relationships to get help. The first interview with Jennifer took place in a residential transition program where she was completing court-ordered substance abuse treatment after leaving jail. Five months later, our second interview was conducted on Skype, following Jennifer's move to a distant state for a job. The jail incarceration in which I initially encountered Jennifer was her first, and Jennifer had no prison experience at all. Jennifer's overall life story featured strategizing talk that took form through evaluation and assessment, indeed, a running analysis of the causes behind her actions, feelings, and attitudes. For instance, although Jennifer began her life story by relating an idyllic-sounding childhood and adolescence, that narration of early life was followed almost immediately with premonitory ruminations on what Jennifer identified in hindsight as sources of future trouble. She explained that what she once considered to be a "normal," carefree childhood she now recognized as lacking in emotional connection, steeped in the heavy drinking of her parents' social circle and the superficiality of bonds among her own friends. As a form of strategizing talk, Jennifer's analysis functioned to control interpretation of her life history by making explicit the connections between causes and effects, between her early life and the later perceptions, choices, and actions that comprised the particular, embedded trauma story in which she lost custody of her middle child. Jennifer's trauma story, "And These Were the Last Strings," followed a "downward spiraling" series of events. After college, Jennifer married and had a child, eventually divorced her husband (he was granted custody of the child), and, weary of the frequent travel associated with her career, accepted a less challenging position that left her idle much of the time. Jennifer explained that this led to boredom, and boredom led to vulnerability to substance abuse. Just before becoming pregnant with her third daughter, Jennifer returned a positive drug test. Then, in what she denominated the one real trauma in her life, Jennifer lost custody of her second daughter and was soon after incarcerated. In her narrative, Jennifer focused on assessing intentions, weighing costs, and guiding her audience's interpretation. Her strategizing talk in "And These Were the Last Strings" highlighted a process of risk assessment, outlining the factors that she, as a drug user and a mother, considered in determining whether to initiate a relationship with an agency that might help her address her addiction. First, Jennifer was challenged to identify the right source of help-whom to entrust with her predicament. The right assistance, she observed, was "not very easily found." She then reviewed the costs of disclosing drug use and asking for help, which included potential loss of a child, breaking up the family, and feeling marked as a drug user. Jennifer's strategizing talk highlighted the extent to which a perfectly rational perception of a punitive function (the state's power to remove children) might impede viability of support functions (the state's provision of help for substance use) in the perception of a help seeker. Jennifer also cited the slightly different cost of managing loss of face (E. Goffman, 1967), because, as Jennifer pointed out, asking for help at all was opposed to the bootstraps view of what she ascribed to herself: "I was brought up to get through it on my own." Jennifer's trauma narrative stressed that, if strategizing is the mode in which a woman imagines herself to manage relationships, a lack of legibility or transparency around support -in this case for drug-abusing women with children-might prove prohibitive. To Jennifer's exasperation, nothing about the process of seeking support before or getting formal help after losing her child was "straightforward," except the threat ("grab it or you're gone") related to what Foucault (1990) might have called an imperative to confess. Jennifer recognized that "grab it [and] you're gone" was equally likely, as confessing drug use and getting help for it could very well mean the dismantling of a family. When asked what she would entitle her overall life story, Jennifer tellingly proposed "What Not to Do," a clause that implies a scene of learning, where persons who encounter her story might strategically equip themselves with lessons based on her experience to improve their own capacity to make choices and achieve desired outcomes. --- Fatalizing talk Fatalizing talk was dominant when women expressed a sense of compromised self-efficacy in making decisions and managing relationships for support in their narratives. In fatalizing accounts, women represented crisis and their own roles in it as either foreordained and out of their control or as radically unpredictable and incomprehensible. In fatalizing talk, women ascribed situations and outcomes to fate, God, or chance and identified themselves as detached, confused, surrendering, and dazed observers. Cicely-A Black woman in her mid-30s, Cicely, was mother to four school-aged children, all living by court order with relatives or adopted out. Cicely reported over 10 incarcerations and one prison stay. In her first interview with me in June and again in the follow-up interview in October, Cicely acknowledged recent use of cocaine, methamphetamine, and phencyclidine. When I saw her in a field encounter a month after the second interview, she claimed to be in hiding, saying that a local dealer had a "hit" out on her. In her overall life story, Cicely made frequent declarations of self in which she seemed intent on establishing agency: no-nonsense, goal-driven, someone who gets stuff done. In the chapter she created around her relationship with the father of her first three children, for instance, Cicely outlined her role as the family supporter, "the breadwinner," and "a gogetter," working two and then three jobs to provide for the household. In a formulation that she repeated in almost identical terms in the two interviews, Cicely stressed her attitude about roles in her relationships: "[I]f you ain't about to go get it, then get out of my face, so I can go get it myself." Based on such self-defining moments, one might expect Cicely's trauma narratives to be dominated by strategizing talk. What instead emerged in both Cicely's life story and her stories of crisis was more aligned with a second cluster of story attributes, those defining of fatalizing talk. Cicely's life story had a reactive, chaotic quality to it, a disjointedness that was especially evident in the recounting of her early years, where she reported moving from mother's and grandmother's homes in one part of the country; to father's, aunt's, and paternal grandmother's homes in another; to foster homes; to a state-administered children's home; to an uncle's home; to being on the run; to living at 16 with a much older male partner. Similarly, her embedded trauma narratives, which were numerous, featured disorder, including lots of motion but not much focused direction. Events appeared to occur at random or as inevitable, and other people's motivations were elided altogether or depicted as inscrutable. Exemplary of fatalizing talk around the management of relationships in Cicely's narratives of trauma was the protracted and circumlocutory account she gave of her relationship with <unk>ngel, a long-term partner with whom Cicely had two children. At the center of that relationship, and arguably an organizing force in her life story as a whole, was the embedded trauma narrative of the death of Cicely's child, James. Cicely volunteered the story of James's death, here titled "'Til Today I Still Don't Understand," but she clearly found it difficult to narrate. She wove her way to the infant's death gradually. Cicely prefaced with a description of <unk>ngel's strange writing in a secret journal that he called his "manifesto," a detailed, daily chronicle of everything he and she said and did. Then, in another instance of what might be called foreshadowing in a work of fiction, Cicely described how during a home visit a social worker or public health nurse told Cicely that her son lagged developmentally, that "when he was about six months, he was probably about maybe four-you know, his level." In the same segment of narration, Cicely recounted having a mysterious premonition or "prevision," in which, overcome by a sudden conviction that her child would not be with her much longer, Cicely told her mother, a month before his passing, "I don't think he's going to make it." Finally, immediately before the events of the morning of James's death, Cicely shared that she believed she had previously witnessed <unk>ngel trying to smother her older child (by another man), when that child was 6 months old. By these means, Cicely created a context of inevitability, retrospectively constructing the infant James's death as an event destined or foretold. Cicely's fatalizing talk became more intense as she narrated the trauma. In telling the trauma story, Cicely portrayed herself as ba

Jail admissions in the United States number nearly 1 million women annually. Many have limited access to public support and must seek assistance from family, friends, and strangers to maintain health and safety after release. This study sought to learn more about how women with a history of interpersonal trauma and criminal justice involvement perceive and manage social relationships. In-depth, story-eliciting interviews were conducted over 12 months with 10 participants who were selected from the convenience sample of an ongoing parent study in a Midwestern urban jail. Embedded trauma narratives were analyzed for self-presentation, form, and theme. The trauma narratives registered a continuum of agency, anchored at either end by patterns of strategizing talk and fatalizing talk. Providers and advocates can improve support for justice-involved women post incarceration by becoming familiar with and responding to patterns of strategizing and fatalizing in their personal narratives.

a work of fiction, Cicely described how during a home visit a social worker or public health nurse told Cicely that her son lagged developmentally, that "when he was about six months, he was probably about maybe four-you know, his level." In the same segment of narration, Cicely recounted having a mysterious premonition or "prevision," in which, overcome by a sudden conviction that her child would not be with her much longer, Cicely told her mother, a month before his passing, "I don't think he's going to make it." Finally, immediately before the events of the morning of James's death, Cicely shared that she believed she had previously witnessed <unk>ngel trying to smother her older child (by another man), when that child was 6 months old. By these means, Cicely created a context of inevitability, retrospectively constructing the infant James's death as an event destined or foretold. Cicely's fatalizing talk became more intense as she narrated the trauma. In telling the trauma story, Cicely portrayed herself as baffled and helpless, unable to control and struggling to piece together the meaning of events. A dreamlike passivity reigned over "'Til Today," marked in Cicely's delivery by a liberal use of the coordinating "and," the substitution of pronouns ("he," "him") for names, and rhetorical cushioning through repetition of the qualifier "kinda." The account was structured less through a replay of its teller's actions than a series of unanswered questions, bookended by the twin suspicions that <unk>ngel tried once to smother her older son and now might be responsible for the death of her younger one. Within the story, Cicely's questions flowed unanswered from one to the next as she queried what <unk>ngel was doing with James on the bed and what he was doing afterward with the bottle in the closet, why her child was pale and motionless, why the child vomited so much during cardiopulmonary resuscitation (CPR), where <unk>ngel was while she was at the hospital, and what the police and/or courts eventually determined to be true about James's death and <unk>ngel's role in it. Even 8 years after the event, Cicely constructed a story of trauma in which the self she narrated moved through events as a perplexed observer-detached, alone, full of wondering horror, noting without comprehension the gaps left unfilled. Cicely's fatalizing talk exemplified vagueness and passivity, qualities that were nowhere as clear as in her final, aggrieved admission of inaction due to uncertainty: "I stayed with him... I was torn. I didn't know." Cicely's fatalizing talk was at odds with repeated and explicit references to herself in strategizing terms ("And I'm a go-getter, and I will make it fuckin' happen, and we ain't got something, we about to go get it"). In her embedded trauma narrative, she created a mood of helplessness and detachment, deploying formal elements such as questions, gaps, digressions, and foreshadowing to call into relief her diminished role as an agent who could manage social ties to get support for herself and her children. When Cicely acknowledged relationships-the nurse/social worker, her mother-they were perceived not as opportunities for support, even failed ones, but instead functioned as part of the story's machinery of inevitability. Through fatalizing talk, Cicely created what might have been emotionally necessary detachment from a memory that continued to be too painful to bear. She could recollect and narrate the scene of trauma but only through an observer's lens of detached bewilderment. --- Gradations I have concentrated on explicating extreme cases of strategizing and fatalizing in the interviews, though more frequent in the interviews were the many gradations between the two. In "It was Just Boom," Neta highlighted how the two modes of talk might be found working in conjunction in a single story. Neta-Neta was a 45-year-old Black woman with multiple jail incarcerations and more than a decade in prison. "It was Just Boom" presents the conclusion of a story in which Neta described moving several 100 miles from her home and extended family network to live with a partner whom she met in drug rehab. The man beat her viciously over several months, policed her every move, and at one point pushed her out a third-story window. In Neta's narrative, the crisis was no random, unpredictable, inscrutable, or fated event. Neta instead began the story with strategizing talk, reasoning from cause to effect, citing her accountability. Neta judged that it was her own transgression of a well-known rule that led to her predicament: "Tried to get clean again. Went to treatment. I did the big no-no." In what followed, however, Neta switched from an agentic approach to explain the abuse (i.e., "my mistake") to a fatalizing approach to narrate its resolution. That is, once the action shifted to escape, Neta minimized cause-effect and motivation ("It was just boom"). She assigned management of events and relationships to her abuser's sister. The action was both sudden and other-initiated-for example, "she asked me," "she put me"; indeed, the sister issued the imperative and gave it urgency: "If you gonna leave, it's time to go now." As did other women in the interviews, Neta performed a mix of strategizing and fatalizing in her storytelling about crisis. She moved along a continuum of agency, from strategizing talk to fatalizing talk when she shifted from blunder (self-attribution) to remedy (other attribution). It is worth remarking that the use of strategizing talk in Neta's story, while possibly reflecting a useful guideline (i.e., avoid hooking up with others in recovery), signaled as well a kind of victim blaming in which Neta assigned to herself responsibility for the abuser's lack of regard for her human rights. Neta's account of trauma not only highlighted the quality of continuousness in the continuum of agency but also undermined the too-easy assumption that strategizing talk, with its assertion of self, is necessarily beneficial and, by implication, that fatalizing talk is necessarily harmful. As summarized in Table 1, the continuum of agency, for which strategizing talk and fatalizing talk map two poles, was reflected in women's narratives through (a) presentations of self, (b) recurring emphases in theme, and (c) patterned structural and the performative elements. If the modes of strategizing and fatalizing talk designated two poles in theory, in practice they more typically overlapped and alternated, often co-occurring in individual stories, with neither mode proving neatly characteristic of all the stories of trauma told by any one woman. --- Discussion This study examined how strategizing talk and fatalizing talk occurred in trauma narratives in which women with history of incarceration described their struggles to access social support for health and safety. In their accounts of crisis, the storytellers constructed selves and organized and gave meaning to events. When participants like Natalie and Jennifer engaged in strategizing talk, they presented themselves as beleaguered fighters who made or at least tried to make accurate assessments about how, when, and from whom to seek support. Strategizing talk marked stories in which women demonstrated a sense of agency in the midst of crisis. In contrast, in trauma stories where fatalizing talk was more dominant, the women's self-presentation highlighted helplessness, a quality exemplified by Cicely, who depicted herself as overwhelmed and passive, moving through events as a dazed observer rather than as an actor. In fatalizing talk, events unfolded along thematic lines that stressed randomness and the unfathomable, unpredictable will of others-or, alternatively, destiny, the unfolding of something preordained and inevitable. As might have been the case with Cicely, fatalizing talk probably reflected the need of storytellers to buffer unbearable memories-functioning to protect women from painful feelings of reexperiencing, regret, guilt, or abandonment-or as a stopgap for the perceived lack of coherence and connection in event, intent, and outcome that researchers have associated with the recall of trauma (Bromberg, 2003;Janoff-Bulman, 1992). --- Feminist Standpoint and Relationality A central premise of feminist standpoint and relational theory is that women tend to prioritize connectedness over autonomy in understanding self and exercising judgment (Comstock et al., 2008;Miller, 1986). Such tendencies are no doubt found among individuals of any gender group, but according to some feminist-relationalist perspectives, historically, women's socialization might have led them to organize identity, action, and ethics around mutuality over autonomy, cooperation over competition, and collaboration over conflict (Gilligan, 1982;Mackenzie & Stoljar, 2000;Miller, 1986;Miller & Stiver, 1997). The disruption, absence, or denial of social bonds that occurs in trauma (and, for that matter, in incarceration) might be especially difficult for women to cope with (Covington, 1998). The embedded narratives of trauma analyzed for this study indicated that women longed for but were often thwarted in establishing socially supportive relationships. Researchers have found that both during and after incarceration, the psychological effects of complex trauma can produce a combined fear of and need for connection that complicates women's ability to access support through social relationships (Fedock, Fries, & Kubiak, 2013;Fuentes, 2014;Kubiak, 2005). Thus, the strategic talk in both Natalie's and Jennifer's narratives was full of accusatory and alienated references to parents and siblings, making clear the women's perceptions of ruptured connection-of abandonment or isolation in times of extreme need. For Natalie, in particular, this seemed to lead to a still more tenacious insistence on self-reliance. In contrast, Cicely employed fatalizing talk in her trauma narrative to express a sense of broken connections and the unfathomability of others' perspectives and motivations. The effect of interpersonal trauma on relationality for women is a basis for the argument of Covington (1998Covington (, 2007) ) and others who have been involved in the gender-responsive corrections movement, which stresses women's prioritization of relationship and susceptibility to retraumatization as a rationale for criminal justice system reform (Bloom, Owen, & Covington, 2002;Harner & Burgess, 2011;Wright, Van Voorhis, Salisbury, & Bauman, 2012). The present study contributes to this conversation by illuminating what health care and service providers outside of corrections might hear when justice-involved women with significant lifetime experience of interpersonal trauma seek to position themselves narratively as agents in quest of support within the community. --- Managing Social Relationships for Support in a Context of Trauma Social support is governed by principles of direct and generalized exchange undergirded by trust, in which members, even in underresourced networks, assume that everyone gives and everyone takes, so that a kind of balance is maintained across relationships (Dom<unk>nguez & Watkins, 2003;Lourel, Hartmann, Closon, Mouda, & Petric-Tatu, 2013;Plickert, Côté, & Wellman, 2007;Uehara, 1990). For women whose material and psychological resources are overtapped, the requirement of exchange means that even support accessed through family and close friends can be perceived as highly restricted, hedged in with caveats and hazards (Hawkins, 2010;Offer, 2012;Offer, Sambol, & Benjamin, 2010;Radey, 2015). In Natalie, Neta, and Cicely's life stories, close social bonds were subject to limitations similar to those cited elsewhere in the literature on underresourced networks, including exhaustion of bonds due to demands that are too frequent or too great (conditions that become more likely when drug abuse, violence, homelessness, and repeated incarcerations are prevalent; A. Goffman, 2014;Harknett & Hartnett, 2011) and resultant feelings of guilt, resentment, and social isolation (Offer, 2012;Ray, 2015). Only Jennifer presented a divergent case. Jennifer's socioeconomic status and absence of reported physical and/or sexual trauma meant that she almost certainly enjoyed structural and situational advantages unavailable to others in the study, including a social support network with more resources on which she could draw for support. Yet, even Jennifer's trauma narrative-delivered in a primarily strategizing modehighlighted how frustrating and at times discouraging the navigation of support could be. --- Implications and Recommendations The women in this study struggled with drug abuse, poverty, and unstable housing. Nearly, all the women carried a disproportionate share of traumatic loss or injury, much of it involving recurrent interpersonal violence. Given the dearth of public assistance available to women so situated and the high unlikelihood of increased programmatic focus on their needs in the near future, it is urgent for health care and service providers to learn how to hear and build on women's own perceptions of and strategies for accessing informal ties for support. Careful attention to women's narratives should provide lessons for more effectively supporting them in what works and helping them find alternatives for what does not. Health care and service providers who practice in community and clinical settings should be educated in and prepared to give trauma-informed care (Harner & Burgess, 2011). They can further understand that health and safety challenges among women in this group are met within a limited horizon of real and perceived choices related to agency and relationship. Bourdieu's concept of habitus is useful here, as Bourdieu explained how the perceptions that shape action (for him, "practice," 1980, p. 54) occur within contexts that are already structured by socioeconomic determinants (Bourdieu, 1980;Bourdieu & Wacquant, 1992). What might look like poor personal decisions from the outside might not have the look of decisions at all from the women's perspectives. To recognize the operations of a continuum of strategizing and fatalizing talk in women's trauma stories is to acknowledge an underlying dialectic of structure and agency in women's attempts to cope (Rutten & Gelius, 2011;Sweetman, 2003). As an organizing tool, the model of strategizing and fatalizing talk can focus attention on how women manage agency and relationship, self and other, in accessing social support. Health care and service providers who meet with women during or after an incarceration, and who hear their stories in the course of assessing or advising women, need to be aware of both modes of talk and the perceptions about agency and relationship that they imply. However, simply listening for and supporting women in recognizing the ways in which strategizing can facilitate and fatalizing can obstruct the management of social ties might not be enough. In some cases, strategizing talk might mean asserting inappropriate personal responsibility when structural contingencies or other people should be held accountable for a situation. Similarly, fatalizing entails inaction, detachment, and displacement of self, but those might be strategies for survival as well as avoidance, and in certain cases they might represent the best strategies for staying safe. Assisting women in meeting their health and safety needs will require something more than administering an instrument-and probably more than careful listening and dialogue at an individual level. Because this is a population with distinctive shared needs, population advocates, lawmakers, service administrators, and program designers should also have familiarity with how justice-involved women perceive structurally constrained choices. Strategizing and fatalizing talk provide a preliminary framework within which to begin thinking about how to push for programs and policy changes to create coordinated services and more supportive environments and opportunities. These will be necessary to help women recast their stories, learn to reconceive options, and, in particular, ensure that access to resources exists to move women beyond simply getting by. --- Limitations In this study, I derived theory through thematic and structural analysis of narrative data from in-depth interviews with 10 women. I sought to achieve credibility and trustworthiness for the model by sharing sizable segments from the narrative data and a fairly detailed demonstration of interpretive process in the findings, thereby giving readers means to judge the applicability of interpretations to other, similar situations (Guba & Lincoln, 1985). To further strengthen trustworthiness, I used repeat interviews and consulted with a team whose members iteratively and independently memoed on the interviews and regularly discussed interpretations over a yearlong period. Regardless, judged by standards of probabilistic research that seeks to make statistically generalizable claims about populations, the small number of cases might be deemed a limitation by some. I would offer that narrative analysis, the holistic reasoning about underlying realities through detailed analysis of languaged data (Patton, 2015;Polkinghorne, 1988), tends to be information rich, requiring a few detailed cases (Malterud, Siersma, & Guassora, 2016). Indeed, given detailed enough narrative data, analysis might be better performed through interpretation of a few cases than many (Creswell, 2014;Maynes, Pierce, & Laslett, 2008). Another limitation of the study would be that the derivation of a strategizing-fatalizing model does not on its own give much specific guidance as to use, a limitation that might be expected in an exploratory study where next steps could take several directions, including the further refinement of constructs and the operationalization and validation of the model for future use in interventions. Finally, to be more complete, this study might have included consideration of the obverse of strategizing and fatalizing talk, the strategizing and fatalizing silences, or spaces where women who have experienced much trauma leave traces of self in relation to other through what is left unsaid. --- Conclusion Strategizing talk and fatalizing talk emerged from narrative analysis of women's trauma stories as two ends of a continuum that mapped perceived agency in accessing support through interpersonal ties. The women who participated in the interviews were complex individuals, every one of them navigating her own set of highly challenging circumstances. In the community and in clinical situations, health care and other service providers working with justice-involved women can learn to recognize patterns of self-other management to better capitalize on the knowledges and aptitudes that women already evince in accessing support through social ties. Where ways of managing those relationships lead to greater health and safety risks for women, a clearer, more incisive understanding on the part of providers might suggest openings where new interventions can be targeted. Of course, the framework presented in this analysis will only be useful if those who are in positions to give support cultivate the ability to listen closely and respectfully for the nuanced ways in which women express strengths and weaknesses through the stories they tell. --- Supplementary Material Refer to Web version on PubMed Central for supplementary material.

Jail admissions in the United States number nearly 1 million women annually. Many have limited access to public support and must seek assistance from family, friends, and strangers to maintain health and safety after release. This study sought to learn more about how women with a history of interpersonal trauma and criminal justice involvement perceive and manage social relationships. In-depth, story-eliciting interviews were conducted over 12 months with 10 participants who were selected from the convenience sample of an ongoing parent study in a Midwestern urban jail. Embedded trauma narratives were analyzed for self-presentation, form, and theme. The trauma narratives registered a continuum of agency, anchored at either end by patterns of strategizing talk and fatalizing talk. Providers and advocates can improve support for justice-involved women post incarceration by becoming familiar with and responding to patterns of strategizing and fatalizing in their personal narratives.

Introduction Prevention of Human Immunodeficiency Virus (HIV) infection is currently one of the most important global health concerns. Globally, in 2018, 37.9 million people had HIV, of which 1.7 million were new HIV infections [1]. Despite recent progress in HIV prevention [2,3], HIV prevalence has an increasing trend in Iran [4,5]. According to the last UNAIDS report, 61,000 people were living with HIV in 2018 in Iran [6]. In recent years, patterns of HIV transmission in Iran have been changing. Although injection drug use remains the main cause of HIV transmission, sexual transmission has been steadily growing (from 7% in 2006 to 33% in 2012, and to 42% in 2016). Prevalence of HIV among female sex workers (FSWs) was estimated to be 4.5% in 2014 [4,7,8]. A national behavioral survey conducted in 2010 throughout 14 cities in Iran reported that HIV prevalence among FSWs was 4.5% (95% CI=2.4 to 8.3), 4.8% (95% CI=2.2 to 9.8) among those who had reported a history of drug use and 11.2% (95% CI=5.4 to 21.5) among those who had a history of injection drug use [9]. Female sex workers are a hidden and under-reported population whose risk behaviors such as not using condoms, having multiple partners, using drugs or alcohol before sexual relationships, and intravenous drug use, can contribute to an increased rate of HIV transmission [10]. They engage in sexual relations in exchange for something else such as money, drugs, gifts, and shelter. They are not only themselves at risk for HIV, but also may be a bridging group that may transmit HIV to the general population. According to a size estimation study, there were 228,700 FSWs (95% CI: 153,500-294,300) in Iran in 2012 [7]. However, there is little information about this vulnerable subpopulation and their HIV risk behaviors, due to the fact that prostitution is considered as taboo by both government and the general population, and goes along with high social stigma. According to previous studies in Iran, most FSWs knew that condom use is a method to reduce risk of HIV transmission, but only few of them used condoms with their paying and non-paying partners [11]. Although there are some studies about FSWs in Iran as well as other countries, most of them have focused on assessing individual characteristics of FSWs such as HIV knowledge, attitude, risky sexual behaviors, and sexually transmitted disease status [11][12][13]. However, the FSW herself represents only half of a sexual dyad. Considering the characteristics of only one member of the dyad does not provide sufficient information about the social context of HIV-related risk behaviors and the quality of interpersonal interactions that may facilitate those behaviors and health outcomes. There is ample evidence that social networks have an important role generally for risk and health behaviors by providing opportunities for social influence, social support, and social engagement [14,15], and also specifically for sexual behaviors [16]. There are several Several studies which were conducted among vulnerable population such as injecting drug users (IDUs) [17], men who have sex with men (MSM) [18], and adolescents or youth who use drug or alcohol [19], showed that perceived norms are associated with risk behaviors. Also, previous studies showed that social support from network members, especially gatekeepers (managers or pimps) and peers, was significantly associated with condom use among FSWs [20,21]. Despite the advances in our understanding about the potential role of social networks for HIV risk behaviors, our systematic review study showed that evidence about effects of social networks on HIV risk behavior of FSWs is scarce [22]. Most existent studies only considered a few network-related characteristics such as frequency of contact, trust, and social support among FSWs [21,[23][24][25][26][27][28]. Therefore, a comprehensive study to clarify the mechanisms according to which social network characteristics of FSWs affect their HIV risk behaviors is greatly warranted. Our study is about sexual behavior of FSWs, which takes place in the sexual network made up by the FSW and her sexual partners including her clients and regular partners. When studying network effects on this behavior, factors at the level of the individual may be distinguished from those at the network level. The aim of this study was to determine the effect of social network characteristics on the extent of condom use by FSWs in Tehran, considering individual characteristics of the FSWs and of their sexual partners, characteristics of their relationship, and of the FSW's personal network. --- Methods --- Setting This study is a cross-sectional ego-centric network survey of 170 FSWs in Tehran city, conducted between January and June 2017. An ego-centric network survey provides more insight into the effects of the social network around the FSW on her condom use by making an inventory of the personal network of the FSW and collecting relevant individual characteristics of the FSW herself as well as of her sexual partners. This purpose is achieved using a so-called name generator for making the inventory of the members of the personal network, and a name interpreter for determining their relevant characteristics and relevant interactions. In our study, the FSW who is the respondent is the focal individual (ego) in the personal network. --- Participants FSWs are a hard to reach population in Iran, subject to social and official taboos and social stigma. However, in recent years enormous governmental and nongovernmental investments and efforts have been made to increase the access of this vulnerable population to HIV prevention programs, such as the creation of dropin centers (DICs) and consultation centers for at risk women throughout the country. These centers provide free services including basic sexual and reproductive health care, educational programs about sexual transmitted infections (STI) and prevention methods, HIV testing, and counselling [29,30]. It is virtually impossible to obtain a random sample from a hidden and stigmatized population such as this. Therefore, participants were recruited through snowball, purposeful, and convenience sampling methods as successful alternatives to non-applicable random sampling methods for recruiting 'hidden' populations [31][32][33]. Eligibility criteria were being over 16 years old, having had sex for money, drugs, and so on in the last year, identifying oneself as a sex worker, and willingness to participate in the study. For the snowball sample, to initiate the chain referral process, six FSWs who satisfied the inclusion criteria were selected as index participants. After interviewing the index participants, we asked them to introduce some other sex workers whom they had named as their network members. The snowball sample proceeded according to the principles of respondent-driven sampling [31]. We gave the respondents some coupons for their peer friends, which included the ID number, aim of the study, amount of incentive, place of visit for interview, and an expiration date. Each participant was given 100,000 Rails (equal to 3 USD) as a small primary incentive for participating in the study and completing the interview; and a secondary small incentive, of the same amount, if they had introduced their peer friends, who did sex work and named as their network members. This process continued until five waves. In all waves the participants introduced few peer friends. Some introduced only one or two peer friends named as network members, some others introduced other FSWs who had not been named as their network members. It was impossible to access all peer members who were named by the participants. We completed the sample with convenience sampling and purposeful sampling methods. Convenience sampling was used by recruiting participants among FSWs who attended DICs and consultation centers. Recruitment of study participants in DICs was facilitated by DIC staff, who were personally contacted at the center by the first author. The DIC staff identified potential participants and introduced them to the interviewer. To improve diversity from all involved sites, we also used purposeful sampling from outreach spots such as team homes, streets, and parks by a peer outreach worker who collaborated with the study via a DIC. To maintain anonymity of the FSWs, we received verbal informed consent from all participants, because sex work is illegal in Iran. We provided an explanation regarding the study purpose to all participants, and informed them about the confidentiality and their right to withdraw from the study at any time during the interview. --- Measurements and variables The data were gathered using an investigator-constructed questionnaire, after assessing content validity, scalability, and reliability. The questionnaire consisted of two parts, individual and network information. Individual information was collected about the demographic characteristics age, educational level, marital status, and place of living, and about the frequency of sex work in the last month, HIV knowledge, and HIV test. For network information, first, a name generator inventory [34] was applied to indicate the FSW's sexual network members. The sexual network was defined as the set of nominated persons with whom they had had sex in the past 30 days. We asked the FSWs to nominate up to 5 persons with whom they had any sexual relationships during the past 30 days. For these nominated persons further information was collected including socio-demographic information, duration of contact, frequency of contact, frequency of condom use, intimacy, social support, and drug or alcohol use before a sexual relationship with him. The data was collected in face-to-face structured interviews by trained peer interviewers, which was helpful to build trust and get honest responses from the participants. Each interview usually lasted around 45 minutes. --- Socio-demographic variables The socio-demographic variables were collected as follows. Age was recorded in years. Educational level was measured in six ordinal categories, coded as 1 to 6: illiterate, able to read and write, primary education, secondary education, high school or diploma, and university education. Marital status was measured in four categories as single, married, divorced, and widowed; for the analysis this was dichotomized into never married and ever married. Place of living was categorized as homeless, living in the home of others, and personal home. The number of people supported by the FSWs was measured as a count; for the analysis it was dichotomized as zero versus more than zero. The variables collected for the network members were age and educational level, measured similarly as for the respondents. --- Social network variables To assess the frequency of contact with the sexual network members, we asked the participants "How many times did you meet him or did you communicate with him in the last month?". To assess duration of the tie we asked the participants "Since how long have you known him?". This was coded in months. The intimacy of the relationship between participants and sexual network members was assessed with a five-point Likert response, with categories very close, close, somewhat close, distant, and very distant. Social support was assessed by an investigator-constructed questionnaire with five items. Reliability of the items comprising the social support questionnaire was pilot-tested prior to final implementation in this study. Cronbach's alpha and intraclass correlation for the scale were.82 and.85, respectively. Mokken scale analysis was used to assess scalability and unidimensionality of the social support questionnaire [35]. The Loevinger H-coefficient for the questionnaire was larger than 0.5, characterizing it as strong scale. Because sexual network members could be nominated for more than one role or interaction (a sexual partner may also have a familial tie and/or be a drug use partner), the multiplexity of their position was determined. This was defined as 1 if they had more than one role in the respondent's network, and 0 if they had just one role in her network (i.e., only sexual partner). The density of each network was also assessed. To assess density, participants were given a matrix with the names of their mentioned sexual network members on both axes, and then were asked to indicate the pairs of network members who knew each other. The density was calculated as the number of sexual network members who knew each other divided by the maximum number possible, given the size of the sexual network. The variable of drug and alcohol use before or with sex was dichotomized into "yes=1", and "no=0". --- Perceived safe sex norm Perceived safe sex norms were assessed by the perceived norm scale (PNS) of the safe sex norm questionnaire (SSNQ) [16]. The PNS assesses perceptions of the FSW regarding attitudes and behaviors about condom use of their peer friends who do similar sex work. It uses 17 questions each on a five-point Likert scale: "all", "most", "about half ", "some" and "none". These were coded as scores from 1= "none" to 5= "all". The reliability of the questionnaire was pilot-tested. The attainable score range is 17-85. Higher scores indicate that the perception of FSWs of the attitude and behaviors of their friends is that most of them have a positive attitude about condom use and frequently use it in their sexual relationships [16]. Cronbach's alpha and the intraclass correlation for this scale were 0.89 and 0.83, respectively. Mokken scale analysis was used to assess scalability and uni-dimensionality of the questionnaire [35]. The Loevinger H-coefficient for the scale was larger than 0.5, characterizing it as strong scale. --- HIV Knowledge Knowledge about HIV transmission was evaluated by a 14-item tool based on an established questionnaire in the Iranian population [36]. The questionnaire was pilottested with 28 participants in Tehran. Scalability and uni-dimensionality of the questionnaire were assessed by Mokken scale analysis [35]. Each item and the whole questionnaire had Loevinger H coefficients above 0.4, which is good. Cronbach's alpha for this scale was 0.86. The sum score was transformed to a scale of 0 to 100, with high scores meaning more knowledge. --- Sexual practice and HIV test For assessing sexual practices, the participants were asked to report about the frequency of sex work in the last month. This was recorded as a count variable. Participants were also asked to report if they ever had a HIV test. This, together with the test result, was recorded as a categorical variable with values "having HIV test & positive result=1", "having HIV test & negative result or don't know=2", and "having no HIV test=3". --- Condom use Frequency of condom use by participants and the sexual network member was the dependent variable, measured on a five-point ordinal scale. It was defined as Yij = 1 if for network member i of respondent j condom use was reported as 'never', Yij = 2 for 'rarely', Yij = 3 for'sometimes', Yij = 4 for 'often', and Yij = 5 for 'always'. --- Statistical analysis Pearson correlations were computed to get a basic insight about patterns of association of all variables. The distributions of the variables were assessed carefully for missing values and outliers, as these might unduly affect the results. Frequency of sex work had three values larger than 20 (two values 25, one 60) which were considered outliers; these were truncated to 20. Taking into account the FSW as well as the sexual partners implies a multilevel structure [37], with the sexual partners, as network members, nested in FSWs as respondent level. The dependent variable, condom use, is a variable at the partner level. As explained below, the analysis proceeds in steps according to a conceptual framework summarizing the role of individual and network characteristics for HIV risk behaviors of FSWs. This framework is based on previous studies of the effects of various factors on HIV risk behaviors [14,[38][39][40] and goes from the more general background characteristics to the social network characteristics. Responses about network members of the same respondent are likely to be correlated. Therefore, we used multilevel analysis [37,41]. The number of respondents (called 'level-2 units' in the terminology of multilevel analysis) was 170; the total number of sexual network members ('level-1 units') was 615. We created a group mean (for groups defined as all sexual network members of a given FSW) for each explanatory variable at the level of the sexual partner, including age of sexual partner, education of sexual partner, frequency of contact, duration of tie, intimacy, social support, and drug or alcohol use before or with sex. This is required to investigate the difference between within-group and between-group regressions [37]. The within-group regression coefficient is the estimated parameter at the partner level, the between-group regression coefficient is the sum of the FSW-level and the partner-level coefficients. Since the distribution of the dependent variable is highly skewed, with five values, we employed a multilevel ordered logistic regression model for ordered categorical outcomes [37]. This was the multilevel proportional odds model, which can be formulated as threshold model with C <unk> 1 thresholds where C is the number of categories of the outcome variable; here C = 5. The mathematical expression of the model is Here Y ij is the observed dependent variable for network member j of respondent i; c is an outcome ranging from 1 to 5; and P indicates probability. <unk> ij is a hypothetical unobserved auxiliary variable, which can be regarded as an underlying continuous variable that is observed after categorization according to thresholds <unk> 1, <unk> 2, <unk> 3, and <unk> 4. The observed outcome is c when <unk> ij is between the two thresholds <unk> c-1 and <unk> c, where the two outer thresholds formally are defined as minus or plus infinity: <unk> 0 = <unk> <unk> and <unk> 5 = +<unk>. The <unk> h are regression parameters; finally, the x hij are the explanatory variables, which cover characteristics of respondents i as well as of sexual network members j; and U 0j is a respondent-level random effect with a standard logistic distribution. We calculated the intraclass correlation coefficient (ICC), a descriptive statistic that measures the proportion of total variance of an outcome that is accounted for by the group level; in this case, the groups in the data refer to the FSWs. In other words, the ICC measures similarity in condom use between sexual partners of the same FSW. It was calculated according to formula (17.26, page 311) in [37], taking the within-group variance equal to <unk> 2 /3=3.29 (the variance of the logistic distribution). The model selection utilizes a conceptual framework based on previous studies of the effects of various factors on HIV risk behaviors [14,[38][39][40] and goes from more general background characteristics to the social network characteristics. This framework distinguishes three groups of independent variables. The first group is composed of the individual background characteristics of the FSW and her sexual partners, as indicated by age, education, and number of supported people. The second group consists of HIV knowledge and the behaviors directly associated with sex work: its frequency, whether it is accompanied by drug or alcohol use, and HIV testing. The third group is composed of social network characteristics and psychosocial mechanisms through which these may affect condom use, regarded as a behavior P Y ij = c = P <unk> c-1 <unk> <unk>ij <unk> <unk> c <unk>ij = h <unk> h x hij + U 0j protecting against HIV risk: personal network density, and the tie characteristics such as duration of the tie, frequency of contact, and intimacy, social support, social norms, and drug or alcohol use before or with sexual relationship. These three groups may be interpreted as reflecting a hypothetical causal ordering, but we use this as a framework guiding the analysis and do not rely on assumptions of causality. In the multilevel ordered logistic regression analyses, a stepwise model selection procedure was employed, in which the groups of variables were entered sequentially. This allowed estimating the effect of social network characteristics on condom use while controlling for individual background characteristics of the FSWs and their sexual partners. It started with the empty model which contained only the dependent variable and the threshold parameters. Micro soft Excel was used for data management (data entry, quality control and cleaning of quantitative data). The data was analyzed by the ordinal package [42] in the R statistical system [43] which allows fitting a variety of mixed effects models for categorical outcomes. p-values less than 5% were regarded as statistically significant. --- Results The mean age of the participants was 34.4 years (SD=7.6). Of the participants, 71 (42%) had a high school or diploma degree. Most participants reported living in their personal home (69, 40%). The mean score of HIV knowledge was 78.2 (SD=19.2). Among the participants, 11 (6%) reported that they had been HIV tested and were HIV positive. The mean age of initiating sex work and the frequency of sex work, after truncating outliers in the latter to 20, were 24.1 (SD=6.6), and 10.2 (SD=6.4), respectively. Table 1 shows the socio-demographic characteristics of the participants in detail. As Table 2 shows, the distribution of condom use, the dependent variable, was highly skewed. For most of the partners, FSWs reported that they never used condoms in sexual relationships with them in the last month (N=346, 56%). Only for nine sexual partners (1.5%) did the FSW always use condoms. The participants named 615 people as their sexual network members. Table 3 gives descriptions of their characteristics. The age of the sexual partners was significantly correlated with the age of the FSWs (Mean=36.7, SD=7.5 vs Mean=34.4, SD=7.6, r=0.64, p <unk> 0.01). Many of the sexual partners like the FSWs had a high school or diploma degree 194 (32%). Many of the sexual partners were married (280, 46%). The mean sexual network size and density were 3.6 (SD=1), and 0.5 (SD=0.3), respectively. Among the sexual partners 175 (28%) had more than one role (e.g., familial tie or drug use partner) in the social network of the FSWs, which indicates multiplexity in the network. The FSWs reported that they did not know the HIV status of 61% of their sexual partners before the sexual relationship. The mean of the perceived safe sex norm was 34.7 (SD=10.4). In the multilevel ordered logistic regression analyses, the intraclass correlation coefficient (ICC) was = 15.51/ (15.51+3.29) = 0.82, signifying high similarity in condom use between the sexual partners of a given FSW. Table 4 presents the results of the multilevel ordered logistic regression analyses according to the three models. The results of the first model, predicting condom use with individual background characteristics of the participants and their sexual partners, showed that FSWs' age, partners' age, and partners' education were positively associated with condom use. Also, the mean education of the FSW's partners was positively and significantly associated with condom use. A one-point increase in the average of partners' education increases by 1.83 the conditional log odds of condom use. However, the mean age of the partners was not significantly associated (Model 1). Model 2, which included variables related to individual sexual practices and HIV test, adjusting for individual characteristics of FSW and sexual partners, showed that frequency of sex work in the last month was negatively associated with condom use. One unit increase in HIV knowledge of the FSWs significantly increases the ). In this model, individual background characteristics of the FSWs and their partners including age of the participants, age and education of their sexual partners, and HIV knowledge, after adjusting for other variables, still showed significant associations with condom use (Model 3). The R 2 values [37] for Models 1 to 3 were 44, 54, and 69%, respectively. --- Discussion An interesting descriptive finding was the similarity between socio-demographic characteristics of the FSWs and their sexual partners (homophilous sexual partners in network terminology), consistent with Newcomb's study among MSM in 2013 [44]. Such a similarity may create feelings of mutual perception, trust, and emotional closeness, which may lead to unsafe sex behaviors. Another important descriptive finding was the marital status of the FSWs and their sexual partners: most of them were married. This finding is consistent with other studies, and highlights the bridging role of FSWs and their sexual partners in transmission of HIV to the general population. Our finding showed that condom use on average was rather low. This finding is consistent with previous studies [45,46] and suggests that not only infection risk, but also pregnancy risk is not a big stimulus to use condoms among Tehran FSWs. A national bio-behavioral survey among Iranian FSWs in 2010 reported that about 35% of the FSWs had a history of lifetime abortion. This abortion figure suggests that the effective protection against pregnancy risk is not very strong [46]. One explanation for this finding is related to the social stigma and negative attitudes from the community towards FSWs which has constructed an environment that is hardly conducive to accessibility of HIV prevention, condoms, and reproductive health services such as contraception [47,48]. According to the final model of the multilevel analysis (Model 3, Table 4), the age and HIV knowledge of the FSW as well as the age and the educational level of her sexual partner had a positive effect on condom use. As regards the behavioral and attitudinal variables, intimacy with the sexual partner, as well as the duration of the tie, and the frequency of contacts, all had negative effects on condom use. For most variables at the partner level, the only variable with a significant effect for the group mean was frequency of contact with the sex partner, for which the within-group regression coefficient was estimated as -0.270 and the between-group regression coefficient as 0.075-0.270 = -0.195. This means that FSWs used condoms less with the clients with whom they had more frequent contact. Also the average condom use for FSWs who saw their clients more frequently was larger than for other FSWs, but this effect was less strong that might be expected based on the differences between clients for any given FSW. As regards other characteristics of the social network as a whole, perceived social support from the sexual partners, and perceived safe sex norms in the network of peers had a positive effect on condom use. One explanation of the effect of age may be that older people have more knowledge and experience especially about consequences of unsafe sex and risk behaviors, and therefore may act more conservatively and more frequently use condoms in their sexual relationships. A related finding by Schick et al. [49] is that American adults over 50 years old who were in situations that posed an increased potential for risk (e.g., an unknown partner history or STI/HIV status) were more inclined to use condoms in their sexual relationships. However, some other studies, especially in the general population, showed contrary results as adolescents and younger adults used more condoms in their sexual relationships than older adults [50]. These contrasts may be due to differences in the studied populations, methodological aspects, and also social and cultural contexts. Iranian society is characterized by specific cultural and political settings, within which religion has an important role; educational programs for adolescents and young people, especially in public settings such as schools, are not usual and meet with various social and cultural sensitivities. Therefore, adolescent and young people may have insufficient information and knowledge about HIV risk behaviors, sexually transmitted diseases, and risk prevention methods, compared to adolescent and young people in other countries. Another reason which is specifically related to the context of our study, the FSWs in Tehran, is that younger FSWs, especially those who have relationships with older FSWs or pimps, may have lower HIV risk perception, because they have not been informed sufficiently about HIV risks and are more strongly controlled by the older FSWs or pimps in their networks [51]. FSWs with more HIV knowledge reported more frequent condom use. However, in spite of their rather high mean HIV knowledge, many FSWs reported to rarely or never use condoms in their sexual relationships. This suggests that having HIV knowledge in itself is not sufficient for safe sex practices among FSWs. The positive effect of knowledge corresponds to the positive effect of the educational level of FSWs' sexual partners. Sexual partners with a higher educational level may have more knowledge about healthy behaviors and consequences of unsafe sex (health literacy), and may better understand the necessity of condom use. This finding is consistent with previous studies [11,12,52] and highlights the importance of educational programs for FSWs' partners to prevent HIV risk behaviors. FSWs with more frequent sex work in the last month were less likely to use condoms. However, the statistical significance of this association was a borderline. Frequent sexual relationships may go along with having multiple sexual partners, which would add to the risk of HIV infection since having frequent sexual relationships may provide greater opportunity for HIV transmission due to increase the likelihood of unsafe sex and exposure to HIV infected partners. This finding is consistent with previous studies [53,54]. Our findings showed that frequency of condom use varied with the sexual partners. The FSWs used condoms significantly less with partners whom they had known for a long time; had more frequent contacts with; with whom they perceived more intimate relationships; and from whom they obtained more social support. These findings are consistent with previous studies [27,[55][56][57][58]. However, only one of these is a quantitative social network study among FSWs with partner-level information [57]. One explanation is that these mechanisms may create mutual trust and increase feelings of emotional closeness relationships between the FSW and her sexual partner [59]. After a long-term and repeated partnership, the FSW may have become familiar with her sexual partner and feel she knows him well. This may lead to accepting his requests for unsafe sex. This may further depend on several reasons such as the socio-economic status of the FSW and her dependence on her sexual partner [57], considering condom use as a threat to the trust in an intimate relationship [58,60], power-gender inequality, inability of the FSW to negotiate with her partner about safe sex, and fear of the consequences of conflicts with him. Further research is needed to better understand the association between duration of tie, frequency of contact, and intimacy with condom use in this at-risk population. Also, in line with previous studies [61,62], our findings indicated that FSWs who perceive that their network members, especially their peer friends, have positive attitudes about safe sex and condom use, reported to use condoms more than those who did not have such a perception. This finding is consistent with Peterson and colleagues [18] who reported that men who have sex with men in the high-risk group, compared with those in the no-risk group, perceived lower positive reactions about condom use among their sexual network members. This confirms that networks are an important context for understanding social norms [63] and suggests that when developing interventions to change behavioral norms, the network context should be taken into account. When trying to promote condom use among FSWs, it may be beneficial to consider not only the attitudes and behaviors of the FSWs themselves, but also the attitudes and behaviors of their network members, especially those of their peer friends, who are colleagues and friends at the same time and may have dual roles as competitors and support providers. Network-based interventions may be useful to modify network relationships so as to create a social environment that can facilitate sexual behavior changes [64,65]. --- Strengths and limitations Our study was conducted on a large sample of FSWs and obtained partner-level information about sensitive behavioral aspects of sexual behavior. We found that condom use on average was rather low. Using a multilevel approach to include variables at the personal, the relational, and the network level, we were able to obtain a high value of R 2 to explain condom use. Variables that stood out in the explanation were associated with age, education, perceived norms in the network, and several variables related to intimacy with the partner. This information may be useful for effective evidence-based HIV prevention interventions. However, our findings should be considered also in the light of three main limitations. First, our study was cross-sectional. Therefore, we cannot draw causal inferences. Second, our findings might be subject to measurement error because our data was self-reported and collected from the focal individual in the personal network about their network members. Finally, our findings cannot be generalized to all FSWs in the country because we recruited the participants using a non-random snowball sampling from one city in Iran. --- Conclusion The findings highlighted that considering only the individual characteristics of FSWs is not sufficient and we should also consider factors at the network level, especially the role of sexual partners, on condom use among FSWs. FSWs who had high-quality interactions with their sexual partners also exhibited more unsafe sex in their sexual relationships. The results suggest that to effectively promote condom use it is necessary to employ combined intervention programs that focus on the FSWs themselves as well as members of their social network. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. --- Authors' contributions ZJ, TS, SAH, HS, and YS contributed in the study design, data collection and had critical feedback on manuscript. All authors assisted in the prepare data for analysis and data analysis, writing, critically reviewing multiple manuscript drafts, read and approved the submitted version. --- Declarations Ethics approval and consent to participate The study protocol was reviewed and approved by the Ethics Committee of Tehran University of Social Welfare and Rehabilitation Sciences (IR.USWR. REC.1394.187). All methods were carried out in accordance with relevant guidelines and regulations/Declaration of Helsinki. Informed consent was obtained from both literate participants and the legal guardian(s) of all illiterate participants. --- Consent for publication --- Not applicable --- Competing interests The authors declare no conflict of interest. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background: Despite the steady growth of sexual transmission of HIV, there is little evidence about safe sexual behavior of FSWs, and social network effects on this behavior, in Iran. Our aim in this study was to determine the effect of social network characteristics on condom use among FSWs, considering individual characteristics of the FSWs and of their sexual partners, characteristics of their relationship, and the FSW's personal network.A cross-sectional ego-centric network survey of 170 FSWs was carried out in Tehran between January and June 2017. A multilevel ordered logistic regression analysis was conducted to examine the effects of individual and relational characteristics simultaneously. Results: Condom use in sexual relationships of the FSWs on average was rather low. Important determinants of safe sexual behavior were found both at the level of the individual FSW and at the level of the sexual partner. The main determinants at the level of the individual FSW were FSWs' age and HIV knowledge. At the level of the sexual partner, age and education of sexual partners, as well as intimacy, duration of tie, frequency of contacts with a given partner, frequency of contact, perceived social support, and perceived safe sex norms were significantly associated with condom use.The findings highlighted that considering only the individual characteristics of female sex workers is not sufficient for effectively promoting condom use. Factors at the network and dyadic level should also be considered, especially the role of sexual partners. Network-based interventions may be useful which modify social relationships to create a social environment that can facilitate changes in sexual behavior.

Sociology Discovers Disasters Suddenly, disasters are everywhere. The social sciences have recently increased their output in disaster writing massively. The world is one big disaster. Crisis looms. The end is near. One way to diagnose this state is by pointing to an actual increase in disasters. This could be called a naturalization of the problem. Another diagnosis is to point to a general catastrophic cultural mood, a Zeitgeist, what we could call a culturalization. The first diagnosis, naturalization, is problematic for two reasons: disaster statistics tell a complex story: roughly speaking, throughout the 20 th century, the number of people killed by disasters has decreased, while the number of disasters and the damages reported has increased. 2 In short, society protects people better, but disasters have become more frequent because people build and live in increasingly disaster prone areas. Moreover, sociologically speaking, discourses need not be in sync with events, as every student of anti-semitism or racism knows. Just because there are more disasters, there need not be more attention to them. Conversely, an increase in perceiving disasters does not necessarily mean that there are more disasters. There can be other reasons, as the forecasting of the now forgotten Millenium Bug showed. The second, the general Zeitgeist argument may be true, but it is unlikely: Why should it hold for many societies on very different paths? Why would we assume its continuity, after the end 2 The number of people killed by natural disasters has decreased from an average of a million or so per year in the first half of the twentieth century to a few hundred thousand per year in the last decades, but the number of disasters reported, the number of people affected, and the reported damage measured in Dollars has increased by several orders of magnitude since the 1960s. For technological disasters, the same holds, except that the number of disasters reported has decreased since ca. 2000. All data are taken from the International Disaster database EM-DAT: http://www.emdat.be/natural-disasters-trends (last visited 2nd of may 2013). time scare of the Millennium Bug faded? Also, a preliminary bibliographical analysis with Google Ngram shows that the general thematizing of disasters did not really increase, while the sociology of disasters increased remarkably since the mid-1990s. 3 The question then is: Why can we observe such an increase in dealing with disasters in sociology and its neighbouring fields, an increase that is way out of proportion compared to the general increase in disaster literature and the actual amount of disasters? Insert Figure 1 here: Google Ngram of books containing the words sociology of disaster, sociology, disaster (relative total amount of books) I will thus attempt a third answer, which we could call politicization, much more pertinent to this volume. The answer proposed here is that disasters emerge because our theoretical 3 Google Ngrams shows the percentage of books containing certain words relative to all books on Google books. It is a relative, not an absolute measure. I have used the ngrams for "sociology of disaster" and "sociology of disasters", because these are a small amount of texts. The Ngram presented here have "sociology of disaster" multiplied by 10000 to make them comparable with the other two Ngrams. In other words, unsurprisingly the word "disaster" occurs approximately 10 000 times more often than "sociology of disaster(s)". The Ngram for sociology is included to show that the increase of "sociology of disaster" is not a function of a general increase in literature containing the words "sociology" (which actually decreases after a peak in the mid 1970s). apparatus makes us more sensitive to them. They allow social science to test various theories and interests that have come to the fore in the last ten or twenty years. This third answer does not necessarily contradict the other two. It sits beside them. This volume contributes to a further calibration of our conceptions of disasters, focusing on how accounts of disasters are produced and the effects they have in the world. This calibration of disasters also allows us to move the sociology of disasters from the applied margins of the discipline, as an ancillary science to the practical concerns of disaster management, to the central concerns of general sociology (Tierney 2007). The problem of naturalization and culturalization is that they both conceptualize disasters without recourse to politics. In both options, disasters increase, or discourses increase, but how this relates to politics remains unclear. In contrast, the various articles in this volume attempt to understand disasters as politics, and politics as disasters. In short, they analyze both the notion of disaster and the notion of politics. --- Two Movements to Conceptualize Disasters This new relationship between politics and disasters can be understood by combining two important movements within the social sciences. The first relates to an interest in breaks and ruptures, rather than continuity and structure. This is closely linked to an attendant idea of politics as problematization of the composition of the world. Disasters as ruptures produce new compositions of the world and they force explications of these compositions. The second movement relates to an interest in reconceptualising nature or the non-human as actors. Disasters, like accidents, are, sociologically speaking, the result of the combination of these two: they radically question the composition of the world, in all its technical, natural and social forms. Before discussing these two movements in detail, it is important to stress that conversely, there are other sociological ways to understand disasters, which eradicate either of the two. To start with, it is possible to understand disasters not as ruptures but as exaggerated continuations of the normal (Woodhouse 2011). From such a perspective, there are no disasters as ontologically different events from other events in the world. To believe that disasters are exceptional is to misunderstand disasters. Disasters are nothing but what happens in the world anyway, just with a different intensity. From such a perspective, there is no need for a different way of analysing and reacting to disasters. There is neither a need for a different methodology, nor for a specific theory, and there is indeed nothing inherently interesting that sociology can learn from disasters. Whatever society is, disasters are part of it. To further elaborate this argument, one could say that to insist on the ontological specificity of disasters implies buying into an ideology of disasters, that uses disasters to legitimate certain political goals (more on this below when discussing the state of exception). Furthermore, in this argument it is possible to understand disasters as purely social events. Indeed, the original sociological attempts to understand disasters first needed to establish disasters as sociological issues, claiming them from the monopoly of the natural sciences. Sociologists introduced the term disaster to differentiate from hazards, understood as a physical events (Perry 2007, p.8 ff.). According to this definition, a hazard turns into a disaster by its social effects. An earthquake happening in a region where no humans live is not a disaster, it is just a trembling of the earth. This definition, then, is not about separating the human from the non-human but instead about trying to account for the fact that some natural events relevant for natural scientists -earthquakes where nobody is harmed -are not relevant for a sociology of disasters. This argument is, so to speak, premised on a negative definition of disaster: a disaster is a rupturing event, specifically one that ruptures human society. Similarly, sociologists helped to establish the idea that disasters are not natural or technical, in the sense of inevitable and uninfluenced by society, not only in their effects, but "man-made" (Turner 1978) and result from organisational routines (Weick 1993;Vaughan 1996). Such a definition was needed to create space for sociology within disaster studies, a field still dominated by the natural sciences. But for disaster studies within sociology such a definition is tautological: once they are thematized by sociology, what else would disasters be if not social events? Rather, the problem for sociology, once it has identified disasters as a proper theme, is how to conceive of disasters without understanding them as purely social events. Or, in other words: the problem for sociology is how to understand something that has its origins (at least in the case of natural and technical disasters) so obviously not within society. Thus both of these negative answers of a sociology of disasters become framed, visible and problematic through the development of two separate trends within the social sciences. The first trend is to conceptualize disasters as ruptures and thus inherently political and second to conceive of them as not within society but still an object of sociology. Taken together these trends demand certain theoretical changes within the general apparatus of sociology. Only once such a general re-orientation of sociology is in place, do disasters start to make sense as relevant objects for a general sociology as opposed to their being simply another object for the expansion of sociology. Disasters, as non-social ruptures, are ideal test cases for these new strands of sociology, precisely because they highlight and enable the discussion of these new orientations. To understand the first movement, it is important to see how disasters are at odds with most sociological theories and the foundational assumptions of social theory. Since Hobbes' Leviathan, the problem for social theory was the problem of order and the explanation of stability: in short, to look at what connects one social instance with another over time, rather than looking at events that punctuate continuity and disassemble one instance from another. 4In this tradition of social theory man is disaster -and politics (or society in general) is what saves men from killing each other. Whether it is the state, or values and the social system (Parsons), fields (Bourdieu), imitation (Tarde) or technology (Latour): the arrow of explanation is always towards explaining what holds society together, what produces stability and predictability, assuming that society has a "natural" tendency to fall apart. This is why early disaster studies needed to legitimate this unusual derivation. When Erikson did his pioneering study of the Buffalo Creek flood, he thus felt compelled to defend his decision to focus on one "unique human event", "a task normally performed by dramatists or historians" (Erikson 1976, p.246).The predilection to explain stability also explains why many studies of post-disaster communities can make the seemingly counter-intuitive claim that after disasters communities do not necessarily fall apart, but re-adjust (Carr 1932, p.213), are inventive or even hold together better (Jencson 2001;Kendra & Wachtendorf 2007). Breaching experiments in ethnomethodology were probably the first attempt to put rupture at the expense of continuity at the heart of social theory and to use rupture to learn about the composition of the world (Garfinkel 1967). The study of controversies has borrowed this insight by understanding that calling into question, critiquing and disagreeing are sociologically productive processes. These perspectives together have informed studies of accidents, breakdown and repair (Heath et al. 2000). As intellectual precursors of disaster studies (as understood here), studies of accidents showed in small scale how the world falls apart and needs to be put together. Disasters considered as ruptures expand these insights and methodologies to a much larger scale and even to the world (see Clarke, in this volume). 5 By shifting the focus to large scale events and the problem of the contingency of the world they allow us to see how the world is composed. Disasters, then, become inherently political events because they pose questions about who should be allowed to re-compose the world and how (see Farias and Tironi). The shift from accidents to disasters is, then, not merely one of scale but of focus: if a disaster cannot be contained within one location, one machine or one organisation, issues of politics, distribution and justice come to the fore. Who should be responsible for action (Farias)? How does society distribute preparedness (see the articles by Easthope and Mort, and Deville et al.)? How should collectives make decisions regarding risks (Weszkalnys and Ellis)? How should the world be rebuilt (Tironi)? To understand the second movement, it is important to understand that the concepts of disasters and of preparedness measures are both at odds with a purely social sociology. If sociology is the analysis of the social world, actual disasters (and not only their aftermath) become difficult to describe for sociology. How to describe a disaster, if not by describing the movement of earth, the masses of water, the falling trees, levees and houses, the rubble and dust, the birds (Rodr<unk>guez-Giralt & Tirado) and contaminated letters (Ellis)? How to explain preparedness and recovery without recourse to bulldozers, dams and bunkers (Deville et al.)? This new sociological interest in materiality has its roots in the laboratory and in new technologies. It emerged from attempts to understand how scientists construct facts and invent new technologies. From these initial questions, it has spread to all kinds of fields, but it is important to keep its origins in mind since, as Clarke has argued, the materialism of Actor-Network Theory (ANT) and its allies is a materialism of divide and control over technological artefacts (Clark 2010, pp.34-40). It usually assumes scientists and engineers construct things in order to control other things or people. In disasters, however, the situation is very often quite different. Disasters are situations when matter is out of control and, compared with studies of accidents, out of control on a massive scale. In fact, if there were a need to tell accidents apart from disasters, it would be the remaining amount of control over actor networks. Accidents are contained and control lapses momentarily and spatially on a relatively small scale -a space ship explodes, a car crashes, a train derails-, but the larger network continues to work (it is no coincidence that the examples here most relate to traffic). Repair, as a counter strategy can focus on the technology that stopped working (Graham & Thrift 2007). In an earthquake, nuclear disaster or flood, there is no such possible focus. Disasters then, would be defined as networks or cosmograms themselves that collapse, which radically poses the question of the composition of the world as a whole. Power and intelligibility ceases not just for one particular part, but for all parts involved. Disasters, even more than accidents, are test cases to understand a world in which the material and the natural are not only an object of concern and control, but the very origin of radical change. But this very wholeness also poses a problem for Actor-Network Theory because the vocabulary of ANT is geared towards disentangling big concepts into micro-processes. But, one might argue, disasters are precisely those events that cannot be disentangled and that act and are experienced as one big entity. As suggested in the articles of this book, it might be argued that the analytical tools of ANT, and sociology in general, are not very well equipped to deal with such things. It is no coincidence then that the articles assembled here do not give an account of disasters as wholes but of what follows before and after. Temporalization of punctual events is the analytical strategy of choice, but we can legitimately ask, whether this does not miss the very object of analysis. In that sense, a true sociology of disasters still remains to be written. Based on the two central movements in the sociology of disasters, described above, the articles in this volume adopt different angles in analyzing disasters. The articles collected here are attempts to look at how disasters reconceptualize politics and how politics reconceptualizes disasters. Thereby they engage in various kinds of symmetry (Law 2003). By symmetry it is implied that the articles shy away from settling too quickly for one side of any hot or strong dichotomy: whether it is truth or falsehood, political or scientific explanations, the attempt at explanation or the attribution of blame (Potthast 2007). Symmetrical approaches rather take as their starting point to inquire how attributions to one side of various distinctions are empirically accomplished: why and when are people blamed or systems explained? When are scientific or political accounts preferred? When are situations explained with symbols and meaning and when with materiality? However, independently of such symmetries, the authors in this collection begin from different starting points and it is therefore worthwhile to point out how these starting points relate to disasters and politics. The following pages thus try to analytically grasp what it means to start with politics or disasters as the vantage point of analysis. This also allows us to situate the contributions in this book within a wider literature on these issues, including some vantage points not covered. Starting from either disasters or politics produces different accounts of both of these concepts with quite different political solutions, even if these remain often implicit. To sketch these options then allows us to also better understand the underlying conceptions of disasters and politics, through a comparison with some alternatives not taken and not covered in this book. Disasters and politics may be conceived in the two following analytical ways: first, as "disasters producing politics" and second, as "politics producing disasters". In the former case, disasters are relevant because they are productive. The disaster itself is noteworthy for its capacity to produce a particular kind of politics. The latter focuses on the problematic role of politics to produce disasters. From this point of view, it is politics itself, as a mode of ordering the world that produces disasters for its own purposes and according to its own rules. Apart from asking how these approaches conceptualize the relationship of disasters and politics, further questions follow: What are their aims of explaining or changing the world? How do they distribute power and blame in the world among actors? Do they seek to distribute it among many actors, or attribute it to one? How do they take into account "nontraditional" actors, such as lay people or nature? How do they distribute an empirical focus between these various actors? --- Disasters as Producing Politics There are at least three versions of "disasters as producing politics" with changing levels of force attributed to disasters. First, there is an approach, which could be termed disasters as prime empirical sites to understand politics. This approach is not particularly attentive to disasters as ontological events, but rather starts from empirical practicalities: disasters, it turns out, are particularly good empirical sites for understanding politics. This approach does not necessarily depart from established sociological research. It also does not necessarily theorize disasters, but takes them for granted as events. What this strand does understand, and historically, has been early to grasp, is that disasters are prime events for understanding politics, simply because, empirically, all kinds of sociologically interesting things happen, in the same way as ethnomethodology understood that in the collapse of interaction sociologically interesting things happen. In the words of Hoffmann and Oliver-Smith: disasters "unmask the nature of society's social structure" (Oliver-Smith 2002, p.9), echoed by Petersen in this volume when she writes that disasters "are messy times when norms... fail" and "make it possible to analytically denaturalize and examine these practices that create norms" (page 5 of manuscript). This insight is paralleled by the idea in technology studies to study accidents and breakdown to understand technology in use (Wynne 1988). According to this view, disasters change not society, but the work of the sociologist: they decompose what is usually difficult to analyze. Disasters are primarily a welcome methodological tool. Considering disasters as material events is different from the first perspective, as it takes its guidance from Science and Technology Studies (STS) and Actor Network Theory (ANT) to analyse disasters as events in themselves. This approach focuses on disasters as events, but unlike the first approach, does not leave disasters intact, instead disassembling them into their constituent parts. A disaster, as a rupturing event, then, does not rupture a social system, but is produced as an event (also see the conclusion by Michael). Historically, this research strand follows the shift from religious to scientific explanations of disasters. If disasters are not punishments by the gods, but natural events, then the naturalness of the events can be analysed. Measuring the location, strength, likelihood and damage of possible or previous floods, earthquakes or nuclear strikes is a scientific problem, for which different solutions This approach takes a different route to the first with regard to the way in which the empirical focus is distributed. The first attempt accepts disasters as big, single events that pose a problem for society and politics. The focus of analysis becomes the political response, its actors and organisations and the decisions they take to answer to the event. The latter breaks the disaster apart: the question becomes rather when and how 6 a disaster as a unitary event comes into being. The standard STS answer is that disasters are a result of techno-scientific processes, rather than natural or political. The analysis may extend into politics. The focus of this approach is to disassemble a disaster and turn it into a problem: there are not hurricanes as disasters in themselves, but techno-scientific processes that produce hurricanes, floods, or nuclear accidents -both as events themselves and as accounts of these events. The naturalness of (natural) disasters gets bracketed, not just because political processes are guilty of producing disasters in the first place, but also because science and technology in themselves are not taken as simply recording the occurrence of disasters, but producing them. This is not to blame science and technology in an anti-modernist impulse, as the environmental movement is prone to do, but to accept both that the very materiality of social 6 See below the section on "Politics as producing disasters as risks", for a discussion of temporality, and particularly how disasters have effects before they happen, qua risk analysis and preparedness. disruption is co-produced by science and technology and that there is no way to account for such material disruption without the help of science and technology. For such an analysis there are thus two levels. First, to study how modern science and technology produces disasters as material events. This may include a study of the building technologies of dams, cars, space shuttles, or of buildings, or, on a broader scale, of ecological change and how such technological advances in themselves create disasters. Second, it is about studying how scientists account for risks and for existing disasters. What are the scientific assumptions, theories and descriptions for distributing blame between levees, engineers and political decision makers in practice? Disaster as cosmopolitics is a combination of the two former approaches. Disaster as cosmopolitics uses disasters to understand how the world is reorganized on multiple levels through and after disasters. From the viewpoint of cosmopolitics a disaster recomposes the world on every level. Cosmopolitics asks what the world is composed of, who is recognized as a legitimate actor (Farias) and what capacities these actors have (Tironi). The "who" implies not only distributions of decisions among scientific experts and policymakers, but also among different entities, such as tsunamis, birds and measurements, or: actants in ANT terminology. Cosmopolitics then ideally does not disassemble politics or disasters but observes the assembly of worlds, with politics and disasters both constituting events within this world. Disasters are particularly amenable to cosmopolitics, as they unravel the foundations and processes of composing the world, and may therefore offer a place "in which the cry of fright or the murmur of the idiot" can be heard (Stengers 2005). One may remember here the movie "Train de Vie" by Radu Mihaileanu, in which a village of Romanian Jews, hearing of the oncoming holocaust, entrusts itself to the idea of the village idiot: they charter a train, half of the village dresses as Nazis and "deports" the other half and they all drive west, through enemy lines into freedom, duping the real Nazis into believing they are fellow Nazis. The looming disaster, and the hopelessness of the situation literally puts the world at stake, making the proposition of the village idiot one worth taking into consideration. The analytical difficulty of a cosmopolitical approach then is how to juggle the assembly and disassembly of worlds. While the two former approaches can each drill in a single direction and probe the composition of either disasters or political processes, the problem of cosmopolitical approaches is that their field of inquiry and analytical focus is potentially unlimited: who or what should be included in the analysis, and to what extent? This becomes a particular problem if cosmopolitics is conceived as a "positive" form of analysis, that looks into the composition of worlds, and not just the decomposition, into an attempt to ask for new forms of disaster planning (Latour 2007 and see Farias, this volume). --- Politics as Producing Disasters The first, and probably most prominent, version of politics producing disasters is the notion of the state of exception or state of emergency. The state of exception is primarily a legal and technical concept that was given originally political prominence by Carl Schmitt (2005), and more recently through the writings of Giorgio Agamben (2005). While none of the articles in this volume are concerned with state of emergency directly, it is a crucial concept to think about the relationship between disasters and politics, because it highlights the political decisions behind what counts as a disaster from the viewpoint of a state. For Schmitt, to proclaim a state of exception is a sign of sovereignty. But the argument can be turned around. Through the lens of the state of exception -but also "smaller" decisions of the state to act in cases of disasters -it becomes apparent that a disaster, for politics, is not an event out there, but a decision to be taken with repercussions for state action and the state's relation to the population. A disaster in this version is an event outside the state, considered as an organisation, but (usually) within its territory, 7 and which the state uses to do something it could not do without a disaster: change laws, make experts produce reports, send recovery organisations (see Easthope and Mort), and control the movements of people and things (Ellis). The Neo-Foucauldian outlook of Agamben and his followers have pointed out in a critical model the ways that states use the state of emergency to impose new forms of politics on the population. The argument implies that the state of emergency is usually a trick to enforce stricter control of the population, a new biopolitics that would not be possible in normal circumstances. But different from Agamben's focus, what matters here more is not the problem of these extensions of state control, but the way how disasters are used to produce these extensions. In a state of emergency, what matters is that disasters become a resource to justify whatever the state of emergency proclaims. This is a different take from the Foucauldian route Agamben and others take. It focuses not so much on the fact that a state of emergency creates an exception to the law within the law, but rather how the state relates to disasters through a state of emergency. For politics the problem then is to show and prove that its solutions follow from, and are connected to, this external event. While the disaster itself is of no great interest in this perspective, the legal and political processes to connect to the events are. How is it that democratic governments use democratic powers to strip the population or individuals of the very rights that democracy grants them? What is the (discursive) power of a terrorist attack or an earthquake fulfils in these strategies? Where the state of exception takes disasters as starting point, the second version of politics as producing disasters shifts to focus on the risk of the disaster, prior to its occurrence. It is one of the most notable features of modern states that they develop all kinds of theories and 7 However, this does not need to be so: After Chernobyl and Fukushima, various pandemics, and also after 9/11 many states embarked on various paths of disaster preparedness, even if their territory was not directly affected. practices to imagine, calculate and protect against future disasters. The terms risk and preparedness have come to embody these practices (Anderson 2011;Lakoff & Collier 2010). To focus on risk and preparedness implies to shift the focus away from actual disasters: risk and preparedness are both concepts that try to deal with disasters before they happen. It is constitutive of both of these terms that they deal with the unknown. They express the fact that various actors, as in the case of a state of exception, use disasters to act in the here and now, assuming that these actions will change the course of eventual disasters. Studies that analyse risk and preparedness, then, are not so much studies of disasters as material events, but studies of how disasters produce effects before they even happen. In the case of risks, such effects are mostly of the calculative sort, in the case of preparedness it ranges from exercises, to food larders (Collet 2010) and bunkers (see Deville et al.). Studies of risk and preparedness are very much studies of dealing with the time of disasters. Like the precautionary principle (Dupuy & Grinbaum 2005) or prevention (Ewald 1986), risk and preparedness move the time of action before the event, and since they are aimed at undoing or at least alleviating the event itself, and nobody knows when an event will happen, they decouple it from the event. The centre of analysis becomes the imagination of the disaster and the consequences various actors derive from these imaginations. Politics becomes a mediator between the imagination and the practices derived from it. As in the cosmopolitical approach, such an analysis is very much an analysis of how we want to live, how we want to compose the world and what we imagine the world is and should be composed of. But rather than looking at how actual disasters recompose the world, this is about how the expectations of disasters recompose the world. Politically speaking, the problem of risk and preparedness is more difficult and worrisome than post-disaster intervention since the historical basis for action is unknown. While disaster as cosmopolitics builds on some known events and seeks to rebuild a community out of it, preparedness needs to imagine these very events (Weszkalnys). 8 The final approach analyses politics itself as (producing) a disaster. While the former two approaches take politics as answering actual or eventual disasters, politics as itself a disaster assumes that the disaster is an effect of political decisions. 9 As with Foucauldian studies of the state of emergency, such studies assume that disasters are not events beyond political control, but instead are produced by politics in the first place. Underpinning these studies is very much a distrust of the notion of "natural" or "technical" disaster. Rather than taking the disasters as the starting point of analysis, they take the political as the starting point in the production of disasters. They operate from a critical perspective, which seeks to re-orient the blame question and de-neutralize it through sociological analysis. By putting political processes centre stage, and very often concentrating on a particular range of political actors (such as elected politicians and government office holders), disasters become thoroughly politicized. These studies become primarily about actions of political actors, and very often, from a critical angle, about their failure to act before, during and after disasters. Such research is deliberately asymmetrical, since it attempts to point to the fact that it was political decisions that caused a disaster and its subsequent effects. This research is the mirror image of the account given of "politics as producing disasters as risk" since it is about the failure to have created, and prepared for the right risks. "Politics as producing disasters" research points either to the failure of producing the right risks, or, having prepared for the right risks, to have failed to act adequately to address these risks. For example, a large part of the discussion on Hurricane Katrina has focused on the question of whether national and local government assessed the risk of a hurricane adequately, whether the material forms of preparedness, the 8 For a new method how to imagine these events see (Guggenheim et al. forthcoming). 9 This view has been popularized by Naomi Klein in her bestseller "the shock doctrine" (Klein 2007). levees, were adequately designed, and whether FEMA and other government agencies acted adequately after the hurricane. Politics as disaster research also points to the fact that the state and its disaster organisations use disasters for a militarization of response and for taking responsibility out of citizens' hands (Tierney & Bevc 2007). Historically, it follows a trajectory of arguing for a de-militarization of disasters and for preferring the improvisational skills of the population over the organized, bureaucratized and militarized work of state organisations (see Mort and Easthope). --- Overview The special issue is organized in three sections. Each section focuses on a different problem of the relationship between disasters and politics. Its starts with a section "Materials: Ontologies" that circles around the issue how to conceive of disasters once we take seriously the non-human and material nature of disasters. First, Nigel Clark sets out to clarify how the notion of the anthropocene challenges our understanding of politics and disasters. Clark radically seeks to shift agency not simply to the non-human as a general category, but to the earth's climate and its long-term human induced changes. For Clark, the main question is how we can politically understand temporal and spatial processes that far exceed normal sociological concepts of human agency and nation state. In the next article, questions of the redistribution of agency are taken from the biggest scale of the anthropocene to the smallest scale of birds. Israel Rodriguez G<unk>ralt, Francisco Tirado and Manuel Tironi look at a toxic spill in the Do<unk>ana National Park in Spain to question the notion of disaster: Rather than being circumscribed in time and space, they show that disasters should be understood as "meshworks". As the authors show, it is the ringing of the birds, and the possibility to trace how they spread toxic spill, that turns the disaster into a meshwork, with birds as moveable and traceable distributors of what was initially a local event. Drawing on the work of Tim Ingold, they argue that disasters are dynamic realities, difficult to localize and always distributed along disparate scales and actors. In their view, disasters have a dual reality, unfolding actually and virtually at once. In the third article, Ignacio Farias uses the example of the failing of the Chilean warning systems to argue that we should analyse disasters not as instances of accidents but as instances of inquiry. For Farias, what is at stake in the case of cosmopolitics is a generalized problem of recognition, as it has been phrased in critical theory: Who recognizes whom as legitimate and irreducible actor? For Farias, the failure of the warning system must be discussed as a failure to recognize tsunamis as actors and based on this first failure, a failure to openly explore the world. In the second part of the book entitled "experiments: governance" we focus on collective political experiments: these articles deal with different ways of deploying heterogeneous technologies for managing disasters, technologies that are in each case contested, challenged and mutable. Hence the section title: disasters enhance uncertainty, yet the technologies aim to stabilise them, but become matters of conflict and experiment. Katrina Petersen in her article "Producing space, tracing authority: Mapping the 2007 San Diego wildfires" takes a look at how technologies of representation produce different kinds of disasters. In the case under discussion, two different kinds of maps produced completely different disasters with different repercussions for how to deal with the fires. The county produced maps, which ended

What is the relationship between politics and disasters and how does this relate to the recent boom in disaster studies? The introduction to this volume argues that the recent interest in disasters is not because there are more disasters, but because of two recent developments within the social sciences: first, a focus on rupture rather than on continuity and second, a focus on materiality. Disasters are the intersection of these changes. Disasters are ruptures of society and thus inherently political. They provide a particular kind of rupture, one which does not simply affect values and norms, but the material backup of society and its material infrastructure. From this starting point, the article discusses two movements of how to relate disasters and politics: disasters as producing politics and politics as producing disasters. The former begins with disasters and considers how they acquire the power to recompose the world. Disasters from this point of view not only produce politics, but a particular kind of (cosmo-)politics that deals with how humans relate to technology and nature. The latter begins with politics and considers how politics produces disasters. Here, as for example in preparedness, risk assessment and state of exception, politics is the productive force and disasters become means to legitimate, produce and arrive at certain politics.

what is at stake in the case of cosmopolitics is a generalized problem of recognition, as it has been phrased in critical theory: Who recognizes whom as legitimate and irreducible actor? For Farias, the failure of the warning system must be discussed as a failure to recognize tsunamis as actors and based on this first failure, a failure to openly explore the world. In the second part of the book entitled "experiments: governance" we focus on collective political experiments: these articles deal with different ways of deploying heterogeneous technologies for managing disasters, technologies that are in each case contested, challenged and mutable. Hence the section title: disasters enhance uncertainty, yet the technologies aim to stabilise them, but become matters of conflict and experiment. Katrina Petersen in her article "Producing space, tracing authority: Mapping the 2007 San Diego wildfires" takes a look at how technologies of representation produce different kinds of disasters. In the case under discussion, two different kinds of maps produced completely different disasters with different repercussions for how to deal with the fires. The county produced maps, which ended at the borders of the county, whereas those produced by a group of local media and local academic organisations with Google My Map tracked the actual fires. The former aligned with the responsibilities of the state organisations responsible for dealing with the fire, whereas the latter followed those of the population affected by the fires. Petersen's main point here is to highlight that the groups that produce the maps are actually overlapping and have similar interests, but it is the different forms of map production that result in different forms of disasters and how to deal with them. In the next article Manuel Tironi asks to what extent disasters, as radical moments of vital indeterminacy, create extended and iterative forms of political experimentalism. Tironi grounds his argument in the case of Talca, Chile. After the 2010 earthquake, two competing participatory experiments coalesced in Talca. On one side stood the idea that the disaster only unveiled the previous deterioration of the city. This endemic problem should be overcome with a masterplan and the role of the Talquinos would be to evaluate proposals. Against this stand the ideas of Talca con Tod@s, whereby the problem of Talca was not its decay, but its rich history of citizen groups, and its social capital: citizens in this view were engaged, and they were experts. These The third and last section entitled "preparedness: anticipation" focuses on the fact that disasters exist and create political (re)arrangements without even happening. Disasters are politically powerful and generative even when absent, looming, or simply being invoked. The section starts with an articleby Ryan Ellis on "Creating a Secure Network: The 2001 Anthrax Attacks and the Transformation of Postal Security." Ellis looks at how the anthrax attacks led to a reconfiguration of the postal network, which eventually favoured large commercial mailers. For Ellis, the anthrax attacks were just spikes in a story which experts had seen coming. The attacks were not so much a disaster, as the events which led to a political reorganisation of infrastructure: re-enforcing problematic power asymmetries by distributing costs unequally across different categories of users, and engaging in the policing of labor in an effort to aid the shift toward temporary labor. Next, Joe Deville, Michael Guggenheim and Zuzana Hrdlickova look at shelters as materialized forms of preparedness. Drawing on the notion of concrete governmentality, they highlight how shelters, as particularly stubborn and stable forms of preparedness, produce a number of surprising effects. First, they compose preparedness by changing the relationship between citizens and the state. They also decompose preparedness by falling out of use. Lastly, they recompose preparedness because they remain, while disaster experts need to find new disasters to justify their existence. In Deville et al's. view, it is not so much the disaster itself, but particular kinds of preparedness that create different relationships between citizens and the state. In the last article entitled "Anticipating oil: The temporal politics of disasteryet-to-come" Gisa Weszkalnys looks at an extreme case of preparedness and what disasters could be. She looks at how the prospect of oil in S<unk>o Tomé and Pr<unk>ncipe is dealt with as a disaster yet to come. Her interpretation closes a circle by linking back to the problem of anthropocene in Clark's opening article. Weszkalnys asks when a disaster starts and she presents a sliding scale, from the timescales of geology to theories of resource curse and ethnographic observation of how the people in S<unk>o Tomé and Pr<unk>ncipe react to the expectation of oil and their attendant ontological conversions, from crude oil to commodity into money. Finally, in the afterword, Mike Michael complements this introduction by looking at the relationship of disasters and politics through the lens of temporality. For Michael, what the diverse articles do is to tie disasters into knots: they turn past experiences into preparations for the future and they turn these preparations into actual events. These articles draw in various ways on the notion of disasters as politics and politics as disasters. As the section titles make clear, some focus on temporal aspects, some on material and others on those of governance. Where they all converge is to turn the relationship between disasters and politics into a problem. The problem we face is not how to react to existing disasters. The problem we face is how to live in this world knowing that we produce innumerable disasters, which ones we want to prepare for, and how we want to live together in the wake of acknowledging these disasters. These articles are but a first step to answering these questions.

What is the relationship between politics and disasters and how does this relate to the recent boom in disaster studies? The introduction to this volume argues that the recent interest in disasters is not because there are more disasters, but because of two recent developments within the social sciences: first, a focus on rupture rather than on continuity and second, a focus on materiality. Disasters are the intersection of these changes. Disasters are ruptures of society and thus inherently political. They provide a particular kind of rupture, one which does not simply affect values and norms, but the material backup of society and its material infrastructure. From this starting point, the article discusses two movements of how to relate disasters and politics: disasters as producing politics and politics as producing disasters. The former begins with disasters and considers how they acquire the power to recompose the world. Disasters from this point of view not only produce politics, but a particular kind of (cosmo-)politics that deals with how humans relate to technology and nature. The latter begins with politics and considers how politics produces disasters. Here, as for example in preparedness, risk assessment and state of exception, politics is the productive force and disasters become means to legitimate, produce and arrive at certain politics.

Although unplanned settlements grew in density and number after Zambia's independence from British colonial rule in 1964, such settlements were a pre-independence phenomenon (Hansen, 1982;Makumba, 2019). Their origins can be traced to colonial urban settlement policies, which forced poor urban settlers to reside in marginal areas of the city that had not been planned as residential areas by city authorities (Nchito, 2007). New residents simply moved into the settlement areas and claimed parcels of land by constructing low-cost housing and other small structures such as shops and bars. The housing structures were relatively cheaper to rent and thus became a pull factor to many rural immigrants into the city looking for low-cost housing. With time, the population of these unplanned settlements expanded and by the early 2,000s accounted for around seventy per cent of the entire population of Lusaka city and comprised 20% of the city's residential land (UN-Habitat, 2007). Having such a large proportion of the city's inhabitants in unplanned areas has brought about a myriad of socio-economic and environmental challenges. The unplanned settlements occupy and encroach on contested spaces in the city and thus continue to be a source of health, environmental, social and moral problems (Taylor et al., 2015;Msimang, 2017;Muanda et al., 2020). Furthermore, the houses were built without any building authorization by the city municipality as required legally. Zambia's population stands at 19.6 million people with a population density of 26.1 persons per square kilometer (Zambia Statistics Agency, 2022). Although the country's area covers 752,614 square kilometers, only twenty percent (150,522 km 2 ) of the land, translating to about 200,000 land parcels, was fully registered and 600,000 parcels were at various stages of processing as of 2014. The rest (about 80%) was unregistered (Tembo et al., 2018). This entails the vast majority of plots and houses are unregistered, illegal, and potentially resulting in diminished property rights and tenure insecurity. Tenure security refers to the degree of confidence held by people that they will not be arbitrarily deprived of their land rights or of the benefits derived from their land (Knight, 2010;Government of the Republic of Zambia, 2021a,b). Several scholars have observed that informality induces land tenure insecurity and has negative consequences on investments and land resources management (Chirwa, 2008;Ghebrua and Lambrecht, 2017). De Soto (2000) argued that the poor in developing countries fail to turn their land into capital because they lack formal mechanisms for protecting their property rights. Their tenure insecurity acts as a demotivating factor. ---.. Formalization of land rights Formalization of property rights involves the provision of legal representation of property in the form of title deeds, licenses, permits and contracts, all of which must receive official sanction and protection from legitimate national authorities (Benjaminsen et al., 2009). Land rights formalization is argued to be especially beneficial for residents of informal settlements, as the illegal status of such settlements presents a level of precariousness not experienced by other residents, be they in customary areas or formal settlements without documentation for their land. Residents of informal settlements can be evicted legally anytime. Some scholars posit that by formalizing their land rights, residents of informal settlements enhance their land rights claims and tenure security (Sjaastad and Bromley, 1997;Zevenbergen, 2000;Holden and Otsuka, 2014;Wily, 2017). Tenure security is associated with access to credit, stimulating entrepreneurship, provision of services and infrastructure, improved health conditions and the realization of the human right to adequate housing (Jimenez, 1984;de Soto, 1989;De Soto, 2000;Payne, 1997;Smith, 2001;World Bank, 2003;Di Tella et al., 2007;Field, 2007;Durand-Lasserve and Selod, 2009;Reerink and van Gelder, 2010). Furthermore, land rights formalization is theorized to facilitate land market transactions which indirectly leads to higher overall investment in land (Besley, 1995;Place, 2009;Holden et al., 2011;Lemanski, 2011;Ghebru and Holden, 2015). However, the empirical evidence from several countries in sub-Saharan Africa that have implemented a wide range of land formalization initiatives since 2000 reveals mixed results. Despite the numerous espoused benefits arising from titling programs, a variety of challenges impede implementation on a large scale. For instance, developing countries within and beyond sub-Saharan Africa lack the necessary finances and capacity to execute the land formalization process because of the high costs required for land survey, title registration and issuance of title on a large scale (Sjaastad and Bromley, 1997;Toulmin, 2009;Bezu and Holden, 2014;Kim et al., 2019). Hence aid agencies and international financial institutions such as the World Bank, USAID, DFID, the European Union among others have financed formalization processes in some countries (Durand-Lasserve and Selod, 2007). Furthermore, land formalization manifests a gender dimension with women at a greater risk of land tenure insecurity despite de jure equality with respect to land access (Ali et al., 2019). The prevalence of unfavorable customary practices and attitudes that restrict women's control over land resources inhibit fair participation in land formalization programs. For example, in Ethiopia, despite the introduction of progressive land reforms that promote rural women's rights to inherit and own land, discriminatory practices such as restrictions on land ownership continue to persist (Tura, 2014). Similarly, in Malawi's matrilineal societies, though women possess inheritance rights to land, men, be it husbands or uncles, exercise leadership in decision making and control over land, thereby disenfranchising women (Andersson Djurfeldt, 2020). Consequently, as men secure their property rights, those of women risk being undermined (Andersson Djurfeldt, 2020). Further, the landless, women, and orphans may not only be unable to take advantage of formal rights to assets but may find avenues of access effectively closed through the price increases that invariably attend formalization (Benjaminsen et al., 2009). Essentially, a danger inherent in formalization is consolidating existing inequalities (Benjaminsen, 2002;Sjaastad and Cousins, 2008). Urban population growth presents challenges to large scale titling programs by triggering illegalities by wealthier individuals who include local residents, elites, politicians and local leaders in attempts to acquire property. This, however, is at the expense of the relatively poor and marginalized individuals. This high demand for urban property often leads to land conflicts and in some instances land grabbing which often overwhelm local government authorities (Chitonge and Mfune, 2015;Lombard and Rakodi, 2016). Other challenges of large-scale titling include increased perceptions of displacements and tenure insecurity among marginalized individuals, lengthy processes, lack of policies to support titling objectives, failure to achieve intended titling goals and high interests in land by commercial entities (Payne et al., 2009;Lawry et al., 2014;Chitonge and Mfune, 2015;Lombard and Rakodi, 2016;Andersson Djurfeldt, 2020). Some scholars have noted that land rights formalization programs result in heightened insecurity due to displacements, especially among the marginalized groups in society, i.e., the poor and widows (Kyalo and Chiuri, 2010;Andersson Djurfeldt, 2020). For example, in Kenya, the process of converting communal pasture into private plots reinforced male dominance in property ownership as titles were issued to the men, hence increasing inequalities. Further, women lost their access to previously available resources such as livestock and other food products which they previously obtained through communal bargaining processes (Kyalo and Chiuri, 2010;Andersson Djurfeldt and Sircar, 2018). In Zambia, the state embarked on an ambitious National Land Titling Program in 2015. The objectives of the program were to regularize ownership of untitled properties in towns and cities and by so doing, promote security of tenure for property owners on state land, reduce displacements, promote internal security and increase the revenue base and investment in the country and thus contribute to socio-economic development (Government of the Republic of Zambia, 2018; Tembo et al., 2018). During the issuance of the first 92 certificates of title to residents of Madido area in December, 2017, the then minister of the Ministry of Lands and Natural Resources explained that the initiative was motivated by the government's resolve to accelerate social and economic development. This was to be accomplished by regularizing ownership of untitled properties in towns and cities, eradicating inequalities in gaining access to land in order to cater for all, and providing citizens with the impetus for access to credit (Government of the Republic of Zambia, 2018;Ministry of Lands Natural Resources., 2021). The Zambia National Land Titling Program had a target of processing and issuing 300,000 certificates of title to landowners in areas where the program was being piloted by 2018. The Ministry of Land and Natural Resources established a unit specifically to deal with land titling under the National Land Titling Program. This is the National Land Titling Center. Later, it engaged a private company called Medici Land Governance to expand the program. However, reality fell short of aspirations, as program implementers faced challenges characteristic of such initiatives. The challenges included determining land ownership for claimants with incomplete or no documentation, determining property boundaries and upgrading paths to access roads to meet statutory standards. Since private infrastructure had been developed with little or no considerations for building standards, concessions had to be made between demolishing properties that were too close to public roads and those without sufficient space for access roads and providing title security (and its attendant benefits) to poor residents with no other opportunities to own land with secure and enforceable land rights. Arguably, this legalization of illegalities was instrumental to achieving the public good of social-economic development through the provision of legal property rights and all the opportunities this presented. Thus, this study sought to determine how land rights formalization had affected land tenure security among men and women landowners in the pilot areas. Furthermore, it explored how social norms on land inheritance, decision making over land and marital trust had been affected by the changed nature of land rights. Lastly, land conflicts, an indicator of threats to land claims was examined to determine the extent to which the land rights formalization had addressed this threat. Land rights formalization in the two study sites started as part of the national land titling program, an initiative by the state to document all unregistered land in urban areas. Once the initial skepticism and fear of demolition of illegal structures and forcible evictions was over, landowners in the two study sites agreed to participate in the program, and mobilized the initial mandatory payments. The initial payment was ZMW 1,260 (USD73) for Madido and ZMW 625 (USD36) for Bauleni. This covered the survey fees and enabled the program implementers to survey the sites and draw sitemaps. The rest of the payments were to be made in monthly installments of ZMW 100 (USD6) for up to 36 months, for a total of ZMW 4,990 (USD286) in Madido and ZMW 3,747.93 (USD215) in Bauleni. This is about half what landowners spent to acquire title deeds through the normal process. Residents that managed to secure payments early were among the first to receive the title deeds. Better resourced households thus benefitted early in the process. ---.. Conceptual framework for the study The conceptual framework is adapted from Doss and Meinzen-Dick (2018). The adaptation includes the addition of men to the analysis and its application to an urban setting. The framework incorporates four broad areas: the context, threats and opportunities, action arena, and outcomes (Figure 1). The context includes both formal and informal institutions (laws, practices, and norms), socio-economy and history. The threats and opportunities to land rights include the catalysts of change, both those that strengthen and those that weaken tenure security for men and women. The action arena includes both actors and action resources. The actors include every-one who influences land tenure security. The action resources are those resources that different actors can use to seek their preferred outcomes, and include money, education, networks and social status. Finally, men's and women's land tenure security is the outcome of interest, and feeds back to shape the context for men and women's land rights in the future. Land tenure security has three components, (i) Completeness of the bundle of rights (ii) Duration, and (iii) Robustness. Completeness of the bundle of rights looks at the extent to which one person or persons hold the various rights. Duration is about whether the rights are held for a short or long-term and /or if the length of time is known. Robustness is an examination of whether the rights are known by the holders, accepted by the community, and are enforceable. --- FIGURE Conceptual framework of factors a ecting women and men's land tenure security [adapted from (Doss and Meinzen-Dick, )]. The rest of the paper is organized as follows. In the next section, the paper describes the study sites and the study methodology. Section 3 presents and discusses the empirical results. The last section concludes the paper and draws out policy implications.. Materials and methods ---.. Study context and background Fieldwork was conducted from two study sites, Bauleni and Madido residential areas (Figure 2) between August and October 2020. The two areas were selected as study sites as they were pilot areas for the National Land Titling Program in the Lusaka city region (www.mlnr.gov.zm) and thus were ideal for an examination of program outcomes. The two study sites are located in Lusaka and Chongwe districts respectively. Of the total area covered by the country (752,614 square kilometers), Bauleni covers an area of about 1.533 square kilometers and Madido about 0.891 square kilometers. The areas experience a humid subtropical climate, and are overlain on uneven depth of folded and faulted schist. Bauleni had a population of 64,000 (Tidwell et al., 2019) while Madido's was at 210, 672 in 2010. Bauleni and Madido have 3,697 and 2,109 households respectively. Bauleni consists of low-income and middle-income households. Most of its residents are in informal employment. The situation is different for Madido with most of its residents in formal employment, and a large majority of them are middle income households. The genesis of the two areas is very different (Sommerville and Tembo, 2019). Bauleni started in the 1970s as a small unauthorized squatter settlement formed by laborers who worked in nearby commercial farms (Cheyeka et al., 2014). With time, its population grew. It was legalized in 1998. Implementation of the National Land Titling Program started in Bauleni in 2018. Madido-also known as Chelstone Extension to the residentsis located on land that previously belonged to a public agricultural college, the Natural Resource Development College. Around 2006, local party officials of the then ruling political party, the Movement for Multi-Party Democracy encroached on the land, illegally demarcated it into land parcels and sold them to interested persons. The population quickly grew and Madido became an informal settlement. The Ministry of Lands subsequently canceled the college's land ownership rights before offering them to the illegal settlers and has since regularized their ownership through the issuance of title deeds (Personal communication, Key informant, October 2020). Implementation of the National Land Titling Program started in Madido 2017 (Sagashya and Tembo, 2022). ---.. Data collection An explanatory sequential mixed methods approach was employed (Bryman, 2012). In order to come up with a sampling frame of households that had obtained title deeds, a mapping exercise was conducted. Every household in the two study sites was visited and a short survey conducted to establish whether it had participated in the land titling program, whether title deeds had been obtained, and if so when. The mapping exercise was conducted with the help of community members that were widely known and well regarded. For Bauleni, it was with the participation of a long-term female resident whose family was the first to settle in the area. The resident was identified with assistance from the local authority officers based at the Lusaka City Council office in the area. The resident had first-hand knowledge of almost all the housing properties and their owners. She was well known in the area as an executive member of several local development initiatives. She was deputized by two females familiar with the area through their parttime work with Lusaka City Council as distributors of water bills in the area. In Madido, the field team was assisted in the mapping exercise by two male residents who had been the central figures as they had been actively involved in the illegal demarcation and subsequent sale of land parcels. Further, When the National Land Titling Program commenced, the two were recruited as resource persons to explain the program to community members and to encourage the community members to participate in the program. The two were thus trusted by the residents, who were otherwise wary of strangers asking about land matters, given the illegal genesis of the residential area. After the mapping exercise, all the households that had title deeds at the time of the survey were interviewed. These were 54 from Bauleni and 248 from Madido. The survey commenced with a pilot study to test the data collection instrument. During piloting, each of the four research assistants interviewed two respondents each, a male and a female from among the households with title deeds. Thus, a total of eight interviews were conducted during the pilot study. The pilot study was conducted within a day, at the end of which a debriefing session was held. Some questions in the questionnaire were modified, while a few were removed as they were found to be redundant, and some new ones were added after the feedback from the pilot interviews. The eight interviewees from the pilot study were excluded from the survey. After the questionnaire survey, two 3-in-1 focus group discussions (FGDs) (March et al., 1999;Umar, 2021) were conducted in the two study sites. The discussants were recruited from the pool of respondents, on the basis of having extensive knowledge of the land titling process. Each of the four enumerators recommended two male, and two female discussants from the batch of respondents each had interviewed from the two study sites. The discussions were held in two phases. In the first phase, discussions were conducted with men and women separately. The men's FGDs were facilitated by a male researcher while the women's FGDs were facilitated by a female researcher. This initial separation into single gender groups was to minimize any influence of unequal gender relations such as the social and cultural superiority of men to women that could otherwise limit participation in the discussions based on gender norms. The men and women's groups were later brought together in a plenary discussion and asked to present summaries of their group deliberations. The plenary phase of the FGDs resulted in co-production of knowledge by men and women discussants through the detailed discussions that ensued from the single gender group presentations. One member of the research team expertly facilitated the discussion while another observed the proceedings and took notes. The facilitator alternated which group shared its results first. Both facilitator and observer paid attention to the verbal reactions and non-verbal communication of the women to the men's answers, and vice versa. These included voice tone, facial expressions and demeanor. These cues were used to guide the facilitator on whether or not there were disagreements between the two groups, and to probe appropriately (Tecau and Tescasiu, 2015). Four (4) key informants were interviewed. These included a representative from the private entity recruited by the state to implement the National Titling Program, and a spatial planner from the Ministry of Local Government and Housing, with extensive experience in planning who had been seconded to the National Land Titling Program. Others were two key informants with extensive knowledge on land related matters in the two study sites. They were both long term residents of their respective informal settlements and their families had been important players in the establishing of the illegal settlements. Both were also involved in the National Land Titling Program as community representatives. This role entailed communicating community concerns to the government, providing information on undocumented land parcels to the state, and updating community members on program activities. Free, prior and informed consent (Hanna and Vanclay, 2013) was verbally obtained from all respondents and key informants, all of whom were adults aged over 18 years of age. Permission to record the focus group discussions using digital recorders was sought from the discussants and granted for all the sessions. Approval to conduct the research was granted by ERES Converge IRB, a nationally accredited research ethics clearance organization. ---.. Data analysis The questionnaire survey data was entered into Microsoft Excel sheets. After the data entry was completed, an accuracy check was conducted by randomly selecting ten percent of the completed questionnaires and comparing them to the data entered about them. The data was then copied to Minitab 18 (Minitab Inc, 2017) and analyzed using basic descriptive statistics such as frequencies, means, standard deviation and two sample Independent T-test. The T-test was used to test the hypothesis that male household heads were older than their female counterparts. The recordings of the FGDs and key informant interviews were transcribed and categorized into themes based on research questions. The themes were sense of tenure security; sense of ownership; decision-making; land inheritance, land related family conflicts; love and trust and threats and opportunities presented by land rights formalization. ---. Results and discussion Overall, the average age of household heads was 56.4+ (s = 9.8) and 48.5(s = 9.1) in Bauleni and Madido respectively. When disaggregated by the gender of the household head, the data reveals that around 20% of the interviewed households were headed by females; the average age of male household heads in Madido was lower than for the female heads (p <unk> 0.005) while in Bauleni, there was no statistically significant difference (p <unk> 0.05) in the mean ages of the male and female household heads. Almost half (46%) of the respondents reported the housing property owned by a married male household head while a quarter (25.2%) reported joint ownership by husband and wife. Joint ownership by husband and wife entails that both the couple's names are included on the title. The rest reported ownership by female headed households (16.3%) and by wives (11.4%). Female household heads are unmarried (divorced, widowed or single). There were no unmarried male housing property owners in the sample. When disaggregated by study site, the general trend was the same with a few nuances; About 49% and 40% of the housing property owners were male household heads in Madido and Bauleni respectively. Joint ownership was reported in 36% and 23% of the housing properties in Bauleni and Madido respectively. Housing property owned by the wife comprised about 12% of the cases in both study sites, while about 15% and 10% of the housing properties were owned by female household heads. Family owned housing was the least common with only 2% of the properties in Bauleni and 6% in Madido. The rest of this section presents and discusses the findings based on themes derived from the research questions and those emerging from the data collected. ---.. Land tenure security is an outcome of land rights formalization A large majority of the respondents in both study sites thought that women's sense of tenure security had greatly increased (82% for Bauleni and 91% for Madido), while a small percentage (8% for Bauleni and 2% for Madido) thought the increase was only moderate (Figure 3). If feelings of tenure security increase greatly, it means someone feels that it is highly unlikely that someone can arbitrarily expropriate their land. Similarly, if an owner's sense of security of tenure is moderate, it means they feel that it is not so likely that anyone can dispossess them of their property arbitrarily although they may have some lingering feeling that someone might actually do so. The women FGDs in Bauleni revealed that title deeds were "witnesses" or testimony to land ownership and that in the future no one would take the land away from them or their children because the title deeds would serve as proof of ownership. Similar views were expressed by the women FGDs from Madido. In the words of one focus group discussant, "before we got title deeds, we were very worried. Now we are safe, after getting the title deeds. Now we can expand our houses. We are very secure now." These views are shared by women that jointly owned their land with their spouses and those from households where only the male household head is included on the title. Noteworthy is that even married women whose names are not included on the title deeds as joint owners enjoy tenure security because of the provisions of the Intestate Act (Chapter 59 of the Laws of Zambia) of 1989 which protects the rights of surviving spouses and children to inherit property. Enhanced tenure security for women after land rights formalization has been reported elsewhere in sub-Saharan Africa. Some interventions in Uganda (Cherchi et al., 2019), Ethiopia (Bezabih et al., 2016) and Nepal (Mishra and Sam, 2016) showed --- FIGURE Sense of tenure security after land titling by women and men in Bauleni and Madido. a notable increase in women's tenure security than was previously held. Such cases notwithstanding, Vi<unk>a (2020) urges caution. She posits that a woman having title deeds is not a sufficient condition for tenure security as titling may not necessarily translate to decision making about and deriving benefits from the land. Over 80% of the respondents in both sites believed that men's sense of tenure security had greatly increased after land titling (Figure 3). During the FGDs with men in Bauleni, a contrast was made between tenure security before and after land titling. The men unanimously agreed that before titling, anyone could lay claims on their land, but after titling, this was not possible. They all felt very secure post-titling. Their counterparts from Madido observed that before titling, they had been afraid that their houses could be demolished. The state did not provide any public services to the area because it was considered an illegal settlement. But after titling, "the government has brought water, sewerage and roads. We are proud now. There is just that sense of pride that this is my property", one of the discussants narrated. These sentiments suggest that land rights offered by title deeds are known to the rights holders and are enforceable. The rights owners knew that their title deeds, and the rights guaranteed therein, were valid for 99 years. This applies for both men and women land rights holders. A large majority of the respondents reported increases in the sense of ownership for both men and women in both study sites (Figure 4) while a small minority indicated that there was a moderate increase of sense of ownership. The stronger sense of ownership was premised on landowners in both Bauleni and Madido having exclusive rights and control over the land parcels they owned. This was in line with the results from the FGDs. The focus group discussants from both Madido and Bauleni (men and women's groups) observed that the title deeds have accorded them ownership rights, which they never had before the issuance of title deeds. They narrated that prior to land rights formalization; anyone could come at any point and grab the land from them. However, with the title deeds issued to them, all focus group discussants confirmed having full ownership rights. They believed that no entity could grab land from them or demolish their properties without compensation. The focus group discussants further noted that land rights formalization has resulted in their empowerment. That is, they have powers to put up and extend immovable structures, because they have secure land rights to their land parcels. Clearly, landowners are able to enforce their rights when under threat as the rights have been legitimized by the state. A similar study from Tanzania conducted by Parsa et al. (2011) reported that most residents with property licenses felt that the municipality was unlikely to carry out demolitions and if conducted they had a better chance of being compensated by the authority. Having addressed tenure security, the paper proceeds to explore the context under which men and women have gained tenure security and delve into which land rights claimants are able to assert their claims to the secured land rights, how they assert their claims and the conditions under which they do so. ---.. The context In this section, the paper presents results on social norms obtaining around property in the context of titling. Social norms are defined as "rules of action shared by people in a given society or group; they define what is considered normal and acceptable behavior for the members of that group" (Cislaghi and Leise, 2020). Social norms change from time to time. It would be interesting to know how titling affects some of the social norms in the study sites. ---... Social norms on inheritance A large proportion (67% and 73%) of the respondents in Bauleni and Madido respectively, and 74% overall claimed that there was an increase in the inheritance rights to property for daughters due to land rights formalization (Figure 5). A minority of the respondents (<unk>20%) indicated that there was no change in inheritance rights to property for daughters post land rights formalization (Figure 5). In the FGDs debate on land rights inheritance by daughters and sons, some discussants maintained that it was not the best idea to put daughters' names on title deeds because once they got married, they could let their husband take over the property to the detriment of the daughter's siblings. In the words of one male discussant from Bauleni, "For daughters, they can get married and let the man control the property." Conversely, another discussant from the same group argued for land rights inheritance by daughters, "It is better for a girl child to get inheritance because as the boy gets married and dies, his widow will inherit the property." This sentiment was echoed by discussants from Madido. One discussant elaborated the following: Girl child should be on a title deed. Even when the girl child gets married and it happens that the marriage does not work out, the daughter can go back to the house unlike the boy child because when he marries and dies, his wife will inherit the house [Focus Group Discussion, Madido, Zambia 20th October 2020]. Nancekivell et al. (2013) shared this view when they contended that a girl child should be on title because even if she were to get married, she could still look after the property and in cases where the marriage failed to work out, she could go back to the property. Almost 90% of the respondents perceived rights to inherit housing property to have increased due to land rights formalization (Figure 6). During focus group discussions, a lot of skepticism was expressed about sons inheriting property rights to land. The discussants averred that sons could sell the land and chase their siblings. This excerpt typifies this sentiment among Bauleni discussants, "It is best not to put [include] sons on a title because they can sell the land and chase their siblings." A similar view from Madido, "A son may even let the wife control and chase away siblings". The focus group discussants from Madido and Bauleni residential areas expressed strong preferences for including all the children on the title deeds so that no single child could change the land ownership. Half of the respondents from Bauleni and forty percent from Madido claimed that there was no change in the inheritance rights to property for nephews after land rights formalization (Figure 7). About ten percent of the respondents in both areas thought there had been an increase while the rest viewed the rights to have decreased. Similar sentiments were expressed for nieces (Figure 8). The explanations for these results were provided during the focus group discussions. The discussants contended that adding nephews and nieces' names on title deeds was problematic because their (the nephews and nieces) parents could later claim the land parcels as theirs. The following verbatim represent this view from a Bauleni discussant, "Putting names of nephews or nieces may result in problems because their parents may come to make claims" Both respondents and focus group discussants noted that only biological children had inheritance rights and should be the only ones included as land rights claimants on title deeds, besides the parents. This norm was a measure to prevent land claims from extended family members. Either one name or more can be on a title deed. The key beneficiaries are the children, biological children, and no one can claim the land from them [Focus group discussant, Madido, Zambia, 20th October, 2020]. Gibson and Walrath (1947) in Iowa of the United States of America also made this observation when they noted that the inheritance of property rights by nephews such as inheriting the house, farmland or plot following the death of the owner of the property was perceived negatively. Normally, when nephews inherit property rights, it is very likely that their biological parents may claim it is their property when in the actual sense it is not (Gibson and Walrath, 1947). In Rwanda, the land registration and titling program, implemented alongside the 1999 Law of Succession, and the National Land policy of 2004 resulted in; (i) increased inheritance rights of daughters similar to sons, (ii) permanent land rights for divorced or widowed women, and (iii) increased ability to resist restrictive customary practices, e.g. polygamy, where wives property rights were not recognized by the state (Ansoms and Holvoet, 2008;Daley et al., 2010;Santos et al., 2014;Kagaba, 2015). ---... Decision-making authority by men and women household heads Decision-making is an indicator of control. Being able to exercise agency over what happens to land suggests an acceptance as part (owner) with rights and /or interest in the property. Decision making over land is influenced by social norms over who is considered a legitimate decision maker. In both study sites, about 32% of the women respondents noted that the acquisition of land titles had greatly increased the decision-making authority of land owners while 15% noted a moderate increase (Figure 9). For the men, 22% perceived decision making authority to have greatly increased among titled land owners, while 17% thought the increase was moderate, in both study sites. About 5% of the male respondents in both study sites asserted that decision making had moderately decreased while <unk>1% of the women thought so, in both study sites (Figure 9). Close to half (45 %) of all respondents did not attribute any changes in the decision-making authority to acquisition of title deeds. Despite few respondents citing increased decision-making authority, FGDs revealed that the acquisition of title deeds facilitates for men and women household heads to acquire financial loans using titles as collateral and enables them to decide who should inherit their property. Further, discussants noted that title deeds provide men and women household heads legal ownership and consequently authority to invest in their properties thereby increasing the monetary worth of the properties. Land formalization programs implemented across sub-Saharan Africa show positive outcomes

This study investigated how land rights formalization had a ected land tenure security among landowners in two informal settlements of Lusaka and Chongwe districts, Zambia. It explored how social norms on land inheritance, decision making over land, marital trust and land related conflicts had been a ected by the changed nature of land rights. Data was collected through a questionnaire survey of all the households that had obtained title deeds at the time of the survey, two -in-focus group discussions and four key informant interviews. Results suggest that land tenure security is now a reality for residents that hitherto lived under constant threat of eviction. Landowners have benefitted from the formalization initiative through land laws and local norms that allow equitable access to land. Land rights formalization has curtailed land rights for secondary claimants such as extended family members, in preference for man, spouse and biological children. A sense of ownership undisputedly increased for men and women in the two study sites. About % of the respondents in both study sites indicated that formalization of land rights had not resulted in family conflicts. At least one-third from both sites reported an increase in love and trust between spouses after land rights formalization. About half of the respondents reported that no change in decision-making authority had occurred for men while % reported an increase. Formalizing land rights in informal settlements has entailed legalizing illegalities as regulations on plot boundaries are set aside by the state to achieve its aspirations of providing land tenure security to poor urbanites who would not otherwise have recourse to legal or regularized land. We recommend that caution be taken in promoting what is unarguably a pro-poor initiative to ensure that such initiatives should not incentivize future land encroachments.

legitimate decision maker. In both study sites, about 32% of the women respondents noted that the acquisition of land titles had greatly increased the decision-making authority of land owners while 15% noted a moderate increase (Figure 9). For the men, 22% perceived decision making authority to have greatly increased among titled land owners, while 17% thought the increase was moderate, in both study sites. About 5% of the male respondents in both study sites asserted that decision making had moderately decreased while <unk>1% of the women thought so, in both study sites (Figure 9). Close to half (45 %) of all respondents did not attribute any changes in the decision-making authority to acquisition of title deeds. Despite few respondents citing increased decision-making authority, FGDs revealed that the acquisition of title deeds facilitates for men and women household heads to acquire financial loans using titles as collateral and enables them to decide who should inherit their property. Further, discussants noted that title deeds provide men and women household heads legal ownership and consequently authority to invest in their properties thereby increasing the monetary worth of the properties. Land formalization programs implemented across sub-Saharan Africa show positive outcomes in securing property rights and upholding equality across both genders. In Rwanda, equal decision-making rights between formally registered spouses to alienate property and rights to earn independent incomes through private property were reported (Kagaba, 2015). Agarwal and Panda (2007) noted that establishing women's property rights empowers them with decision making authority and enhanced control over resources and ensures the welfare of their households. Titling, however, must be supplemented with ancillary empowering interventions for women (see Monterroso et al., 2019). As Vi<unk>a (2020) avers, focusing on titling alone "without addressing the persistent barriers faced by women, not only misses the mark, but could also end up being counterproductive" Some scholars have argued that the presence of a title does not guarantee access to financial credit to residents, i.e., men and women especially in low-income areas citing low value of most properties as well as unwillingness by financial institutions to offer loans (Rakodi, 2014). However, these negative outcomes are unlikely to apply to residents of our study sites as they are in the city and title is for individualized housing units unlike the case of rural communal pasture or bargaining. ---.. Threats and opportunities presented by land rights formalization Focus group discussants from both study sites articulated that tenure insecurity among informal settlement landowners without formalized land rights is high, with evictions and demolitions pervasive threats. They asserted that the land reforms to regularize land ownership presents an opportunity for informal area residents to secure their land rights and make them enforceable and easily transferable. Scholars have observed that land titling is not without threats. Informal settlements expand unexpectedly and ultimately lead to a change in the use of space and structure of activities, in ways not in conformity with land use planning and legal requirements and may cause contradictions and conflicts (Dadashpoor and Ahani, 2019). ---.. Land conflicts, love and trust among family members Intra-family tensions and contestations are reported over bequest, usage, or sharing of land (see Wong, 1998;Kouamé, 2011;Gyapong, 2021). As Wong notes, disputes among spouses or family members can arise from deteriorating family relations such as a marriage breakdown or from third parties making counterclaims to the property. This study therefore sought to find out, in part, how land tilting had affected internal family relations in terms of conflicts related to land. The conflicts manifested in a number of ways in the two study sites including verbal quarrels and cutting of family ties. Just over half (54%) of the respondents in Bauleni indicated that land rights formalization had not influenced landrelated family conflicts. This is compared to below half (43.5%) of Madido respondents. Interestingly, only <unk>10% (9%) of respondents in Bauleni indicated a decrease in land related conflicts among family members while about a third (31%) of Madido said land related family conflicts had decreased. In both study sites, <unk>10% reported that land related family conflicts had increased following titling; seven and eight per cent for Bauleni and Madido respectively. On whether land related family conflicts had increased moderately only 11% and 10% responded in the affirmative for Bauleni and Madido, respectively. Overall, 18 and 19% of respondents from Bauleni and Madido, respectively, reported a moderate or great increase in land related family conflict following titling of their land. While tenure for agricultural land for women has been associated with women empowerment and reduced gender-based violence in India, the scarcity of land has resulted in tensions between spouses in Kenya over prioritization of consumption crops or commercial crops (Andersson Djurfeldt, 2020). According to Rukema and Khan (2019), family conflicts in Rwanda relating to land are sparked by polygamy with competing inheritance claims from the various wives and their children. However, in Zambia, polygamy is illegal under statutory law and though legal under customary law, polygamy is rare in urban areas. In both Madido and Bauleni, no polygamous marriage was reported by respondents. The other causes of land-related conflicts Rukema and Khan (2019) cite are illiteracy and ignorance of the law empowering women with land ownership. Our study revealed a case where the husband had deserted the wife and children to go and live with another woman carrying the land title with him. One woman in Bauleni, a teacher by profession, reported contributing toward the land title. However, the husband had been elusive, giving contradictory claims that he had received the title and denying this when contacted by the wife during the interview. Love and trust between spouses can potentially be affected by titling with land being more marketable or being used as collateral, for instance. Furthermore, titling could mean adding both spouses on the title deed as co-owners. Over half (52%) of the respondents in Bauleni and only 37% in Madido said the love and trust for their spouse had not changed after obtaining title deeds. In both areas very few reported a decrease in their love and trust toward a spouse, that is, <unk>1%. A third of respondents in Bauleni stated that their love and trust for their spouse had increased modestly or greatly while 41% said so in Madido. Overall, the vast majority of respondents in both areas indicated an improvement or no change (Madido 81%; Bauleni 89%) in their trust and love toward the spouse. Interestingly, seventeen respondents in Madido claimed not to know how or whether titling had affected their spousal relationship compared to none being not sure in Bauleni. ---.. Perverse incentives for illegal land allocations There is a risk that legalizing and formalizing land rights to illegally settled land could provide perverse incentives for new land encroachments. In Bauleni, authorities bypassed regulations about plot size, and distance from public infrastructure such as roads and water pipes in numerous cases (Figure 10) during the surveying and subsequent titling process. Furthermore, the legalization of settlement on land that previously belonged to a public college in the case of Madido could motivate future illegal settlements of public land in the expectation of future regularization. Chitonge and Mfune (2015) cited the illegal allocation of idle and vacant public or private land by political party cadres as important in the creation of informal settlements in Lusaka city. ---.. Visibility for residents Landowners are now "visible" to the state, in that they can use their title deeds as proof of residence, a requirement in accessing numerous services provided by both the public and private sector. For example, in order to open a bank account with a formal banking institution, proof of residence is required. Before the acquisition of title deeds, residents had no way of providing this proof. Land rights formalization has spurred increased participation in local development initiatives, such as through ward development committees. This increased participation is positive for local area development as residents are able to articulate issues of interest to them. ---.. Actors and action arenas Local political players were cardinal in the mobilization of residents. In Madido, the clique of ruling party officials that had appropriated land from the college and sold it were engaged to help the technocrats liaise with the community. Due to the illegal way in which the land had been obtained, residents had lived with the threat of eviction and were very apprehensive about any land related discussions. The land sellers were instrumental in providing confirmation of landowners in the numerous cases where proof of sale was missing. They worked hard to assure the community members that the initiative was genuinely meant to provide title deeds to them and was not an eviction exercise. Local political elites have been influential in illegal land allocations in Zambia (Chitonge and Mfune, 2015). During fieldwork, one of them admitted that, "it is not possible for the state to remove people from illegally occupied land, once the land has been allocated by political cadres. That would not auger well for the ruling party. All the state can do is provide public services such as water, schools, clinics and roads". Residents tended to publicly align themselves with party officials of the ruling party or those they knew to be influential in local community development structures as a way to protect their interests. This has engendered patron-client relationships between residents and local political party officials on one hand, and local party officials and higher level politicians and technocrats. Local party officials have been known to usurp the authority of local development officials and technocrats in matters of land administration. This usurpation has been demonstrated in the collection by political cadres of tax that should be collected by local authorities as revenue for service delivery (Beardsworth et al., --- FIGURE Titled houses situated less than a meter from a public drainage channel and road, Bauleni Lusaka. --- 2022 ). There is a risk that the national land titling program may be overtaken by political elements if this clientelism is not addressed. Our research indicates that the issuance of title deeds is faster since Medici Land Governance became involved than was the case with the pilots under the Ministry of Lands and Natural Resources. Interviews with key informants revealed that Medici Land Governance is more efficient because it makes use of geographical information systems and uses block chain technology and is able to get title deeds issued in batches, unlike the traditional Ministry of Lands and Natural Resources system which provides approval per parcel. This is an opportunity for streamlined issuance of title deeds with potential to handle the land volumes expected once the program is fully fledged. ---. Conclusion This study set out to examine how land rights formalization has affected land tenure security, sense of ownership, decision making, land conflicts, love and trust among men and women landowners in two study areas in Lusaka, Zambia. Our research findings show that the ongoing land rights formalization program in Zambia has provided land tenure security for residents of informal settlements that previously lived under constant threat of eviction from their land. Both men and women have similarly benefitted from the formalization initiative through land laws and local norms that allow equitable access to land and land inheritance. Ownership rights and decision making has also been enhanced among both men and women landowners in the two study sites as they can easily alienate their property. Land rights formalization has in some instances curtailed land rights for secondary claimants such as extended family members, in preference for man, spouse and biological children. This is in line with the majority of the respondents in both study areas who were of the view that only the spouse or biological children's names should be on the title deed and have the right to inherit the property. The process of formalizing land rights in informal settlements has entailed putting aside regulations on plot boundary specifications and plot locations; essentially the legalization of illegalities to achieve the states goals of providing land tenure security to poor urbanites who would not otherwise have recourse to legal or regularized land. The study commends the initiative as a propoor initiative that is enabling socially marginalized groups to access legal land documentation and become visible in urban landscapes that have historically not catered for their land and housing needs. As the National Titling Program is expanded to other districts, implementers should develop robust mechanisms for keeping track of the payments made by program beneficiaries through their community municipality offices such as a short messaging system to send alerts whenever payments are made. The messages should include information on the amount paid and balance remaining. Program implementers should also continue to improve on the time between initial payments and issuance of title deeds. Policy makers are cautioned not to incentivize illegal land allocations by not extending the initiative to areas illegally occupied after the start of the program. --- Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. --- Author contributions Conceptualization, visualization, project administration, funding acquisition, and methodology: BB. Validation: JK, KK, LS, and DM. Formal analysis, investigation, and writing: BB, JK, KK, LS, and DM. Supervision: BB and JK. All authors have read and agreed to the published version of the manuscript. --- Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. --- Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

This study investigated how land rights formalization had a ected land tenure security among landowners in two informal settlements of Lusaka and Chongwe districts, Zambia. It explored how social norms on land inheritance, decision making over land, marital trust and land related conflicts had been a ected by the changed nature of land rights. Data was collected through a questionnaire survey of all the households that had obtained title deeds at the time of the survey, two -in-focus group discussions and four key informant interviews. Results suggest that land tenure security is now a reality for residents that hitherto lived under constant threat of eviction. Landowners have benefitted from the formalization initiative through land laws and local norms that allow equitable access to land. Land rights formalization has curtailed land rights for secondary claimants such as extended family members, in preference for man, spouse and biological children. A sense of ownership undisputedly increased for men and women in the two study sites. About % of the respondents in both study sites indicated that formalization of land rights had not resulted in family conflicts. At least one-third from both sites reported an increase in love and trust between spouses after land rights formalization. About half of the respondents reported that no change in decision-making authority had occurred for men while % reported an increase. Formalizing land rights in informal settlements has entailed legalizing illegalities as regulations on plot boundaries are set aside by the state to achieve its aspirations of providing land tenure security to poor urbanites who would not otherwise have recourse to legal or regularized land. We recommend that caution be taken in promoting what is unarguably a pro-poor initiative to ensure that such initiatives should not incentivize future land encroachments.

Introduction 1.Importance of Workforce Diversity in Cancer Disparities Research The underrepresentation of women and individuals identifying as members of racial/ethnic minority groups in cancer disparities research and cancer care is well known [1,2], as are the potentially linked racial and ethnic disparities in cancer risk, cancer incidence, and cancer mortality. For example, African American/Black (hereafter, Black) and Hispanic/Latinx (hereafter, Hispanic) men and women bear an unjust burden of incidence and mortality from several cancers (e.g., breast, lung) relative to their non-Hispanic White counterparts [3]. Relatedly, Black and Hispanic adults also experience disparities in cancer risk behaviors, whereby they have greater incidence of physical inactivity, a poorer diet, greater rates of overweight/obesity, and more difficulty quitting tobacco than non-Hispanic White adults [4,5]. Prior studies have found that nearly half of cancer incidence is attributable to environmental and lifestyle factors, with tobacco use, excess body weight, and alcohol intake accounting for 29%, 7%, and 4%, respectively, of approximately 9000 male and 7000 female deaths annually [6][7][8][9]. These are modifiable behaviors/conditions that-if targeted in most at-risk and general populations-may prevent cancer incidences that are not solely hereditary in origin. Moreover, in some cases, women bear more of the burden of cancer risk behaviors relative to men (e.g., higher rates of severe obesity [10], greater smoking relapse rates [11]), making their inclusion in the field critically important to better understanding and addressing these disparities that can ultimately impact cancer incidence and mortality. There are many benefits of training a diverse workforce to address cancer and cancer risk disparities. For example, minority and women researchers may better understand and address the needs of their communities through their unique perspective and awareness of issues that are critical to solving disparities in their communities [12]. This understanding includes how social determinants, or conditions in the places where people live, learn, work, and play [13,14], affect cancer risk and risk behaviors [15][16][17][18]. Additionally, it is important to have researchers and medical professionals from minority populations involved in cancer research to foster population representativeness in clinical trials to better achieve cancer health equity [17,19]. Moreover, research has shown that diverse teams outperform homogenous ones when working together advantageously on incorporating innovation and distinct perspectives [20]. However, further work is needed to address the current research gaps regarding how the social determinants of health disproportionately contribute to cancer health disparities experienced by Black and Hispanic groups [15], as well as how to develop and prepare a diverse workforce to lead this work. Recognizing the importance of training minority and women scientists in order to more ideally address cancer health disparities, the National Cancer Institute (NCI) and other funders have invested heavily in the development of research and training programs [21]. However, more programs focusing on cancer behavioral risk factors such as diet, exercise, and weight management, and those specifically dedicated to developing women and racial minority populations for this work are needed [17,19,22,23]. There is also a need for better knowledge dissemination and translation between researchers and diverse communities [24,25]. Knowledge translation is particularly important in prevention as communities that have greater awareness and understanding of healthy lifestyle behaviors to reduce cancer risk are likely to have improved lifestyle behaviors, treatment options, patient advocacy, and patient compliance. Thus, it is extremely important to address these gaps in training programs to provide a more comprehensive approach in addressing cancer disparities with a more diverse workforce. --- UHAND Partnership Program The "UHAND" Program is a collaboration between the University of Houston (UH) and The University of Texas MD Anderson Cancer Center (MDA), funded by the NCI under a Feasibility Studies to Build Collaborative Partnerships in Cancer Research (P20) initiative (PAR-16-084). The purpose of this collaboration is to bring together a minority-serving institution and a comprehensive cancer center to create a comprehensive research/educational training program that would provide opportunities for underrepresented student "scholars" to develop careers in behavioral cancer prevention and cancer disparities research. Although not a requirement of the funding mechanism, both institutions are located close to one another and in Houston, Texas, the 4th largest metropolitan area in the US [26]. The UH is the second most ethnically diverse major research university in the US and is designated as a Hispanic-Serving Institution and an Asian American and Native American Pacific Islander-Serving Institution by the US Department of Education, Office of Postsecondary Education [27]. MDA has been named as one of the top two US cancer hospitals in the U.S. News & World Report's "Best Hospitals" survey annually since 1990 [28]. The UHAND Program's overarching goal is to ensure that scholars preparing to join the scientific workforce have the necessary skills and capacity to eliminate cancer inequities through the reduction of social and physical risk factors among disproportionately affected groups. The UHAND Program provides support for developing scholars and early-stage investigators (ESIs) to better understand and conduct intra-institutional research projects that address the social determinants of cancer risk behaviors that predispose Black and Hispanic groups to disproportionate cancer risk. The UHAND Program provides integrated education, broad stakeholder engagement, and targeted approaches to involve women, Hispanic, and Black scholarsin cancer research (see Figure 1). cancer hospitals in the U.S. News & World Report's "Best Hospitals" survey annually since 1990 [28]. The UHAND Program's overarching goal is to ensure that scholars preparing to join the scientific workforce have the necessary skills and capacity to eliminate cancer inequities through the reduction of social and physical risk factors among disproportionately affected groups. The UHAND Program provides support for developing scholars and early-stage investigators (ESIs) to better understand and conduct intra-institutional research projects that address the social determinants of cancer risk behaviors that predispose Black and Hispanic groups to disproportionate cancer risk. The UHAND Program provides integrated education, broad stakeholder engagement, and targeted approaches to involve women, Hispanic, and Black scholarsin cancer research (see Figure 1). The UHAND Program incorporates best practices of successful training programs, such as the use of a variety of learning methods [29][30][31] and the continuous evaluation of data on programmatic impact and outcomes [30]; likewise, it features several relatively unique characteristics, such as collaboration between researchers and advocates [24,25], and between institutions [32]. Additionally, the UHAND Program prioritizes diverse minority researchers and women for participation, focuses on behavioral risk factors for cancer, and includes knowledge dissemination and translation between researchers and community through social media and other means. Altogether, these components and the collaborative partnership resulted in the Program's funding by the NCI at first submission. --- UHAND's Major Goals The UHAND Program has four specific goals. The first goal was to develop a research and educational partnership between the UH, MDA, and local community-based organizations to stimulate collaborative cancer disparities research related to tobacco use, poor diet, and physical inactivity among Black and Hispanic individuals. The second goal was to support the development of ESIs through a rigorous Pilot Research Program that facilitates direct-experience proposing, conducting, and leading cancer disparities research, with the support of a larger UHAND Program and its undergraduate, graduate, and postdoctoral scholars. The third goal was to increase the number of underrepresented racial/ethnic minority scholars and faculty engaged in cancer disparities research by providing them with research training, mentorship, and service-learning experiences. Finally, The UHAND Program incorporates best practices of successful training programs, such as the use of a variety of learning methods [29][30][31] and the continuous evaluation of data on programmatic impact and outcomes [30]; likewise, it features several relatively unique characteristics, such as collaboration between researchers and advocates [24,25], and between institutions [32]. Additionally, the UHAND Program prioritizes diverse minority researchers and women for participation, focuses on behavioral risk factors for cancer, and includes knowledge dissemination and translation between researchers and community through social media and other means. Altogether, these components and the collaborative partnership resulted in the Program's funding by the NCI at first submission. --- UHAND's Major Goals The UHAND Program has four specific goals. The first goal was to develop a research and educational partnership between the UH, MDA, and local community-based organizations to stimulate collaborative cancer disparities research related to tobacco use, poor diet, and physical inactivity among Black and Hispanic individuals. The second goal was to support the development of ESIs through a rigorous Pilot Research Program that facilitates direct-experience proposing, conducting, and leading cancer disparities research, with the support of a larger UHAND Program and its undergraduate, graduate, and postdoctoral scholars. The third goal was to increase the number of underrepresented racial/ethnic minority scholars and faculty engaged in cancer disparities research by providing them with research training, mentorship, and service-learning experiences. Finally, the fourth goal was to develop a robust community outreach program focused on engaging community members in cancer disparities education, research, and clinical trials. --- UHAND Program Components There are three main components to the UHAND Program: an Education Program, a Pilot Research Program, and a Community Outreach Program. The Education Program (EP) serves the undergraduate scholars, doctoral scholars (hereafter referred to as graduate scholars), and postdoctoral fellows who represent diverse backgrounds and who were invited to participate in the UHAND Program through a rigorous application and selection process. The EP's goal is to enable scholars to develop the attitudes, knowledge, and skills necessary to conduct research on reducing cancer disparities, with a focus on lifestyle behaviors (e.g., tobacco use, poor diet, physical inactivity) that increase cancer risk for Black and Hispanic adults. This is done by providing the scholars with mentored research projects, seminars in cancer disparities and career development, interactive and community-based service-learning experiences, and summer research experiences, all guided through individual development plans (specifically myIDP [33], one of two IDPs recommended by the NIH) and executed between scholars and their mentors. Scholars also receive at least 8 h of ethics and responsible conduct of research training. All scholars are paired with university research faculty mentors-from either UH or MDA-who have expertise in cancer risk, social determinants of health, clinical and population cancer research in Black and Hispanic populations, and student mentoring. UHAND Program participation was designed to span a 2-year training period. The Pilot Research Program (PRP) is comprised of ESIs who work on pilot projects with senior mentors and an experienced investigative team. The PRP provides support to new investigators in developing innovative and impactful research conducted in community and clinical settings within cancer disparities research, particularly within tobacco-and lifestylerelated disparities research. The UHAND Program grant included funding to support two pilot projects: one on the stress-based biological and behavioral cancer risks among Mexican immigrants, and another on a lifestyle intervention for Black prostate cancer patients on active surveillance and their partners. The PRP was later expanded to include additional projects through Administrative Supplements that extended the scope of the initial two pilot projects. The extended scope incorporated the role of physical activity perceptions and barriers in Mexican immigrant cancer risks, and an evaluation of skeletal muscle strength and function among Black prostate cancer survivors in order to reduce the risk of developing cardiometabolic diseases and ultimately, to improve the quality of life in survivorship. UHAND scholars participated in these projects, assisting with data collection and analyses. The Community Outreach Program (COP) is integrated in the cancer health equity development of scholars and in the execution of pilot research projects to increase the real-world impact of these efforts. The COP comprises UHAND team members with community experience and hosts UHAND Programmatic events, informs community members about the UHAND Program, coordinates opportunities for scholars to attend health fairs and seminars, provides scholars with opportunities to present research to community representatives for feedback, works with scholars in fostering presentation skills to community audiences, and offers scholars unique opportunities to connect with the local community. The COP includes a Community Partners Network of local health-based organizations that provide service-learning opportunities for scholars, and Community Mentors who work with scholars on mentor plans and meetings, provide research feedback from the community perspective, enable access to at-risk populations, and identify local opportunities for sharing research findings. The COP team also works closely with the Community Advisory Board (CAB, elaborated below). The overall aim of the COP is to enable bi-directional communication between the community and the UHAND scholars to enhance the translation and potential impact of UHAND research. --- UHAND Structure and Guiding Boards The UHAND Program was funded through two grant awards (P20CA221696 and P20CA221697): one given to a Principal Investigator (PI) at MDA, and another to a PI at UH. Additionally, the components of the UHAND Program (e.g., EP) each have co-leadership and support staff from both MDA and UH, facilitating a true partnership between institutions. The UHAND Program is guided by a CAB, an Internal Advisory Committee (IAC), and an External Advisory Board (EAB). The CAB comprises prominent community leaders across a range of sectors in the greater Houston area that work collaboratively with the UHAND team in providing training, outreach, and research activity support, in addition to providing guidance for the dissemination of project findings in the community. The IAC comprises institutional leaders and researchers with equal representation from UH and MDA, and assists with scholar selection, institutional support, and program sustainment. The EAB comprises nationally renowned researchers and provides broad scientific direction to and drives innovation in the UHAND Program's research activities, in addition to keeping the UHAND team aware of new and innovative research and educational practices in the cancer disparities field. --- The Current Report The purpose of this report is to present information on the UHAND Program's first cohort of scholars and lessons learned that may inform similar training programs. Our objectives, delineated more fully in the Methods section, were broadly to enhance the scholars' research self-efficacy and academic output through competent mentorship and their sustained participation in the UHAND program, which would ultimately lead to interest in, and be an evidence of, the pursuit of higher education (and/or faculty positions) in the cancer disparities field. Our results are presented in relation to these objectives. This work may provide a comprehensive, evidence-and need-based model for other educational training programs with similar goals for improving cancer health inequities by creating more cancer research career opportunities for women and minorities. --- Materials and Methods --- Participants and Procedures Table 1 describes Note: <unk> Underrepresented racial and ethnic groups in health-related sciences, individuals with physical or mental disabilities, individuals from low-income families, and individuals from inhibiting educational environments [19]; * Scholars who are the first generation from their families to attend college; Postdoc: Postdoctoral Fellow; Grad: Graduate Scholar; UG: Undergraduate Scholar. --- Scholar Eligibility Criteria The eligibility criteria for admission to the UHAND Program were the following: (1) undergraduate or doctoral student enrolled in, or recently accepted into, a full-time degree program at the UH; (2) having at least 2 years remaining in their degree plans prior to graduation at the start point of the UHAND programming; (3) to have backgrounds in, or be interested in, behavioral and social sciences, social work, communications, biomedical sciences and related public health disciplines, or any other disciplines relevant to studying cancer health disparities; and (4) to be a US Citizen or permanent resident (a requirement of the funder). The eligibility criteria for the postdoctoral fellow included the need to have successfully earned a doctoral degree by the start point of the UHAND programming, and interests in cancer health disparities complementary to those of the intended mentor. In the case of the 1st UHAND Program cohort, that was the UH PI. It was not a requirement to be underrepresented in the sciences to be eligible for UHAND Program participation, but demographic data were collected at the point of application and women and/or individuals from racial/ethnic groups underrepresented in the sciences were prioritized for admission. --- Recruitment of Scholars Undergraduate and graduate scholars were recruited through the UHAND website, UHAND social media accounts, free-standing displays and recruitment materials at UH events, presentations in UH classes, word of mouth from UH faculty members to students, and interactive UH events such as UH Honors College summer research recruiting event and UH career fairs where UHAND team members presented the program [34]. The postdoctoral fellow was recruited through a national search through a job ad to facilitate a competitive pool of applicants more quickly than a local recruitment-only approach may have allowed. The job posting was shared with colleagues and on professional listservs, and distributed via targeted emails at the partnering institutions. --- Selection of Scholars Undergraduate and graduate scholars were selected from a variety of UH majors through a 3-level review process. First, the UHAND EP screened application materials and ranked candidates. Next, the UHAND IAC reviewed application materials and ranked applicants. Finally, the UHAND PIs reviewed application materials, received and reviewed rankings, and interviewed candidates. Applications were reviewed for demonstrated academic accomplishments, potential for academic and scholarly success (as indicated from letters of recommendation and transcripts), interest in cancer/health/social disparities, and prior research and/or working experiences. All information was considered by the UHAND PIs in making final admission selections. Out of 5 complete graduate scholar applications and 17 complete undergraduate scholar applications, 3 graduate and 6 undergraduate scholars were invited to join the program. The proportion of undergraduates versus graduate scholars invited to join the first cohort was based on selecting the best candidates of both pools and accounting for the maximum number of scholars we could afford on the budget, balancing that undergraduate inclusion was more affordable than graduate inclusion, based on respective pay rates. As part of their application materials, scholars indicated the top three mentors they would like to work with from a list on UHAND's website. Potential mentors (n = 10 MDA, n = 9 UH) were selected by the UHAND PIs based on their engagement in social/behavioral cancer disparities work and their agreement to participate. Scholars were matched to top-choice mentors, with consideration for ensuring distribution across potential mentors. All approached mentors (4 from UH and 4 from MDA; some mentored >1 scholar) accepted the scholar as a mentee. Postdoctoral applicants applied to the UH and were interviewed by the UH PI, with whom the selected candidate would directly work. In this case, the successful applicant was not previously associated with either of the partnering institutions. Undergraduate and graduate scholars were paid an hourly rate and committed 20 h per week to work with their research mentors and participate in the UHAND Programming (i.e., Cancer Disparities Seminars, Cancer Prevention and Control Grand Rounds, Community Presentations, Writing Sessions, Research Webinars, and other seminars). For undergraduate scholars, UHAND participation also included participation in the UH Summer Undergraduate Research Fellowship (SURF) program, where they learned about various aspects of research-from topics in data science and responsible conduct of research, to learning how to develop effective resumes and poster presentations, applying for awards and graduate school, and managing expectations and challenges in research [35]. They were also given an opportunity to formally present their research project results. The post-doctoral fellow was selected and appointed as a full-time scholar in the UHAND program with a designated faculty mentor. She also participated in all UHAND programming. --- Scholar Outcome Measures 2.2.1. Research Self-Efficacy (Goal 1) Surveys with questions on research self-efficacy were administered at baseline (2018), mid-program (2019), and exit (2020). Postdoctoral fellow and graduate scholar responses were combined to facilitate confidentiality. A modified version of Forester et al.'s research self-efficacy scale was used [36]. The original 33 items were slightly modified and reduced to 14 relevant items regarding successfully accomplishing research related tasks (e.g., writing a research paper, collecting data, formulating hypotheses, etc.), with response options ranging from no confidence (0) to total confidence (9). Two additional investigator-generated items asked about interest and likelihood in pursuing a career in cancer disparities research, with response options ranging from not at all (0) to extremely (4). Our explicit goal was to increase research self-efficacy over time among UHAND scholars from baseline to exit. The research self-efficacy scale demonstrated strong reliabilities (2018: 0.92; 2019: 0.95; 2020: 0.91) in this sample. --- Academic Products (Goal 2) The number and category of academic products were assessed by scholar report throughout the duration of the program. Additionally, an annual survey following program exit captured academic products, noting UHAND support that came to fruition following their enrolled period in the UHAND program. Due to the commonly experienced lag between submission and presentation delivery/publication, our timeframe to assess adherence to our explicit goal was extended to the end of the 2020 calendar year (i.e., 7 months following the end date of Cohort 1 s intended enrollment period). Our explicit goal was to engage UHAND scholars in research, yielding peer-reviewed academic work products such that 2 presentations would be achieved by each undergraduate scholar, 1 presentation and 1 publication would be achieved by each graduate scholar, and 2 presentations and 2 publications would be achieved by the postdoctoral fellow. Goals in excess of this could be pursued by the mentor and mentee as part of the IDP. It is notable that the MDA and the UH PIs were each involved in organizing local conferences to which the scholars could and did submit research for presentation: "Eyes Have Not Seen, Ears Have Not Heard: Breakthroughs in Cancer Research" (Fall 2019), and the "Inaugural HEALTH Research Institute conference" (Winter/Spring 2020). Other presentation opportunities included the annual "Undergraduate Research Day" sponsored by the UH Honors College, and a host of national conferences to which scholars could submit abstracts for presentation. --- Program Retention (Goal 3) Scholar program retention was measured by the comparison of the number of scholars from the time of enrollment to program completion (exit). Our explicit goal was to retain 100% of scholars in the 2-year program by program completion. --- Mentee Evaluation of Mentors (Goal 4) Surveys with questions on scholar evaluations of their mentors were administered at mid-program and at exit. The postdoctoral fellow and graduate scholar responses were combined to facilitate confidentiality. Fleming et al.'s mentor competency assessment was used [37]. This 26-item assessment enables mentees to evaluate 6 mentor competencies in the topic areas of maintaining effective communication, aligning expectations, assessing understanding, fostering independence, addressing diversity, and promoting professional development. Each competency had 2-6 questions asking how skilled the mentee feels the mentor was in the mentioned areas (e.g., active listening, setting research goals, building confidence, etc.) on a scale of not at all (1) to extremely (7). Our explicit goal was to obtain overall scholar ratings of their mentors as at least "moderately" skilled for each mentor competency assessed. The scholar evaluations demonstrated strong reliabilities in all subscales, in both years (2019: 0.91-0.98; 2020: 0.95-1.00). --- UHAND Program Feedback and Strategies Learned (Goals 5 and 6) Exit surveys on program feedback and strategies learned were administered to scholars. The postdoctoral fellow and graduate scholar responses were combined to facilitate confidentiality. Scholars were asked to rate their experiences in the following: 10 categories of UHAND educational seminars and sessions on a 5-point Likert scale, from very poor (0) to excellent (4), with an option to select N/A; UHAND programmatic aspects via 12 items on a 5-point Likert scale from very poor (0) to excellent (4); and general program satisfaction in 5 different programmatic activities on a 5-point Likert scale, from very poor (0) to excellent (4), with an option to select N/A. Face valid, investigator-generated items were used to assess confidence in using strategies learned from UHAND trainings, workshops, and books. These questions asked scholars to rate their confidence in time management, conflict resolution, productivity in writing and completing writing projects, engaging in difficult conversations, career exploration, being resilient in academic rejections, and being resilient in discriminatory/unfair experiences in academia/training on a 5-point Likert scale, from not confident at all (0) to completely confident (4), with an option to select N/A. Our explicit goals were to obtain scholar ratings of the UHAND Program and programming, from "good" to "excellent" (i.e., 3-4), and of confidence in using strategies learned in the UHAND Program, from "fairly confident" to "completely confident" (i.e., 3-4). --- Post-Program Progression (Goal 7) Scholar post-program progression was assessed through surveys that scholars completed monthly during their participation in the program and bi-annually thereafter. Our explicit goal was to have scholars pursue progression in their education and career. Specifically, to have 100% of undergraduate scholars pursue admission to a graduate program (e.g., master's programs, doctoral programs, medical school), 100% of graduate scholars pursue a postdoctoral training or faculty position, and the postdoctoral fellow obtain a tenure-track faculty position following program completion. --- Analyses Descriptive statistics, including means (and standard deviations, SDs) and frequencies (and percentages), were calculated for continuous (i.e., research self-efficacy, mentee evaluation of mentors, and UHAND Program feedback) and binary variables. To account for the small sample size, the Wilcoxon signed-rank test was used to analyze data between matched subjects for differences in distribution and for outcomes of interest over time. Alpha was set at 0.05. All analyses were conducted using SAS 9.4 [38]. --- Results --- Research Self-Efficacy (Goal 1) Table 2 presents the means and standard deviations for each of the 14 research selfefficacy items and for the total research self-efficacy score of all scholars across years and by groups (undergraduates vs. graduates/post doc). The means of total research self-efficacy scores across all scholars were 93.4 (SD = 18.75), 88.9 (SD = 27.8), and 106 (SD = 16.63) for 2018, 2019, and 2020, respectively. The total research self-efficacy score for all scholars significantly increased from 2019 to 2020 (88.9 vs. 106, p = 0.0273) and from 2018 to 2020 (93.4 vs. 106, p = 0.0293), but not from 2018 to 2019 (93.4 vs. 88.9, p = 0.5566). Thus, our explicit goal of increasing the research self-efficacy of scholars from baseline (2018) to exit (2020) was achieved. A closer examination by scholar group indicated that the research self-efficacy of undergraduate scholars increased significantly from 2019 to 2020 (74 vs. 102.83, p = 0.0313), but not from 2018 to 2019 (83.67 vs. 74, p = 0.2188), or from 2018 to 2020 (83.67 vs. 102.83, p = 0.0625). No significant differences were found in graduates/post docs in research self-efficacy scores across years. --- Academic Products (Goal 2) Table 3 displays the academic presentation and publication goals for scholars. Regarding academic presentations, 100% of scholars exceeded presentation goals. Regarding publication goals, the postdoctoral fellow, 1 graduate scholar, and 1 undergraduate scholar met or exceeded goals; however, 2 graduate scholars failed to achieve this goal. Thus, only 60% (3/5) achieved/surpassed publication goals. However, it is important to note that 1 of these graduate scholars achieved multiple publications over this time span, though not with her UHAND mentor (not shown in Table 3), and the other graduate scholar has manuscripts in progress with her UHAND mentor (not shown in Table 3). Overall, we partially achieved academic product goals during the assessed time span. --- Program Retention (Goal 3) Five of six undergraduate scholars completed the 2-year program (83.33%), with 1 scholar completing only 1 year due to an opportunity to graduate early. All 3 graduate scholars completed the 2-year program (100%), and the postdoctoral fellow left after a year to take a tenure-track faculty position (0%). Overall, program retention was 80%. As such, we did not achieve our goal to retain 100% of scholars in the 2-year UHAND Program by program completion. However, none of our scholars dropped out of the program while still enrolled in/employed by UH full time; thus, we retained them in the program as long as they remained eligible for support as UHAND scholars (Table 3). --- Mentee Evaluation of Mentors (Goal 4) Descriptive statistics for different aspects of the mentee evaluation of mentors by scholars in the 2019 and 2020 surveys are presented in Table 4. There was no significant difference between 2019 and 2020 on any aspect of the scholar evaluations of mentors. All mean ratings were above the threshold for at least "moderately skilled" (Table 4), indicating that we had achieved this program goal. Note: Responses for each item ranged from 1 to 7, where 1 = Not at all; 4 = Moderately; and 7 = Extremely. <unk> A total of 11 mentor-mentee dyads were assessed across 10 scholars; 1 undergraduate scholar switched mentors after Year 1 (and thus had 2 mentors). --- UHAND Program Feedback and Strategies Learned (Goals 5 and 6) Table 5 shows scholar feedback on the quality of UHAND educational seminars and sessions, the quality of other UHAND programmatic aspects, general program satisfaction, and confidence in using strategies learned in the UHAND Program. All ratings were above the scale of 3 (i.e., "good" for UHAND Program seminars/sessions, programmatic aspects, and general program satisfaction; and "fairly confident" for confidence in using the strategies learned) except for 2 ratings: community service-learning experience with community partners (Mean = 2.6, SD = 1.07) and community service-learning experience with UHAND staff/health educators (Mean = 2.78, SD = 1.09). Thus, we failed to achieve our goal of obtaining scholar ratings from "good" to "excellent" (i.e., [3][4] for UHAND programming, and ratings from "fairly confident" to "confident" (i.e., 3-4) for using strategies learned in the UHAND Program, particularly in the case of 2 of 34 (5.89%) UHAND Program seminars/sessions, both in the area of community service-learning. 3.80 (0.42) I have increased my ability to be an independent researcher as the result of UHAND. 3.60 (0.52) The UHAND Program was important to my career development. --- Post-Program Career Progression (Goal 7) All 6 undergraduate scholars pursued admission into health science graduate (4 scholars) or medical school (2 scholars) programs (100%), 5 of whom have reported program acceptance to date. It is too early to assess the 3 graduate scholars' program progression, as they are all currently in good standing in graduate school. The postdoctoral fellow secured a tenure-track faculty position. With follow up, we will be able to evaluate the launch of the 3 graduate scholars in their career progression. --- Discussion This report described the implementation of the UHAND Program, a UH and MDA collaboration to create a comprehensive career development training program for scholars underrepresented in the fields of behavioral cancer prevention and cancer disparities research, funded by the NCI's Partnerships to Advance Cancer Health Equity initiative. The UHAND Program incorporated best practices of successful training programs, such as integrated education (diverse professional teams working together, cross-training, exposure to a broad range of disciplines, an integrated curriculum with different subjects, and a variety of learning methods) [29][30][31]; professional development [29]; collaboration between researchers and advocates [24,25]; collaborative partnerships between institutions [32]; broad stakeholder engagement [25,29]; targeted approaches for at-risk groups [25]; training programs at predoctoral, doctoral, and postdoctoral levels [30]; continuous evaluation data on programmatic impact and outcomes [30], while uniquely focusing on research and training in behavioral risk factors for cancer-prioritizing diverse women and minority researchers for training-and facilitating knowledge dissemination and translation between researchers and community. This report presents outcomes relative to the 7 goals of the UHAND Program's work with scholars and information that may be helpful for the design of future training programs that share similar goals. The first goal to increase scholars' research self-efficacy from program baseline to exit was achieved. The UHAND Program provided scholars with research experiences, didactic coursework, career development seminars, and interactive community-basedlearning experiences guided through IDPs with their mentors. Scholars were also involved in presenting research to community representatives and audiences, as well as professional/academic audiences. For our undergraduate scholars, they also participated in the SURF program at the UH, which has several didactic research and ethics lectures, tours of labs on campus, and research-building skills over a 10-week summer term. These types of experiences were designed to help build and develop their research self-efficacy skills through mastery via skill-building (e.g., through hands-on mentored research projects), vicarious learning (e.g., through engagement in research seminars given by established researchers), and support (e.g., through fellowship activities, EP check-ins, PI meetings with mentees and mentors). Overall, research self-efficacy significantly increased over the 2-year program, although this was driven by the undergraduate scholars and followed an overall decrease in self-efficacy from the pre-program to the mid-program assessments. The latter may be explained by the Dunning-Kruger Effect, whereby inexperienced people tend to have high confidence and falsely and unknowingly rate their performance highly [39]. Over time and with gained experience, they realize how much they do not know, and this may cause a sharp decline in self-confidence [39]. One report has shown that less competent junior physicians tended to rate their self-efficacy higher than what it was, while competent junior physicians, especially women, tended to rate their self-efficacy lower than those who were less competent [39]. Suggestions for prospective programs would be for the program directors to be aware of the Dunning-Kruger Effect, and particularly that self-efficacy may not translate directly to performance [39]. Tailoring the assessment to the education level of the scholar would be advisable (e.g., undergraduate vs. graduate scholar) by setting clear expectations, measures, and providing factual feedback on skills for undergraduate scholars [39]. Another suggestion is to have assessments and feedback from multiple sources to accurately capture the scholars' progress in the program [39]. Additionally, identification of research self-efficacy skills that received lower ratings relative to the others could be used to develop programming for future scholars. This may be particularly important for programs that are designed to be less than 2 years in duration, given that there may be a sharp decline in self-efficacy over time that should be monitored and addressed in real-time as the scholars enter the research arena. Moreover, it is also worth noting that the more educated group (graduate scholars and the postdoctoral fellow) did not experience statistically significant increases in self-efficacy over time, in contrast to the undergraduate group. Nevertheless, while self-efficacy ratings were higher among this group of scholars relative to the undergraduates to begin with, ratings generally rose from pre-training to exit (from 108 in 2018 to 110.75 in 2020) and were quite high on average, given that the scale range was from 0 to 126. Their smaller group size and higher starting point relative to the undergraduates may have also affected the inability

Black and Hispanic adults are disproportionately affected by cancer incidence and mortality, and experience disparities in cancer relative to their White counterparts in the US. These groups, including women, are underrepresented among scientists in the fields of cancer, cancer disparities, and cancer care. The "UHAND" Program is a partnership between institutions (

providing factual feedback on skills for undergraduate scholars [39]. Another suggestion is to have assessments and feedback from multiple sources to accurately capture the scholars' progress in the program [39]. Additionally, identification of research self-efficacy skills that received lower ratings relative to the others could be used to develop programming for future scholars. This may be particularly important for programs that are designed to be less than 2 years in duration, given that there may be a sharp decline in self-efficacy over time that should be monitored and addressed in real-time as the scholars enter the research arena. Moreover, it is also worth noting that the more educated group (graduate scholars and the postdoctoral fellow) did not experience statistically significant increases in self-efficacy over time, in contrast to the undergraduate group. Nevertheless, while self-efficacy ratings were higher among this group of scholars relative to the undergraduates to begin with, ratings generally rose from pre-training to exit (from 108 in 2018 to 110.75 in 2020) and were quite high on average, given that the scale range was from 0 to 126. Their smaller group size and higher starting point relative to the undergraduates may have also affected the inability to achieve statistically significant increases in overall self-efficacy. It is also worthy of note that overall interest in a cancer disparities career declined over time in the program, and the likelihood of pursuing a career in cancer disparities experienced a mid-program dip, similar to scholar self-efficacy. We believe that this pattern may be similarly explained-greater experience with something may reveal nuances and challenges that dissuade further pursuit and alter original intentions based on new information gained through experience. However, it is important to note that these changes were not statistically significant for any group, sample sizes were very small such that a single rater could significantly influence averages, and all average ratings were <unk>3.0 in every year and every scholar group, equivalent to "very interested" and "very likely" to pursue a career in health disparities (except for the undergraduates' mid-program dip to 2.83). Nevertheless, we failed to completely advance interest and intention to extreme interest/likelihood over time for our scholars; this suggests more work is needed to better understand scholars' responses and thus develop programming to heighten interest/intention, perhaps by further enhancing self-efficacy, addressing imposter syndrome, and/or focusing on scaffolding approaches that clearly indicate pathways to such careers beyond the training program itself. Qualitative feedback, potentially gathered by a researcher not affiliated with the training program, may be helpful in better understanding these patterns and may be an advisable evaluation method to build into similar programs a priori so that issues can be addressed in real-time, as applicable. The second goal was to achieve peer-reviewed academic work products amongst the undergraduate, graduate, and post-doctoral scholars. While all scholars achieved the minimum number of presentations, just over half of the scholars achieved publication goals with wide variability amongst scholars (e.g., one undergraduate scholar gave 8 presentations and a graduate scholar gave 11 presentations by mid-program; one undergraduate scholar published 2 empirical manuscripts). Variability in achieving publication goals was at least partially attributable to some scholars working on new projects in the design or data collection phase, whereas others worked with existing data that facilitated faster manuscript development. Additionally, the publication of scientific articles often takes time, which may partially explain why some scholars did not achieve the publication goals during the 2-year program. Overall, results inform approaches for program leadership to take when setting expectations with research mentors, who could in turn set specific publication goals with attainable timelines in scholars' IDPs. This may include recommendations that mentors provide scholars with access to datasets for secondary analysis and manuscript development while they are involved with ongoing research data collection. Other similar programs could make specific recommendations for academic work product outcomes or provide access to datasets for scholars with mentors who do not have secondary data to work with. The third goal-attaining complete 2-year program retention-was not achieved, but not for undesirable reasons. Some scholars obtained outstanding career opportunities for which they left the training program early, such as our postdoctoral scholar, who was offered a tenure-track faculty position. In UHAND, applicants to the program were required to have at least two years remaining prior to graduation so that they could participate in the entire 2-year program; however, training programs need to be flexible for changing circumstances or opportunities over time that may affect a scholar's original plan. It is noteworthy that apart from the postdoctoral scholar who left the program early due to a job offer, all other scholars remained in the program until completion or (unanticipated early) graduation; this positive result partly supports the high program satisfaction expressed by participants. However, it is important to note that UHAND graduate scholars reported becoming overburdened by UHAND program requirements in their second fellowship year, as they also began their required practicum clinical work in the community, which required additional time and travel that affected their availability for UHAND seminars and trainings. Additionally, two of the three graduate scholars in the UHAND Program were paired with a mentor other than their counseling psychology doctoral mentor, which may have increased the pressure on these scholars to conduct research with UHAND in addition to that with their doctoral program mentor. Thus, the design and duration of future training programs should take into account how scholars' degree program requirements may vary over time, and how the training program can be adapted in kind to ensure academic success and prevent scholar burn out. Additionally, we recommend that graduate scholars in research training programs similar to UHAND be paired with their doctoral mentors for training program research in order to reduce the burden associated with having multiple research mentors. The fourth goal was to obtain a mentee evaluation of mentors of at least "moderately skilled" in the competencies desired of mentors for the UHAND program, specifically in clear communication, setting expectations, being understanding, fostering independence, respecting diversity, and guiding professional development. We achieved this goal, and thus our program may provide a suitable example for mentor and mentee collaboration. The PIs of the UHAND Program were both department chairs at their respective institutions and had a deep understanding of the faculty across departments doing work in cancer disparities. As the PIs personally knew the mentors and/or supervised them in these roles, they were able to successfully match them with scholars and guide them in this process. Generally, we recommend that other training programs work with individuals who have vast institutional knowledge/experience/connections when selecting mentors for program inclusion. In the absence of such knowledge, or perhaps in any case, programming could also be provided to support and develop the mentors in the execution of their roles. In the UHAND program, we routinely communicated with the mentors about external opportunities (e.g., locally held workshops) to enhance their mentorship skills, but we offered no formal training ourselves nor did we assess or monitor training that they may have received. In retrospect, offering and assessing the results of mentor training may have further enhanced competencies in this area and might be particularly appropriate for enhancing training with women and minority scholars, who may not be of the same sex/ethnicity/race as their mentors. The fifth and sixth goals were to achieve ratings from "good" to "excellent" on UHAND Program Feedback, and ratings from "fairly" to "completely confident" on strategies learned by our scholars. The programming included a range of seminars designed to enhance time management, facilitate competency in academic writing, explore career options in cancer disparities, and develop skills in presenting oneself and one's research in both academic and community settings. "Good" to "excellent" scores were achieved for each program element, with the exception of the two tapping into community servicelearning experiences. Feedback provided by the scholars indicated that they were, at times, assigned tasks at community agencies that were rather mundane (e.g., filing), which contributed to dissatisfaction. Feedback from the community agencies in which scholars were placed indicated that they were used to working with more substantial scholar time commitments than what this portion of the UHAND Program allowed (i.e., about 30 h a semester) and thus struggled with how to involve our scholars in substantive and meaningful work. Hence, training programs with service-learning components should explore methods that would enable more time at community agencies, particularly in ways that do not increase overall programmatic time requirements. To give an example, this might be achieved by assisting scholars to get course credit for work with community agencies (e.g., internship experiences) or degree-required clinical practice hours in these settings. Despite challenges, program satisfaction was high, and scholars indicated they were "fairly" to "completely confident" in using many of the strategies and skills they were taught in practice (e.g., engaging in difficult conversations, being resilient in handling discriminatory or unfair experiences in the academy). The seventh goal was to track post-program progression, with undergraduate scholars pursuing graduate program admissions, graduate scholars pursuing postdoctoral or faculty positions, and the post-doctoral scholar obtaining a tenure-track faculty position. We achieved this goal with the post-doctoral and undergraduate scholars, the latter of whom went to a combination of master's degree, doctoral degree, or medical degree programs. It is too early to assess the 3 graduate scholar's program progressions as they remain in good standing in graduate school. With an ultimate program goal of supporting scholars in pursuit of careers in cancer disparities research, it is important to have programs build processes for continuous monitoring and guidance of scholar achievements that extend beyond graduation and the program funding period. To this end, we will continue to follow our scholars over time to track their academic career achievements and ultimate employment as trained professionals. Mechanisms to bolster continued engagement with scholars over time (e.g., periodic newsletters and mailings, "reunion" dinners or virtual events) will need to be considered. While the UHAND Program achieved many goals, lessons were learned that could enhance similar training programs in the future. Although the UHAND Program represents a model for educational training programs focused on reducing health disparities, particularly those focused on cancer, there are limitations that need to be considered in the evaluation of our goals. These include the fact that lessons may not generalize to other cohorts, institutional collaborations, or mentors. This may especially be the case given the low representation of men (n = 1) in our initial cohort of scholars. Future programs may wish to determine a desired sex distribution a priori and select scholars accordingly, inasmuch as a balance between the sexes is desirable. Additionally, all the graduate scholars were in counseling psychology, which is likely a reflection of the UH PI's affiliation with that program and the program's focus on health-oriented, community-engaged work and social justice, which complements the UHAND mission. However, the UHAND Program serves as a model of a grant-supported training program that spanned three diverse educational levels at a minority serving institution, achieved excellent collaboration with a comprehensive cancer center that gave scholars access to many opportunities unavailable at their home institution, and provided diverse scholar access to the cancer center faculty for research training. --- Conclusions The UHAND Program may inform other educational training programs that aim to reduce inequities in cancer, and in health more broadly, by increasing the number of underrepresented racial/ethnic minority student scholars through the provision of research training, mentorship, and service-learning opportunities. The UHAND Program also provides an example of successful engagement and collaboration between institutions, and with community via community outreach. On the whole, the UHAND Program may provide a comprehensive, evidence, and need-based model for other educational training programs with similar goals. --- Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and confidentiality concerns given the very small group of mentees and mentors and the ability to link the two from the data alone, which could affect dynamics of ongoing mentoring relationships in unknown ways. --- Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Black and Hispanic adults are disproportionately affected by cancer incidence and mortality, and experience disparities in cancer relative to their White counterparts in the US. These groups, including women, are underrepresented among scientists in the fields of cancer, cancer disparities, and cancer care. The "UHAND" Program is a partnership between institutions (

Introduction --- D uring the first wave of the coronavirus disease 2019 (COVID- 19) pandemic in spring 2020, 19 countries in the EU/EEA and the UK implemented social distancing or'stay-at-home' recommendations for risk groups or vulnerable populations. 1 One of these countries was Sweden, where the Public Health Agency issued a nonmandatory recommendation for individuals aged 70 or over, i.e. the most vulnerable population group concerning severe COVID-19 disease, to avoid contact with persons outside the household and in crowded places (e.g. stores, public transportation). 2 Systematic reviews of the literature on non-pharmaceutical interventions (NPIs) against COVID-19 suggest that social distancing recommendations and'stay-at-home' orders were moderately effective at reducing the incidence of COVID-19 disease. 3,4 The tendency to comply with NPIs against COVID-19 seems to increase with age, 5,6 and survey data from Sweden suggest that many followed the age-specific recommendation. 7,8 However, it remains unclear how effective the policy was in preventing COVID-19 disease among older adults in Sweden. In this article, we employ a regression discontinuity design (RDD) 9 to isolate the additional effect of Sweden's age-specific recommendation on social distancing behaviours and disease outcomes during the first wave of the COVID-19 pandemic, beyond the effects of general recommendations that were present at the time. Agency in Sweden issued a specific recommendation that individuals aged 70 years or older should avoid crowded places and contact with people outside the household, 8 which was in effect until October 2020 8 (just prior to the second wave). The same recommendation was given to individuals younger than 70 years if they had at least one of the following risk factors: high blood pressure, heart disease, lung disease, obesity, diabetes or receiving immunosuppressant treatment. --- Social distancing outcome measures We used data from COVID Symptom Study Sweden (CSSS), 10 an app-based study that collects data for epidemiologic surveillance and prediction of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection via daily self-reports of disease symptoms. [11][12][13] On their first use of the app, participants self-reported their year of birth, sex, height, weight and postal code. They also completed a health survey with questions about pre-existing health conditions. Participation was voluntary and anyone aged 18 or above living in Sweden could download the CSS app and participate after providing informed consent. The app received considerable attention in the national and local press in the areas surrounding the two founding universities (Uppsala and Lund). Overall, participants were more healthy, less likely to live in disadvantaged areas and less likely to be smokers than the general population. 10 As the first wave occurred during the spring of 2020, we considered individuals who were 70 years of age at the end of 2019 (i.e. born in 1949) to be exposed to the age-specific social distancing recommendations. From 7 May to 29 September 2020, the app also included a weekly question about the levels of isolation during the last seven days. The respondents were asked: (i) 'In the last week, how many times have you visited somewhere with lots of people (e.g. groceries, public transport, work)?', (ii) 'In the last week, how many times have you been outside, with little interaction with people outside your household (e.g. exercise)?' and (iii) 'In the last week, how many times have you visited a healthcare provider (e.g. hospital, clinic, dentist, pharmacy)?'. As described further below, our analysis focuses on individuals close to the 70-year threshold. However, it requires data from younger and older individuals to model the relationship between these social distancing measures and age. We decided a priori to include period-specific averages of the social distancing measures for individuals born before 1980 (i.e. age 39 at the end of 2019) in the study, and there were too few participants born each year before 1940 (79 years) to be included in the analysis of the social distancing data. Participants also had to have at least one observation of isolation data from the period when the social distancing questions were asked up until the end of the first wave of the pandemic (7 May-31 July 2020) (n 1<unk>4 96 053). We averaged the three social distancing measures for each respondent to form a weekly average during this period. Supplementary table S1 contains an overview of characteristics of the entire sample and for individuals close to the age threshold for the recommendations (65-69 years, 70-74 years). Due to the isolation policy's aim to reduce visits to crowded places, measure (i) was our primary measure of social distancing. Going outdoors with limited physical interaction was fine according to the recommendations. Measure (ii) should therefore not be affected. It was less clear what to expect for measure (iii). It was recommended that a courier (such as a younger relative) collect prescriptions from pharmacies. Measure (iii) did, however, include in-person healthcare visits, for which postponement could be considered an adverse effect. --- Disease outcomes We also investigated population-level effects on severe cases (hospitalizations or deaths attributable to COVID-19). We obtained national data on all individuals born before 1980 and coded a binary indicator for whether they had at least one inpatient COVID-19 disease episode or had died due to COVID-19 disease during the first wave (16 March-31 July 2020; n population 1<unk>4 5 396 837; n severe cases 1<unk>4 21 804). The inpatient data were retrieved from the National Patient Register 14 and mortality data from the Cause of Death Register 15 (see Supplementary material for a detailed description). The retrieved data also contained information on year of birth, home address postal code and sex. As a secondary disease outcome, we used the number of confirmed infections by polymerase chain reaction (PCR) testing obtained from the SmiNet database at the Public Health Agency (n confirmed cases 1<unk>4 48 984). It was mandatory for all clinical laboratories in Sweden to report PCR tests positive for SARS-CoV-2 to SmiNet during the COVID-19 pandemic. 10 Tests were highly selective during the first wave, and positive cases represented mostly people who either needed treatment or were being tested because they worked in the healthcare industry. Thus, absolute effects should be interpreted with caution. Nevertheless, as explained in the following section, our design compares individuals aged just above and below 70 years. Therefore, relative estimates can still be meaningful, assuming that testing probabilities were equal close to this threshold. --- Regression discontinuity analysis We relied on a sharp RDD to estimate the effect of the recommendations on social distancing behaviours and disease outcomes as a discontinuous function of age, in years at the end of 2019, at the 70year threshold. The design has, e.g. been used to estimate the effects of early antiretroviral therapy for HIV patients 16 and other agespecific policies (e.g. minimum drinking age laws and co-payments in healthcare). 17,18 Causal effects can be estimated in observational data without controlling for confounders by exploiting changes induced by arbitrary thresholds, such as an age limit. 16 If no other causes of the outcome changes discontinuously at the policy threshold, the RDD estimates will reflect causal effects at the threshold (i.e. for people who are precisely 70 years of age). 19 We are not aware of any other policies that might have affected social distancing or COVID-19 disease at the 70-year threshold. The ability to isolate may be affected by retirement, but 65 is the most common retirement age in Sweden, and retiring at 70 is rare. 20 Hence, retirement should not bias the results by causing a discontinuity at the 70-year threshold. Our implementation follows the RDD estimation and reporting guidelines outlined by Athey and Imbens, 21 Hilton Boon et al. 9 and Gelman and Imbens. 22 While our primary interest is in individuals just above and below the 70-year threshold, RDD estimation requires fitting models to estimate the relationship between outcome variables and age. This estimation is usually performed within a small age window around the threshold (also known as bandwidth). The outcome-age relationship is not of primary interest but helps capture the effects of confounding variables that develop smoothly with age. However, we have to use appropriate model specification and bandwidth to avoid model misspecification bias. 22 Complex model specifications in RDD analyses are prone to overfitting, and Gelman and Imbens 22 caution against using models with high-order polynomials (greater than linear or quadratic). We therefore used local linear and quadratic regressions to estimate the jump in the outcomes at the threshold. In each analysis, we used a data-driven bandwidth selection method to identify the mean squared error optimal window around the 70-year threshold. 23 The larger the bandwidth (i.e. the age window used in the analysis), the more individuals are included, which increases the precision of the effect estimates. However, the risk of model misspecification bias also increases. The data-driven procedure aims to identify the largest possible window in which the relationship between the outcome and age is approximately linear (or quadratic, depending on the model). The analyses were performed using the rdrobust package (version: winter 2020) for Stata (version 16.1). 24 Further details are provided in the Supplementary material. As the pandemic did not affect all regions equally, we conducted subgroup analyses by geographical area (Stockholm, which was hit particularly hard in the first wave, versus the rest of Sweden). We also stratified results by sex to investigate how the underlying risk affected the effect of the recommendation. In the social distancing data, which contained information on medical risk factors, we also considered two additional subgroups: those without any and those with at least one of the following six risk factors communicated by the Public Health Agency in May 2020: obesity (body mass index! 30), diabetes, lung disease, cancer, heart disease or on immunosuppressant medication. --- Sensitivity analyses We performed recommended sensitivity, balance and falsification checks to assess the risk of bias. 9 We present these analyses in the Supplementary material. In summary, analyses with alternative bandwidths are similar to the main results. The data also passed standard falsification and balance checks (e.g. no evidence of sorting or discontinuities on covariates). --- Ethics approval The Swedish Ethical Review Authority has approved CSSS and the collection of the register data used in this study (DNR 2020-01803 with addendums 2020-04006, 2020-04145, 2020-04451, 2020-07080 and 2021-02316). --- Results --- Principal findings The study participants went to crowded places 5.4 times a week, outdoors with limited interaction 8.8 times a week, and to healthcare providers 0.5 times a week on average during follow-up (see Supplementary table S1 for details). Figure 1 shows how these behaviours varied by age, alongside the fitted values from local linear (figure 1A) and quadratic (figure 1B) regressions estimated within the optimal windows around the 70-year threshold. The analysis suggests that the policy threshold is associated with a sharp decline in the average number of times older adults visited crowded places (e.g. stores) during the first wave of the pandemic [<unk>0.47 (95% confidence interval, CI: <unk>0.89, <unk>0.05) times less per week in the entire sample, which corresponds to a 13% reduction; table 1; figure 1]. We found no evidence of discontinuities at the 70-year threshold on being outside with little interaction or visits to healthcare providers (table 1; figure 1). Figure 2 shows estimated effects on the incidence of severe COVID-19 disease and all confirmed cases per 1000 population at the national level during the first wave of the pandemic, and table 2 contains the effect estimates expressed as incidence rate differences and rate ratios. Overall, it appears that the recommendations may lead to a reduction in COVID-19 disease at the age threshold compared to a scenario without the age-specific recommendations (table 2; figure 2). The local linear estimates indicate a 16% reduction in both severe COVID-19 cases [incidence rate ratio (IRR) 1<unk>4 0.84 (95% CI: 0.73, 1.00)] and the number of confirmed cases [IRR 1<unk>4 0.84 (95% CI: 0.69, 1.08)] at the 70-year threshold, although the CI for confirmed cases overlaps the null (table 2). For severe cases, the estimate was slightly larger in the quadratic specification [IRR 1<unk>4 0.78 (95% CI: 0.64, 0.99)]. Our calculation in Supplementary Box S1 uses these numbers to estimate the impact of the recommendations, assuming that the relative effect is the same for everyone older than 70 years. The results imply that the policy prevented 1803 (95% CI: --- Subgroup analyses The results from the subgroup analyses are presented in table 1 (social distancing outcomes), table 2 (disease outcomes) and Supplementary figures S1-S5 (RDD plots). These estimates were generally imprecise, and the observed differences between subgroups should therefore be interpreted with due caution. Nonetheless, we found an indication that the effect on visits to crowded places was larger among individuals without risk factors than those with at least one risk factor, larger among men than women and larger in Stockholm County than in the rest of Sweden (table 1; Supplementary figures S1 and S2). Exploring these subgroups further, we found that the effect on visits crowded places appeared to be limited to men (independent of risk group status) and women without other risk factors. We found no indication of an effect among women with other risk factors (Supplementary figure S3), but they also isolated themselves more than the other groups even at younger ages (Supplementary figure S3). For the disease outcomes, stronger absolute effects were suggested among men than among women (table 1; Supplementary figures S4 and S5), which is consistent with the social distancing results. However, no effect on COVID-19 disease was observed for Stockholm County, where results were inconclusive (table 2; Supplementary figures S4 and S5). --- Discussion The results suggest that Swedish 70-year-olds isolated themselves more than those just below 70 years, implying that at least parts of the population adhered to the non-mandatory, age-specific recommendations communicated by the Swedish Public Health Agency. The results were generally in line with expectations. In particular, we found that the effect was limited to visits to crowded places, which is the social distancing outcome we assumed would be affected most by the recommendations. The impact on social distancing also seems to have caused a drop in disease outcomes at the 70-year threshold. We were unable to draw firm conclusions from our subgroup analyses, however. Results were inconsistent and inconclusive for Stockholm County, where the pandemic hit particularly hard during the first wave in Sweden. Statistical uncertainty aside, our data suggest also that men may have experienced larger disease risk reductions from the age-specific recommendation than women, while the impact on behaviours seems to have been roughly equal. It seems reasonable that men would benefit more due to their higher disease risk. However, previous research suggests that women tend to comply with NPIs against COVID-19 to a greater extent than men, 25 which we did not find evidence of for this particular policy. People with other risk factors (especially among women) also appeared to be more willing to self-isolate even at younger ages, which could--at least in part--be a consequence of the recommendations aimed at people with other risk factors. Our study adds to the body of knowledge about the effectiveness of NPIs for the control of novel viruses. Previous evidence regarding the effectiveness of social distancing recommendations and stay-athome orders indicates that they were moderately effective in reducing disease transmission during the COVID-19 pandemic, 3,4,26 which is in line with our results. We are not aware of any other empirical studies evaluating the effects of age-specific restrictions or recommendations. Our study therefore provides new insights into how populations may react to age-specific social distancing policies. The notion of higher risks among older people most likely became widespread among the public early during the pandemic, which probably had a general effect on social behaviour across age groups Social distancing recommendations for older adults in Sweden during COVID-19 801 Figure 1 Regression discontinuity plots for the impact of Sweden's age-specific isolation recommendations on social distancing behaviours at the 70-year threshold with binned means (grey dots) and fitted values (black lines) from local linear (A) and quadratic (B) regressions estimated within mean squared error optimal bandwidths around the threshold, for three social distancing measures: (i) mean weekly visits to crowded places, (ii) mean weekly outdoor episodes with no or limited interaction and (iii) mean weekly visits to healthcare providers Note: Additive estimates reflect bias-corrected effects on the difference scale (where 0 1<unk>4 null effect) estimated within mean squared error optimal bandwidths, with 95% Eicker-Huber-White heteroskedasticity-robust confidence intervals from the rdrobust package for Stata in parentheses. Relative estimates reflect ratios (where 1 1<unk>4 null effect) computed using the additive estimates (see Supplementary material for details). Note: Estimates reflect bias-corrected incidence rate differences per 1000 population (IRD, i.e. absolute effects where 0 1<unk>4 null effect) and incidence rate ratios (IRR) (i.e. relative effects where 1 1<unk>4 null effect) estimated within mean squared error optimal bandwidths, with 95% Eicker-Huber-White heteroskedasticity-robust confidence intervals from the rdrobust package for Stata in parentheses. Social distancing recommendations for older adults in Sweden during COVID-19 803 irrespective of recommendations. This notwithstanding, the discontinuities we observe suggest that the age-specific recommendation had an effect in addition to that general effect on from the pandemic and other policies. A modelling study conducted by the Swedish Public Health Agency estimated that the age-specific recommendation prevented between 2100 and 3600 hospitalizations and 750-1312 deaths during March-September 2020. 8 Their study is based on assumptions about the reduction in the number of contacts. Our study provides direct empirical support that the recommendations helped control the outbreak, with impact estimates that are slightly smaller but close to the simulation study results (Supplementary Box S1). The Swedish response to the COVID-19 pandemic was relatively lenient compared to most countries and mainly included nonmandatory recommendations to the public during the first wave of the pandemic. 27 Part of the strategy was to shield vulnerable population groups while keeping society as open as possible. The agespecific recommendation was an important aspect of this strategy, and it is conceivable that the effects are dependent not only on the acceptance among those targeted but also on which other population-level measures (such as limiting the size of gatherings and restrictions directed towards non-essential businesses) that were implemented during the same period. 26 The Swedish public also has high levels of social trust and trust in its government, 28,29 which may have played a role in the success of the age-specific recommendations. 30 However, data from other countries suggest that individual psychological factors (e.g. beliefs about efficacy of the recommendations) may have a larger effect on compliance with NPIs against COVID-19 than institutional trust. 31 Our results should also be interpreted in the light of concerns about adverse effects on mental health. 7,[32][33][34][35][36] In fact, age-specific recommendations were withdrawn in October 2020 due to these concerns. 8 Investigating potential adverse effects is therefore an important avenue for future research. --- Strengths and limitations Our study relied on an RDD, which allows for causal effect estimation in observational data under relatively weak assumptions. 9 Other policies that use the same threshold may, however, bias the results. 19 Sweden had no other relevant policies using a 70-year threshold during the COVID-19 pandemic. The observed discontinuities were also isolated to the expected outcome variables, suggesting causality. The validity of our estimates also depends on appropriate modelling of the age-outcome relationship. We followed the current best practice recommendations, which is to fit simple models (linear or quadratic) within a data-driven bandwidth (age window) around the threshold. 21,22 A typical concern is that the conclusions may depend heavily on the selected bandwidth, 9 but our results are robust to other reasonable bandwidth choices as shown in the Supplementary material. A limitation is that RDD can only be used to estimate effects for persons who are exactly 70 years old. The estimates may not generalize to older parts of the targeted age group, and the calculations in Supplementary Box S1 should, therefore, be interpreted with caution. In addition, while urging older adults to isolate themselves seems to have been a better alternative than encouraging no one to isolate, our data do not permit us to explore what would have happened if the policy had been aimed at a broader age group. A key strength of our study was the availability of detailed and complete register data for severe COVID-19 disease, which most likely limited the extent of outcome misclassification, together with repeated assessment of social distancing during the study period. However, there are some noteworthy limitations to our social distancing data. First, the social distancing measures were selfreported and could therefore be prone to bias if respondents feel pressured to provide a socially acceptable response. 37,38 While the overall levels of the isolation data may be affected, this would only be a problem for the validity of the effect estimates if persons just above 70 years falsely reported greater levels of isolation as a consequence of the policy. Second, participants in the app study were healthier and less disadvantaged than the general population. 10 Thus, the social distancing effect estimates may not generalize to the Swedish population if socioeconomically advantaged groups comply more with non-mandatory recommendations, as suggested by data from Norway and the USA. 39,40 Reassuringly, none of these problems affect the disease outcome data, and the fact that we find an effect in both datasets suggests that our overall conclusions are valid. Another limitation to our study is that social distancing data was only available after 6 May 2020, and thus presents the latter part of the first pandemic wave. Our study was further limited by the selective PCR testing strategy in Sweden during the spring of 2020, which meant that we could not quantify effects on infection rates in absolute terms. Moreover, since we only had access to data on year of birth and lacked data on cohabitation with persons above 70 years, our estimate may suffer from exposure misclassification bias. In both cases, we believe that the misclassification would lead to an underestimation of the true effect. Another limitation was that we could not stratify effects on disease outcomes by medical risk factors, as such register data were not available for the present study. --- Conclusion The age-specific social distancing recommendations appear to have had an additional impact on disease risks and social distancing behaviours beyond the general recommendations that were present at the time. This suggests that non-mandatory social distancing recommendations targeting risk groups may reduce disease transmission during a pandemic, protect against severe disease and save lives. --- Data availability The data supporting this article constitute sensitive personal information and can only be made available to researchers with an approval from the Swedish Ethical Review Authority. Please contact the corresponding author for further details. The analysis code can be found in the Supplementary appendix. --- Supplementary data

Background: This article investigates the impact of a non-mandatory and age-specific social distancing recommendation on isolation behaviours and disease outcomes in Sweden during the first wave of the coronavirus disease 2019 (COVID-19) pandemic (March to July 2020). The policy stated that people aged 70 years or older should avoid crowded places and contact with people outside the household. Methods: We used a regression discontinuity design--in combination with self-reported isolation data from COVID Symptom Study Sweden (n ¼ 96 053; age range: 39-79 years) and national register data (age range: 39-100þ years) on severe COVID-19 disease (hospitalization or death, n ¼ 21 804) and confirmed cases (n ¼ 48 984)--to estimate the effects of the policy. Results: Our primary analyses showed a sharp drop in the weekly number of visits to crowded places (À13%) and severe COVID-19 cases (À16%) at the 70-year threshold. These results imply that the age-specific recommendations prevented approximately 1800-2700 severe COVID-19 cases, depending on model specification. Conclusions: It seems that the non-mandatory, age-specific recommendations helped control COVID-19 disease during the first wave of the pandemic in Sweden, as opposed to not implementing a social distancing policy aimed at older adults. Our study provides empirical data on how populations may react to non-mandatory, age-specific social distancing policies in the face of a novel virus.

INTRODUCTION The positive association between inequality and homicide is well-established (Blau and Blau, 1982;Bailey, 1984;Wilson and Daly, 1997;Daly, 2016) and holds at social scales from communities all the way to countries (Levitt, 1999;Messner et al., 2002;Ouimet, 2012;Rufrancos et al., 2013;Harris and Vermaak, 2015;Daly, 2016;Di Matteo and Petrunia, 2019). As a matter of definitional clarity, we adopt the U.N. definition of homicide as intentional homicide, which is unlawful death purposefully inflicted on a person by another person, excluding unintentional homicides and deaths due to armed conflict (UNODC, 2019). Despite the pervasiveness of the relationship between inequality and homicide, scientific consensus to explain it remains elusive. Kelly (2000) argues that inequality creates stresses and erodes social norms among the poor, making homicide more likely. Using the U.N. Human Development Index (HDI), Ouimet (2012) demonstrates that inequality is associated with homicide in countries with medium to high human development. In impoverished low HDI countries, economic factors such as poverty and inequality interact with high proportion of youth, decreasing the effectiveness of the criminal justice system, which in turn leads to increased homicide. Daly (2016) argues that the concentration of wealth and status among wealthy males causes poorer males to compete more lethally over their dwindling supply of social status. Other researchers have proposed that the association between inequality and homicide is spurious, noting that homicide rates tend to be higher in hotter climates and on hotter days (Cheatwood, 1995;Anderson et al., 1997;Mishra, 2015;Heilmann and Kahn, 2019). As with inequality, there is debate over the causal mechanisms that underly this association (Miles-Novelo and Anderson, 2019). Some researchers argue that it is a function of heat stress, which weakens impulse control (Anderson et al., 1997). For instance, recent research indicates that heat interferes with serotonin reception to reduce inhibition, making homicide more likely (Tiihonen et al., 2017). Alternatively, the routine activity hypothesis proposes that homicide is more common in warm weather simply due to the fact that people intermingle more in warm weather, providing more opportunities for violence, and spend more time outdoors where they are less protected, placing themselves at greater risk (Cheatwood, 1995;Rotton and Cohn, 2003;Miles-Novelo and Anderson, 2019). Coccia (2018) importantly demonstrates that there is a strong correlation between hot climate and inequality that confounds the relationship between homicide, inequality, and temperature. The association between heat and homicide is reinforced at deep historic and prehistoric time scales (Hsiang et al., 2013), and growing climatic warming and associated climate change has created fears that violence will increase (Rotton and Cohn, 2003;Mares and Moffett, 2016;Van Lange et al., 2016;Miles-Novelo and Anderson, 2019). Anderson et al. (1997) pioneering study of the 50 largest U.S. metropolitan areas established a strong positive relationship between temperature and homicide through time. In another early study, Rotton and Cohn (2003) found that temperature was associated with assault, rape, robbery, and burglary through time, but not homicide in a state-level study of the United States, and in a study of U.S. counties, found that temperature had a strong bivariate effect on all major categories of crime. Recent empirical studies have largely corroborated the association of climate change and homicide, but the results are mixed and nuanced. Climate change appears to have an indirect effect on homicide, mediated by other factors. For instance, Barlett et al. (2020) provide a path model based on country-level correlations from 1961 to 2015 that connects global warming to extreme weather events that threaten clean water supplies, which they infer creates resource stresses that motivate homicide. Similarly, Pe<unk>aherrera-Aguirre et al. ( 2019) conducted a 25-year moving average study demonstrating that climate change exacerbates resource competition and inequality, which is mediated through proposed evolutionary influences based on latitude, cultural norms, and climate change. Furthermore, some studies find mixed or no association between temperature and homicide. A study of nine U.S. cities from 2007 to 2017 found that the positive association between temperature and homicide held only for Chicago and New York (Xu et al., 2020). Finally, a study of New York and London covering the years 1895-2015 found that including per capita GDP in a statistical model eliminated the temperature effect on homicide, and correcting for serial autocorrelation eliminated all relationships (Lynch et al., 2020). These recent studies indicate that further research is necessary to establish a clear link between climate change, notably global warming, and increased homicide. In this paper, the relationships between homicide, inequality, inequality-driven risk sensitivity, and temperature are explored on a global country-level data set from 1960 to 2019. Inequality is examined with several different metrics. The Gini coefficient measures overall patterns of inequality, the percent population below the poverty line measures the effects of absolute poverty, and a new measure of inequality based on wealth and status distributions provides an examination of these relationships for poor, middle class, and wealthy segments of society. The fundamental finding is that inequality is the prime driver of homicide rather than temperature, although inequality-driven risk sensitivities of poor, middle-class, and wealthy segments of society interact with temperature. These interactions have potential policy implications and deserve further scrutiny. --- INEQUALITY, SOCIAL STATUS, AND RISK SENSITIVITY The association of inequality and homicide suggests further investigation regarding how inequality and class differences are related to lethal violence. Friedman and Savage (1948) suggested that unequal distributions of wealth and social status impact individuals' utility functions (satisfaction derived from status), which in turn influence an individual's willingness to take or avoid risks. Despite their focus on private individual subjective utility functions, their copious footnotes, almost equal to the text of the article, provide data and arguments suggesting that the publicly observable distribution of wealth influences individuals' utility functions and sensitivity to risk. The use of violence, especially among peer-competitors who have the same access to the means of violence, is by definition highly risky behavior and so should be influenced by wealth and status differences (Wilson and Daly, 1997;Kuznar, 2007;Daly, 2016). Inequality is measured many ways including percentage of wealth owned by the top x percent, percent of a population living in poverty, or the commonly used Gini Coefficient, which is the difference between the Lorenz curve, defined by percent wealth of each percentile of a society, and the line of total equality, in which each percentile of society shares equally in society's wealth (see Kovacevic, 2010 for a full explanation). Each of these measures provides insight into how wealth is distributed in a society, but each obscures variation in inequality between classes. An alternative way of describing inequality records the wealth of each percentile of society against the rank of each percentile in wealth, as suggested by Friedman and Savage (1948). This reveals abrupt increases of wealth as one moves from the poorest to the wealthiest ranks in a society; wealth class boundaries are defined by relatively sharp increases in the curve. Figure 1 represents a wealth distribution curve typical of most societies; it has a low tail for the very poor, followed by a sharp increase in wealth that is fairly level and defines a middle class, which is then followed by an extremely sharp increase that continues to the wealthiest individuals in a society. Mathematically, this curve has an initially concave upward section (the poor), followed by a concave downward segment (a middle class), followed by a strongly concave upward section (the wealthy). The curve reflects the fact that wealth is typically concentrated at the top of most societies, and research demonstrates that this pattern is found in societies as varied as small tribes to ancient kingdoms, modern states, and even the entire world economy (Kuznar, 2001(Kuznar,, 2002(Kuznar,, 2007;;Lewis, 2004). The practical utility of wealth is obvious; it can be used to purchase goods and services people need and desire. However, wealth also signals social status. Tokens such as metal or shell armbands in ancient and current tribal societies are classic examples (Mauss, 1967;Malinowski, 1985). Industrial societies are no different. Conspicuous consumption (Veblen, 1994) by the wealthy is abundantly present in modern society; they build elaborate mansions and buy luxury cars whose cost far exceeds what is necessary to satisfy basic needs of shelter and transportation. The wealthy are not the only people interested in tokens of status. A study of social media discussions among the U.S. general public found a positive correlation between inequality and a desire for status goods (Walasek and Brown, 2015). Wealth has much greater significance than purchasing power, it signals one's position in society and consequently one's social worth and status. Wealth has material and social value and therefore one would expect people to compete for it. However, not everyone is equally motivated to compete. People whose utility functions are concave downward are expected to avoid risk and competition because more status can be lost vs. what could be gained; risk aversion is a widespread human tendency (Bernoulli, 1954;Cashdan, 1985). One would expect people to accept risks to achieve status when potential gains can exceed potential losses, represented by the concave upward sections of Figure 1 (Friedman and Savage, 1948;Markowitz, 1952). This pattern of risk-taking behavior has been confirmed across an incredibly wide array of cultures including hunting and gathering bands, tribes, ancient kingdoms and modern states (Pryor, 1976;Kuznar, 2001Kuznar,, 2002)). Risk taking to gain status can take many forms. Legal and socially accepted forms of risk taking include investing in the stock market or starting a legal business. However, people may engage in unsanctioned or illicit forms of risk taking as well. Violently challenging rivals for status is by definition risky. For instance, political science research has identified relative differences in wealth as a core motive for lethal political risk taking, including revolutions (Gurr and Moore, 1997;Besancon, 2005), terrorism (O'Neill, 2000;Kuznar, 2007;Kuznar and Lutz, 2007), and mass protest movements (Midlarsky, 1988). Pratt (1964) provided a measure of risk sensitivity for an individual at different levels of wealth. Subsequent work by Arrow (1974) reinforced his research and it is known as the Arrow-Pratt measure of risk aversion. The measure is calculated as: Equation 1. Arrow-Pratt Measure of Risk Aversion. --- METHODS AND DATA r (x) =- U(x) ′′ U(x) ′, where U(x) is a utility function that measures satisfaction for differing levels of wealth, x. The measure can range from -<unk> to +<unk>; negative values indicate the degree of risk acceptance and positive values indicate the degree of risk aversion. To the extent that wealth is a measure of social status, the distribution of wealth therefore creates a function for the utility of social status, which was implied in the original proposition for measuring utility by Friedman and Savage (1948). Therefore, by fitting a curve to a wealth distribution, one can measure wealth's utility for conveying social status, and the Arrow-Pratt measure can be applied to determine the risk sensitivity of an individual at any level of wealth. The function fitted to a wealth distribution is called the expo-sigmoid function (Kuznar, 2007) because typical wealth distributions are generally exponential (wealth concentrates at the top), but exhibit sigmoid (S-shaped) oscillations that define wealth classes (Figure 1). The function is: Equation 2. Expo-sigmoid utility function. S rank =e a+b(rank)+csin(rank )+dcos(rank), where S is the expo-sigmoid fit to the wealth distribution, and rank is the rank from poorest to wealthiest in the society. A full description of the curve fitting method is presented in Kuznar (2007). The World Bank provides publicly available data on homicide, Gini coefficients, percent population below the poverty line (a measure of absolute poverty), and wealth distributions for all countries from 1960 to 2019, enabling measurement of wealth distributions and Arrow-Pratt measures and testing the relationships between all of these variables. These data were used to create wealth distribution curves for the 173 countries with 2019 populations over 300,000, which excludes small tropical Island states who are often outliers in a variety of social measures. Expo-sigmoid curves were fit to their wealth distributions and Arrow-Pratt measures were calculated for each centile. The Arrow-Pratt measure was averaged over these centiles, providing an overall measure of a country's risk sensitivity. Nearly every country exhibited an upward concave poor class, a concave downward middle class, and the wealthy were uniformly strongly concave upward in every case. In order to provide a finergrained analysis of risk sensitivity and homicide in a society, the average Arrow-Pratt measure was calculated for each of these population segments. The World Bank provides homicide rates from the U.N. Office on Drugs and Crime International Statistics, which is recognized as a source of reliable homicide data (Ouimet, 2012). The distribution of the homicide rate by country is heavily skewed toward the lower end, consequently we used its natural log to make it more "normal" and thus more appropriate for parametric statistical analysis. Due to missing data, the sample consisted of 679 country-years. Data imputation methods were employed to fill in missing data, but the resulting analyses provided the same qualitative results as the raw data. Therefore, only the original 679 observations were used in order to remain as true to the original data as possible. Countries have different histories of homicide due to serial autocorrelation (rates in 1 year tend to carry over to the next), and varying levels of social control, policing, and cultural norms surrounding violence. The logged homicide data show breakpoints between low homicide countries (LnHomicide <unk> -0.23, raw homicide rate <unk> 0.8/100,000, number of cases = 124), high homicide countries (LnHomicide > 2.46, raw homicide rate > 11.7/100,000, number of cases = 98), and medium homicide countries in-between (number of cases = 457). These countryspecific effects should be taken into account; panel regression is a common method that takes into account within-group variance on key variables within a linear model (Wooldridge, 2010). Therefore, panel regressions were conducted using panels of low, medium, and high homicide countries. Because the inequality metrics were highly correlated (Table 1), separate models were run for each of the inequality metrics (Gini coefficient, absolute poverty, and the risk sensitivity measures) while controlling for temperature. This allowed each of the inequality metrics to compete statistically with temperature in the association with homicide, avoiding multicollinearity between the inequality metrics and providing insights into (a) the relative importance of inequality and temperature, and (b) varied insights provided by the different inequality measures. --- RESULTS Table 1 presents the raw Pearson correlations between the independent variables based on inequality, poverty, and temperature, and the dependent variable homicide. These data confirm the first order effects of each of the independent variables on homicide. However, the correlations between the independent variables are very high, requiring an account of the relative effects of each independent variable in relation to one another. Ordinary least squares models were employed, using standardized coefficients to examine the statistical significance and relative strengths of the relations between homicide and the inequality, controlling for temperature (Table 2). Separate models were run for each inequality or poverty measure, and always included temperature in order to test for the relative effects of inequality vs. temperature. The Gini and Absolute Poverty models have the highest adjusted R 2 values (0.114, 0.113, respectively), indicating that they perform best. They demonstrate highly statistically significant and substantially higher standardized coefficients for inequality than temperature; absolute poverty has over four times the effect on homicide and the Gini coefficient is 50 times more associated with homicide than temperature. By these measures of inequality, temperature loses its association with homicide world-wide, and inequality appears to be the real driver. Average risk sensitivity is also more associated with homicide than temperature, about twice as much so, but the independent effect of temperature is also statistically significantly related to homicide, although the model explains less than half the variance of the Gini and absolute poverty models. The value of taking a risk sensitivity approach emerges when examining how inequality and temperature operate within classes. The risk sensitivity of the poor is not statistically related to homicide rates, but temperature statistically is and by three times as much as risk sensitivity. The opposite relation holds for the middle class; as they become more risk acceptant homicide rates increase strongly (4.4 times stronger) and temperature is not statistically related. The wealthy present a counter-intuitive result for inequality. As their risk acceptance decreases, homicide rates increase, and the effect of temperature is slightly stronger. --- DISCUSSION Overall, in this study temperature performs poorly when it has to compete with measures of inequality for explaining homicide as previously demonstrated by Coccia (2018). However, the risk sensitivity measure, disaggregated by social class, provides insights into the conditions when temperature may be an important factor. The model based on the risk sensitivity of the poor shows no statistically significant association between homicide and their risk sensitivity, but a strong association with temperature. This indicates that relative poverty may expose people more to the effects of temperature, which can lead to homicidal behavior. For instance, Cheatwood (1995) notes that access to technological means of mitigating heat may lower the likelihood for committing homicide, the very means the poor typically lack. Furthermore, Heilmann and Kahn (2019) argue that extreme heat decreases policing efforts and therefore social control mechanisms in poor neighborhoods, which contributes to higher homicide rates in impoverished areas. Contrasting the results for the poor, homicide rates are strongly influenced by increased risk acceptance among the middle-class but temperature has no discernable effect. This could be due to the fact that middle class people most likely have means with which to mitigate the effects of increased temperature, leaving risk acceptance as the dominant influence on homicide rates. Additionally, a strongly risk averse middle class may lobby for and support more aggressive social control in order to protect themselves and their assets from violence, and as their risk aversion decreases they may be less supportive of these measures, in turn raising homicide rates. The relationships between homicide, inequality, and temperature among the wealthy are more challenging to explain. As the wealthy become more risk averse and as temperature rises, homicide rates increase. This may be a function of the fact that poorer countries tend to be located in hotter regions of the world (Coccia, 2018). It is possible that the poorer a country is, the less wealth differences exist between the wealth, which would decrease their risk acceptance. As demonstrated in the risk acceptant poor model, the poor are particularly susceptible to the effects of heat because of their absolute poverty, raising They argue that exposure to heat stress lowers the economic productivity of the poor whereas the wealthy are able to move to cooler parts of the country, exacerbating wealth differences in developing countries. However, decreased productivity of laborers decreases overall GDP for that country. In all cases, researchers are careful to stress that they can comment only on associations since the specific causal mechanisms are as yet unknown. However, these results reinforce the associations between poverty and homicide in warmer regions of the world, regardless of the risk sensitivity of the wealthy. This research points to several implications for homicide reduction. First, simple equations between homicide and inequality or temperature are not nuanced enough to support effective policy decisions concerning homicide. Inequality is the overarching dominant influencer of homicide, but the impact of temperature appears to be related to class. The data presented in this article indicate that, contrary to Daly (2016), if the poor risk violence to get ahead they will neither gain nor lose much in social status. Therefore, the poor have little rational reason to use violence to get ahead because the cards are stacked against them. Increased temperatures, however, appear to have a disproportional impact on the poor, leading to increased interpersonal homicide. Addressing their basic needs for shelter and relief from heat may a more effective way to decrease homicide rates. The middle class is typically the most peaceful segment of society; they fear loss and so are less likely to take risks and compete, especially in violent ways. However, this study's results indicate that when their status is threatened, that is when their risk aversion decreases, homicide rates can strongly increase. This result invokes loss aversion, in which the fear of loss switches a normally risk averse decision maker to one acceptant of risk taking (Kahneman and Tversky, 2000). Whether this is because they become more violent or because their support for social control decreases with their risk acceptance is unclear and requires further study. Finally, the wealthy present a challenging case. The patterns between wealthy risk sensitivity and homicide appear to represent a complex interaction between their level of risk sensitivity and its impact on middle and lower classes. The results of this study suggest further research. Exactly who kills whom is unclear in available country-level statistics. Data on the class of perpetrators and victims is needed to test the relationships between class, risk sensitivity, and homicide suggested by this analysis. Furthermore, the indirect mechanisms that might link risk sensitivity levels in one class to homicides in another need to be explored. The relationship between elite inequality and homicide also needs to be explored further. People of means not only have alternate and non-violent means of competing, but they also compete in lethal ways that are not classified as interpersonal homicide, as in coups and leading rebellions (Brinton, 1964;Braithwaite et al., 2019), which disproportionately take place in warm weather countries of the developing world. Therefore, research into the linkages between inequality, temperature, and political violence should also be explored. Finally, much work needs to be done to explore other causal factors that operate on the country level. Inequality and temperature are clearly pieces of the puzzle, but not the whole picture. --- DATA AVAILABILITY STATEMENT Publicly available datasets were analyzed in this study. This data can be found at: Harvard Dataverse https://doi.org/10.7910/ DVN/OWYI1B. --- AUTHOR CONTRIBUTIONS LK was the primary author of this article and developed the theory and methodological approach. JD provided technical support for the data analysis. All authors contributed to the article and approved the submitted version. --- Conflict of Interest: Both authors were employed by the company NSI, Inc. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Furthermore, the views of the authors do not represent the official views of either Purdue University Fort Wayne or NSI, Inc. Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Researchers debate the causal connections between homicide, inequality, and temperature. This study examines these relationships globally based on country-level data. A new measure of inequality is introduced that provides a more granular measure of inequality patterns than commonly used metrics. The approach allows estimation of risk sensitive decision-making that helps to explain how class impacts violence under different climate conditions. The results indicate that homicide rates are higher when poorer segments of populations are disproportionately influenced by temperature, middle class segments are influenced by inequality, and the wealthy are influenced by middle and impoverished class dynamics.

The word Kol is derived from the Mundari word, 'ko' which means 'they and others' 16. They are mainly concentrated in Central India and regions of Deccan plateau (Fig. 1A). Kols claim themselves to be descendants of epic Ramayana character Savari or Sheori, calling her "Mother of all Kols", and also believe they once inhabited the hills of Rajasthan with another prominent tribe Bhils and helped Rana Pratap, Rajput King of Mewar Rajasthan, in his struggle with the Mughal invaders 17. The linguistic association of Kol is conflicting 6,11,12,16, therefore we undertook this study to dissect a fine-grained genetic structure of them. We used large number of autosomal and mitochondrial DNA markers to investigate the incompatible association of Kols as well as their inter and intra population affinities (Supplementary Tables 1 and2). --- Results and Discussion Caste and tribal affinities in South Asia are factors known to have played a vital role in shaping the genetic landscape of the subcontinent 4,18,19. In our attempt to understand this genetic complexity, we have assessed the ancestry and geneflow pattern of the major tribal populations of South Asia 7,20,21. In present study we evaluated the genetic affinities of the Kol population, which, as the third largest tribal population of South Asia, comprises <unk>1.7 million people (Fig. 1A). In conducting our genetic study, we first ascertained the classical ethnographic work, which has suggested Kol as an Austroasiatic (Munda) speaker 17. As observed previously, the Austroasiatic speakers in India fall out of the South Asian cline due to their Southeast Asian genetic affinity [22][23][24][25]. Therefore, we expected to see their (Kol) clustering with the Munda speakers. However, in the principal component analysis (PCA), both of the studied Kol groups aligned along the South Asian cline with clusters formed by a large number of Indo-European and a few Dravidian speakers (Fig. 1B). Although, the Kols are geographically immediate neighbours of Mundari and Transitional populations, they remarkably exhibit no attraction towards Austroasiatic or Transitional populations (Fig. 1B). At the intra-population level, both of the Kol groups were distinct from each other, suggesting their long-term separation or a possibility of assimilation of different neighbouring tribal groups into a single ethnolinguistic unit called Kol. More specifically, we see three sub-clusters in the vicinity of both of the Kol groups (Fig. 1B). Kol1 and Kol2 fall in the subclusters1 and 2 respectively. Kol1 falls in the subcluster1 with Meghwal, Kurmi, Dharkars, Kanjars (Indian Indo-European), and Lambadi (Dravidian) populations, whilst Kol2 was found to be in-between subclusters 2 and 3 harbouring Dravidian (Sakilli and North Kannadi) and Indo-European (Harijan) populations (Fig. 1B). It is noteworthy, that both of the Kol groups largely share a closer genetic relationship with the majority of the Scheduled caste populations living to their north, speaking Indo-European languages. In order to understand the genetic component sharing of Kol with the other Indian populations, we have plotted various ancestry components inferred from ADMIXTURE analysis (Supplementary Fig. 1). The log-likelihood estimate was in favour of best K value as K = 12 (Supplementary Fig. 2). Apart from two major components prevalent in South Asia, we also see other minor and population-specific ancestry components (Fig. 2). The majority of these minor components were either sporadic or present among some specific language groups 5,8,25 e.g. the Southeast/East Asian components among Mundari and Tibeto-Burman speakers 5,25. However, we also see a South Indian component which was nearly fixed in Irula and is geographically widespread amongst other South Asian populations with a frequency gradient from east to west or south to north (Fig. 2). Amongst both of the Kol groups, all these three components (two major and one minor) were substantially visible. Except for a single sample, none of the Kol individuals showed any East/Southeast Asian specific component significantly (two tailed p value <unk> 0.001), which is otherwise abundant among their geographic and linguistic neighbours (Transitional and Mundari speaking populations). This finding ruled out their recent common ancestry with the Austroasiatic (Mundari) speakers. Hence, together with the PCA, ADMIXTURE analysis also suggested a non-Austroasiatic connection of these 'Kol' groups. We further investigated one outlier sample of Kol which showed high level of East/Southeast Asian ancestry. In the PC analysis, this Kol individual (Kol outlier) aligned along the Trans-Himalayan cline 5 (Fig. 1B). In terms of population-wise affinity, this individual clustered with the Tharu population of Uttarakhand. In the ADMIXTURE plot (Fig. 2), this individual also showed Tharu like ancestry pattern, confirming the PC analysis result. We retraced our steps from sampling to genotyping of this particular sample, and learnt that the Kol samples were processed in the lab together with the Tharus, and it is likely that one of the 'Tharu' sample was mislabelled as 'Kol'. For further population based analysis (f3 and D statistics) we omitted this sample from the pool. For shared drift analysis of Kol groups, we performed the outgroup f3 test (Supplementary Fig. 3). The result was consistent with the PCA in terms of their closer affinity with extant South Asian populations (Fig. 1B). Both of the Kol groups showed a significant level of allele sharing with other South Asian populations, particularly with Harijans. Populations who were closer to the Kols in the PCA also showed higher shared drift with the Kols. When we compared the alleles shared with East vs. West Eurasian populations, we observed an inverse affinity of Kol1 vs. Kol2 with the East and West Eurasian populations. Kol1 shared more drift with the West Eurasians, whereas Kol2 shared greater drift with the East Eurasians (Supplementary Fig. 3). In the allele frequency based analysis, the Kols exhibited a closer genetic affinity with the Indo-European scheduled castes and tribal populations, rather than with Austroasiatic or Dravidian populations (Figs. 1B and2 and Supplementary Figs. 1 and3). To gain a deeper insight into the extent of genome sharing between the Kols and other South Asian populations, we applied haplotype-based ChromoPainter 26 and fineSTRUCTURE analysis 26. On the basis of haplotype sharing amongst the studied groups, we compared the mean chunk counts donated by Eurasian populations with Kol groups (Fig. 3). As expected, Kols received majority of the chunks from South Asian populations when compared with other Eurasians. Amongst the South Asians, the Indo-European scheduled caste population Harijan was the major chunk (chunklength as well as chunkcounts) contributor for both of the Kol groups (Fig. 3). The chunk donation of Austroasiatic (Mundari) populations was significantly lower (two tailed p value <unk> 0.0001). The distinct ancestry of one Kol sample can be also seen in this analysis. The Maximum Likelihood (ML) tree obtained from the fineSTRUCTURE analysis placed both of the Kols together with the Indo-European populations (Supplementary Fig. 4). Kol1 and Kol2 fell in to two distinct clusters. Together with other populations, Kol1 is distributed in to two sub-clusters, whereas Kol2 form their five largely own sub-clusters, where one was shared with the Harijans (Supplementary Fig. 4). To see, if the Kol1 and Kol2 belong to same pan-Kol ancestry, we computed D statistics asking if there is any population which share more alleles with either of these (Table 1). When we filtered the top 10 D values for Kol populations, we didn't find any population which shared significantly more alleles than Kol1 shares with Kol2. Thus both of the Kol groups share a more recent common ancestry. To investigate further the inbreeding and relatedness among both the Kol groups, we analysed Runs of Homozygosity (RoH) in the populations [27][28][29] (Supplementary Fig. 5). In an inbred populations RoH tend to be longer and recent in time as recombination doesn't get enough time to break the identical-by-descent segments. Conversely, shorter RoH segments are considered to be older. Both of the Kol groups showed lower RoH segments when compared with the Austroasiatic (Mundari) speaking populations, suggesting their different population history as well as high effective population size (Ne). In order to gain information about their maternal ancestry sharing, we analysed mitochondrial DNA (mtDNA) sequences of the HVS-I (hypervariable segment I) and selected coding regions. Both of the Kol groups shared M2, M3, M18, M30 and R5 haplogroups (Supplementary Fig. 6 and Supplementary Table 2). Our previous study has identified haplogroup R7 as highly frequent haplogroup among North Mundari speakers 30. However, we didn't find any sample of Kol belonging to haplogroup R7 (Supplementary Table 2). The mtDNA haplogroups of Kols were quite distinct from the general trend of Mundari populations 10,22,30,31. We utilised haplogroup frequencies to calculate the principal components. We have used geographic labels in one plot and linguistic labels in another plot (Supplementary Fig. 7). In the geographical placement, the pattern followed the isolation-by-distance model. The Uttar Pradesh/Madhya Pradesh Kol (Kol1) clustered with Uttar Pradesh and Madhya Pradesh populations, whereas Maharashtra Kol (Kol2) clustered with the neighbouring Andhra Pradesh populations (Supplementary Fig. 7a). In terms of linguistic affiliation, Kol1 clustered closely with populations speaking Indo-European languages, whereas Kol2 cluster with Andhra Pradesh Dravidian speakers (Supplementary Fig. 7b). Therefore, their maternal ancestry also precludes their Austroasiatic (Munda) affinity. Yet, previous studies have identified the Austroasiatic language communities of South Asia as the result of a gender biased linguistic intrusion, with resulted from the spread of the language by male speakers who introduced the predominant Munda paternal lineage along with a small but recognisable Southeast Asian autosomal component 26. However, because of the absence of Y chromosomal haplogroup information from the Kol groups, we are unable to test their paternal affiliation. In conclusion, contrary to what is suggested by their name, we found no recent common genetic ancestry of these two Kol groups with the Austroasiatic (Mundari) speakers. The genetic structure of these Kols is more akin to the North Indian Indo-European scheduled caste population known as the Harijan. This finding matches our recent finding that Harijans and Kols shared short IBD (identical by descent) segments with Indian Mundari speakers 25 rejecting any recent geneflow or common ancestry. Our analysis also discards a case of recent language shift, as none of the Kol carried the signal of Southeast Asian ancestry that is present in Austroasiatic (Mundari) populations. Thus, our detailed analysis on one of the major South Asian tribal populations, support a deeply rooted endogamy, which not only exist among caste populations, but also present among tribal populations. Particularly in this case, our sampled Kols lived side-by-side with the Mundari populations. Our finding leaves us with the question as to whether the sampled 'Kol' populations could represent the remnant of ancestral Kol before the ancestors of Munda were linguistically assimilated by incursive Austroasiatic speakers. Since antiquity and even in modern times, in the social climbing process, entire ethnic groups and language communities have been known to pass themselves off as another caste or linguistic group that happens to rank higher in the caste hierarchy 26. The present study presents what appears to be the first genetic evidence for such a collective ethnolinguistic identity reassignment. --- Materials and Methods To sample Kol population, in the first phase, we surveyed 566 individuals from 12 villages covering three major states of their settlement (Uttar Pradesh, Madhya Pradesh and Maharashtra). It was striking that, in our survey to the sampling regions (Fig. 1A), we did not find a single Kol individual, speaking or having knowledge of Mundari languages. All of the individuals surveyed were fluent in the local Indo-Aryan languages instead, i.e. Bhojpuri-Bagheli in Uttar Pradesh and Madhya Pradesh, Marathi in Maharashtra. Since all early anthropological and linguistic studies on Kols unanimously established their linguistic affinity as speaking Ho or other languages of the Kherwarian cluster within the Northern Branch of the Munda subgroup within Austroasiatic [7][8][9], in the case of the linguistically assimilated young Kols whom we sampled, we double-checked their ethnolinguistic identity with linguistic expert involved in the study. Since language shift has previously been reported amongst Central Indian tribes 10,32, we presume that this is also the case with the Kols sampled in the present study. However, we note that a similar model did not appear to apply to the Gond in our previous studies 7,20. Therefore, in this study we used large number of autosomal and mitochondrial DNA markers to investigate the conflicting association of Kols as well as their inter and intra population affinities (Supplementary Tables 1 and2). We finally collected blood samples of the Kol population from 55 unrelated individuals with informed consent. We avoided people related up to three generations. The first group of Kol (Kol1) was sampled from the geographic borders of Uttar Pradesh and Madhya Pradesh states and second group (Kol2) was collected from Maharashtra state (Fig. 1A). Both of these sampling points were from the places where the Kol are highly concentrated. The DNA was isolated and quantified from standard protocol 33. We further selected 17 high-quality samples (seven Kol1 and ten Kol2) and generated Illumina 650 K genotype data. This data was released in our earlier publication 23. All the 55 samples were sequenced for the mtDNA HVS-I region (Supplementary Table 2). We first classified them in their tentative haplogroups, based on the HVS-I mutation and further confirmed these findings by genotyping for coding region mutations (Supplementary Table 2). This study was approved by the ethical committee of the Banaras Hindu University, Varanasi, India. All methods were performed in accordance with the relevant guidelines and regulations. For autosomal data we used PLINK1.9 34 for quality control and data management. We merged the data of the 17 Kol samples with the 1756 samples belonging to 119 world populations (Supplementary Table 1). Similar to our previous studies, SNPs with more than 3% missingness across individuals or with a minor allele frequency less than 10% were removed 23,35. We have also removed SNPs deviating from Hardy-Weinberg equilibrium 36. After all quality control measures, we obtained 258311 high quality SNPs, which we used for all our analyses. We classified Indian populations according to their language group. For the populations having conflicted linguistic affiliation, we followed Kumar and Reddy 32 and classified them as 'Transitional'. To remove background linkage disequilibrium (LD) that can affect both principal component analysis (PCA) and ADMIXTURE, we thinned the data set by removing one SNP of any pair in strong LD r2 > 0.4, in a window of 200 SNPs (sliding the window by 25 SNPs at a time). We performed PC analysis using the smartpca programme of the EIGENSOFT package 37 with the default settings to capture genetic variability described by the first ten components. We ran unsupervised ADMIXTURE v1.3 38 with a random seed number generator on the LD-pruned data set 25 times from K = 2 to K = 15. The best supported clustering was shown at K = 12 21,23. Given the result of the PC and ADMIXTURE analysis, we removed one outlier sample from the Kol2 group for further population-based analysis. The outgroup f3 statistics 39 were calculated as f3 = (Kol1/Kol2, X; Yoruba), where X was any other population and Yoruba served as an outgroup. To investigate the pan-Kol ancestry, we performed D statistics by taking African Yoruba as an outlier D = (Yoruba, Kol1; Kol2, X), whereby X was the other Indian populations. For haplotype-based comparison ChromoPainter v1 26 and fineSTRUCTURE v1 26 were used to perform an MCMC iteration, using 10 M burning runtime and the same MCMC iterations. We first phased our samples with Beagle 3.3.2 40 and modelled haplotype sharing among studied individuals by using ChromoPainter. The ChromoPainter creates a co-ancestry matrix where each and every individual share chunkcounts and chunklength with each other 26. Thereafter, fin-eSTRCUTURE algorithm cluster them in to subgroups based on the pattern of co-ancestry matrix. The data of fineSTRUCTURE were used to construct the maximum likelihood (ML) tree using MEGA 41. Runs of homozygosity (RoH) were performed to investigate the inbreeding and ancestral homozygous component sharing. For RoH estimation, we applied window size 1,000 kb, a minimum of 100 SNPs per window allowing one heterozygous and five missing calls per window 27 --- Author contributions --- competing interests The authors declare no competing interests.

Both classical and recent genetic studies have unanimously concluded that the genetic landscape of South Asia is unique. At long distances the 'isolation-by-distance' model appears to correspond well with the genetic data, whereas at short distances several other factors, including the caste, have been shown to be strong determinant factors. In addition with these, tribal populations speaking various languages add yet another layer of genetic complexity. The Kol are the third most populous tribal population in India, comprising communities speaking Austroasiatic languages of the Northern Munda branch. Yet, the Kol have not hitherto undergone in-depth genetic analysis. In the present study, we have analysed two Kol groups of central and western India for hundreds thousands of autosomal and several mitochondrial DNA makers to infer their fine genetic structure and affinities to other Eurasian populations. In contrast, with their known linguistic affinity, the Kol share their more recent common ancestry with the Indo-European and Dravidian speaking populations. The geographic-genetic neighbour tests at both the temporal and spatial levels have suggested some degree of excess allele sharing of Kol1 with Kol2, thereby indicating their common stock. Our extensive analysis on the Kol ethnic group shows South Asia to be a living genetics lab, where real-time tests can be performed on existing hypotheses. The Indian subcontinent is renowned for the cultural, linguistic and genetic diversity of its inhabitants 1,2 . This diversity has mainly arisen, in part, through long term human settlement, social customs and genetic drift 3- 5 . Broadly, Indian populations can be categorised as the castes, tribes, linguistic and religious communities. Presently, India counts hundreds of tribal groups, belonging to four major language families; Austroasiatic, Dravidian, Indo-European and Tibeto-Burman 6,7 . Kol is one of them, with their major concentration in Central India (Fig. 1A). Kol is another name for Ho, whose language is a member of the Kherwarian cluster within the Northern Branch of the Munda subgroup of Austroasiatic language family 7-9 . In fact, the language family came to be known as 'Mon-Khmer-Kolarian' when Francis Mason first identified that Kol and the other Munda languages were related to the Mon language of eastern Burma and Thailand in 1854. He suggested that these Munda or 'Kolarian' languages of India and the 'Mon-Annam' languages of Southeast Asia, collectively belonged to one and the same language family 10 . The language family was given its current name ' Austroasiatic' in 1904 by Wilhelm Schmidt 11-15 .

of the French telecommunications operator. Our results are presented in the form of a narrative (Ewick & Silbey, 1995, 2003) that renders the polyphonic and dialogic dimensions of the processes under study (Belova, King, & Sliwa, 2008;Humphreys & Brown, 2002). On the basis of in-depth interviews with key actors of the resistance, a wealth of secondary data, and the detailed reconstruction of the chronology of events that unfolded during the crisis, we produce a narrative of hegemonic transformation in three phases: (a) the rise of a new "hegemonic despotism" steering workers' resistance within the firm, (b) the emergence of a counter-hegemony reaching out to the broader civil society, and (c) hegemonic transformation via state intervention. Using this approach, our contribution to the literature is twofold. First, our narrative enriches the literature on resistance by uncovering the "cascade effect" (Levy & Egan, 2003) through which workers' resistance can escalate to reach civil society and the state apparatus and produce hegemonic transformation in the country. Second, we contribute to neo-Gramscian studies on hegemonic transformation by showing the role of a "hybrid space" through which resisters produced this cascade effect. In our neo-Gramscian perspective, a "hybrid space" is formed of micro-and meso-organizational processes that enable resisters to share, develop, and leverage both discursive and material resources across the three main sites of hegemony, that is, the firm, civil society, and the State. Inspired by Moje et al.'s (2004) work in educational studies, this notion serves to capture the cross-institutional dynamics that give resistance the potential to become broadly transformative. The rest of the article is organized as follows. The first section places our work at the intersection of the literature on workplace and civil society struggles and discusses how the concept of hybrid space could bridge these two literatures in a neo-Gramscian perspective. The second section provides an explanation of our methodology. Empirical results unfold as a narrative of hegemonic transformation in the third section, followed by an exploration of the role of the hybrid space of resistance in this cascade effect. We discuss these results by elaborating on the capacity of hybrid spaces of resistance to induce systemic transformations in the contemporary hegemony, before offering a brief conclusion. --- A Neo-Gramscian Reading of Resistance Critical studies of resistance in organizations have emerged from the observation of social struggles at the workplace, most notably in the Marxist tradition of labor process theory. Resistance was understood as an attempt by workers to regain control and autonomy in the production process, against a background of inherent antagonism between labor and capital (Burawoy, 1979;Knights & Willmott, 1990). Various forms of resistance were studied, from collective upsurges and the rise of the labor movement to informal tactics to subvert managerial rules. However, the key tenets of labor process theory were criticized for offering an overdeterministic account of social struggles at the workplace. Some critics argued that the possibilities for workers' emancipation had been excessively downplayed since resistance ultimately sustained capitalist relations of domination, whereas others pointed out that more subjective forms of resistance had not been acknowledged (Courpasson, Dany, & Clegg, 2012;Mumby, 1997). An alternative line of research subsequently drew on the work of Foucault and others to explore the discursive construction of subjectivities at the workplace and the micro-politics through which employees might be able to resist managerially imposed subjectivities, including outright rejection, feigned acceptance, or pragmatic negotiation (Knights & MacCabe, 2000;Thomas & Davies, 2005). The ascendance of such "radical pluralism" was in turn critiqued for celebrating "localized and heterogeneous struggles" (Carrol & Ratner, 1994, p.7), or a form of "anemic" (Mumby, 1997), "decaf" (Contu, 2008) resistance reduced to the "clever tricks of the weak within the order established by the strong" (De Certeau, 1984, p. 40). A number of recent contributions have sought to overcome such limitations by arguing that a combination of practices, both hidden and public, individual and collective, could allow resisters to effectively counteract managerial power, thus enabling "productive" (Courpasson et al., 2012;Courpasson, Dany, & Delbridge, 2017) or "impactful" (Courpasson, 2016) resistance to reverse managerial decisions. However, focusing on the workplace meant that the role of external civil society and state actors, although acknowledged for tilting the balance of power between workers and the firm in empirical accounts of the struggles, was not theorized as a lever of successful resistance. Hence, a broader theoretical perspective is needed to embrace the capacity of workers' resistance to produce outcomes not just within but also beyond the firm, by engaging civil society and the State in a broader process of hegemonic transformation. Several features of a neo-Gramscian perspective suggest that it may be suitable to such an enlarged reading of resistance. First, Gramsci (1971) conceived of a social order or "hegemony" as spanning a "historical bloc" allying "the coercive and bureaucratic authority of the State, dominance in the economic realm, and the consensual legitimacy of civil society" (Levy & Egan, 2003, p. 806). Second, neo-Gramscian hegemony refers to a "process of struggle rather than an existing state of consensual domination" (Mumby, 1997, p. 365), whereby social actors draw on discursive, material, and organizational resources in ways that are constrained but not predetermined by established forms of power, thus opening spaces for contestation and hegemonic transformation under the influence of resisting agents (Levy & Egan, 2003). The concepts of hegemony/counter-hegemony are both "sensitive to the material moment of practice, yet also inclined toward the discursive issues that surround the securing of consent and the advancement of an alternative vision" (Carrol & Ratner, 1994, p. 7). Our choice of a neo-Gramscian approach is thus justified by the possibility it offers of reading power and resistance as dialectical dynamics spanning the firm, the State, and civil society, while also acknowledging the specific historical conditions of the hegemony to be resisted. --- Workplace Struggles Against Neoliberal Hegemony Key tenets of a neo-Gramscian perspective have been mobilized at the macro level to account for the rise of a "global neoliberal hegemony" (Gill, 1997) and the related financialization of Anglo-American and European economies (Bieling, 2013). At the level of the firm, this trend has been traced by following the diffusion of a shareholder value ideology (Ezzamel, Willmott, & Worthington, 2008;Faulconbridge & Muzio, 2009) that entails massive workplace restructuring via cost cutting, downsizing, externalization, and the weakening of collective forms of workers' organization (Cushen, 2013;Forsberg & Stockenstrand, 2014;Hirsch & De Soucey, 2006;Salento, Masino, & Berdicchia, 2013;Saltorato & Benatti, 2017). The shift was significant enough to question whether the new order of domination was still based on hegemonic consent or rather relied on "managerial despotism" (Burawoy, 1985), that is, the arbitrary application of coercion. Gill (1997) suggested that "a less consensual order was emerging, one based increasingly on the politics of supremacy and coercion rather than built from broad-based popular legitimacy" (p. 7). Burawoy (1985) qualified the new regime as "hegemonic despotism" whereby the interests of capital and labor continue to be concretely coordinated, but where labor used to be granted concessions on the basis of the expansion of profits, it now makes concessions on the basis of the relative profitability of one capitalist vis-à-vis another-that is, the opportunity costs of capital. (p. 150) Workers' resistance to neoliberal hegemony has been studied in varied organizational settings such as the airline, engineering, and artistic professions (Cushen, 2013;Forsberg & Stockenstrand, 2014;Fraher & Gabriel, 2016), multinationals' plant shutdowns (Contu, Palpacuer, & Balas, 2013;Erkama & Vaara, 2010;Vidaillet & Gamot, 2015), or the privatization of public services (Spicer & Fleming, 2007). This body of work has departed from neo-Gramscian readings by focusing on the everyday practices of "resisting subjectivities" (Cushen, 2013;Forsberg & Stockenstrand, 2014;Fraher & Gabriel, 2016), the discursive struggles between "competing narratives" whereby workers collectively engage in resisting (Erkama & Vaara, 2010;Spicer & Fleming, 2007), or workers' capacity to "reintroduce a symbolic authority" in a psychoanalytical reading of resistance (Vidaillet & Gamot, 2015, p. 987). These theoretical contributions thus addressed micro-practices of resistance that did not openly challenge or unsettle the dominant order and/or left aside a broader context enabling workers' resistance to become more impactful (Fleming, 2016;Thompson & Harley, 2013). By contrast, a neo-Gramscian perspective would acknowledge the role of an enabling context that took the form of a state-based legal apparatus that workers were able to mobilize in their struggle against neoliberal restructuring at the French factory studied by Vidaillet and Gamot (2015), or the "fundamentally different" funding and governance conditions that allowed professional workers to preserve collective skills and identities in the cases studied by Fraher and Gabriel (2016, p. 171). Hence, workers' capacity to draw on levers and resources located beyond the firm may prove instrumental to challenging neoliberal hegemony, as shown in the neo-Gramscian study of Contu et al. (2013), where the decision to shut down a plant was successfully reversed, thanks to the ideological, organizational, and material support that workers received from the broader civil society and the tribunals where they repeatedly took action. These findings suggest that "productive" resistance might critically depend on understudied processes through which workers can mobilize and develop a variety of tools and resources in and beyond the corporation. Along such lines, Spicer and Böhm (2007) pondered whether "the historically separated realms of workplace politics and civil society have become increasingly blurred" (p. 1683) and envisioned a process of "escalation" whereby workplace movements could "take their struggle into the broader realms of civil society." The authors see escalation as likely to occur when resisters "find few spaces in a workplace to voice their grievances, and issues of the broader legitimacy of management discourses are at stake" (Spicer & Böhm, 2007, p. 1687), a situation that may arise from acute forms of hegemonic despotism. --- Civil Society Movements and Hegemonic Transformation The capacity of resistance to alter contemporary forms of domination has more often been studied outside the workplace. In a neo-Gramscian perspective, broader social movements have been cast as central forces of "counter-hegemony," understood as the "creation of an alternative hegemony on the terrain of civil society" (Pratt, 2004, p. 332), which conveys "an alternative ethical view of society" (Cohn, 2016, p. 113). The most systematic attempts to integrate discursive, material, and organizational dimensions into the study of counter-hegemony have been made by Levy and co-authors in transnational fields where civil society actors pressed multinationals to intervene on social or environmental issues such as climate change (Levy & Egan, 2003), access to AIDS drugs in developing countries (Levy & Scully, 2007), economic inequalities in global production networks (Levy, 2008), or standards for corporate social responsibility (Levy, Brown, & de Jong, 2010) and sustainable coffee (Levy, Reinecke, & Manning, 2016). Yet, counter-hegemonic struggles linking the workplace to civil society have remained understudied, with the significant exception of anti-sweatshop campaigns where activists in Europe and North America have pressured large corporations to support workers' rights at subcontracting factories in the Global South. For Spicer and Böhm (2007), such campaigns epitomize the escalation of workplace struggles that spill over into civil society. However, southern actors-both civil society and workers-have been mainly involved as "secondary subjects" in these transnational campaigns where northern activists tend to play a leading role (Wells, 2009). As a result, we still know little about processes of hegemonic transformation where workers take the lead in driving broader forces of resistance via civil society, or even via the State. The latter has remained largely off the radar of neo-Gramscian organizational studies, which have focused either on the firm or on transnational struggles led by civil society movements. --- Exploring "Hybrid Spaces" of Resistance As previously discussed, the studies of workplace resistance to neoliberal hegemony have mostly focused on micro-practices located within the firm and have thus overlooked the levers and resources that resisters could successfully mobilize in the spheres of the State and civil society. On a broader level, hegemonic transformation has mainly been studied through the prism of movements and struggles among broad groups of players-that is, nongovernmental organizations (NGOs), corporations, governmental institutions-an approach that ignores the more micro processes through which resisters could launch new forms of struggle either within or across such groups. The spillover of workers' resistance into civil society and the State could thus be seen as situated "in between" these two perspectives, in the microand meso-organizational processes whereby resistance could span the three main sites of neoliberal hegemony and induce transformations in these sites. The notion of in-betweenness has attracted interest in the "Third Space" literatures where it serves to define a "hybrid" space created through the encounter of already existing spaces and where new forms of knowledge, discourses, and identities could emerge. These literatures are typically concerned with the processes at play between a predominant "first" space and a "second" less apparent or more marginal space, giving rise to a "third" space. In Soja's (1996) work on political geography, the third space serves to capture the construction of contemporary humans as "intrinsically spatial beings" via the encounter between physical and socialized spaces. From the postcolonial, discursive perspective of Bhabha (1994), the third space symbolizes a place where multiple meanings, appropriations, and translations of the same linguistic signs and cultural symbols may occur, thus challenging the privileged position of the colonizer's ways of knowing. In educational studies, Gutiérrez, Baquedano-Lopez, Alvarez, and Chiu (1999) see the third space as offering access to a multiplicity of meanings and knowledge that students can use as a bridge, or scaffold, between their community's or home-based discourses and school-based discourses, so as to develop stronger understandings of the natural world. It is in the latter field that Moje et al. (2004) set out to integrate these varied contributions by conceptualizing the third space as a "navigational space" in which to gain the skills and expertise to negotiate different discourses and to "cross discursive boundaries"; a space where different knowledge and discourses will "coalesce" to generate new knowledge and "expand the boundaries" of official discourses; and a bridge, or "supportive scaffold," between marginalized and dominant discourses, helping students from disadvantaged backgrounds to develop stronger understandings and strengthen their future social and cognitive development (also see Calabrese-Barton & Tan, 2009). Such perspective may offer some guidance in exploring the spillover of resistance throughout the three spaces of the firm, civil society, and the State, to assess how resisters may develop a hybrid space whereby to "generate," "navigate," and "leverage" not just new knowledge and discourses, as analyzed in Third Space studies, but also material and organizational resources, in line with our neo-Gramscian perspective, to challenge and reshape the dominant hegemony. --- Research Setting and Method The resistance movement at the French Telecom company (hereafter FT) lends itself particularly well to a single case study approach, designed to explore the complex processes through which resistance may spill over from the workplace to civil society and the State. The exceptional reach of resistance further makes this case "unusually revelatory" (Eisenhardt & Graebner, 2007), casting FT as a fairly "unique exemplar" (Gehman et al., 2017) of workplace-led hegemonic transformation. While such features may induce limitations in the replicability of the transformational achievements to be observed, we believe that the theoretical insights derived from this casebased research can be useful to practitioners and scholars of resistance alike, as they offer clues on the question, "What are the activities you actually have to engage in overtime to produce it [transformational resistance]" (Langley, p. 6, in Gehman et al., 2017). We started analyzing the social crisis at FT in December 2009 as it was unfolding in the media. Our initial interest was in identifying the managerial policies that might have contributed to the rise of the employees' ill-being stigmatized by the series of suicides. We observed that FT had undergone the kind of transformation we had studied in other major French multinationals where financialization and a neoliberal regime had deeply unsettled work identities and social relations at the workplace (Palpacuer and Seignour, 2012). FT had experienced a particularly stretching shift in status: From a state administration in the late 1980s, it had become one of the largest publicly traded corporations on the French financial market, with state ownership reduced to 27% at the onset of the crisis in 2008. The corporation had grown abroad while massively restructuring and downsizing in France, cutting more than 60,000 jobs or 40% of the national headcount in a decade (Minella, 2009). FT could thus be seen both as exemplary of the shift toward financialization and managerial coercion occurring in the new hegemonic regime, and as a radical manifestation of this same shift due to the acute form of transformation that the firm had experienced. The media crisis generated an abundant flow of information and discourses in the press and on the Web, and no fewer than seven nonacademic books about the crisis and its premises (Decèze, 2004;Diehl & Doublet, 2010;Du Roy, 2009;Champeaux & Foulon, 2012;Dervin, 2009;Ledun & Font Le Bret, 2010;Talaouit & Nicolas, 2010). We were able use this material to trace the transformation of the firm via financialization, restructuring, and neoliberal rhetoric. During this first phase of research, our attention was drawn to an important source of information on work-related issues within the firm, the Observatory of Stress and Forced Mobility, which some of FT's labor unions had set up as a nonprofit association. Preliminary interviews indicated that the Observatory (referred to as the "Obs" by the resisters) was the organizational arm of an innovative resistance initiative, playing a key role in feeding and steering the national debate. This prompted us to conduct a series of 26 interviews in 2011 and 2012 with 18 actors who had been involved in the resistance and its outcomes (Table 1). These retrospective interviews were conducted at a time close enough to the crisis to allow respondents to produce vivid accounts of factual events, while a postcrisis context made them more inclined to reflect on what had been a rich experience of resistance, but also an intense, politically sensitive, and humanly trying one for most of them. Most interviews lasted from 1 to 3 hr and followed an exploratory, open-ended approach so as to allow interviewees to recall in their own words and perspectives what had happened and how they had been involved in the events under study. Our sample comprised key people involved in launching resistance within FT, spreading the debate in the broader civil society, and acting within the State in response to the media crisis. We identified them either incrementally from the initial interviews or from their visibility in the media. We were also careful to include enough diversity in our sample to allow for contrasted perspectives to be recorded on the initiative under study. For instance, we interviewed labor representatives who had refused to take part in the Observatory, as well as civil society outsiders who were critical toward this initiative. We used additional interview data published in nonacademic sources to complement our primary data and, when relevant, for the purpose of triangulation. We chose to construct our results as a narrative, inspired by scholars who claimed that resistance could be "enabled and collectivized, in part, by (...) narrating moments when the taken for granted social structure is exposed and the usual direction of constraints upended, if only for a moment" (Ewick & Silbey, 2003, p. 1329). In this vein, narrative scholarship is seen as "overtly political" in its capacity to "give voice to the subject" (Ewick & Silbey, 1995, p. 199), in this case embodied by the resisters. This choice further allowed us to account for the polyphonic and dialogic dimensions of the resistance initiatives under study (Bakhtin, 1984), which left space for a diversity of voices to be heard in our rendering of the story (Belova et al., 2008;Humphreys & Brown, 2002). Borrowing from Todorov (1968), our story captures the three typical phases of a narrative: out of an initial situation of relative stability, a disruptive event occurs-a tipping point in the history of the firm, here embodied by the restructuring plan NExT (New Experience in Telecommunication); this triggers actions-in the form of workers' resistance, culminating in broad civil society debates; after which calm is restored via a transformation of the initial situation-the CEO is dismissed and regulatory tools are established to detect and prevent psychosocial risks at work in the country. Inspired by a neo-Gramscian perspective, our account highlights relations of force unfolding through the corporation, civil society, and the State, by successively giving prominence to one of these sites and its key players in the sequences of the narrative. The three phases of (a) "hegemonic despotism" within the firm, (b) "counter-hegemonic resistance" in civil society, and (c) "hegemonic transformation" via state actions are closely interwoven, as schematized in Figure 1. These three sequences were further substantiated by systematically coding our interview material so as to characterize the discursive, organizational, and material aspects of the actions and interactions under study in each sequence of the narrative. Drawing on abundant secondary data, including the press releases from the French Press Agency (AFP), we first elaborated a precise chronology of events occurring at or around FT before and during the crisis (Abbott, 2001). This led us to record more than 100 events between the beginning of FT's transformation in 1990 and the postcrisis announcements made by its new CEO in July 2010, 22 of which are shown in Figure 1. The sequential structure of the narrative is particularly well-suited to capture the interplay of agentic forces and broader contextual elements in the processes under study, enabling the examination of "how the actions of one period lead to changes in the context that will affect action in the subsequent period" (Langley, 2009, p. 919). Further iteration among our narrative and the neo-Gramscian literature led us to conceptualize hegemonic transformation as a "cascade effect" of resistance. We then set out to explore in more detail how this cascade effect had been produced by resisters, which led us to mobilize the concept of "hybrid space" to highlight the cross-cutting dynamics of resistance throughout the firm, civil society, and the State. The literature on hybrid space inspired us to locate the source of the cascade effect in resisters' capacity to share, develop, and leverage discursive and material resources across the three spheres of Gramscian hegemony. --- A Three-Stage Process of Hegemonic Transformation The Rise of Hegemonic Despotism Within FT The transformation of FT from a public service into a global corporation was part of a broader shift toward a French form of neoliberal hegemony based on financialization. The French State stimulated the growth of financial markets and encouraged the entry of foreign investors through two waves of reforms in the 1980s and 1990s (Coriat, 2006), whereas the managerial elite endorsed a shareholder value-oriented ideology (Goyer, 2006;Morin, 2000;Schmidt, 2003). Top executives retained significant autonomy vis-à-vis financial markets and continued to collaborate closely with the government elite (Clift, 2004), a pattern that Lubatkin, Lane, Collin, and Very (2005) describe as "centralization based on personal relationships." Three laws were passed between 1990 and 2003 to allow for the progressive privatization of the telecommunications firm (Du Roy, 2009). FT engaged in the kind of international growth that most French multinationals were pursuing at the time, reaching a Number 2 position on the European market. Meanwhile, the firm drastically downsized and reduced the share of employees with public servant status from 90% to 70% of the workforce (Minella, 2009). In this setting, Didier Lombard took over as CEO in 2005 to launch a strong international acquisition policy under the Orange brand and a restructuring plan (NExT) designed to shift FT from its technical focus on being a "network access provider" to the commercial orientation of a "service access provider." On the financial side, Lombard established a low target of €7 billion of annual cash flow in the period 2006 to 2008, of which 40% to 45% would be distributed to shareholders. This material turn in the distribution of wealth was backed by strong adhesion to the ideology of shareholder value and by cultivating close relationships with the financial markets (Chabrak, Craig, & Daidj, 2016). The implementation of NExT was to play a central role in the 2009 crisis. Downsizing objectives were particularly ambitious with a target of 22,000 job cuts over the period 2006 to 2008. This was coupled with a large-scale mobility plan aimed at switching people from technical jobs to commercial jobs in Orange boutiques and call centers. Tight procedures were established and harsh pressure was exercised by intermediate and human resource managers to push people through mobility procedures. These organizational forms of coercion were discursively manifested in the address given by the CEO and HR Director to top executives at a mid-term review of NExT in October 2006, when the program was found to be below targets. Didier Lombard announced that in 2007 he would "implement the departures, one way or another, either through the door or through the window" and that he strongly backed the "crash program" presented by HR Director Olivier Barberot to accelerate mobility via "systematic identification and compulsory registering of people at the development space." Such stringent measures induced widespread bullying of FT workers, causing suicides to proliferate in a context of deep disruption of FT's historical culture and social regulation (Chabrak et al., 2016). NExT was implemented in an unusual context as 70% of FT employees retained their civil servant status, that is, subject to administrative laws that differed significantly from the labor code applied to employees under private contracts. This situation resulted from labor union negotiations to preserve employees' statutory protection when FT was privatized in 1996. This had created a legal vacuum in which the top management could turn a blind eye to legal social constraints and early signs of the social crisis. "They had a feeling of total impunity," recalled a labor inspector (STA2). At a national meeting with occupational physicians in 2008: It was surreal. He [HR Director] comes in, sits down, puts his feet on the table and starts by saying "labor doctors, what's that for?" (...) We were in full crisis, it had not yet come out in the media but it was terrible. And (...) he didn't care at all. (OC3) Labor unions also played a role in the rise of managerial abuses by failing to act as a counter-power during the restructuring plan. In fact, the unions did not form a unitary front against the "despotic" managerial pressures, as they were undermined by a series of restructuring programs, the decline of the workers' collective identity based on technical skills, and deep internal divisions during the privatization process. The institutions of private labor law, such as Enterprise Committees (ECs), Health and Safety Committees, labor inspections, or even private employment contracts, had only recently been introduced, if at all, within the firm, and the unions lacked the skills and knowledge to operate them. Collectively bargaining over the mass departures under the mobility plan would also have required symbolically acknowledging that the protective civil servant status-for which the FT unions had fought hard over two decades of privatization-had become more of a myth than a reality. Mirroring top managers' denial of the early signals of the crisis, the unions refused to recognize the ideological shift of the corporate elite and its organizational consequences ("We could not imagine that our leaders would start acting like minions of Wall Street," UN-CGC1). --- Emergence of a Counter-Hegemonic Front While NExT unfolded, taking the "drunken boat" (STA1) of FT to "insane" heights (OC4) of managerial violence, a number of resistance initiatives were launched. These were mostly spearheaded by the labor union SUD (Solidaires, Unitaires, Démocratiques) set up at FT in 1989 in opposition to the firm's privatization. As a founding member of the French alter-globalization movement, SUD had been created as a platform of unaffiliated unions in the early 1980s with the aim of renewing the French labor movement. It gained in strength during major strikes and public demonstrations in the 1990s around themes such as the defense of public services, solidarity, democracy, and work enrichment (Biétry, 2007). Concerned about the limits of established forms of union militancy, SUD had initiated critical debates and actions on the question of work organization and suffering at work within the firm in the early 2000s (Decèze, 2004). The union had also started to solicit labor inspectors, who went on to play a key role during the crisis. The most influential resistance initiative at FT would prove to be the Observatory of Stress and Forced Mobility (hereafter the Observatory, or the Obs) established by unionists from SUD and the CGC (Confédération Générale des Cadres), a union that had seen new leaders emerge during FT's start-up acquisitions in the late 1990s (Delmas & Merlin, 2010;Du Roy, 2009). The Obs brought together CGC constituents, mostly managers and engineers under private employment contracts, and SUD members, that is., activist public sector technicians. This rapprochement came about mainly, thanks to an original entente forged by the unions' central delegates-Patrick Ackerman at SUD and Pierre Morville at CGC-who acted as spokespersons for the Observatory. Via the Obs, two labor groups with distinct identities formed a unified front against managerial abuses, financialization, and CEO Didier Lombard, who had come to embody these drifts ("The top management was amazed that we became allies," UN-CGC1). Backed by the relatively abundant material resources of the two unions-a legacy of generous labor union endowment at FT-and a dedicated team of about 20 people, the Obs sought to assess and expose the managerial violence unfolding at FT through systematic analysis and communication. To do so, it drew on the expertise of a scientific committee comprising academics from a variety of disciplines-mainly sociology-to produce a large amount of surveys and research ("We wanted to involve academics in order to legitimize work on which the labor unions were not at ease," UN-CGC2). The diffusion of information was simultaneously geared toward FT employees, labor unionists, and an outside audience that was reached through the Obs' website, conferences, meetings, and sustained interactions with the media. 1 The rise of employee suicides triggered internal debate on whether the phenomena should be addressed by the Observatory, and how. A decision was taken to cast suicides as an extreme manifestation of a broader phenomenon of the employees' ill-being caused by a pathogenic form of management. This framing acknowledged "new forms of suicide that appeared at FT, political suicides.... These suicides were signed and put in relation to work" (UN-SUD6). However, not all labor unions supported this surge of resistance. Neither the CFDT (Confédération Française Démocratique du Travail) nor the well-established CGT (Confédération Générale du Travail) joined the initiative. Interunion rivalries played a role in this resistance to resistance, as did tensions between this innovative and the more traditional forms of labor action embedded in FT's historical bloc. Union representatives who opposed the Observatory favored a posture of "negotiation," that is, making specific collective demands to the FT's management, rather than a more radical but open-ended posture of "denunciation" on topics new to labor unions ("We always said that we wanted to open negotiations. The priority was to talk with the firm in order to change things and not just to contest," CGT2). By framing and spreading critical views of managerial practices at the workplace, this resistance paved the way for the major crisis of 2009: The suicide that provoked the shift happened on July 13th in Marseille. It was summer, news was scarce, and the regional TV was covering the event, especially when the guy left a letter that the family insisted on reading aloud at the funerals. (UN-CGC1) The Observatory was suddenly in daily contact with the press, radios, and TV channels, playing a key role in feeding the media with information and analyses. Considered a legitimate source on FT suicides, it regularly confirmed or denied, via its union constituents, the work-related nature of employee suicides that continued to hit the news. Under the impulse of the Obs, the opposition between managerial and resisting actors inside the firm spilled over into the broader civil society through heated debates during

In the summer of 2009, a major media campaign was unleashed in France concerning a series of suicides committed by employees of the main French telecommunications operator. Headlines reported a suicide on 14 July-the 18th since early 2008-which the victim explicitly linked to growing managerial pressures at the company. The case took a broader political dimension when the government intervened in September 2009 to summon the top management of the corporation. Meanwhile, additional employee suicides continued to hit the news. A national debate developed from fall 2009 until spring 2010 on the causes and significance of these suicides, related to changing work conditions, management practices, and broader trends toward privatization and financialization in the national economy. The case stirred the vocal participation of many diverse civil society actors such as professional associations, labor unions, political parties, intellectuals, and religious organizations. This led to the launch of governmental plans, collective framework agreements on stress and violence at work, and a judicial investigation of the legal responsibility of the firm and its top managers. Such widespread media attention, controversies, and regulatory changes did not occur as a spontaneous reaction to the radical actions of individual employees. They resulted rather from the workers' organized resistance, which reached beyond the corporation with the objective of making the issues of work pressures and deteriorating social conditions within the firm visible, recognized, and acted upon at a broader level. In this article, we ask how a workers' resistance movement could gain such a transformative capacity, escalating from the firm into civil society and the State so as to produce long-lasting changes in workplace practices and regulation across the country. We argue that the literature on resistance does not adequately deal with this question as it tends to focus either on the workplace or on civil society as the main locus of the struggle. As a result, it overlooks the processes through which resistance is able to span these different spaces to achieve systemic changes. To capture these understudied processes, we adopt a neo-Gramscian perspective that allows us to think of social transformation as occurring through what Gramsci (1971) sees as the three pillars of a social order or "hegemony," that is, the firm, civil society, and the State (Levy & Egan, 2003). Seen from this angle, hegemony is neither complete nor stable, but rather continuously challenged and transformed via "relations of force" mobilizing discursive, material, and organizational resources that we seek to analyze in our study of the resistance initiative. We show how counter-hegemonic forces emerged and successfully opposed the neoliberal hegemony that materialized in the privatization, financialization, and heavy restructuring 846408J MIXXX10.

s management, rather than a more radical but open-ended posture of "denunciation" on topics new to labor unions ("We always said that we wanted to open negotiations. The priority was to talk with the firm in order to change things and not just to contest," CGT2). By framing and spreading critical views of managerial practices at the workplace, this resistance paved the way for the major crisis of 2009: The suicide that provoked the shift happened on July 13th in Marseille. It was summer, news was scarce, and the regional TV was covering the event, especially when the guy left a letter that the family insisted on reading aloud at the funerals. (UN-CGC1) The Observatory was suddenly in daily contact with the press, radios, and TV channels, playing a key role in feeding the media with information and analyses. Considered a legitimate source on FT suicides, it regularly confirmed or denied, via its union constituents, the work-related nature of employee suicides that continued to hit the news. Under the impulse of the Obs, the opposition between managerial and resisting actors inside the firm spilled over into the broader civil society through heated debates during the fall of 2009. Neoliberal views that public sector employees were too "fragile" and had been overprotected by the State clashed with denunciations of rising managerial violence at the workplace. Representatives of leftist parties demanded the resignation of the CEO of FT and the opening of a governmental mission on suicides at work, pointing to broader systemic issues of management and restructuring practices in the country. Polemics developed around whether the motives for the suicides could be linked to the company and whether suicides could be considered statistically more significant at FT than in the broader national population. A dense sequence of events followed until the spring of 2010, closely relayed by the media and punctuated with new suicides and suicide attempts in the firm's workforce. --- Hegemonic Transformation Via State Intervention The social agitation triggered several responses from the State. First, strong media coverage prompted the government to intervene to "stop the crisis." Faced with a radically new situation, the Department of Labor was "in panic," "nobody knew what should be done" (STA2). A key player in this situation was Minister of Labor Xavier Darcos, who regularly intervened in the media to push for greater action on the part of FT and other large corporations in the country. Summoned by the government, the CEO announced a number of managerial measures, including the end of forced mobility, which compelled employees to systematically change job every 3 years, and the launch of a vast internal survey on work conditions. Several interim reports were released between December 2009 and March 2010 with large-scale media coverage, revealing that the majority of employees felt under pressure or distressed at work and had experienced deteriorating work conditions in recent years. The survey confirmed the Obs' findings and was instrumental in shifting public opinion in favor of the workers' view that the firm's responsibility was involved in the crisis. Meanwhile, the government pushed for Lombard to resign and flanked him early October 2009 with a spin doctor, Stéphane Richard, a former director of the Private Office of the Minister of the Economy and Finance, who took over as CEO in March 2010. Richard drastically changed the communication style of top management. Several antistress and work management plans were announced in the weeks following his appointment. The shift was acknowledged by resisters: "Richard is not Lombard and his arrival has put an end to the crisis" (UN-SUD2); "his discourse centers on the human side, the public mission, recruitment and the end of massive downsizing plans" (UN-SUD1); "the situation is totally different" (OC3). SUD and CGC came back to the negotiating table, and in the following months Richard successfully concluded collective agreements on stress at work with most FT unions-but not SUD. The consensus was incomplete and, at best, reluctantly embraced. Some protagonists of the historical bloc continued to side with Lombard as a symbol of the technical and engineering tradition of the firm: "I always say: Lombard knows what a phone is, Richard knows what a bank is" (CGT2). The State also took a second line of action. An informal collaboration emerged between labor inspectors covering FT's 450 sites in France and the General Directorate of Labor (DGT) of the French Ministry of Labor. This led to the inspectors agreeing to communicate field information to Hervé Lanouzière, the DGT's technical counselor in charge of work conditions. This pooling of information shed light on critical aspects of FT's human resources management, which Sylvie Catala, the labor inspector in charge of FT headquarters, was able to compile and analyze in a comprehensive report on psychosocial risks at work within the firm. The report qualified the implementation of NExT as pathogenic and documented the link between work conditions and 15 employee suicides. The inspector framed a discourse on psychosocial risks at work by building an analogy with asbestos exposure: Psychosocial risks, it's a bit the same. You take people, you impose restructurings on them, they lose their bearings, they are denigrated... and out of ten, one will commit suicide, four will get depressed, and the others will make it through. (STA1) Submitted to the criminal court in February 2010 and widely commented on by the media, the report supported a claim under Art. 40 of the French penal code that FT management was "endangering others by implementing forms of work organization capable of producing severe damage to workers" health. SUD filed a complaint against FT and the three senior officers-the CEO, HR Director, and COO-who had been in charge of the NExT plan, prompting the launch of a judicial proceeding by the tribunal of Paris in April 2010. Over the course of the following months, other FT unions joined SUD in the legal battle. This type of judicial inquiry was a first in France, where the responsibility of a firm, let alone its top managers, had never before been under legal scrutiny for pathogenic management. A third line of government action addressed the crisis at a broader level. In October 2009, Darcos launched an emergency plan to push forward the national collective agreement on stress at work that major employers and labor organizations had signed in 2008, as late adopters of the 2004 European framework agreement on work-related stress. Darcos aimed to have the 1,500 largest corporate employers in France engaged in its implementation by February 2010. At the suggestion of Lanouzière, he adopted a "name and shame" tactic and published a list of companies on the government's website under green, orange, and red tags depending on the progress made in launching an internal plan. The "black list" caused strong reactions in corporate circles and was withdrawn within 24 hr, and Darcos resigned a month later. His push to strengthen the regulation of work conditions in the country nevertheless led to the signature of a national agreement on harassment and violence at work in March 2010, as a transposition of the European framework agreement of 2007. It resulted in a societal and legal recognition of employees' exposure to psychosocial risks and sent a signal to the managerial elite that they should "add a supervisory system in their cockpit" (EXP2) to detect and prevent these risks, a move made by most large corporations in the aftermath of the crisis. The crisis symbolically ended in March 2010 with the arrival of Richard at the head of FT and the signature of firmlevel and country-level collective agreements. Although cases of employee suicides continued to hit the news, their frequency greatly diminished. The media coverage of FT declined and the Observatory slowed down its activities. Table 2 sums up the overall process of hegemonic transformation by highlighting key events, major sites, lead actors, and the main ideological, material, and organizational dimensions of these three sequences. --- Explaining the Cascade Effect of Resistance To explain how workers were able to trigger changes in work practices and regulation both within and beyond the firm, we need to further characterize what can be seen as a "cascade effect" of resistance. The image is borrowed from Levy and Egan (2003) who reflected on hegemonic change: "small perturbations can often be absorbed and accommodated with little impact on the overall structure. Periods of relative stability, however, are punctuated by discontinuity and change, as fissures split open and cascading reactions lead to major system-wide reconfiguration" (p. 811). In our case, the "fissure" opened when FT was privatized while employees retained the status of civil servants. This created an "outlaw zone" between public and private labor regimes, that is, a space where arbitrary managerial coercion could unfold without meeting collectively defined limits. Cascading reactions occurred when actors outside the dominant alliance, that is, newly formed or recently transformed labor unions, organized a counter-hegemonic front that extended beyond the firm via the Obs and the media into the broader civil society, where heated debates on work conditions and employee suicides destabilized the established consent structures in society. A systemic change was achieved when the State reentered the loop to reestablish hegemonic consensus by setting limits on managerial abuses and by restoring workers' rights via collective agreements, the dismissal of FT's CEO, and a legal inquiry into managerial responsibility in the FT crises. However, this cascade effect was in no way mechanistic or structurally predetermined. Rather, our narrative highlights the key role of resisting agents in launching innovative forms of resistance across the firm, civil society, and the State. The threads of this cascade effect can be pulled together by thinking of the Observatory as a "hybrid space" through which workplace resisters could escalate their struggle into civil society and the state apparatus. We elaborate on this perspective to suggest that the Obs developed three dynamics of resistance enabling workers to combine, develop, and leverage not just discursive, but also material and organizational resources, to escalate their struggle into civil society and the state apparatus. --- Escalation Into Civil Society First, the Obs offered a space to navigate through the knowledge and discourses produced by distinct social groups located inside the firm-SUD and CGC-and in the broader civil society-for the most part, social scientists. As the Obs was set up on the terrain of civil society, outside the direct control of dominant groups and under the legal status of association, it was able to reduce the barriers between the workplace and other spheres. As such, the hybrid space was constitutive of a capacity for resistance, allowing the resisters to mobilize a broader spectrum of resources-the scientific expertise of academics, contacts with journalists, labor inspectors, and union members, and funding provided by SUD and CGC. Importantly, it also enabled them to develop navigational skills across discursive boundaries. This was notably the case in the discussions about workers' suicides, when the boundaries between private and public spheres were revisited in view of political action. The hybrid space was itself an outcome of the navigational skills that it came to support collectively, as observed in the union leaders' capacity to launch and sustain the Obs over several years. Pierre Morville, an atypically left-wing unionist at CGC, and Patrick Ackerman, a SUD activist who considered allying with a managerial union, were both aware of the relevance of academic knowledge to the workers' struggles. Second, the hybrid space generated new knowledge, discourses, tools, and acts of resistance. Research participants recall witnessing "the collective elaboration of a grammar and a discourse adapted to the situations experienced at FT" during the Obs' national meetings, thus redefining the "borders between what can and cannot be morally and tactically said on sensitive issues" (Delmas & Merlin, 2010, p. 40). The scientific capacity to "observe" was mobilized not simply to develop a new understanding of the situation experienced by workers but also to act upon it, as expressed in the slogan Observer, Comprendre, Agir (Observe, Understand, Act), devised in response to criticisms by hostile unions that the Obs lacked the capacity to engage with management. Hence, the Obs produced surveys, reports, meetings, as well as training sessions for labor unionists. Third, the hybrid space was subversive. Academic activists depicted a Foucauldian struggle on the battle field of power-knowledge. On one side, the camp of law, order, and truth, with all its privileges, its technologies, its global discourses and their power effect. On the other, the camp of small, dispersed, confiscated, silenced knowledge. The position asserted by the Obs was to detect, decipher, buried knowledge in order to put them in insurrection against the despotism of managerial knowledge. (Burgi & Gojat, 2009, p. 2) The Obs was thus to "serve as a tool for labor unions and enable workers to reclaim the overall project of a true work collective, enable them also to rebuild self-esteem and give meaning to their harrowing experiences" (Burgi & Gojat, 2009, p. 6). Having common counter-hegemonic purpose did not mean that the hybrid space was devoid of internal tensions, as observed, for instance, between the pragmatic expectations voiced by labor unionists and the analytical orientations inherent to academics' contributions (Delmas & Merlin, 2010). --- Escalation Into the State The hybrid space extended into the state apparatus, primarily via the labor inspectorate where connections were forged with key players such as Sylvie Catala, who received early evidence of the situation at FT and also mobilized her expertise for the Obs' educational activities. The navigational space trickled up the State's hierarchical lines, notably via two former labor inspectors-including Lanouzière-who acted as technical counselors of the DGT Jean-Denis Combrexelle and Minister of Labor Xavier Darcos. Neither of these high-ranking officials had the same social and educational background as the French elite and both were deeply concerned by the series of suicides ("Darcos was an important figure and a strong support," UN-CGC3). The pooling and integration of knowledge across government, civil society, and corporate borders also fueled an intense productive activity in the hybrid space in the form of a decisive labor inspection report, the ensuing court case, and the "name and shame" initiative inspired by Anglo-Saxon tactics. For the state resisters, reestablishing employee protection against arbitrary managerial pressures meant opposing the neoliberal logic promoted by the Ministry of the Economy and Finance, representing the State in its shareholder role. Hence, the hegemonic struggle redeployed itself within the state apparatus through the formation of a "diarchy" (CGC1) between the Ministry of Labor and the Ministry of the Economy and Finance. This formation was not monolithic, but rather infused with continuous movement and political maneuvering as, for instance, when Hervé Lanouzière built a "cordon fence" around Sylvie Catala, which allowed her to move forward with her report under Art. 40 of the French Penal Code. Social gains were obtained by the Labor arm of the diarchy in a context of continuing influence of the shareholder State and aligned interests in corporate and civil society circles, as illustrated by Richard's choice to replace Lombard, the quick halt of the name and shame tactic, and the slow progress of the legal procedure launched against FT and its top managers. --- Hybrid Spaces of Resistance as Levers for Systemic Change "There is nothing inherent in a social situation that will automatically prompt resistance" (Fleming, 2016, p. 107). Even though management practices were acknowledged to be particularly destructive in the case under study (Chabrak et al., 2016), workplace bullying has become pervasive and work-related suicides are estimated to be on the rise in major industrial economies (Waters, 2017), underscoring the brutal "logics of expulsion"-from jobs, houses, land, and in our case, life itself-which Sassen (2014) sees as a defining feature of our times. It is rather due to the characteristics of the resistance per se that this case could stand out as exemplary. What is remarkable is the way in which the workplace struggle escalated beyond the firm and gained leverage in civil society and the state apparatus, bringing about hegemonic transformation in the form of collective agreements on stress and violence at work, the launch of new corporate procedures on psychosocial risks, and a pending court case engaging the penal responsibility of the firm's management for harassing workers. A neo-Gramscian perspective allowed us to account for the "cascade effect" of this resistance, which we further explained by highlighting the multifaceted role of a hybrid space, situated at a meso-level between individuals and dominant institutions, and spanning the three sites of hegemony/ counter-hegemony. We highlight three processes of "sharing," "generating," and "leveraging" not only discursive but also material resources, which occurred in what we see as a hybrid space of resistance. To do so, we drew inspiration from the notion of hybridity offered by Moje et al. (2004). We recast this notion and extended it to a neo-Gramscian perspective where it serves to highlight the cross-institutional nature of the resistance across the firm, civil society, and the State. As a result, our perspective on hybridity could account for a "fourth" rather than a "third" space, as it has the capacity to integrate actors and resources across the three main sites of hegemony/counter-hegemony. Such features distinguish our neo-Gramscian perspective on hybrid spaces from the related notion of "free space" (Evans & Boyte, 1986), which has been mobilized in studies on workplace and civil society resistance (Courpasson et al., 2017;Polletta, 1999). Free spaces are seen as small-scale settings located outside the control of dominant groups, which are "voluntarily participated in, and generate the cultural challenge that precedes or accompanies political mobilization" (Polletta, 1999, p. 1). The fact of being located outside the firm's managerial control, in the realm of civil society, was certainly important to the Obs' capacity to launch and sustain a counter-hegemonic movement. Nevertheless, our neo-Gramscian take on hybrid space does not presuppose the informality and marginality of a free space vis-à-vis hegemonic institutions. It rather signals the capacity to mobilize resources and launch actions across the core institutions constituting the hegemony. By doing so, it provides for two broad types of contribution to the literature. First, in Gramscian terms, civil society is seen as the main field of "interest articulation and social struggle," where the alignment of hegemonic forces can be destabilized through political contestation (Carroll & Ratner, 1994, p. 6). From the perspective of neo-Gramscian studies, a core contribution of our research is thus to show that political contestation may lead to systemic change when-and if-it is prompted from within a hybrid space that interconnects resisters not only within civil society, but also in the spheres of the State and the corporation. Our account of the resistance at France Telecom underscores the alliances that resisters had to build across the main sites of neoliberal hegemony, as well as the tensions they faced within each of these three spheres, particularly-but not exclusively 2 -in the firm and the State, so as to achieve hegemonic transformation. Our findings support the assumption that "counter-hegemony has to start from that which exists, which involves starting from 'where people are at'... [and] involves the'reworking' or'refashioning' of the elements which are constitutive of the hegemony" (Hunt, 1990, p. 136, in Ewick & Silbey, 2003, p. 1335). We emphasize the cross-institutional hybridity of the micro processes that enabled such counter-hegemony to become broadly transformative. Second, by offering a narrative of the hegemonic transformation that people were able to induce through a hybrid space of resistance, we help to bridge the gap between workplace and civil society resistance studies. Our rendering of the hybrid space acknowledges features that have been observed in both sites insofar as the "meaning" it served to elaborate was not "innate and predetermined" but remained "fluid and multivalent" over the course of the struggle, which itself took multiple forms: "a struggle to form, a struggle against power, a struggle that involved power, and an internal struggle" (Hardy, 2016, pp. 103-104). Our contribution adds to these perspectives by offering a practical theory of resistance based on the three core processes of "sharing," "generating," and "leveraging" discursive and material resources across the three sites of hegemony/counter-hegemony: (a) "sharing" or "navigating through" diverse cultural, procedural and discursive resources, in our case among labor unionists, civil servants, academics, journalists and other resisters; (b) "generating" new resources, that is, new political views on suicides and psychosocial risks at work, new legal and managerial procedures at the workplace; and (c) "leveraging" these resources to transform dominant discourses and practices as conveyed, in our study, by the Ministry of Finance and the firm's top management. We suggest that this scheme could offer a blueprint for new practices of-and research on-resistance, if and when such resistance is geared toward achieving systemic change. --- Concluding Comments In troubled times, when new forms of hegemonic despotism are becoming pervasive in the workplace and the concentration of wealth and power continues unabated in global patterns of production and organization, how can resisters make the invisible, visible, and the unspoken, a subject of debate, so as to drive concrete political changes in the living conditions of workers and more broadly, of people and other living beings who are affected by these changes? We believe that our case study of resistance at France Telecom offers relevant theoretical and practical insights on this question, by tracing the sequences of hegemonic transformation that occurred within this firm and beyond, and by highlighting the hybridity of the space of resistance that served to curb hegemonic despotism and to establish new norms and practices at the French workplace. Cutting across the three main pillars of hegemony, that is, the firm, civil society, and the State, was instrumental to the transformative capacity of the hybrid space, as was the engagement of resisters in launching new ways of sharing, generating, and leveraging material and discursive resources, both from within, and by stretching beyond, their established role and position in the three sites of neoliberal hegemony. --- Notes 1. National conferences were organized in December 2007 andNovember 2008, gathering 200 to 300 participants for workshops, debates, and information dissemination. A network of journalists specialized in social issues was formed around the initiative, who commended the work of the Observatory. 2. We gathered anecdotal evidence that such tensions also occurred within civil society, most notably in the media where some journalists faced pressures to stop divulgating-and as a compromise, to euphemize their rendering of-the suicide events at French Telecom company (FT). For instance, in the midst of the social crisis, it was decided at a major French media that the practical protocols by which FT employees killed themselves-which could be, at times, very spectacular-would not be specified in further suicide announcements. --- Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

In the summer of 2009, a major media campaign was unleashed in France concerning a series of suicides committed by employees of the main French telecommunications operator. Headlines reported a suicide on 14 July-the 18th since early 2008-which the victim explicitly linked to growing managerial pressures at the company. The case took a broader political dimension when the government intervened in September 2009 to summon the top management of the corporation. Meanwhile, additional employee suicides continued to hit the news. A national debate developed from fall 2009 until spring 2010 on the causes and significance of these suicides, related to changing work conditions, management practices, and broader trends toward privatization and financialization in the national economy. The case stirred the vocal participation of many diverse civil society actors such as professional associations, labor unions, political parties, intellectuals, and religious organizations. This led to the launch of governmental plans, collective framework agreements on stress and violence at work, and a judicial investigation of the legal responsibility of the firm and its top managers. Such widespread media attention, controversies, and regulatory changes did not occur as a spontaneous reaction to the radical actions of individual employees. They resulted rather from the workers' organized resistance, which reached beyond the corporation with the objective of making the issues of work pressures and deteriorating social conditions within the firm visible, recognized, and acted upon at a broader level. In this article, we ask how a workers' resistance movement could gain such a transformative capacity, escalating from the firm into civil society and the State so as to produce long-lasting changes in workplace practices and regulation across the country. We argue that the literature on resistance does not adequately deal with this question as it tends to focus either on the workplace or on civil society as the main locus of the struggle. As a result, it overlooks the processes through which resistance is able to span these different spaces to achieve systemic changes. To capture these understudied processes, we adopt a neo-Gramscian perspective that allows us to think of social transformation as occurring through what Gramsci (1971) sees as the three pillars of a social order or "hegemony," that is, the firm, civil society, and the State (Levy & Egan, 2003). Seen from this angle, hegemony is neither complete nor stable, but rather continuously challenged and transformed via "relations of force" mobilizing discursive, material, and organizational resources that we seek to analyze in our study of the resistance initiative. We show how counter-hegemonic forces emerged and successfully opposed the neoliberal hegemony that materialized in the privatization, financialization, and heavy restructuring 846408J MIXXX10.

Introduction Colonias are unincorporated and unregulated peri-urban settlements along the United States-Mexico border that are home to primarily Mexican-origin populations [1]. Colonia residents face unique health challenges due to poverty, lack of access to health care, inadequate infrastructure, and environmental threats. The physical health of colonia residents has been previously examined and researchers have documented significant public health challenges [2][3][4]. In spite of this, the mental health of colonia residents remains largely understudied. Westway, the site of this study of mental health, is a colonia 1 in El Paso County (Texas). In this county in 2010, 86,472 residents lived in 321 communities defined as colonias [5]. Westway is home to about 4000 of these individuals. The population is 97 % Hispanic and 45.5 % are foreign-born, and of those foreign-born residents, only 22.2 % are naturalized US citizens [6]. Westway is located next to Interstate 10, across the ArcelorMittal Vinton steel plant and proximate to other polluting industries such as a scrap metal recycling plant. In Westway, as in other colonia environments, a confluence of challenging socio-environmental conditions contributes to a difficult quality of life for residents and, as a consequence, the potential for high rates of mental health issues and barriers in coping with them. As such, we have two objectives: (1) characterize Westway residents' rates of mental health outcomes and access to mental health care and (2) assess intraethnic disparities (i.e., based on level of environmental concerns, nativity, language acculturation, comorbidities, gender, health insurance, and stressful life events) in mental health outcomes within this impoverished, Hispanic population. --- Health and Environmental Concerns in Colonias Colonias tend to lack basic community infrastructure such as paved roads, sewer system, electricity, gas, clean water, and health care services [1,3,7]. The lack of infrastructure and poverty combines to foster the spread of diseases such as hepatitis, dysentery, and tuberculosis [1]. Poverty also leads to food insecurity and poor nutritional health, which can cause obesity and diabetes, which are common along the US-Mexico border [8]. Chronic illnesses, such as diabetes and asthma, are also prevalent in colonias [3]. Unsafe environmental conditions (e.g., presence of junk yards) have been linked with physical illness in colonias. Underdevelopment and the presence of empty lots sometimes leads both colonia residents and others to dispose of household and industrial wastes in the neighborhoods, creating physical health risks [9]. Ramos et al. [3] found that residents living closer to junk yards or dry cleaners in the Cameron Park colonia in east Texas had significantly higher risk of developing a respiratory illness than those living farther away. In terms of mental health, the focus of this paper, Anders et al. [4] found high rates of mental illness in an El Paso County colonia (not Westway); 20 % of adults had been diagnosed with depression and 17 % had been diagnosed with anxiety. It may also be that the environmental conditions in colonias are correlated with chronic stress and psychological distress, which can permeate the physical and psychological well-being of individuals [10]. Downey and Van Willigen [11] argued that environmental stressors (as chronic stress sources) can have long-lasting effects on the psychological well-being of residents living in industrialized areas. While not studying colonias, Downey and Van Willigen [11] found that Illinois residents living near industrial activity reported higher depressive symptomatology, a greater sense of disorder, and more feelings of powerlessness than those not living close to industrial facilities. Similar findings relating environmental concerns to mental disorders were found in national-level datasets in Spain [12] and Portugal [13]. --- Mental Health and Access to Care for Hispanics Given that colonia residents are Mexican-origin Hispanics, it is relevant to this study that Hispanics have lower rates of mental disorders than non-Hispanic whites [14]. Within the Hispanic population, there are disparities in mental health outcomes, based on three important characteristics: nativity, language acculturation, and chronic physical illness. In terms of nativity, immigrants tend to have lower rates of mental disorders than their US-born counterparts [15,16]. Hispanics that have acculturated to white middle-class norms are more prone to develop mental health disorders. For example, Englishspeaking Hispanics have higher rates of mental disorders [17] than Hispanics who are predominantly Spanish speaking. Those suffering from chronic physical illnesses (e.g., diabetes) are more likely to develop a mental health condition [17]. This fact is concerning in Hispanic populations because they show an early onset and a high risk of developing conditions such as diabetes, kidney disease, and hypertension [18] which often co-occur with depression [19]. Even though Hispanics have lower rates of mental illness, the US Department of Health and Human Services has identified critical mental health disparities that apply to Hispanics, which include reduced access to mental health services, decreased likelihood to receive needed services, and poor quality treatment [20]. In terms of reduced access to needed services, foreign-born, primarily Spanish speaking, recent immigrant, first generation, and uninsured Hispanics showed the lowest rates of access to mental health services [21]. Hispanics who did not speak English received needed mental health services at lower rates than proficient English speakers [22]. Mexicanorigin people with mental health challenges have very low utilization rates for mental health services even when compared to other Hispanic subgroups, like Puerto Ricans [21]. The lack of access to quality care experienced by Hispanics makes them more prone to chronicity and higher levels of impairment due to mental illness [18]. Poor quality of care is reflected in the fact that Hispanics and other minorities' main source of mental health care is primary care physicians rather than specialists such as psychiatrists [22]. --- Other Influences on Mental Health There are other influences on mental health besides the Hispanic ethnic features and industrial concerns previously discussed. These include gender, lack of health insurance, and stressful life events. In terms of gender, women have higher rates than men of major depressive disorder, anxiety disorders, posttraumatic stress disorder, and eating disorders in the United States [23]. Lack of insurance coverage is an important barrier when in need of mental health services, and it might cause a deferment of treatment. Finally, stressful life events (e.g., loss of loved one) are associated with poor mental health because these events can "lead to stress by adversely altering the meaning of persistent life strains" [24]. --- Materials and Methods --- Study Area Westway (see Fig. 1) is home to a low-income population: 49.2 % live below the poverty line, 15.4 % have an income of less than $10,000 per year, and 35 % received public assistance during the last 12 months. Only 55.4 % of Westway residents are in the labor force, mainly in service occupations, sales, construction, maintenance, and transportation [6]. Ninety-eight percent of residents are Hispanic and only 9.4 % report speaking only English at home. Over 64.3 % of Spanish speakers speak English less than very well [6]. For additional demographic information, see Table 1. Westway residents are exposed to numerous environmental hazards. Just across Interstate 10 is the steel plant, which was built in 1962 and emits zinc, lead, and manganese into the air [25]. As such, the elementary school located in Westway was ranked in the second percentile nationally in terms of bad air quality in a 2009 report [26]. Strong seasonal winds in the area provoke dust storms that spread dust from unpaved roads and lots and contaminants from industry. Residents' concerns about the environment led them to request that a team of university researchers conduct a health screening survey in their community [27]; this paper reports some of the results from that effort. --- Data Collection and Participants The first author and an assistant collected the data through door-to-door surveying between March and August 2012. Every ninth household (out of the 1120 lot units in Westway) was asked to participate. If the residents were not home, the neighbors on the left side were asked to participate. We made contact with an adult over 18 at 127 households; 23 declined and one withdrew for an 81 % response rate (n=103). For the analysis, we excluded the one non-Hispanic respondent and four relatively affluent respondents (i.e., those earning more than the county per capita average for income2 ) for a final N of 98 low-income respondents. The survey, translated by the first author and an assistant, both of whom are native speakers of Spanish, was offered in English and Spanish. The survey was designed by the research team through a community-based research process led by the third author. Community leaders were involved in the research process including making revisions and additions to the questionnaire. This study was approved by the Institutional Review Board at the authors' home institution. --- Measures and Analysis The first research objective involves descriptive statistics for a suite of mental health variables and access to care variables. The mental health outcomes include three sets of variables: diagnoses of specific conditions, psychological symptoms [28], and physical symptoms [29]. We utilized a physical symptomatology scale because Hispanics might tend to report physical symptoms in lieu of depression, anxiety, and/or stress [14]. Table 2 shows all variables used in the first research objective, along with information about the survey questions, coding, and descriptive statistics. The second research objective involves correlation and regression analyses. Table 3 reports descriptive statistics and details for how the independent variables (i.e., industrial concerns, nativity, language acculturation, comorbidities, gender, health insurance, and stressful life events) were constructed. We used "industrial concerns" as our measure of environmental concern given the many industrial sources of pollution in the neighborhood and our knowledge of residents' concern about these activities in their neighborhood. Table 3 reports the same information for the four dependent variables. (1) "Mental health diagnosis" is coded 1, if the respondent said "yes" to a diagnosis of anxiety, depression, and/or another mental health illness; it is coded 0 if the person had none of those diagnoses. ( 2) "Stress and excess worry" is coded 1 if the respondent said "yes" to either or both of the questions about ever experiencing stress and ever experiencing excess worry and 0 if the person has never experienced either. ( 3) "Psychological symptoms in the last 4 weeks" was created by factor analyzing (Principal Axis Factoring) six Likert items (see Table 3). Descriptive statistics for the Likert items contained in this factor (and the next one) are included in Table 3, along with the component loadings. ( 4) "Physical symptoms in the last 1 week" was created by factor analyzing (PAF) seven Likert items (see Table 3). After running bivariate correlations between the independent and dependent variables, we ran four regression models including the seven independent variables detailed in Table 3. We used binary logistic models for the dichotomous dependent variables and ordinary least squares (OLS) modes for the continuous dependent variables. Collinearity diagnostics [i.e., variance inflation factors (VIF) and tolerance] revealed that there were no issues with multicollinearity; all independent variables met the standard of a VIF below 2.9 and a tolerance value greater than 0.4. --- Results --- Describing Mental Health Outcomes and Access to Care One third of surveyed Hispanic residents have been diagnosed with depression and one quarter have been diagnosed with an anxiety disorder; 36 % had been diagnosed with at least one of the following mental health issues: depression, anxiety, and posttraumatic stress disorder or had attempted suicide. Nearly 60 % reported ever feeling stressed and excessively worried. In terms of access to care for mental health issues (see Table 2), 77 % of individuals with a diagnosed mental health condition have received health care for it. Of those receiving care for a diagnosed mental illness, 71 % receive care from their primary care physician for their diagnosed condition and 56 % take a medication. Only 7 % have seen a specialist for their diagnosis and 17 % have received a form of psychological therapy. More generally in the sample, 46 % of all respondents did not have health insurance (see Table 3). Of the total respondents, 18.6 % had insurance through their workplace, 20 % had Medicare, and 7 % had Medicaid. Eight percent had other forms of health insurance (e.g., Mexican health insurance) (not shown in a table). Assessing Disparities in Mental Health Outcomes Table 4 presents the results from the correlational analysis between the independent and dependent variables. The following statistically significant correlations were found: Language acculturation was positively correlated with stress and excess worry. Comorbidity was positively correlated with diagnosis of mental health condition and psychological symptoms in the past 4 weeks. Being female was significantly correlated with all dependent variables. Having more stressful life events variable was positively correlated with the diagnosis variables as well as psychological and physical symptoms variables. Finally, industrial concerns were significantly correlated with stress and excess worry and psychological symptoms of distress in the past 4 weeks. Table 5 presents results from the regression models. For the logistic model predicting stress and/or excess worry (Table 5, A), there were three significant (p<unk>0.05) findings and one finding that approached significance (p<unk>0.1). A one-unit increase in the language acculturation scale (which is 1 standard deviation) leads to a 2.5 times increase in the odds of a respondent experiencing stress and/or excess worry. Likewise, females are 2.6 times more likely to feel stressed and/or worried than males. Individuals with one or more comorbid conditions are 2.9 times more likely to experience stress and/or excess worry. A one-unit increase in the industrial concerns scale results in a 2.5 times increase in the odds of a respondent experiencing stress and/or worry. US-born individuals are 0.8 times less likely to experience stress and/or excess worry than foreign-born individuals. For the logistic model predicting a diagnosis of a mental health condition (Table 5, B), there were two statistically significant (p<unk>0.05) findings. Females and those who experienced one or more life events in the past year are both 6.9 times more likely to be diagnosed with a mental illness. For the OLS model predicting the respondent's psychological symptoms of distress in the last 4 weeks (see Table 5, C), four variables were significant at the p <unk>0.05 level. For individuals with at least one comorbid condition, there is a 0.4 increase in the psychological symptoms scale. For females, there is an increase of 0.5 in the scale when compared to males and there is an increase of 0.7 for those who experienced one or more life events in the last year. A one-unit increase in the industrial concerns scale results in an increase of 0.255 in the psychological symptoms scale. For the OLS model predicting the respondent's physical symptoms of distress in the last week (Table 5, D), there were three significant predictors (p<unk>0.05). Females had a score on the physical symptoms scale that was 0.5 higher than males and those who experienced a life event also had an increase of 0.4 in the scale. A one-unit increase in the industrial concerns scale resulted in a 0.2 increase in the physical symptoms scale. --- Discussion In terms of rates of mental health problems, Westway residents have rates of diagnosed depression (29 %) that were more than three times the rates for Texas residents (9 %) and the US population (7 %) in any given year [30,31]. Similarly, their anxiety rates (25 %) are higher than the national rate of 18 % in any given year according to the National Institute of Mental Health [32,33]. Thirty-six percent of Westway residents report being diagnosed with at least one mental illness contrasting with the 18 % of individuals over 18 with any mental illness in Texas and the 20 % of people with any mental illness in the United States [33]. Locally, Westway residents also showed greater rates of diagnosed mental health conditions than the residents of the San Elizario colonias (also in El Paso County, Texas) as previously reported by Anders et al. [4]. They found a 20 % depression rate and a 17 % anxiety rate compared to our rates of 29 and 25 %, respectively. Elevated rates of mental illness in Westway are concerning because if left untreated or undertreated, they can create unnecessary disability and further the economic challenges faced by residents [34] exacerbating inequality. Like the rest of the US population, most Westway residents diagnosed with a mental illness treat their conditions through primary care physicians. Seventy-one percent of those receiving care for depression or anxiety receive it from a primary care provider, which is slightly lower than the national statistics as 74 % of Americans who seek help for depression go to a family doctor rather than a specialist [35]. Twenty-three percent of Westway residents have not received mental health care since their diagnosis of anxiety or depression. It is probable that a lack of health insurance is related to this lack of treatment. Westway residents' most proximate source of health care services is La Fe Clinic, which is located at the north edge of the community. The clinic is part of the Centro de Salud Familiar La Fe Inc., a non-profit organization dedicated to provide health care to underserved individuals in El Paso County. Residents of El Paso County can receive a variety of services at affordable prices from this clinic, which are determined on a sliding scale according to income. The clinic also accepts Medicare or Medicaid. However, La Fe does not have mental health specialists and psychiatry is not included in their list of specialty referrals according to their website. This likely contributes to the lack of specialty care observed among surveyed residents, given that we know that a good proportion of Westway residents, especially the uninsured and those with state-sponsored insurance coverage, use the La Fe clinic for their health care needs. The lack of access to specialty mental health care is a nationwide problem. At the local level, El Paso County has only 823 mental health professionals (and only 27 of these are psychiatrists), which is 107 for every 100,000 people [20]. In the state of Texas, only 34 % of people with mental illnesses receive needed care. The situation is worse in Westway where only 7 % of people with an anxiety or depression diagnosis have seen a specialist. Texas is the state with the least treatment dollars per capita of all states and its mental health care system has received a D grade [36,37]. The United States system also has a D grade. Nationwide, mental health care often lacks a focus on health and wellness; the data on mental health is inadequate and funding is scarce; the mental health workforce is not sufficiently maintained nor developed; and there is a deficiency in cultural competency among providers and a culture of disrespect for the mentally ill [36,37]. In terms of assessing intra-ethnic disparities in mental health outcomes, we found that Westway residents with higher levels of industrial concerns were significantly more likely to experience three of the four mental health outcomes tested here, and the fourth (diagnosis) approached statistical significance. While not the subject of many studies (see only [12][13], the relationship between industrial concerns and mental health problems has been found in diverse social groups ranging from the population of Spain [12] to Westway, Texas. Westway residents have joined efforts with community organizations to address environmental issues in their community. They have participated with the Texas Industrial Areas Foundation (IAF) in El Paso to affect policy in a number of issues including environmental health. The newer IAF group in the region, Border Interfaith (BI), has worked with leaders in Westway to make the Texas Commission on Environmental Quality (TCEQ) aware of their environmental concerns. After several years of pressure, TCEQ has increased their oversight, albeit on a small scale (e.g., additional hours per year) of ArcelorMittal Vinton, one of the industries near Westway [27]. Counter to the literature [14][15][16], being born in the United States is associated with having less stress and excess worry (p<unk>0.05) and a lower likelihood of a mental illness diagnosis and fewer psychological symptoms in the last 4 weeks (both p<unk>0.10) among this colonia population. However, previous studies [14 -16] did not examine excess stress and worry specifically and instead focused on mental illness diagnoses. The association between being foreign-born and higher levels of stress and worry could be attributed to the fact that Westway is located near the US-Mexico border and that there are routinely law enforcement and border protection officials in the area. This police presence could cause stress and concern for foreign-born residents that are unauthorized to be in the United States [38]. Additionally, cartel violence occurring just across the border in Ciudad Juárez during the data collection may have contributed to stress and worry, especially for foreign-born residents, who likely maintained strong ties with relatives in Juárez. The strong presence of law enforcement officials and the cartel violence in Mexico as factors influencing foreign-born border residents to experience stress and worry while living in the US are hypotheses that should be investigated in future research. We found that higher scores on the language acculturation scale predicted experiencing increased stress and excess worry (p<unk>0.05). This is surprising in light of the finding that foreign birth is a risk factor for stress and excess worry, although it aligns with the literature on the health-damaging effects of acculturation [17]. As expected, US-birth and language acculturation are correlated at 0.351 (table not shown). It means that holding constant the effect of nativity (and the other covariates), English speaking was associated with greater stress and excess worry. This reflects other studies finding that greater acculturation is associated with higher rates of mental illness [17]. In Westway, language acculturation approached statistical significance (p<unk>0.11) as a predictor of higher rates of actual diagnosed mental illness. In a study of mental health outcomes post-disaster in El Paso County, researchers also found that language acculturation was a risk factor for a mental health problem (i.e., experiencing depression or posttraumatic stress) post-flood among those whose homes were damaged [39]. Comorbidity was a significant predictor of psychological symptoms in the last 4 weeks, and it approached statistical significance for the stress/worry and diagnosis variables (p <unk> 0.10), aligning with the strong relationship between chronic physical illness and mental disorders previously observed [18]. Given that Hispanics are at increased risk for these chronic conditions that commonly co-occur with depression, this finding is concerning from a public health perspective. Being a female was a significant predictor (p<unk>0.05) of the diagnosis, psychological symptoms, and physical symptoms variables, and it approached significance for stress/worry (p<unk>0.10). This means that poor Hispanic women in this colonia, like women elsewhere, are more likely to experience mental health challenges [23]. Life events were also significant predictors (p<unk>0.05) of diagnosed mental illness, psychological symptoms, and physical symptoms of distress. --- Limitations The main limitation of this study is the small sample size (N= 98 individuals in 98 different households), which is reflective of the small community under study (N=1120 homes). A larger sample could allow for a more sophisticated statistical analysis. Secondly, the Center for Epidemiological Studies-Depression (CES-D) measure [40], which is one of the most widely used instruments to measure depressive disorders, was not used in this study, limiting its comparability. Third, while beyond the scope of this community-based participatory project, we did not collect information about when the person was diagnosed with their mental illness and his/her healthcareseeking experiences. Last, studies with Hispanics of more diverse backgrounds could also provide more information on how ethnicity impacts mental health outcomes. --- Relevance Despite the numerous challenges facing individuals in colonias, their mental health has remained understudied and unaddressed by policy makers and activists. Mental illnesses can be debilitating and they further risk for physical illness complications [41] and economic decline [34]. Hispanic residents of this colonia had high rates of mental health challenges and few resources to deal with them; residents' per capita income was one quarter of the US average and half lacked health insurance. As the Hispanic population in the US continues to grow, it is becoming increasingly important to understand the complexities of Hispanic health [39,42]. This case study in a US colonia contributes to that goal by illustrating the health challenges and disparities present within this group of largely Mexican-origin, low-income Hispanics. Even within this socially marginalized group, mental health disparities were present with respect to environmental concerns, nativity, language acculturation, comorbidity, gender, and life events. These findings underscore the importance of considering intra-ethnic disparities in Hispanic health outcomes. The findings also highlight deficiencies in mental health care in the US, especially for low-income, peri-urban residents, such as those in Westway, and point to the need for increased access to mental health care at charity clinics serving poor populations. --- Conflict of Interest Guadalupe Marquez-Velarde, Sara Grineski, and Kathleen Staudt declare that they have no conflict of interest. Informed Consent All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000 (5). Informed consent was obtained from all patients for being included in the study. No animal studies were carried out by the authors for this article.

Unregulated residential settlements along the US side of the US-Mexico border, often called "colonias", are mainly populated by low-income Mexican-origin Hispanics. Colonia residents face numerous social, environmental, economic and public health challenges. Despite this, the mental health of individuals living in colonias has remained largely understudied. Drawing from a survey (N=98) conducted through a community-based participatory research project in one colonia suffering from numerous environmental and social challenges, this study analyzes residents' mental health outcomes and access to mental health care with a focus on intraethnic disparities based on environmental concerns, nativity, language acculturation, comorbidity, gender, health insurance, and stressful life events. Data were analyzed using descriptive statistics, correlation, and regression. More than one third of the residents have been diagnosed with a mental health condition and over half reported stress and excess worry. In terms of mental health care, 77 % of individuals diagnosed with a mental health problem have sought additional help mainly through a primary care provider despite the high levels of uninsured individuals. Comorbidity, being female, recent negative life events, and high levels of environmental concerns were significant predictors of negative mental health outcomes. This study contributes to the understanding of the complex health dynamics of the US Hispanic population. It also highlights the need for additional research and resources devoted to the mental health of low-income minorities in isolated communities. Keywords Mental health . Colonia . Hispanic . Environmental concerns . Mental health care . US-Mexico border 1 Officially, Westway is a collection of seven colonias (Westway 1-Westway 7; see http://www.sos.state.tx.us/border/colonias/reg-colonias/ elpaso-6.shtml), although residents treat it as one contiguous community.

Background Scaling up of malaria control strategies has achieved a remarkable reduction in the burden of malaria worldwide [1][2][3][4]. However, only 25 countries among 109 reached the pre-elimination/elimination stage [4]. Despite worldwide strategies to control malaria in middle and low incomes countries, ongoing epidemic dynamics are currently observed in certain malariaendemic area. In South Africa, Plasmodium falciparum accounts for the majority of malaria cases and the main vector, Anopheles arabiensis, seems to be gradually giving way to new species [5]. Malaria remains a concern in three South African provinces bordering Zimbabwe and Mozambique, namely the Limpopo, Mpumalanga and KwaZulu-Natal provinces, where annual resurgence of malaria cases is observed [6]. However, even in the districts closest to the border, where most of the malaria burden is observed, malaria transmission intensity is highly heterogeneous and associated with low socio-economic status [7]. Currently, in South Africa, imported malaria cases are estimated to represent 47% of the total reported cases [8]. In South Africa, the Vhembe District Municipality (VDM), located in the north of Limpopo Province, has the highest (and increasing) number of malaria cases and deaths (3.79 cases per 1000 inhabitants-year during the 2017-2018 transmission season) compared to the other two malaria endemic provinces [9]. Furthermore, concerning, the Eastern Limpopo border in 2014, last available official data show more than 60% of the national cases were reported in the VDM. From 1998 to 2007, 65.6% of cases reported in Limpopo were located in VDM. More specifically, Mutale, which is one of the four municipalities in VDM, reported a total 15 739 cases from January 1998 to May 2017, this accounts for 27.1% of the total cases reported in VDM during the same period [10]. The VDM District has remained relatively unchanged compared to other provinces. Over the period 2002 to 2019, the estimated population growth of the VDM was 260 000 to a current estimate of 1.4 million people, thereby increasing the number of people at risk. The poverty headcount in Limpopo increased from 10.1% in 2011 to 11.5% in 2016. Among the 5.8 million residents in the Limpopo province, the 2016 census estimated that 389,151 residents moved to the province from elsewhere [11]. A better understanding of persistence of malaria in this specific area of South Africa has become a public health challenge. Malaria control policies face the emergent need to adapt to global goals and targeted interventions reaching elimination. Single intervention approaches based on health behaviours and health education, such as repellents and bed net use have been used previously, however, malaria cases still remain high in endemic areas. More complex studies reported on the importance of measuring multilevel factors, such as parasite genetics, societal organization, individual behaviours and geographic context, associated in dynamics of malaria in other countries. In literature, these multilevel factors are investigated using mixed medializations developed by the society-behaviourbiology approach based on Glass and McAtee's works. The integration of natural and behavioural influences aid in modelling how and why individual health could be influenced at different levels and may change across the life course [12]. Indeed, if the health of populations is mainly the result of human activities including both individual behaviour and social organization, explaining the persistence of malaria requires to consider all the dimensions of this phenomenon. This study focused on individual health behaviours, i.e. behaviours that have deleterious or protective consequences on individual health, and this complexity forces us to study human exposure to malaria risks at three interconnected levels of people living conditions: (i) characteristics and changes in nature and climate; (ii) social contextual factors such as societal organization, cultural environment and access to health-related goods, and (iii) individual intention associated behaviours. In line with this approach, this community-based survey aimed to analyse the association of these factors with having at least one malaria episode and relate these findings to housing condition and health behaviour, among inhabitants in a malaria endemic area in the east Limpopo River Valley. --- Methods --- Design A community-based cross-sectional survey combining both household and individual data collection has been conducted at three selected sites of the east Limpopo River Valley (LPV) using face-to-face questionnaires and in-situ field notes. Due to unavailability of current census data, area frame sampling was based on mapping geographical units. This was augmented with the aerial photographs to identify all household units' location and ensure optimal coverage of the targeted population. Bende Mutale, Nkotswi and Doreen farms were chosen as investigation sites due to the particular high malaria incidences and known main characteristics of inhabitants, which include seasonal border mobility of workers or migrants. Bende Mutale and Nkotswi are two villages, with a majority of local population, along the Mutele River. Doreen Farms encompasses a number of farms 10 km along the Nzhelele River east of where the Doreen road crosses the R508 (10.3 km from the Tshipise Forever Resort and 22.1 km from the Musina Nature Reserve). Common residence rules (de jure rules) defined household unit as group of inhabitants usually living in the same housing unit, which is a separate living quarter. All housing units have been systematically visited during the two field investigations. Data collections were planned in two different time periods to maximize contact with mobile and/or hidden population groups such as migrant (border) workers. A high contact rate mechanically increases the response rate of all eligible household members. Low-level equipment, such as landline or mobile phones, coupled with the lack of census data, prohibited the use of alternative multiple frame sampling processes. --- Data collection During 2019, two data sets were collected and the household data was collected through face-to-face questionnaire with the head of household (older adult present during the investigation). With help of field notes (in-situ observations), additional information captured included structures of families living in the housing unit, incomes and goods, housing conditions, the bedding practices of all members and malaria history of household members. The individual level data was also collected through another face to face questionnaire on Knowledge, Attitudes and Practices toward malaria, administered to all household members aged over 18 years and in-situ during the inquiries visit. Having at least one malaria episode during the previous years (from January 2018) was used as the response variable. --- Statistical analyses To assess household social profiles, household unit data was analysed by unsupervised classification using the hierarchical ascendant classification on the multiple correspondence analysis results. The study determined the most homogeneous household characteristics groups, using v.test to describe how each variable influences each category. The household resulting classification was then analysed using the following approach, similarly to other variables [13]. The description of the household profiles and KAP characteristics has been made after applying a weighting procedure, to ensure that the data were representative of the local population in relation to age and gender. Applied weights have been calculated as regard as Bende-Mutale inhabitants' characteristics drawn out national census in 2011. When comparing those having reported a malaria episode during the previous years (from January 2018) to those who did not among KAP respondents (22), all descriptive analyses have been also weighted as well as aetiologic ones. Chi-squared tests and Student T-tests were used in univariate analyses. To identify the associated factors, multivariate logistic regression model was used. During the univariate analysis, eligible variables for the multivariate analysis was selected with both 20% significance threshold and supporting evidence from literature. However, considering the small size of the sample, candidate variables were selected avoiding interactions between covariates especially in case of qualitative collinearity (redundant variables controlling the same underlying factor). A multivariate logistic regression was then performed, using a step-by-step procedure based on the Akaike Information Criteria. The classification procedure was performed using the R 4.0.0 software (R foundation for statistical computing, Vienna, Austria) with the <unk>Fac-tominer<unk> package. Univariate and multivariate analysis were performed using the IBM SPSS Statistics for Windows, Version 27.0 software (IBM Corp. Armonk, NY). The final test results were interpreted applying a fixed threshold at <unk> = 0.05. --- Results --- Survey location The Vhembe Municipal District, Limpopo Province, is the area most heavily impacted by malaria in South Africa. The VDM borders Zimbabwe and Mozambique and is characterized by substantial trans-border movement of people, including temporary migrant workers. The Limpopo River valley has a high annual malaria incidence. The study sites we used for sampling were Bende Mutale, Ntokswi and the worker village at Doreen Farms (Fig. 1). --- Recruitment During the two-field investigations administered during working hours, between August-December 2019, a total of 398 household units were investigated with 1,681 members. Among them, 949 were aged 18 years and older and eligible to participate in the KAP survey. Only 439 participated in the KAP data collection with a crude response rate of 46.3%, which ranged from 34.1% in Nkotswi to 60.3% to Doreen Farms (Table 1). When comparing eligible household members respondents to nonrespondents, women were significantly more prone to participate in the survey than men (57.6% versus 31.5%; p <unk> 0.001). Mean age of respondents were also significantly older than non-respondents (39.1 [standard error (se) = 15.7] versus 35.7 [se = 15.8]; p <unk> 0.001). --- Household social profiles Classification was performed including data collected by the household questionnaire which included household members' and housing unit's main characteristics. Among the 1681 targeted by the survey, 12 were excluded during the classification process and 1 669 household members were grouped in three main classes: class 1 = 1202, class 2 = 109, and class 3 = 358. When comparing these classes, with regard to social characteristics of household, housing conditions and malaria history (Table 2), three specific profiles can be defined. Major household characteristics collected became operational indicators pragmatically tailored to build social profiles contrasted that consider the specificity of the local population (poor area with lot of seasonal workers). Two major axes organize those indicators related the economic status that ranges from poor to less poor, and the family structure composing the household opposing workers and/or migrants to traditional family units (parent with children): Level of richest of the household unit was indicated by level of incomes, access to goods, housing equipment and quality of the house construction or proxy such as education level. Worker/migrant or family unit composing household was defined mainly through Profile 1: the first profile contained all members pooled in class 1, designated as relatively poor family (RPF). A household unit was composed by a significantly higher part of single-parent families (woman living with at least one child aged under 12 years) with several social vulnerability indicators such as education, incomes and available goods. Furthermore, housing conditions were more deleterious regarding the size of the house and the quality of the construction. Malaria exposure of the members was more important, considering bed net use and more strongly attested by impact on infection based of reporting of malaria episode during the past 5 years (Table 3). Profile 2: the second class, named as worker/migrant profile (WMP), was significantly composed by more single men and small family units than the two other profiles. Despite a high level of incomes, members have a low level of education and low access to permanent goods, such as electricity, TV/radio, animal, or house equipment (water barrel, electricity, animal enclosure). Housing conditions reflected family structure with small homes but low occupancy rate. Housing conditions were not deleterious compared to the first class but was deleterious compared to access to permanent goods. Concerning malaria prevention, use of bed nets were limited, but significantly fewer members declared having a malaria episodes during the past 5 years. Other specific elements confirmed the suitably of this profile regarding hidden characteristic or mobility. A large part of seasonal workers or migrants living in Doreen Farm refused to give nominal information (name and/or surname). Furthermore, more than one of five household members of profile 2, who declared having malaria episode, reported that they were not at Doreen Farm, but half were from other countries such as Zimbabwe. Field notes indicated that men's clothes differ significantly in this group, as men wear long trousers, long sleeve and socks, which indicate regularly wearing working clothes. Profile 3: The third profile differed significantly from the first based on all economic indicators, access of good and housing conditions. Malaria prevention was better known as well as impact of the disease. However, the family structure of the household unit was similar to those grouped under the first profile and, differs significantly from those labelled as profile 2. By contrast, this profile was named as Relatively Rich Family (RRF). --- Knowledge, attitudes and practices toward malaria Concerning specific knowledge on malaria and malaria prevention, 83.9% of the survey respondents identified mosquitoes as the main vector of infection and for 56.2%, fever as the main symptom of the disease. However, if 75.7% of respondents knew about the Health District actions against malaria, few of them have been able to cite one specific intervention is currently conducted, such as rapid and free access to care (2.8%), bed-net (6.6%), house spraying (25.0%), provision of coils (2.4%) and educational training (6.0%). A wide majority of respondents (78.7%) declared being worried by malaria, but only 44.9% of them thought that malaria could kill, less than one of ten (28.3%) cited malaria as a major health problem for local population, ranged after diabetes and HIV/ AIDS. We observed that 15.2% said having never heard anything about malaria before the investigation. Many of the respondents reported night activities (85.9%). Among those nocturnal actives, 75.2% were back at home before 9:00 pm and were outside basically for socializing (77.8%). Fewer of them were outside for eating (20.2%), cooking (18.1%) or working (1.1%). When outside, respondents who were active at night reported specific preventive behaviours such as the use of spray (11.4%), fan (1.4%), coils (8.9%), fire (6.8%), specific beverage (1.1%) or an area cleaning (1.1%). It is noteworthy to mention than wearing long clothes at night was the most frequent preventive behaviours described (25.1%). A large proportion of respondents (49.4%) did not report any specific preventive behaviours during nighttime activities. When inside, only 5.2% of respondents reported using specific measures to prevent malaria such as closing windows and doors. Focusing on sleeping behaviours, despite a good opinion toward bed nets shared by nearly all the respondents (90.6%), less than a quarter of them (22.9%) declared using it even though three quarters of the respondents were sleeping in a bed (75.4%). Among those who reported sleeping outside when too hot (59.6%), 77.0% did not use a bed net. --- Malaria history Almost a third of respondents (31.9%) reported a malaria episode occurring during the previous 5 years, but in only 15.2% did the last episode occur from 2018 (as considered occurring during the previous year). Among them, 36.3% declared several episodes and almost all were at home for the last one (91.2%), had access to medical care (97.0%), and received treatment in the 3 days (55.2% during the two first days after symptoms). When comparing respondents that have reported having a malaria episode during the past years to others, no significant statistic differences appeared regarding social characteristics age, gender, education level or employment; including attitudes or knowledge measured or the night-time activities described and linked prevention behaviours. By contrast, as shown in Table 3, living in Nkotswi was significantly associated with reporting a malaria episode as well as living in a household unit with the RPF profile. Concerning attitudes, respondents that did not consider housing conditions as a main living problem, were more prone to declare having a malaria episode. Concerning preventive behaviours, if a tendency differs respondents declaring outside activities at night to staying inside, those using spray declared significantly more often a malaria episode occurring during the past year. After multivariate adjustments (Table 3), determinants associated with a higher propensity among respondents to declare having a malaria episode during the past first years were: sex, profile of the household, location sleeping behaviours and an opinion on living conditions. Being a man, living in Nkotswi, living in a poor family and sleeping outside when too hot were factors positively associated to a reported past malaria episode. In the opposite way, paying attention to housing conditions stayed significantly associated to no malaria episode report. --- Discussion This study used a community-based survey which was aimed at analyzing the association of individual behaviour and social organization factors with having at least one malaria episode and relate these findings to housing condition and health behaviour, among inhabitants in a malaria endemic area in the east Limpopo River Valley. --- Water proximity and quality The proximity of water generated by rainfall, water courses or due to human activities could partly explain the housing risk level due to house locations. In addition to the impact of the proximity to water bodies on malaria exposure, ongoing growing number of publications involved other factors associated to the water type or quality [13] even in drought period [14]. At least, ecological changes also impacting the distribution of Anopheles mosquitoes should be considered especially considering new species within the area [15]. --- Migration and seasonal workers: a pathway infection? The results identifying a migrant/seasonal worker as a major conclusive social profile underlined the importance of investigating cross-border workers as a specific group with their own lifestyle habits that could mainly affect their health status [16]. This is consistent with health policy investigations aimed at targeting emerging health phenomena associated with invisible social networks in a cross-border context in order to call for more cooperative interventions at the international level [17]. Cross-border labor mobility, partly driven by economic conditions, is a major challenge for public health strategies, particularly in South Africa, where migrants and rural workers are already considered as a high-risk group for infectious diseases, including HIV [18]. A recent study has shown that the specific social characteristics of these hidden labor population groups and their associated mobility resulted in an epidemiological pattern generated by particular infection pathways called "corridors" [19]. In the context of malaria, migration remains a major problem for control strategies [20]. --- Lessons from the results Focusing on housing conditions and health behaviours, and migrant workers, the relationship between vulnerability to infection and poor housing conditions did not seem to be as obvious or specific as already reported in studies along the Amazon River [21,22]. Housing interventions are mainly used to prevent malaria in any target population [23] and the fragility of frontier workers is more complex. The area method framing of the survey implied de facto the inclusion of many population groups and the comparative approach using a hierarchical selection of social profiles showed that deleterious housing conditions were not only associated with migration but also with the income level of sedentary families. Mobility [24], especially cross-border mobility [25,26], emerged as a more effective and endogenous determinant to characterize seasonal/migrant workers as a homogeneous population group radically different from others. Following the results obtained, the under-reporting of malaria episodes among migrants/seasonal workers, which remained significantly lower than among the poorest families, suggests that understanding malaria risk among them should be approached in two complementary and more specific ways using the above-mentioned "corridor" concept. Focusing on the level of individual risk, exposure to malaria may be part of the self-vulnerability principle of members. Thus, mobility affecting routine practices and health behaviours generates a relapse into prevention. For example, bedding-related behaviours, such as access to a bed net or a good, quality room, become emblematic changes in health prevention behaviours that generate new infections [27]. Given the level of risk and vulnerability of the community, another perspective could be put forward to address the question of asymptomatic transport of the parasite by mobile and hidden groups. This is to identify the epidemic pathway related to human mobility that could partly explain the persistence of malaria in the particular area. More broadly, because of changes in mobility associated or not with climate change or economic crises, infectious pathways are geographical gateways for the emergence and spread of malaria or other infectious diseases. The "corridor" concept could help to adapt the survey to show what is happening in a hidden world and also to implement specific interventions. --- Fundamental causes or health behaviour? In the last decade, social epidemiology, driven by the current craze for the theatrical theory of root causes developed by Link and Phellan, has strongly placed the social hierarchy caused by economic relations between groups at the center of the mechanism of health inequalities, which fully explains epidemic trends [28]. A critical point of view could be supported by the results of the study, which renew the interest in the relationship between external, social and collective constraints and individual freedom and subjectivity in health behaviours. The twostep analysis, carried out at the level of the household unit and the individual, allowed for the admittance of the effective and massive weight of socio-economic status on individual vulnerability to malaria, if one considers that the members of the household relatively grouped in the poorest profile declared to be the most exposed. Moreover, combining the KAP data in a single survey allowed for being limited to a structural and economic view and not to exclude individual subjectivity. Despite the growing influence of the structural approach, from the 1980s onwards there was a large body of literature arguing for the effectiveness of health promotion intervention based on health education by assuming a direct relationship between health knowledge and health behaviour. The current critical position of researchers, supported by the persistence of the epidemic and risk behaviours in the population despite years and years of implementation of health education programmes and policies, cannot undermine all the successes already achieved and the evidence for this type of action [29,30]. However, this success of health education does not exclude, as this study showed, that the risk behaviours that remain in the group and population regularly studied and trained are not negligible, for example sleeping behaviour or the use of prevention during the night. Other explanations could be discussed, including interindividual or community dimensions. Ongoing social interactions in the local cultural context should also be considered as potential factors directly or indirectly influencing individual health behaviour. Community beliefs, which were not assessed in this survey, need to be further investigated in order to identify potential causes for the persistence of risk behaviours. Indeed, comprehensive interventions to reduce the incidence of malaria in the general population should not rely exclusively on one approach. In addition to targeted health promotion programmes, actions at an intermediate level, such as social networks and communities, as well as at a broader level focusing on the economic wellbeing of society and the quality of life of the population should be pursued with a view to achieving complete eradication of malaria. However, a published paper promotes another possible way forward based on stricter global control [31]. The balance between human rights and individual freedom on the one hand, and centralized, strict control on the other, introduces a new aspect of weighing up the pros and cons of interventions, without forgetting the central concern along the way in choosing to change people's lifestyles, or more simply to enable them to reduce risk without the ambition to change their lifestyles. --- Gendered differences in life style, behaviours and powered relationship Other things being equal, a significant independent association between gender and malaria showed that men were more exposed than women. In contrast to previous report that women are more exposed than men [32], the findings may be attributed to men in the study area being frequent travelers due to border migration for work, increasing their exposure. However, social desirability bias and possible under recruitment of men during sampling cannot be excluded. Furthermore, apart from these exceptions, other types of explanation can be put forward and highlight the role played by masculinity in risky health behaviours from the perspective of the social construction of gendered identity over the course of a lifetime [33], assimilating male attitudes and stereotypes from the first stage of the construction of the self [34]. Recently, gendered health behaviours were observed during the COVID-19 pandemic showing that men in many countries were less likely to wear a mask in public [35]. More specifically, with regard to malaria transmission, it has been documented for years that in rural areas, men's lifestyles, especially night-time activities, contribute to deleterious health behaviours and increase exposure to the risk of becoming infected [36]. --- Limits and strengths In this study, collection of data was based on a day-time investigation, during working hours, which was a limitation as it drastically reduced the availability of part of the population and generated an imbalance on gender participation to the survey. Additionally, malaria history was only collected through reported experiences. Lastly, the lack of official data on local population makes it difficult to define quality of the sample in terms of representability. Conversely, originality of this survey provided updated information on a vulnerable South African population towards malaria still less surveyed, indeed unknown. Despites these limitations, significant differences of impact of malaria based on respondent's reports highlight the importance of global life conditions including location of housing, mobility related to working status, and economic level linked to housing conditions. The findings do not exclude individual characteristics as possible cause of a higher exposure to malaria. Results of multivariate modeling designated also subjectivity through health behaviours or gender. Furthermore, the design, which combines in one analysis two distinguished levels of factors, yielded a more complete interpretation of linkage between individual behaviours and surrounding context. --- Conclusion The malaria elimination agenda is a high priority for malaria endemic countries such as South Africa and the United Nations Sustainable Development Goals (specifically SDG 3.3). Thus, in context of endemic malaria, the findings of this paper suggest the importance of considering social and contextual factors in assessing malaria risk. In addressing malaria elimination in South Africa and accounting for the level of vulnerability and risk of the community, considering a potential 'corridor' could be an approach to inform targeted malaria interventions and studying asymptomatic carriage in migrant workers in malaria endemic areas. Using the Fundamental Causes Theory, this study proposes that these interventions should include the integration of malaria control policies and health behaviour prevention to reinforce the malaria elimination agenda. • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year --- • At BMC, research is always in progress. --- Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research? Choose BMC and benefit from:? Choose BMC and benefit from: --- Availability of data and materials The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. --- Code availability The codes are available from the corresponding author on reasonable request. --- Author contributions JG & RB, as leaders of the "GeoPal" project, elaborated the research question and planned the investigation. MKBD lead the questionnaire constitution, with the contribution of BH, JG, SP, TK, MR, TdJ and RB. Additionally, BH, SP and TK performed the field data collection. MKBD and JG performed the analysis. MKB.D wrote the first draft, with the contribution of SP, JG and RB. All authors read and approved the final manuscript. --- Declarations Ethics approval and consent to participate All procedures performed in studies involving human participants were in accordance with the Ethical International Standards of the Faculty of Health Sciences Research Ethics Committee, University of Pretoria. Ethics Reference No: 710/2018. --- Consent to participate Informed consent was obtained from each participant included in the study. --- Competing interests The authors declare no conflict of interest. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background Over the past decade, implementation of multiple malaria control strategies in most countries has largely contributed to advance the global malaria elimination agenda. Nevertheless, in some regions, seasonal epidemics may adversely affect the health of local populations. In South Africa, Plasmodium falciparum malaria is still present, with the Vhembe District experiencing an incidence rate of 3.79 cases/1000 person-years in 2018, particularly in the Limpopo River Valley, bordering Zimbabwe. To elucidate the complexity of the mechanisms involved in local regular malaria outbreaks, a community-based survey was implemented in 2020 that focused on the relationship between housing conditions and malaria risky behaviours.The community-based cross-sectional survey was conducted among the population of three study sites in the Vhembe District, which were selected based on malaria incidence rate, social and health characteristics of inhabitants. The household survey used a random sampling strategy, where data were collected through face-to-face questionnaires and field notes; to described the housing conditions (housing questionnaire), and focus on individual behaviours of household members. Statistical analyses were performed combining hierarchical classifications and logistic regressions.In this study, 398 households were described, covering a population of 1681 inhabitants of all ages, and 439 adults who participated in community-based survey. The analysis of situations at risk of malaria showed that the influence of contextual factors, particularly those defined by the type of habitat, was significant. Housing conditions and poor living environments were factors of malaria exposure and history, regardless of site of investigation, individual preventive behaviours and personal characteristics of inhabitants. Multivariate models showed that, considering all personal characteristics or behaviours of inhabitants, housing conditions such as overcrowding pressures were significantly associated with individual malaria risk. † Sean M. Patrick and Marc-Karim Bendiane are contributed equally. † Riana Bornman and Jean Gaudart share the last author position as they contributed equally.

experiences or circumstances caused by lack of power and privilege (Bhui, et al., 2019;Sederer, 2016;World Health Organization, 2021). During COVID-19, the definition of disparities remains true however they are exacerbated, and more students are experiencing them (Dorn et al., 2021). Effects of disparities include difficulty regulating emotions and decision making, threatening safety and a sense of control (National Scientific Council on the Developing Child, 2014). Parental stress, mental health, and poverty are predictors of child abuse (Crouch & Behl, 2001;Nair et al., 2003), behavioral problems (Child Welfare Information Gateway, 2019), poorer academic performance (Bick & Nelson, 2016), limited social connectedness and lower self-esteem (Herrenkohl et al., 2017), increased anxiety, depression, post-traumatic stress disorders (Sege et al. 2017), substance abuse, self-injury, and suicidal behaviors (Kaplow & Spatz-Widom, 2007). The long-term impact of COVID-19 induced disparities on all these factors is unknown, however the delivery of school counseling services can mitigate these disparities (Villares et al., 2012;Whiston et al., 2011, ASCA 2020). Therefore, it is crucial to understand how school counselors continued to implement their school counseling programs during the COVID-19 pandemic and what strategies they found beneficial and may want to continue. According to the National Academy of Education (2020), there is evidence to suggest that the closing of schools due to COVID-19 coupled with the transition to online learning has had negative impacts on students' academic performance, as well as their physical and behavioral health. Although, arguably the impacts on students' physical and behavioral health were expected, vulnerable children who regularly relied on services such as free and reduced meal programs and behavioral health resources were disproportionately impacted (Pattison et al., 2021). Catalano and colleagues (2021) also reported that students learning in high-needs districts This is an author-produced, peer-reviewed version of this article. The final, definitive version of this document can be found online at Professional School Counseling, published by SAGE. Copyright restrictions may apply. https://doi.org/10.1177/2156759X221105797. were significantly more likely not to complete their assigned work throughout online learning. Although COVID-19 has had an impact on every American, the African American population has also been disproportionately affected. African Americans had a higher rate of contracting and dying from the Coronavirus, more likely to be working as an essential worker which increases their risk of contracting the virus. Furthermore, when focusing on K-12 education, in January of 2021, more than half of all Black, Hispanic, and Asian fourth graders were learning in a fully remote environment whereas only a quarter of their white counterparts were learning fully remote (Camera, 2021). Of those students who were receiving remote education, 25% of fourth graders were only receiving two hours or less of live instruction and 5% of fourth graders were receiving no live instruction during their time as a remote learner (Camera, 2021). All students faced a challenge when schools closed and traditional learning methods were forced to readjust, but unfortunately the students that were impacted the most were students of different racial groups (Peek et al., 2021). The following study provides deeper insights into what COVID-19 induced disparities school counselors recognized and how they addressed them. More specifically, given that research related to COVID and school counseling is developing, we wanted to investigate the differences of school counseling programs and classification of schools and how they addressed these disparities. Therefore, the purpose of our study was to understand how school counselors adapted their comprehensive school counseling programs to meet K-12 students' academic, social-emotional, and career development. Additionally, we examined differences between ASCA aligned non-ASCA aligned schools and Title I and non-Title I schools. Our primary research questions were: (1) What COVID-induced disparities did school counselors identify?; (2) How did school counselors adjust their school counseling services due to COVID-induced disparities?; and (3) On average, did school counseling activities differ during the pandemic between ASCA-aligned schools and non-ASCA-aligned schools?; and (4) On average, did school counseling activities differ during the pandemic between Title I and non-Title I schools? --- Methods We utilized a convergent mixed methods design (Kroll &Neri, 2009), following the university's institutional review board (IRB), to answer our research questions. We selected this to obtain a more complete understanding of our data and to corroborate the results from different data collection sources. To gain an in-depth understanding of how school counselors identified and adjusted services in response to COVID-induced disparities, we utilized a convergent parallel design, a type of mixed methods design. In this design, the researcher concurrently conducts the quantitative and qualitative elements in the same phase of the research process, weighs the methods equally, analyzes the two components independently, and interprets the results together (Creswell & Pablo-Clark, 2011). With the purpose of corroboration and validation, the research aims to triangulate the methods by directly comparing the statistical results and qualitative findings. We conducted thematic analysis to analyze patterns in the open-ended survey responses (Boyatzis, 1998;Braun & Clarke, 2006) and to understand what disparities and how school counselors responded. Additionally, we used t-tests to examine if there was a difference between types of schools and their school counselor activities to further explore the who/where aspect of our data. --- Participants A national representative sample of school counselors (N = 589) belonging to the American School Counselor Association participated in the current study. The sample primarily identified as female (71%), White (84.4%), between the ages of 25 to 55 (74.4%), earned a This is an author-produced, peer-reviewed version of this article. The final, definitive version of this document can be found online at Professional School Counseling, published by SAGE. Copyright restrictions may apply. https://doi.org/10.1177/2156759X221105797. master's degree (78.4%), and had between two and ten years of experience (50.6%). Half of the participants reported having formal training of the ASCA model; however only 205 participants stated their school is somewhat aligned with the ASCA model and 178 indicated mostly. Table 1 provides additional information about the sample demographics. --- Research Team & Positionality The research team consisted of a tenured Professor and Associate Professor, and two doctoral students from two Southeast universities with Council for the Accreditation of Counseling and Related Educational Programs (CACREP) approved programs. All four researchers have worked as professional school counselors and are members of the American School Counselor Association (ASCA). Prior to beginning our work together, we discussed our positionality related to the topic of our study. Positionality refers to a researchers' values and beliefs related to a topic and how that may influence the research process (Holm, 2020). Considering three of us were previous school counselors and one of us is a current school counselor, our position centers around valuing the role of the school counselor as a mental health professional in the schools and the importance of implementing a comprehensive model. We also reflected on what we think we would've done and what we are currently doing in our role. We anticipated that participants would provide innovative ways to deliver their school counseling models that may be useful even after COVID-19, and we hoped that participants were able to counsel students and not only focus on other duties. We recognized that our values and beliefs may influence our coding process and discussed how we valued looking at the data from a qualitative and quantitative lens. --- Data Collection This is an author-produced, peer-reviewed version of this article. The final, definitive version of this document can be found online at Professional School Counseling, published by SAGE. Copyright restrictions may apply. https://doi.org/10.1177/2156759X221105797. In November of 2020, a nationally representative random sample of professional school counselor practitioners (4,910) belonging to ASCA received an invitation to participate in the current study along with an electronic link to complete the electronic SurveyMonkey survey. The eligible participants received follow-up requests for participation two weeks and four weeks after the original invitation. The response rate was 12%. We collected qualitative data using four open ended questions in the survey. These questions asked to briefly describe, Counselor Activity Scale (SCARS; Scarborough, 2005). The SCARS is a self-report measure that determines types and frequency of counseling services (i.e., including individual and group counseling, curriculum lessons, coordination, consultation, and other activities) being delivered in the school setting. Frequency is measured on a 5-point Likert rating scale, ranging from 1 -(never) to 5 (routinely). We asked counselors to answer the questions using the following prompt, "Please select the rating that best indicates the frequency you performed each service since the COVID-19 school closure in Spring 2020." The demographic questionnaire, developed by the first two authors, and included questions about the school counselors' age, gender, ethnicity, highest degree earned, school level, program characteristics, alignment with and training in the ASCA National Model, caseload, Title 1 status, and methods for communicating and learning during the 2020-2021 school year (see Table 3). --- Data Analysis & Trustworthiness Qualitative Data Analysis. The research team followed the six stages of thematic analysis (Braun & Clark, 2008) to analyze the data: (1) familiarize yourself with the data, (2) generate initial codes, (3) search for themes, (4) review themes, (5) define and name themes, and ( 6) produce a report. We first familiarized ourselves with the data by reading all of the responses to the short answers on the survey. Next, we generated the initial codes (i.e., one word or a short phrase description of each part of the data). We developed 356 initial codes, and then collapsed this list to a second version of initial codes which consisted of 90 codes. Next, we grouped the codes to develop themes. We created 17 initial themes related to the research questions and reduced them to six themes and three sub themes after discussing the themes with the auditors. Throughout the process we revisited the raw data to ensure it was represented accurately. We then developed descriptions of each theme and we report our findings in the following section. To ensure trustworthiness, two of the researchers conducted the thematic analysis while the other two served as auditors. The external auditors reviewed the initial codes during step two and the themes during step five. Additionally, we reviewed data independently and then discussed our coding collectively to develop a consensus of findings. We acknowledged our biases regarding our views (i.e., our positionality) prior to reading the open-ended question response and acknowledged them throughout (Creswell, 2007;Hays & Wood, 2011), as part of the first step of thematic analysis. Quantitative Data Analysis. Descriptive and inferential statistics were run using SPSS Version 27. We conducted independent t-tests to compare participants' responses on counseling activities among those at (or not at) a school adhering with the ASCA National model. We also performed independent t-tests to compare participants' responses on counseling activities among those at (or not at) a Title I school. Levene's test was used to confirm equality of variances. Effect size statistics were evaluated using Cohen's (1988) criteria in which d 3 <unk>0.20 is interpreted as a small effect, d 3 <unk>0.50 is interpreted as a medium effect, and d 3 <unk>0.80 is interpreted as a large effect. --- Findings We describe the findings using the six themes and three sub themes we identified from the data analysis process and examined the results from the SCARS. All descriptive data of the SCARS subscales can be found in Table 3. --- Qualitative Findings Our first research questions is, What covid induced disparities did school counselors identify? Three themes and one sub theme connect to this research question: (a) difficulty accessing and connecting with students and caregivers (b) limited direct school counseling services (sub theme: highlighted lack of role understanding by administration), and (c) increased student mental health concerns, and (d) lack of role understanding by administration. The first theme difficulty accessing and connecting with students and caregivers, describes how COVID-19 made it challenging for school counselors to communicate with students on a consistent basis. More specifically, participants emphasized that students and caregivers had limited or no access to internet limited their points of contact with them because a lot of interactions were being done virtually due to various learning models. Participant 9 stated "I feel that my job consists mostly of putting out fires this year--there is so much transition with quarantine, hybrid, online, and in-person school, planning is difficult, as we never know if/when we will be in school in person or at home, and which students will be in person/at home, as well as who will have internet and who won't at any given moment" Additionally, since all meetings with students had to be scheduled due to them being virtual or to follow face-to-face protocol, school counselors described the inability to have informal check-ins with students that they had prior to the pandemic. For example, a school counselor (participant 166) said when it comes to students "not being with them in person is just not the same. I can't really get to know them or talk with them the way I used to. I used to like to go visit them in the cafe or in a class, but I can't now". Theme two limited direct school counseling services (theme two), includes how students received less direct services (i.e., instruction, appraisal and advisement, counseling). Participants described how they needed to focus on crisis as opposed to being able to deliver a preventative comprehensive program. Participant 136 describes how they've navigated student needs during COVID-19 "There is no time for lessons this year. Students are in crisis -their families are falling apart, they are failing classes (kids who have never failed before), they no longer see friends...Their anxiety, grief, and fear is real. They are not afraid of the virus; they are afraid of how life is changing. The want normalcy". Furthermore, participants shared that decreased opportunities for school counselors to offer direct services was due to an increased workload of non-counseling duties. For example, participant 285 said "My delivery has already been forced to change due to new policies preventing many of my usual services from happening (no group counseling; no group or individual counseling for virtual students; extra non-counseling duties), and I feel extremely frustrated about it. It feels as if my district neither understands nor values my job or the services I provide to students and families." The sub theme within theme two, highlighted lack of role understanding by administration. Participants articulated that COVID-19 made it clearer that some administration did not understand appropriate duties of a school counselor. For instance, participant 151 said "The principal delegates non-counseling duties that take me away from student contact and therefore not enough time in each day to implement counseling responsibilities". Furthermore, participant 99 states "I am asked to do many non-counseling duties. When there are counseling related issues that arise, I am often unavailable because I am covering for someone else in my building. This has minimized the importance of my role and I am finding that staff members overlook the counseling department when we are actually needed. However, without the support of supervisors and administration, I don't see any of this changing soon ". The third theme is, increased student mental health concerns. Participants reported that due to COVID-19 students exhibited higher levels of social/emotional distress. Participant 405 discusses the concerns they have for students "We have only done face to face learning and the anxiety levels, mental health crises, and suicidal intakes are extremely high. It's possible we may need to go remote as our COVID numbers are increasing. I'm not real sure how that will look for counseling those individual students that are struggling". Furthermore, participant 34 speaks to how being fully remote is difficult; "students are still 100% remote and they are struggling significantly with mental health and motivation". This is an author-produced, peer-reviewed version of this article. The final, definitive version of this document can be found online at Professional School Counseling, published by SAGE. Copyright restrictions may apply. https://doi.org/10.1177/2156759X221105797. For our second research questions, How did school counselors adjust their school counseling services due to COVID-19 induced disparities?, we were most interested in action oriented ways school counselors altered their delivery and if these adjustments would be Both have made it possible to communicate with students who are distance learning. I have also done several presentations to parents using the technology that would normally done in person" and participant 370 echoes the same sentiment "Technology has been my friend. I have been able to use Zoom, Google Meet to have small groups and individual sessions as well as parent and colleague meetings. Using Google Slides, Google Forms, and Loom video to help deliver content and provide resources to students and families. "Additionally, school counselors stated how they utilized new forms of technology to improve their services, participant 203 states "I have found new technology for recording lessons in creative ways! Loom and screencastify are helpful. I also plan to experiment with other virtual platforms. Google Meet has also been a valuable resource for reaching remote students". The first sub-theme within theme four is, increased flexibility of virtual classroom guidance consists of ways participants adjusted their delivery of This is an author-produced, peer-reviewed version of this article. The final, definitive version of this document can be found online at Professional School Counseling, published by SAGE. Copyright restrictions may apply. https://doi.org/10.1177/2156759X221105797. guidance lessons using technology. Participant 525 stated "I actually feel that having more of an online presence is making some aspects of my job (delivering information, providing extra guidance lessons beyond the mandated ones) makes my job easier and more efficient.". The second sub-theme, implementing creative ways to access students and caregivers, describes strategies school counselors used during COVID-19 to communicate with their students and the student's caregiver. Participant 282 discusses the ways in which they engaged with students and their families "For those students that have access to tech and internet, I am able to maintain services and support. I am able to participate in meetings with parents via zoom, I communicate with most of them through email, and I created a Seesaw classroom for parents". The fifth theme, increased collaboration with teachers, consists of how school counselors and teachers collaborated to provide academic and mental health support. Participant 387 mentions how they work closely with their teachers, "Classroom teacher & advisement teachers have been my eyes and ears. They communicate any and all student concerns with me. I also collaborate a lot with other support staff". The final theme, prioritized social/emotional health and identified as the expert, describes how the participants and others recognized the importance of social/emotional health and that school counselors were recognized as the expert. Participant 388 states "Administration is relying more on my input as a school counselor and realizing that the areas I've been pointing to for a few years are actually in need of attention", participant 452 says "I feel as though my administrators are starting to understand how much work counselors actually do which is a positive impact on our relationship" and lastly, participant 10 says "Administration has been essential. Their support before I started as well as during my time in the school has made delivering the school counseling program easy". --- Quantitative Results School counselors were asked the degree to which their school adopted and adhered to a comprehensive, developmental school counseling program that aligns with the ASCA National model. Participants (n = 263) responding not at all or somewhat were coded 0 (indicating none to little alignment), whereas participants (n = 230) responding mostly or completely were coded 1 (signifying a high degree of alignment). Levene's test confirmed equal variances could be assumed for counseling (F =.10, p =.752), consultation (F = 3.43, p =.065), curriculum (F =.70, p =.403), and coordination (F = 1.63, p =.203) but not other activities (F = 7.78, p =.006). Results from independent t-tests show that counselors at schools adhering to the national model spend significantly more time counseling (t( 491 4. Participants were also asked whether their school was a Title I school (no = 266, yes = 244). Independent samples t-tests were used to determine if there were significant differences in counselors' activities based on Title I status. Levene's test show that equal variances could be assumed for all subscales of the SCARS: counseling (F = 1.72, p =.190), consultation (F = 0.11, p =.736), curriculum (F = 1.76, p =.185), coordination (F = 2.56, p =.110), and other activities (F =.26, p =.614). Results from the t-tests illustrate that counselors at non-Title I schools engage in significantly more curriculum activities (t(508) = 3.24, SE D =.10, p =.001, d = 1.14). No other results were significant (see Table 5 for details). --- Discussion Our study shows how school counselors identified COVID-induced disparities and pivoted their services and activities to meet the needs of teachers, students, and parents during the pandemic. Our qualitative results reveal several ways school counselors adjusted and took action to address these disparities. We found that embracing and learning to utilize new software and technology while capitalizing on systems they were familiar with, school counselors could minimize student challenges to complete their work. Similar to Catalano et al., 2021 our findings support the differences amongst types of schools and the transition to online instruction. We found that school counselors increased their collaboration with teachers and other professionals within their building, which may have helped teachers recognize the role of school counselors. Although not conducted during COVID-19, Limberg et al, 2021 also found the importance of collaboration has on role recognition. Considering the effects of poor mental health in children (Kaplow & Spatz-Widom, 2007;Sege et al. 2017), school counselors addressed social-emotional and mental health needs of students creatively through virtual classroom lessons, counseling, and ways in which they accessed students and parents. Additionally, throughout the pandemic, school counselors were recognized for their expertise as a mental health professional and mental health was prioritized. Nonetheless, school counselors were assigned additional duties related to COVID-19 such as following up with students who have not participated in virtual classes or have not returned since schools reopened, attendance/check-ins, and an increased responsibility regarding social-emotional learning (SEL) implementation (ASCA 2020). Overall, our qualitative themes show school counselors increased their counseling, consultation, curriculum, coordination, and other out-of-role activities to address COVID-induced disparities. Our quantitative results offer some context to the qualitative findings by revealing school characteristics of the school counselors providing more services. On average, school counselors that identified their school as ASCA aligned spent more time doing appropriate duties as determined by ASCA. Conversely, school counselors that identified their school as not ASCA aligned spent more time on average doing incongruous activities. We found similar school counselors at Title I schools did less curriculum activities on average, which is concerning considering the classroom guidance was the primary way to address social-emotional concerns. --- Implications Our study has implications for both current and future school counselors and the school counseling profession overall. For many participants, COVID lead to them being recognized as a mental health expert for the first time. Therefore, current and future school counselors and school counseling researchers can continue to focus their attention on the short-term and long-term mental health impact COVID-19 had and continues to have on student's social-emotional development. Furthermore, since school counselors placed their focus and attention on student's social/emotional development, academic gaps and career development need to be reconsidered. The school counseling continues to be a point of confusion, but our participants described how COVID led to more role understanding and collaboration. Therefore, our findings highlight the necessity for school counselors to be assigned to appropriate school counselor duties at their respective schools. Additionally, current and prospective school counselors should continue to promote and facilitate collaboration with teachers and administrators but also other helping professionals within the school building and other community stakeholders, energizing each professional's role for the benefit of the students. The implementation of the ASCA National Model through remote learning and the utilization of technology calls for continued innovation by school counselors to address student's social/emotional, career, and academic development. School counselors' abilities to move their comprehensive school counseling programs onto virtual and hybrid platforms accentuates the opportunity to expand comprehensive programs through the utilization of online platforms and other software and technologies. Overall, given the little to no preparation school counselors had to take a comprehensive school counseling program into remote learning, school counselors responded with resilience as they relied on their training and preparation to meet the needs of their students through the COVID-19 pandemic. We found that school counselors that aligned with ASCA were more likely to be counseling, consulting, implementing curriculum, and coordinating during COVID which is supportive of the ASCA model. But as the mental health needs of student increases the role may need to be revisioned because it may not be possible for school counselors to continue to do all activities effectively. Although COVID-19 research related to school counseling is in its early stages, we recommend that the impact of COVID-19 and the emphasis of mental health can not be overlooked when the ASCA model is revised --- Limitations and Directions for Future Research There are limitations in our study that warrant discussion. First, our data was cross-sectional, meaning causation cannot be determined. As such, we also did not have the ability to ask follow-up questions regarding open-ended responses on the survey. We recommend future researchers conduct a follow-up survey post COVID-19 and/or conduct a longitudinal study to determine COVID-19 induced disparities long term impact on students. Another limitation is the use of self-report data. Research has demonstrated that self-reported measures may have method bias, as self-report often acts as a primary source of measurement error (Podsakoff et al., 2003). Future research should aim to extend this work with a longitudinal design and/or by using observational or behavioral measures, and measures from other reporters, when examining the relations in our study. For example, employing rubrics to measure quality of school counselor activities may be helpful. Another limitation is that we dichotomized ASCA-alignment from a 4-point ordinal scale. In the current study we simply wanted to examine average differences in the type and amount of activities school counselors reported doing during the pandemic. However, future research needs to better quantify the relationship between ASCA-alignment and school counselors' activities using regression or latent variable methods. Furthermore, other school characteristics (e.g., case load, grade level) may have stronger relationships with the activities school counselors perform and, as such, should be explored. Another limitation involved our sample, which represented 12% of the original random sample of professional school counselors. As such, results may not generalize to all school counselors. The COVID-19 response has been fluid and evolving with little consistent response across school districts, which has impacted school counselors' capabilities and actions. This likely contributed to some inconsistencies in the responses.

We utilized a mixed methods design to understand how a national sample of 589 school counselors adapted their approach to address K-12 students' academic, social-emotional, and career development during the COVID-19 pandemic. More specifically, we examined how COVID-19 induced disparities influenced school counselors' delivery of services. We identified six themes using thematic analysis and conducted a t-test to further understand what services and strategies counselors were actively using six months after the onset of COVID-19. Our findings suggest that school counselors continued to adjust their comprehensive school counseling programs to address COVID-induced disparities.

Global health research and practice have been-and are increasingly-criticised for their reproduction of a hierarchy of knowledge that subordinates especially rural populations in low-income and middleincome countries (LMICs) to Western biomedical logic and to local medical elites. 1,2 For example, a report by Horton 3 about a 2013 workshop on neocolonialism included positions that 'western imposed (psychiatric) diagnoses, which ignore local understandings of distress, are "what imperialism is all about"', while a recent Nature editorial commented that the inclusion of traditional Chinese medicine as a chapter in WHO's International Classification of Diseases is 'likely to backfire (and) risks legitimising an unfounded underlying philosophy'. 4 Examples like these are reminiscent of critiques of modernism and neocolonialism in international development. 5 More than 30 years ago, Arturo Escobar described that, 'types of power and knowledge are being deployed (through Western disciplinary and normalising processes) in the Third World which try to insure the conformity of its peoples to a certain type of economic and cultural behaviour'. 6 Considering the salience of health in the 2030 Sustainable Development Agenda, which aspires to be a more inclusive representation of global development than its predecessor the Millennium Development Goals, 7 the persistence of a hierarchy between Western and biomedical knowledge on the one hand, and local and non-biomedical notions of health in LMICs on the other hand, would be problematic. Antimicrobial resistance (AMR) is a global health topic that exemplifies this tension and the persistent hierarchy between 'global' (ie, Western biomedical) and 'local' knowledge. A top priority item on the global health agenda, the WHO Director-General has declared AMR (or 'drug resistance') as 'one of the most urgent health threats of our time'. 8 AMR involves the evolution of microbes like bacteria and viruses to withstand the medicine that humans use to treat them, thereby making them increasingly 'drug resistant' and the medicine less effective. This is in principle a naturally occurring process, but humans accelerate it by using antimicrobials (antibiotics, antivirals, antifungals, etc) in human and veterinary medicine, in agriculture, and through their leakage into the environment. The global health response to AMR mirrors the biomedical interventionism with which postcolonial medicine has been characterised. 9 Global policies to address AMR define it 'as a global threat emerging from LMICs' 10 and foreground individuals' behaviour as one of the principal problems of a subject that connects humans, animals and the environment. 11,12 Remedial action focuses thus on awareness and education campaigns to change population behaviour especially in LMICs, 13,14 implying that knowledge and practices that deviate on February 17, 2024 by guest. Protected by copyright. original research from a Western biomedical rationale-for instance, care from traditional healers during an illness-are problematic and require rectification. [15][16][17] The global health response to AMR therefore continues to champion biomedical knowledge and to neglect or otherwise subordinate the knowledge, beliefs and identities of rural communities in LMICs. At the same time, the growing emphasis on 'public engagement' (a form of 'patient and public involvement') among health researchers and medical research funders has in principle the potential to break down or at least undermine hierarchical relationships between medical elites and local populations. [18][19][20][21][22] However, rather than establishing dialogue and challenging this hierarchy, public engagement activities have thus far primarily been instrumentalised to impose global health agendas on local populations (eg, through theatre plays 23 ; see 'Background' section for further explanation). This need not be the case. The medical humanities intersect with the popular practice of public engagement, emphasising in particular creative co-production research that uses methods including theatre, storytelling or artistic production (we related to the medical humanities in this paper especially through creative, participatory and dialogical approaches from within the visual arts and social sciences/cultural studies branches of the medical humanities). 24,25 Medical humanities scholarship thereby shares an interest in the role of experience and expression of illness and healing, [26][27][28][29] and it has an important role to play in broadening debates away from reductionist towards a holistic and contextsensitive understanding of complex medical phenomena like AMR. [30][31][32] At the same time, explorations building on artistic and creative processes do not exist in opposition to but rather enable dialogue with established medical notions and conventions, giving them new perspectives and interpretations. [33][34][35][36] Building on this logic, this paper presents a case study of public engagement under the umbrella of the medical humanities in which we examine new and locally grounded perspectives on the sociocultural context of AMR and its related topics of medicine use and health systems. Our research question is, 'Can medical humanities approaches challenge hierarchies and promote engagement in global health research on antimicrobial resistance?' To answer this question, we employed a case study design to illustrate the diverse ways in which public engagement can reveal perspectives and forms of knowledge that are not typically incorporated into AMR research and policy-in particular cultural knowledge, personal experience and 'lay' sensemaking surrounding health systems and healing (including medicine use). [37][38][39][40] We focus on these aspects of knowledge as they are typically replaced by biomedical assumptions in AMR policies, whereby our case relates in particular to the sociocultural context of human antibiotic use (a major determinant of AMR). We consider in the context of northern Thailand co-production workshops, storytelling and dialogue surrounding artistic display as vehicles for a bottom-up process for knowledge generation under the umbrella of participatory (rather than instrumental) engagement. 41 The objective of this paper is thereby not to influence global and local AMR policy directly, but to offer a case that demonstrates how the inclusion of medical humanities methods in global health research (via public engagement) can open up new and locally grounded perspectives for thinking about the complex issue of AMR and its related and seemingly established topics of medicine use and health systems. We thereby contribute to a small but growing body of medical humanities research that relates directly to AMR and the broader interface of humans and health systems. [42][43][44][45][46][47] --- BACkgRound Research on population health behaviour in AMR mobilises conventional public health research methods. We review these methods in this section and argue that they risk reproducing a hierarchical relationship which subordinates local medical knowledge and traditional forms of healing in LMICs to the biomedical model of health that is prominent in high-income countries and among local medical elites-which underscores the important role of the medical humanities in this space. 48,49 Standard forms of AMR knowledge generation with particular relevance to the current case are public awareness surveys and knowledge, attitude and practice (KAP) surveys. For example, one of the most influential documents in the context of awareness-related global AMR policy is the WHO's Antibiotic resistance: multi-country public awareness survey. 50,51 Based on online and face-to-face surveys in 12 countries and using a range of knowledge-testing questions, the survey argues that, 'it is critical that people understand the problem (of drug resistance), and the way in which they can change their behaviour'. 52 KAP surveys are similarly prominent in the field of public health AMR research, 53 including (with a focus on antibiotics) contexts as diverse as the studies by Belongia et al 54 Public awareness and KAP surveys as mainstream tools for global health knowledge generation typically conclude that awareness needs to be raised, and call on individuals' responsibility to change antimicrobial-related health behaviours. [57][58][59][60][61] A major problem of these approaches is that problematic antimicrobial use is framed in terms of knowledge and attitudes, and the notions of what constitutes 'desirable knowledge' are typically imposed by health researchers who implicitly assert superiority of modern over local and traditional forms of knowledge. 62 Yet, such studies devote little if any concern towards the social and ethical antecedents of current behaviour (and the corresponding consequences of intervention) in LMICs-for instance, the historical role of drug promotion, the precarious balance between antimicrobial 'access and excess' or culturally specific notions as to what constitutes 'good care'. [63][64][65] In contrast, recent social sciences and interdisciplinary research on AMR has pointed out non-individual components of antimicrobial use that reflect on the broader sociocultural context of AMR. For example, Chandler 66 describes the interconnectedness of AMR across the domains of human, animal and environmental health and the social role of antimicrobials as 'infrastructure' that contributes to the functioning of market economies; Hinchliffe et al 67 indicate how Bangladeshi shrimp and prawn farmers adapt their antimicrobial use in response to economic uncertainty and perceived disease risks and Chuengsatiansup and Limsawart 68 analyse the tensions between administratively defined borders and their history, enactment and continued negotiation in the control of drug-resistant tuberculosis in the border area of Thailand and Myanmar. However, the global health discourse around AMR has not yet been infused with these perspectives and continues to portray a dominance of Western high-income countries' priorities and solutions, LMICs as source of a global problem and individuals' knowledge and behaviour as critical targets for intervention. 69,70 Alternative forms of knowledge generation could broaden the global health discourse around AMR. --- original research into the sociocultural context of AMR and thereby provide an avenue to challenge the mainstream framing of AMR and the implied hierarchy of medical knowledge and practice. However, qualitative research in public health often remains limited to examining people's attitudes and knowledge akin to quantitative public awareness surveys. [71][72][73] Similarly, 'participatory methods' or 'public engagement' in public health research are typically instrumental means with an emphasis on health education provision, on'mobilising' communities to change their health behaviour, and/or on building trust and legitimacy of health research locally. [74][75][76][77][78][79][80][81] Global health scholars have argued that qualitative research and public engagement involving the co-production of knowledge with inputs from target populations can broaden understanding and open new directions for debate. 82,83 Yet, in public health and global health research, these methods have a tendency to retain biomedical assumptions, to fall short of their potential to challenge hierarchies of knowledge and even reproduce neocolonial relationships in global health. 84,85 This persistent challenge underlines the important role of the medical humanities. 86,87 Aside from historical analysis or literary interpretations, 88,89 common methods in the medical humanities have included narration, artistic means such as theatre and photography and also directed qualitative approaches like focus group discussions to interrogate the nature of illness, healing and people's relationship to health systems. [90][91][92][93][94][95] Unlike common public health research methods and instrumental forms of public engagement, a medical humanities approach thereby helps reveal subjective truths that are often overlooked in biomedical perspectives, and thus challenge hierarchies of knowledge between clinical 'experts' and non-medical 'lay' people. [96][97][98][99] For example, Cole and Gallagher 100 argue in the context of clinical neuroscience that the medical humanities, in the form of firstperson narratives, 'can complement the clinical third-person approach, and in some cases lead to better understanding and point towards further empirical work itself '. Our focus on the medical humanities in the present study thereby builds on the understanding that the joint consideration of medicine, participatory research and artistic forms of expression enables us to generate new and critical knowledge about complex global health phenomena like AMR. [101][102][103] --- MATeRIAlS And MeThodS --- Study design We employed an embedded case study research design, in which we focused on three public engagement components of an interdisciplinary research project on rural health behaviours and drug resistance in northern Thailand and southern Lao PDR (each component is a'subcase' embedded within the larger public engagement case study). 104,105 Case studies in the medical humanities help illustrate conceptual and methodological applications, present close-up experiences of (often otherwise neglected or marginalised) research participants and open up new perspectives on medical topics. [106][107][108][109][110][111] These designs also often combine different methodological approaches within the same analysis (eg, Hume et al, 112 who combine insights from historical, ethnographic and creative research). In relation to the research question, our starting assumption for the case study analysis is that co-produced local inputs can (and should) broaden the understanding of the sociocultural context of AMR. We included the three subcases to illustrate the diverse ways in which public engagement can reveal perspectives and forms of knowledge that are not typically incorporated into AMR research and policy-in particular cultural knowledge, personal experience and 'lay' sensemaking. [113][114][115][116] We situate the case study research design firmly within the medical humanities. Aside from using medical humanities methods of knowledge generation (see section 'Data collection and analysis' for further explanation), our public engagement activities operationalised the phenomenological and intersubjective approach of the medical humanities through qualitative and participatory techniques that treat participants as expert informants and, where possible, let them choose the terms of the conversation. 117,118 At the same time, the case study considers the dialogue with the medical profession by relating qualitative interpretations to quantitative patterns of health behaviour and to the global health discourse around AMR, and by considering the costs and risk of incorporating our approach into public engagement practice. 119 Our mixed-method approach is thereby compatible with both the case study research design and its application in the medical humanities. 120 Several recent examples have adopted approaches similar to our current study: for instance, Macnaughton and Carel 121 use case examples to describe how a critical medical humanities approach can help understand the phenomenon of breathlessness beyond its clinical dimensions-based, among others, on reflections from collaborative meetings between medical researchers, artists and social scientists. Barbieri et al 122 present a qualitative case study of patients in Italian paediatric wards, whereby the production and interpretation of semifictional autobiographic narratives offered child patients a channel to communicate to clinicians their personal experience of illness and healthcare needs. A case of HIV-related public engagement in South Africa by Treffry-Goatley et al 123 further uses a combination of qualitative (focus group discussions, observation) and quantitative methods (surveys) to understand the role of digital storytelling in promoting health literacy. We therefore build on an established body of work in the medical humanities. --- Case overview The public engagement components in this project involved workshops in three villages and the collection and exhibition of photographic stories of healing in Chiang Rai province in northern Thailand. The workshops took place in three villages in Chiang Rai province, namely Chiang Rai village, Chiang Khong village and Mae Fah Luang village (pseudonyms named after the districts in which they were located). The two main objectives of the workshops were, first, to share with villagers some ideas and concepts about antibiotics and drug resistance (based on material from the WHO 124 ), without assuming that their current knowledge and behaviours were in any way deficient; and, second, to enable our research team to learn from the villagers about the local context of medicine and healing and how the antibiotic-related information has been received. The half-day workshops involved 20-35 adults per village, who were recruited in a combination of purposive and snowball sampling to ensure spatial and ethnic diversity of the workshop participants (however, all of the attendees had Thai language abilities, which limited the representativeness of the workshops). 125 The workshop activities involved, in chronological order: The workshops took place alongside larger health behaviour surveys in Chiang Rai. Feedback from our field research team also revealed that our survey questionnaire did not capture important aspects of local healing. As the project surveyed 72 villages in Chiang Rai, the team shared experiences of herbalists curing broken bones and spiritual healers summoning ghosts. What was the meaning and significance of these practices, and what would the survey category 'traditional healer' mean for villagers? To investigate these questions further, the research team revisited some of the villages to document local stories of healing from the perspective of traditional healers (with their permission). The resulting narratives were exhibited in the 'Tales of Treatment' photo exhibition series in Bangkok (Art Gallery g23), Chiang Rai (Tai tea shop and bar), Oxford (Green Templeton College) and Coventry (Warwick Arts Centre) between July 2018 and March 2019 (note that the international exhibition was not intended to 'validate' local knowledge through urban elites but to encourage dialogue and reflection about what healing is on an international scale with diverse audiences). 127 The content of the exhibitions varied slightly by location (considering available space and logistics; see figure 1 for illustrations) and included: <unk> Fifteen photographic stories with Thai/English captions and guided tours by the research team (all four exhibition sites). <unk> Exhibits of pharmaceuticals and medicinal plants (Bangkok, Chiang Rai, Oxford). <unk> 'Medicine wall' of pharmaceutical images and local notions of medicines (Bangkok, Chiang Rai, Oxford). <unk> Programme booklets and souvenir postcards (Coventry). <unk> Research fieldwork photographs (Bangkok, Oxford). <unk> Research infographics, word clouds and/or animated presentations (Bangkok, Chiang Rai, Oxford). --- data collection and analysis We used primary qualitative and quantitative data to document and explore how knowledge co-production challenged our own expectations as survey researchers and contributed to new perspectives on the sociocultural context of AMR. We will present our research according to subcases representing the three medical humanities methods contained in this case study: participatory co-production, photographic storytelling and dialogue between researchers and the public. The groups with whom we engaged included Chiang Rai villagers, traditional healers in Chiang Rai and urban audiences of the photo exhibitions in Thailand and the UK. The data sources and involved groups are described in the remainder of this section, and a summary is provided in table 1. 128 In the first subcase, we collected observational data to document the inputs from workshop participants and to formulate locally grounded hypotheses about medicine use (documented in digitised fieldnotes and audio-visual material including photographs and video). To illustrate the applicability of the co-produced knowledge for understanding the sociocultural context of AMR, we tested these hypotheses using rural health behaviour survey data from the main research project (for details on sampling strategy, please refer to the published study protocol referenced here 129 ; survey data collection was through electronic questionnaires on tablet computers in face-to-face survey interviews). In brief, the rural health behaviour survey data used in this analysis had two components, namely: a. Two rounds of complete census surveys in the three workshop villages. As complete census surveys, the entire adult population of the three villages was invited to participate in both survey rounds (total adult population est.: 694), with an average response rate of 91%. The workshops took place in between the two survey rounds but were not intended to be an 'intervention'-rather, they were a public engagement activity and were therefore not implemented as a (quasi-) experiment. b. A provincial-level representative rural survey. This survey used a three-stage stratified random sampling design to represent the rural adult population of Chiang Rai province (522 000 adults according to census data). 130 We selected five districts purposively (stage 0; selected for diversity within the province), within each of which we selected six primary sampling units randomly (stage nearest town), followed by the random selection of at least 30 households per primary sampling unit (stage 2; interval sample and stratified by village segments), and by the simple random selection of one per every five household members (stage 3). The total sample comprised 1158 adults. As shown in the questionnaire (see online supplemental material), the surveys collected data on the individual level (eg, demographic attributes, antibiotic knowledge and attitudes), illness level (eg, healthcare choices) and the step level (eg, medicine use at each step of the treatment-seeking process). We tested the hypotheses using descriptive statistical analysis, comparing responses across groups and, where appropriate, performing Pearson's <unk> 2 tests whether these differences were statistically significant. We first applied the hypotheses to the village(s) where they arose, then to all three workshop villages (using the first or both survey rounds depending on whether data analysis took place on the individual or illness level) 131 and subsequently to the representative sample of rural Chiang Rai province. For the second subcase, the storytelling component involved stories from traditional healers in northern Thailand, whereby we considered photography and oral accounts as interlinked elements of a narrative in line with McKechnie. 132 The emphasis on these stories as 'tales' implies that the activity was not aimed to uncover an objective reality but rather a systems of experiences, relationships and subjective truths that can help challenge entrenched and simplistic outsiders' perspectives on such topics as health systems, medicine use and behaviours linked to AMR. 133,134 Health systems thereby refer to pluralistic landscapes of in-/formal healthcare providers (including public and private doctors, unregulated medicine sellers, traditional healers or carers within a family network), 135,136 which global health researchers and policy makers in AMR often simplify as organised and formal healthcare provision. 137,138 The narratives were captured by a Thai photographer (PT) and a chronicler (KW), who were both members of the research survey team (ie, data were collected via audio-visual material and digitised written narratives). Based on the experiences of the survey team, we revisited 15 traditional healers and received their permission to document their tales through written narratives and audio-visual records of the people, artefacts, process and context of traditional healing in any way the healers wished to present these aspects (yielding 61 GB of visual material). The exhibition curator and survey team manager (NC) translated the narratives into English and edited them for consistency. Out of the 15 narratives (which can be accessed at https:// tinyurl. com/ talesoftreatment), we reflected in this subcase on three that related directly to AMR, medicine use and health systems. The photographic narratives were presented at the 'Tales of Treatment' exhibition. Understanding that display plays an integral role in the dialogue between art and public audiences, 139 we drew in this third subcase on verbal and written feedback from the photo exhibitions and reflected as a team on audience reactions and the potential impact of the public engagement activity (quantitative feedback was digitised in a spreadsheet, written feedback was digitised via word processing software or as photographs of guest book entries). 140 In summary, the study team reviewed and analysed descriptively the qualitative and quantitative material against the objective to illustrate the diverse ways in which public engagement can reveal perspectives and forms of knowledge that are not typically incorporated into AMR research and policy. The primary analytical focus was on how co-produced local inputs could broaden the understanding of the social context of AMR (ie, constant critical comparison with the global health AMR discourse), which was directed by the principal investigator and supported by all coauthors (local Thai social researchers). Note that none of the data collection and analysis methods presented here constitutes a formal evaluation of the public engagement activities. However, based on experiences in the medical humanities literature and our previous public engagement projects, we were conscious that unstructured interactions with narratives and artistic products could entail unintended and potentially adverse interpretations and behavioural responses, which we explicitly incorporated as analytical angles as well. [141][142][143] --- Patient and public involvement statement This publication reports a case of public involvement for informing global health research. The broader project was further based on preceding qualitative research with local northern Thai patients and the general public, which revealed the need for more grounded sociomedical studies that respond to participants' viewpoints and cultural context. Members of the Thai public were involved in this project through cognitive interviewing to inform the survey design and the interpretation of the data, Thai participants were involved in local workshops during the study period to improve our understanding of local medicine uses and health behaviours, Thai villagers shared their stories about traditional healing on their own terms to broaden the scope of the standardised survey and Thai and UK publics were involved in the photo exhibitions where they were able to relay feedback and their interpretations of the photo stories. Note that this study did not specifically focus on patients but on members of the general public. --- ReSulTS We report separately on the knowledge co-production workshops and storytelling activities, using observations from co-production and engagement activities, primary survey data and event feedback. The results illustrate how insights and reflections sparked by the direct input from research populations and through the engagement of the public can broaden debates and viewpoints within the field of global health. However, the results also hint at the limitations and potential risks of a co-production approach, which we address separately in the subsequent 'Discussion' section. --- Subcase 1 (co-production): participatory workshops Using the example of the medicine pile sorting activity, we exemplify in this section how interactions between the research team and the workshop participants helped shape our understanding on February of medicine use in rural Chiang Rai. The pile sorting session was one of three workshop activities in which the participants were directly asked to educate the research team, and it lent itself for this illustration due to its focus on the meaning and uses of medicine in the context of AMR (the mapping activity focused on the healthcare landscape, and the poster making activity focused on participants' interpretations of the workshop content). The sorting activity thereby enabled at least three hypotheses about the relationships between the local social context, notions of medicine and treatment-seeking behaviour-which we outline and test below (for details on local notions of 'antibiotics', see a report of the main research project referenced here). 144 Theme 1: 'antibiotics that you can buy' Our first example involved workshop participants in Mae Fah Luang village. The participants described how they categorise different types of antibiotics into the groups "you can buy this medicine over the counter" and "you need a prescription from a doctor to obtain this medicine". These categories related directly to global health awareness campaigns, as for instance, WHO advocates that antibiotics should only be used 'when prescribed by a certified health professional'. 145 Based on the input from the villagers, we therefore hypothesised that, H1: Villagers' attitudes towards buying antibiotics over the counter differ depending on the types of antibiotics that they recognise. Our survey data contained the names with which people referred to antibiotics. 146 While the survey did not ask respondents to classify antibiotics into the categories of 'can buy' and 'cannot buy', we could at least learn from this data whether they were familiar with common colloquial names for antibiotics as 'anti-inflammatory medicine' ('<unk> <unk> <unk>' or 'yah kae ak seb'). The survey further asked knowledge and attitude questions corresponding to antibiotic awareness-raising material from the WHO, including whether there are situations in which the respondent would buy antibiotics over the counter-'desirable' responses being those that fell in line with the WHO position (not to judge whether their behaviour was inappropriate or unjustified), meaning that the respondent would not buy this medicine without a prescription. 147 If the hypothesis holds, then we would expect to see different attitudes to over-the-counter antibiotic purchases depending on how the respondents referred to the medicine. Table 2 shows the most commonly mentioned names of the three antibiotic images presented to the survey respondents (all respondents were shown the same images). The colloquial name 'anti-inflammatory' dominated the local notions of antibiotics. Owing to the ethnic diversity of Mae Fah Luang village, several other local language descriptions unbeknownst to us circulated alongside notions like 'germ killer', capsule medicine, cough medicine, pain reliever or vernacularized generic antibiotic names like 'amoxi' (for amoxicillin) and 'colem' (for chloramphenicol). The column '"desirable" attitude' indicates whether people's attitude aligned with WHO positions, depending on how the respondents described the medicine presented to them. Because the respondents could mention several different names, and because the mentioned names were likely correlated with respondents' personal characteristics (eg, ethnic background, language ability, education), these data should be interpreted with caution. However, a pattern emerged in which the technical term 'antibiotic' was associated with a relatively high share of 'desirable' attitudes. Consistent with hypothesis (H) Although the patterns were indicative rather than conclusive, the data provided circumstantial evidence in support of H1, namely that different names given to antibiotics were linked to different attitudes about antibiotic purchases. Future research could incorporate this aspect more systematically to understand which antibiotics villagers may be more inclined to procure over the counter-regardless of whether they have a biomedical understanding of antibiotic medicine. Theme 2: 'prescription medicine for children' Our second example relates to another insight from the pile sorting activity that we encountered in Mae Fah Luang village and Chiang Rai village. We learnt that villagers categorised medicine into'medicine for adults' and'medicine for children'. According to the workshop contributions, people would be extra careful with'medicine for children' (follow instructions closely, and indeed only receive it against prescription), whereas the workshop participants would buy'medicine for adults' for themselves over the counter. Antibiotics fell into both categories, which led us to hypothesise that, H2a: If children receive antibiotics, these antibiotics are more likely to originate from formal healthcare providers. H2b: If children receive antibiotics, these antibiotics are more likely to be used in accordance with their instructions. The surveys elicited healthcare pathways during an acute illness or accident within the 2 months prior to the survey interview-both for the respondents and for children under their supervision (children were defined in the survey as anyone below 18 years of age; adults would thereby report the illness episodes of 'children' under their supervision). At each step of the process, the respondent could indicate whether any medicine was received, whether it was taken in line with the instructions received, and whether the medicine was finished. 148 To operationalise these data for the hypothesis, we considered (a) illness episodes involving at least one dose of antibiotics, (b) whether these antibiotics originated from formal (public or private clinics/hospitals/pharmacies) or informal sources (unregulated healthcare providers including, grocery stores, traditional healers), (c) whether at least one set of antibiotics remained unfinished and (d) whether respondents took at least one set of antibiotics as recommended. We examined these factors initially for the two workshop villages where these statements originated (focusing on the first survey round prior to the workshop), and then expanded the analysis to the full sample of illness episodes in both the workshop villages and the provincial survey. To test whether these differences were statistically significant, we performed Pearson's <unk> 2 tests. Table 3 reports that adults consumed antibiotics in 12.2%-19.2% of all recorded illness episodes, whereas children's antibiotic consumption ranged from 13.2% to 24.5%. Within these episodes, the sources of children's antibiotics were systematically more likely to include formal healthcare providers, whereas adults were more likely to use antibiotics from informal sources. The Pearson's <unk> 2 tests indicated that use of antibiotics from formal sources was statistically significantly different between adults and children at least at the 10% level (Mae Fah Luang and Chiang Rai: p=0.070; all workshop villages: p=0.041, provincial level: p=0.083). The difference in informal antibiotic use, too, was statistically significant, except in the provincial data (Mae Fah Luang and Chiang Rai: p=0.070; all workshop villages: p=0.088, provincial level: p=0.235). In contrast, none of the differences in completing antibiotic courses or adhering to instructions was statistically significant for any of the three samples. However tentative, these findings help broaden the understanding of antibiotic use (and potentially the identification of priority or high-risk target groups) in different segments of the population. Children were indeed more likely than adults to receive antibiotics from a formal healthcare provider, which is consistent with H2a. In contrast, the limited survey evidence did not indicate that the distinction between'medicine for adults' and'medicine for children'to antibiotic use instructions (H2b), but it is noteworthy that adult illness episodes (n=125) across Chiang Rai province had a more than 10-percentage-point lower rate than children (n=31) in terms of non-completion of an antibiotic course (36.5% vs 48.6%). --- Theme 3: 'assertive youth' Our last example again relates to demographic differences in medicine use. In Chiang Rai village, workshop participants reported that young adults would more commonly engage in arguments and assert their position vis-à-vis figures of authority, like doctors or elders. This may be generic judgement of older towards younger generations, 149 but we could argue that older people-when they were younger-had experienced a different health system and different social hierarchies than today's youth. This raised the question whether age gradients may reflect different patient-health system relationships across generations, and with them different patterns of antibiotic use. We therefore hypothesised that, H3a: Younger adults are more likely to source antibiotics from informal healthcare providers. H3b: Younger adults are less likely to use antibiotics in accordance with their instructions. To test these hypotheses, we again examined first the initial survey round from Chiang Rai village, followed by the complete workshop village sample and provincial survey. We used the same analysis categories as in the previous section (a: illness episodes involving at least one dose of antibiotics, b: antibiotics from formal/informal sources, c: unfinished antibiotic courses, d: adherence to antibiotic instructions) and analysed the differences across five age groups, namely 18-24, 25-34, 35-44, 45-59 and 60+ years. We used Pearson's <unk> 2 tests for differences across age groups. Figure 2 presents the results of the group comparison (see online supplementary appendix table A1 for detailed results including Pearson's <unk> 2 tests). Because of the small sample of 17 illness episodes in Chiang Rai village, we focused the analysis on the larger workshop village sample (100 episodes) and the provincial survey (125 episodes). Among the workshop villages, the age group 35-44 years exhibited the highest degree of formal antibiotic use (84.2%; sample average: 75.0%) coupled with the lowest incidence of informal antibiotic consumption (15.8%; sample average: 26.0%), the lowest incidence of leaving antibiotics unfinished (21.1%; sample average: 40.0%) and the highest rate of instruction adherence (78.9%; sample average: 67.0%). Both younger and older age groups indicated higher informal use and less strict adherence to antibiotic regimes. However, only the difference in unfinished antibiotic courses was statistically significant at p=0.020. While the provinciallevel age group differences were statistically significant in the cases of formal antibiotic use (p=0.001) and instruction adherence (p=0.006) (informal antibiotic use: p=0.418; unfinished antibiotics: p=0.110), the patterns across age groups were distinctly different from the workshop village sample: the age group 25-35 years had notably below-average formal antibiotic use (42.4%; sample average: 83.6%), above-average informal antibiotic use (14.2%; sample average: 4.6%) and below-average unfinished antibiotics (13.1%; sample average: 36.5%) and adherence to instructions (31.3%; sample average: 72.2%). The younger age group of 18-24 years did not follow this trend and mostly corresponded to

Global health champions modernism and biomedical knowledge but tends to neglect knowledge, beliefs and identities of rural communities in low-income and middle-income countries. The topic of antimicrobial resistance represents these common challenges, wherein the growing emphasis on public engagement offers a yet underdeveloped opportunity to generate perspectives and forms of knowledge that are not typically incorporated into research and policy. The medical humanities as an interdisciplinary approach to illness and health behaviour play a central role in cultivating this potential-in particular, through the field's emphasis on phenomenological and intersubjective approaches to knowledge generation and its interest in dialogue between medicine, the humanities and the broader public. We present a case study of public engagement that incorporates three medical humanities methods: participatory co-production, photographic storytelling and dialogue between researchers and the public. Situated in the context of northern Thailand, we explore subcases on co-production workshops with villagers, tales of treatment shared by traditional healers and dialogue surrounding artistic display in an international photo exhibition. Our starting assumption for the case study analysis was that co-produced local inputs can (and should) broaden the understanding of the sociocultural context of antimicrobial resistance. Our case study illustrates the potential of medical humanities methods in public engagement to foreground cultural knowledge, personal experience and 'lay' sensemaking surrounding health systems and healing (including medicine use). Among others, the engagement activities enabled us to formulate and test locally grounded hypotheses, gain new insights into the social configuration of treatment seeking and reflect on the relationship between traditional healing and modern medicine in the context of antimicrobial resistance. We conclude that medical-humanities-informed forms of public engagement should become a standard component of global health research, but they require extensive evaluation to assess benefits and risks comprehensively.

the lowest incidence of leaving antibiotics unfinished (21.1%; sample average: 40.0%) and the highest rate of instruction adherence (78.9%; sample average: 67.0%). Both younger and older age groups indicated higher informal use and less strict adherence to antibiotic regimes. However, only the difference in unfinished antibiotic courses was statistically significant at p=0.020. While the provinciallevel age group differences were statistically significant in the cases of formal antibiotic use (p=0.001) and instruction adherence (p=0.006) (informal antibiotic use: p=0.418; unfinished antibiotics: p=0.110), the patterns across age groups were distinctly different from the workshop village sample: the age group 25-35 years had notably below-average formal antibiotic use (42.4%; sample average: 83.6%), above-average informal antibiotic use (14.2%; sample average: 4.6%) and below-average unfinished antibiotics (13.1%; sample average: 36.5%) and adherence to instructions (31.3%; sample average: 72.2%). The younger age group of 18-24 years did not follow this trend and mostly corresponded to the remainder of the sample. The mixed patterns across the samples suggest caution in supporting or rejecting the hypothesis, but the data did suggest that antibiotic use behaviour was likely to have an age dimension. Further qualitative research would allow us to investigate whether these patterns are systematic, and, if so, if they related to different age groups' assertiveness (eg, driven by formal education) as argued in the workshops, or whether they were a result of different meaning of medicines across generations. In summary, the three workshop themes presented here illustrate how the public engagement activity helped co-produce locally grounded hypotheses relating to medicine use as a specific aspect of the sociocultural context of AMR. --- Subcase 2 (storytelling): of healing and treatment in northern Thailand In this second subcase, we exemplify how storytelling techniques as part of public engagement helped shed light on the broader sociocultural context of AMR. The stories narrated in the 'Tales of Treatment' exhibition illustrate insights about local healing, health systems and reflections on global health that would not otherwise have emerged from the broader research project. The narratives did not intend to present superior or effective forms of treatment but rather to document disappearing perspectives and practices of healing in Chiang Rai, following the bottom-up process of participatory engagement in which traditional healers chose the focus of their stories. Figure 3A presents such a narrative from a traditional 'ghost doctor' (a spiritual healer) in a Mien village. The tale told of sacred books of chants in traditional Chinese, which in their entirety were often only accessible to ghost doctors who learnt their craft over generations. However, minor chants and small ceremonies were not reserved exclusively to the ghost doctor-it was a common skill in the village, applied for instance when teenagers sought forgiveness from their parents traditional healer and spiritual care might involve. How could practices like asking for forgiveness be incorporated into a standardised survey instrument on treatment seeking, and how might the omission of, for example, pastoral dimensions of care distort the representation of local realities? Second, the fluid interpretation of who was a ghost doctor in a village (ie, potentially everyone) undermined our initially binary distinction between the general population versus medical providers. The second example from the Tales of Treatment exhibition involved a traditional treatment adapted from 'gua sa' (<unk> <unk>), which was common in Thailand, China and Southeast Asia more broadly. Also known as gua sha in Chinese (<unk>), or'scraping'/'coining' in English, 150 gua sa involved scraping the skin to stimulate blood circulation until bruises appear. The tale relayed in figure 3B told of local adaptations of this practice that involved pulling rather than scraping the skin-locally known as 'dueng sa' (<unk> <unk>). Like the spiritual chants in the previous narrative, dueng sa was a common skill among villagers, and its superior effectiveness over gua sa was explained by the pain it inflicted on the recipient. The insights generated by this tale did not only involve the local adaptation of international yet non-Western medical practices of gua sha and the (for us) unexpected interpretations of how people assessed the quality of dueng sa-very much unlike conventional Western interpretations of what 'quality of care' would entail. One of the main surprises from this story was also the idiosyncrasy of medical practice. Aunt Porn's village performed a version of gua sa that was different from local medical practice in neighbouring villages. This begged the question, 'What does "traditional healing" mean at all, and how can we usefully categorise it?' (This is of course not a fundamentally new question as medical anthropology and medical humanities have been engaging with such topics for decades.) The final narrative (figure 3C) related to the broader discourse of AMR as a global health issue. Grandma Kaew was among the last traditional healers in her village, applying knowledge passed down to her from generations ago. Fellow villagers received her herbal treatment for symptoms like headaches and indigestion, and steady demand had required her to process these herbs more efficiently. As she explored methods to store herbs for convenience and longer shelf life, she begun sun-drying them, blending them into fine powder and apportioning them into small ziplock bags. She also filled bitter-tasting herbs like 'fah talai jone' ('<unk> <unk>' or andrographis paniculate) into capsules so that children or patients who did not like taking medicines could use them as well. The significance of Grandma Keaw's story rested in the fact that modern Thai health policy had begun advocating, among others, the treatment of uncomplicated conditions like sore throats with traditional Thai herbal medicine. The purpose of this development had been to respond to healthcare providers 'who feel pressured by patients' expectations' for antibiotics and therefore reduce the reliance on antibiotic treatment in human medicine. 151 The tale of Grandma Keaw's 'at-home medical unit' underlined the irony of this proposal: herbal and non-medicinal alternatives for antibiotic treatment had been practised for centuries, but were over the past decades crowded out increasingly by the modernisation of medicine. 152,153 On reflection, one could argue that modern medicine had sown the seeds of its own demise through the pharmaceuticalisation of care (ie, reducing the idea of healing to the transaction of capsules), and now depends for its survival on the traditional medicine that it had been displacing. At the same time, critical academic voices wonder whether the modern Thai health policy approach incorporating herbal medicine capsules does, yet again, reduce holistic traditional treatment to a transactional relationship. Although this might be a valid concern, Grandma Kaew herself had been producing and administering herbal medicines in capsules-for pragmatic reasons, and without obviously adhering to a biomedical agenda. The practice of recording narratives alongside our survey therefore enabled our research team to perceive illness and treatment beyond the survey questionnaire. Examples of local medical practice challenged our conceptualisation of care in rural northern Thailand-for example, the dichotomy between population and healthcare providers, the spectrum of conditions that deserved a traditional healer's attention and the fluidity of its performance-but it also added nuance to our understanding of modern health policy and its critiques in the context of AMR. --- Subcase 3 (dialogue): reactions and reflections from the photo exhibitions 'Tales of Treatment' was a mechanism to capture narratives from northern Thai villages, and to acquaint the urban public interested in photography, culture and alternative systems of medicine with this material. The overall >500 visitors across our four venues engaged enthusiastically with the exhibits, the stories and the exhibition hosts, typically spending 45-60 min at the exhibition. The dialogue between researchers and the public thereby enabled a further expansion of the understanding of the Interactions between the research team and the exhibition attendees revealed how the exhibition stimulated reflection and recall of personal treatment histories. For example, some of our Thai audience, including those from northern Thailand, said that they had seen their parents or grandparents follow the practices shown in the photographic stories, but they had never experienced herbal or spiritual healing themselves. UK and US audience members related the content to the role of complementary medicine in their respective home countries, and drew parallels between Thailand and Western countries in terms of sensemaking about the body, illness and healing techniques. Written feedback also suggested that the exhibition sparked reflection. Attendees related the content to their personal experiences growing up in families where modern medicine was unpopular ("My dad never liked modern medicines so I've experienced (traditional and alternative forms of healing) a lot! Acupuncture, power therapy, psychotherapy, [...]"), or their encounters with traditional medicine in other Southeast Asian contexts ("[...] In Vietnam, we have a practice called cao gióvery popular for'scratch(ing) out the wind' from a cold/fever [...]"). Together with the attendees, we reflected yet further on cross-cultural comparisons of behaviour and possible research avenues about the co-evolution and global spread of drug resistance and local forms of healing. During the latest iteration of the exhibition at Warwick Arts Centre, we collected more formal feedback in addition to verbal and guestbook testimonies. At a response rate of 32.9% (23 out of 70 visitors, all of whom were university students or staff), 95.7% agreed that they learnt'something new' during the exhibition (100% of the responses agreed that the event was 'worthwhile'). Among the explanations of what had been learnt, the attendees indicated, for example, <unk> '"Alternative" treatments in other parts of the world'; <unk> 'The popularity of using the supernatural'; <unk> 'The interconnectedness of Thai, Chinese medicine'; <unk> 'So much! In particular the pulling and pinching (gua sa, dueng sa)'. Yet, not everyone was equally impressed. A subset of the attendees in all exhibition sites also expressed doubts both about the content of the stories (eg, narratives about medicinal plants functioning as fever absorber) and the photographs themselves (eg, concerns about animal cruelty where ghost doctors used tiger claws during treatment). Specifically with regard to antibiotics and drug resistance, some attendees in Bangkok also enquired about the subject, behaving as if the team were medical specialists. Although such feedback and reflections only arose in conversation with the attendees rather than in writing, some also indicated that they had 'never realised how effective these treatments can be'. The exhibition stated explicitly that its intention was not to advocate a particular treatment method nor to suggest the superiority of traditional healing-rather, to relay stories from the field. Nevertheless, we as hosts may have on occasion been misinterpreted as medical specialists, and some interpretations of the content may have potentially entailed unintended behavioural outcomes of the public engagement event. --- dISCuSSIon Summary The case study exemplified that public engagement under the umbrella of the medical humanities-that is, not focused on instrumental awareness raising and community mobilisation-can complement and contribute to the understanding of local health practices and global health priorities. Our activities involved knowledge co-production, storytelling and dialogue between researchers and the public, and they emerged partly in response to the limitations of a health behaviour survey. Overall, the three subcases enabled: <unk> a better understanding of local conceptualisations of medicine (subcases 1, 2, 3); <unk> new insights into the social configuration of treatment seeking (subcases 1 and 2); <unk> for us otherwise invisible idiosyncrasies of traditional healing across villages in northern Thailand (subcase 2); <unk> new perspectives on the relationship between 'the general population' and 'traditional healers' (subcases 2 and 3); <unk> reflection on the relationship between modernity and tradition in AMR (subcase 2). At the same time, not all points raised in the co-production workshops could be supported by our quantitative survey data, and participation in the exhibitions appeared to have created misleading impressions of our purpose and messages among a small group of attendees. 154 Despite its seeming value for challenging thought and research in global health, we should therefore not underestimate the consequences of intervening in a social system through co-production and bi-directional communication-however well-meaning it might be. Our case study contributes to the practice of public engagement in global health research as an important element of sustainable development and the empirical understanding of the sociocultural context of AMR as a global health priority. As opposed to mainstream community engagement activities in global health and AMR in particular, 155 the case study suggested how medical humanities methods can help researchers to learn from their target populations instead of instrumentalising 'engagement' to change communities along biomedical ideals. The importance of bi-directional communication highlighted in our work indicated that global health researchers indeed require local inputs to formulate hypotheses and ground analytical categories, and also to define the research problem itself-similar to arguments surrounding the practice of patient and public involvement in Western medical research. 156,157 By depicting new and for biomedical researchers often invisible subjective truths, our case also supports positions in the medical humanities that knowledge co-production, stories and dialogue based on artistic production can yield new and practically important perspectives on complex problems 158,159 -for example, healing and the nature of health systems-that have bearing on AMR. One often neglected aspect of healing is for instance the role of spirituality, 160 as the tales of treatment (subcase 2) have powerfully brought to the fore (spirituality may play a role in considering non-pharmaceutical solutions for population health and well-being, eg, in the form of pastoral support and meaning-making). At the same time, the documented risks of the unintended consequences of engagement also expand the recent argument by Abimbola 161 (in the context of community health committees) to steer global health researchers and practitioners away from an unrealistically optimistic 'a priori bias' in community engagement. The insights provided by our workshop participants (subcase 1) and exhibition attendees (subcase 3) further added to debates and empirical knowledge in the field of AMR. For example, knowledge co-production in the workshops highlighted the varied relationship between antibiotic conceptions and attitudes towards over-the-counter purchases and related to the literature on language and local conceptions of antimicrobials. 162,163 Other locally grounded research hypotheses demonstrated how on February 17, 2024 by guest. Protected by copyright. http://mh.bmj.com/ Med Humanities: first published as 10.1136/medhum-2020-011894 on 18 September 2020. Downloaded from original research antibiotic usage differed across generations, which contributed to understanding the determinants of antibiotic use and the values that underlie antibiotic choices in Thailand and other LMICs. [164][165][166] In addition, biomedical writing often portrays traditional healers as an unqualified source of antimicrobials or as a healthcare solution that could delay access to biomedically trained healthcare providers. 167,168 Rather than pitching traditional against formal healthcare, stories of healing and treatment (subcase 2) demonstrated the fluidity of traditional healing in Chiang Rai and, together with the dialogue with urban publics, enabled reflections on its relationship to AMR. Considering that 'AMR is understood as a threat to health, to economies, to security and to modernity itself ', 169 traditional healing may ironically play a role in'saving' modern medicine by limiting dependence on pharmaceutical treatment in the case of uncomplicated minor ailments such as muscle pains or sore throats (as the tale of Grandma Kaew in subcase 2 illustrated). Overall, the inputs from workshop participants, traditional healers and exhibition attendees in our project challenged assumptions and expectations among the international research team, helping to expand understanding incrementally and to challenge geographically and disciplinarily defined hierarchies of knowledge in global health research. --- Costs and risks of knowledge co-production Overall, our analysis suggested that there were complementarities between the co-production of knowledge on the one hand, and the data collection methods and the interpretation of health behaviour research on the other hand. However, these activities also produced costs and risky outcomes that we discuss briefly in this section. Short of immersive ethnographic research, cross-sectional qualitative research could have similarly helped to generate knowledge about local behaviours and medicine use, and to inform the development of a structured questionnaire. Qualitative pretesting of the survey instruments-for instance, through cognitive interviewing 170 -can help uncover unforeseen categories and refine quantitative data collection as well, although this often happens at a stage when research design and hypotheses are already relatively fixed. We applied both these techniques in this study, but the workshop setup helped to complement these qualitative approaches. Although activities like medicine pile sorting are not specific to a workshop setting and could in principle be also incorporated into semi-structured interviews and focus group discussions ('participatory' methods like pile sorting exercises have long been incorporated in development survey research 171 ), the wide range of media and activities employed during the workshops helped generate a more open and engaging atmosphere and a greater degree of bi-directional knowledge exchange than could be achieved in the more structured data collection settings of face-to-face interviews or focus group discussions. The monetary costs of the workshops themselves amounted to £450 per workshop for consumables and eight facilitating staff plus approximately £3000 for consumables and staff costs for the development and trialling of the workshop format. Similarly, gathering and exhibiting photographic narratives from our field sites was an opportunity for the project to cultivate and benefit from the talent of the Thai research team, and to learn about healing and treatment from the perspective of local residents. The narratives enabled us to explore perspectives that especially the non-Thai project collaborators would not have considered otherwise. The visual component of the narratives thereby offered additional space for reflection compared with, for instance, a solely text-based semi-structured interview, and it opened a pathway to engaging the broader public interested in photography, culture and traditional healing in our project. The latest exhibition at Warwick Arts Centre also paved the way for closer collaboration between the research team and the creative industry. However, knowledge co-production through visual methods and storytelling served primarily a supplementary purpose in our project-for our research objectives, it would have not have sufficed as an alone-standing research and knowledge production method (which comes with its own methodological and ethical challenges, compounded by interdisciplinary frictions in ethical review committees [172][173][174][175][176] ). As the feedback from the photo exhibitions showed, presenting health-related practices could also potentially influence people's health behaviour even if the research team explicitly distanced themselves from advocating any particular practice. The collection and preparation of the material and hosting the four photo exhibitions also required a budget of approximately £8000. These costs and risks mean that knowledge co-production has to be weighed against alternative qualitative and quantitative modes of generating global health knowledge. As a complement to conventional research methods, however, they can usefully inform a project during its design phase, aid the interpretation of its results, and make the dissemination of its findings more effective. The costs and risks of these methods should therefore be assessed on the basis of complementing conventional global health research. --- limitations The primary limitation of this research was that the co-production activities did not involve an independent evaluation. Being embedded in the research and positioning ourselves as 'learners' vis-à-vis villagers and the public prevented the research team from carrying out a formal independent assessment of the consequences of the activities. A parallel research team not involved in the project or its design would have ideally worked alongside our group to add additional depth on the unintended consequences and potential (and actual) benefits and harms arising from the co-production activities. Despite our best attempts to be mindful of alternative interpretations of our work and the negative outcomes of the activities, there remained thus a residual risk that our position as social researchers invested in this project unconsciously biased us towards a particular interpretation of the data and participants' responses. The mixed insights from the quantitative analysis further indicated the shortcomings of using a prespecified survey instrument to assess locally emerging research hypotheses. Especially where the quantitative findings did not support the hypotheses, the question remained whether this was because the hypothesis could not be supported, or because the questionnaire and research design were not suitable to investigate the respective point. Although the questionnaire was developed with the help of prior qualitative research, field pilots and cognitive interviewing, the new insights provided by the participatory workshops would have required further iterations to accommodate workshop participants' inputs. An additional limitation of the scope of this case study is that the public engagement team comprised social scientists and no medical specialists. As a result, our focus is confined to sociocultural aspects of AMR, in particular local knowledge and intersubjective truths on health systems and healing. 177 However, a tension also remains between the benefits of co-producing knowledge and the risk of unintended consequences from public engagement and the presence of external research teams. Ours is not the first study to problematise hazards of potentially detrimental interpretations of narratives and arts-based engagement with the public, 178,179 misinterpretation of roles and competences of researchers 180 or issues of preproducing hierarchical relationships to the point of oppressing local communities. 181 The knowledge to evaluate such mixed consequences of public engagement and participatory research is yet limited and requires further methodological research. [182][183][184][185] Once evaluation frameworks and guidelines have been established, varied applications of process, ex post and impact evaluation (both qualitative and quantitative) would enable us to map the consequences of knowledge co-production and to assess their costs and benefits more comprehensively and pragmatically-even if the costs of an evaluation itself mean that such assessments can only be conducted on a sample of research projects. Overall, our analysis leads us to conclude that knowledge co-production and medical-humanities-informed forms of public engagement should become standard secondary objectives of global health research to prevent misrepresentation of local realities and to more effectively ground the interpretations of its findings in the local context. One precondition of this strategy to succeed is to frame global health research more actively as a learning exercise and embed the agenda to 'decolonise' global health more firmly in research education and international health policy circles. An international commission-led by interdisciplinary researchers from LMICs-could further legitimise this practice by establishing formal ethical guidelines for global health research to be more receptive to local voices, rather than merely instrumentalising the rhetoric of public engagement for public health interventions. 146. The questionnaire did so by showing all survey respondents three images of common antibiotic capsules being used in Chiang Rai (green-blue: amoxicillin; red-black: cloxacillin; white-blue: azithromycin-see questionnaire page 10 in the online supplementary material). Respondents were asked to name what they saw, and all their answers were recorded (field-coded and as free text). 147. The 'desirability' of the responses was field coded by the survey team. Sample responses (as instructed through the survey manual) for 'desirable' answers included, for example, "Only if the doctor says that I should". Sample responses for 'undesirable' answers included "Yes, you can buy it in the shop over there!" The variable should be interpreted as 'the fraction of respondents who uttered a 'desirable' response'-the inverse is the fraction of responses that could not be deemed 'desirable' (eg, 'do not know' or 'no opinion'). 148. Because recalled descriptions of medicine tend to be ambiguous, we limited our analysis to medicines where we had a high degree of certainty that they were an antibiotic. This was specifically the case if survey respondents mentioned common antibiotic descriptions such as 'anti-inflammatory', 'amoxi' or 'colem', if they indicated explicitly that they know what 'anti-inflammatory medicine' is (noting that the term describes antibiotics unambiguously in Thai), and if they subsequently mentioned any of the previously mentioned antibiotics during their description of an illness episode (conversely, we excluded cases were the medicine could not be confirmed as either antibiotic or non-antibiotic, including descriptions like 'white powder' or 'green capsule'). 149. Aristotle (1954) --- original research --- https:// creativecommons. org/ licenses/ by/ 4. 0/. oRCId ids Marco J Haenssgen http:// orcid. org/ 0000-0002-5849-7131 Nutcha Charoenboon http:// orcid. org/ 0000-0002-6975-3524

Global health champions modernism and biomedical knowledge but tends to neglect knowledge, beliefs and identities of rural communities in low-income and middle-income countries. The topic of antimicrobial resistance represents these common challenges, wherein the growing emphasis on public engagement offers a yet underdeveloped opportunity to generate perspectives and forms of knowledge that are not typically incorporated into research and policy. The medical humanities as an interdisciplinary approach to illness and health behaviour play a central role in cultivating this potential-in particular, through the field's emphasis on phenomenological and intersubjective approaches to knowledge generation and its interest in dialogue between medicine, the humanities and the broader public. We present a case study of public engagement that incorporates three medical humanities methods: participatory co-production, photographic storytelling and dialogue between researchers and the public. Situated in the context of northern Thailand, we explore subcases on co-production workshops with villagers, tales of treatment shared by traditional healers and dialogue surrounding artistic display in an international photo exhibition. Our starting assumption for the case study analysis was that co-produced local inputs can (and should) broaden the understanding of the sociocultural context of antimicrobial resistance. Our case study illustrates the potential of medical humanities methods in public engagement to foreground cultural knowledge, personal experience and 'lay' sensemaking surrounding health systems and healing (including medicine use). Among others, the engagement activities enabled us to formulate and test locally grounded hypotheses, gain new insights into the social configuration of treatment seeking and reflect on the relationship between traditional healing and modern medicine in the context of antimicrobial resistance. We conclude that medical-humanities-informed forms of public engagement should become a standard component of global health research, but they require extensive evaluation to assess benefits and risks comprehensively.

Introduction The past several decades have witnessed rapid fertility decline in the People's Republic of China. The total fertility rate (TFR) dropped from 5.76 in the 1960s to 2.71 in the early 1980s. By the 1990s the TFR had further dropped to below-replacement level, and the low fertility regime has persisted thus far (Gu et al. 2007;Lavely and Freedman 1990;Morgan, Guo, and Hayford 2010). There is a large body of literature examining fertility decline in China. However, findings are inconsistent regarding the nature and extent of the factors that are contributing to the fertility decline. For example, scholars disagree on the major driving forces behind the fertility decline (Cai 2010;Lavely and Freedman 1990;Li, Zhang, and Zhu 2005). Results from quantitative analyses based on publicly available data do not entirely match with those from in-depth fieldwork studies in terms of the roles that specific variables (such as education and economic development) play in affecting fertility behaviors (Greenhalgh and Winckler 2005;Harrell et al. 2011;Peng 2010;Qian 1997;Short and Zhai 1998;Zhang 1999Zhang, 2007)). One possible reason for the conflicting findings is that the effects of the driving factors on fertility vary spatially because of differences in local contexts and characteristics. The underlying fertility decline processes are context-specific and vary across space. Uneven socioeconomic development levels, coupled with variation in the implementation and enforcement of family planning policies, could lead to fertility variation across space. In fact, existing theories of fertility decline suggest that fertility behaviors vary according to how they are influenced by economic structure, social interaction, culture, and policy implementation strategies (Axinn and Yabiku 2001;Bongaarts and Watkins 1996;Cleland and Wilson 1987;Hirschman 1994;Montgomery andCasterline 1993, 1996). Thus, the relationships between fertility and social, economic, cultural, and policy-related factors may not be readily summarized by a single one-size-fits-all model. Successful policymaking needs to go beyond the search for a universal explanation of fertility decline and requires a contextual understanding of not only local fertility patterns but also the unde rly ing factors that drive these patterns. However, the possible spatial variation in the effects that the socioeconomic and policy factors have on fertility decline has not been studied. The goal of this study is therefore not to search for universal relationships but to investigate the spatially varying relationships between county-level fertility rates and a set of social, economic, and policy factors in China. In particular, we aim to examine whether and how relationships between the fertility level and its commonly identified driving forces vary across space. In the study we apply geographically weighted regression (GWR) models to analyze geocoded 2010 county-level census data. The GWR models allow different relationships to exist at different points in space by calibrating multiple regression models using spatial weights (Brunsdon, Fotheringham, andCharlton 1996, 2008;Fotheringham, Brunsdon, and Charlton 2003;Fotheringham, Charlton, and Brunsdon 1996). Our results reveal a complex picture of spatial variation in the relationships between fertility and its driving factors in China, and thus highlight the importance of incorporating spatial heterogeneity into fertility research. --- Background and literature --- Fertility policy, socioeconomic development, and fertility decline The decline of fertility rates in China is often seen as a special case, as it is usually characterized as the result of a combination of socioeconomic development and family planning policy implementation (Winckler 2002). Although it is helpful to differentiate the effects of fertility policy and of socioeconomic development, fertility policy cannot be isolated from the broader social and economic context. Existing studies agree that it is useful to draw a conceptual distinction between the impacts of fertility policy and socioeconomic development; in practice, however, the two forces operate to reinforce each other. The level of socioeconomic development may be a precondition for the implementation of family planning programs, and fertility policy may in turn further reinforce existing social and familial norms (Merli, Qian, and Smith 2004;Tien 1984). This section provides an overview of the evolution of China's fertility policy and socioeconomic development. We focus particularly on existing evidence as to how these two forces interact with each other to influence variation in China's fertility rates at subnational levels. Chinese family planning policy has long been recognized as one of the largest population control exercises in the modern world. Fertility policy has been an integral part of the nation's development strategy. After the charismatic rule of Mao Zedong came to an end in the mid-1970s the central government initiated a set of socioeconomic reforms to create a transition from a centrally planned to a market-oriented economy and to hasten the country's industrialization and modernization. The fertility policy that was initiated in the late 1970s aimed to manage China's population in a planned,'scientific' way (Greenhalgh 2008). Starting in the mid-1980s, the policy went through a set of decentralizing processes (Greenhalgh 1986;Gu et al. 2007;Short and Zhai 1998). Marked by the central government's release of Document 7 in 1984, the creation and implementation of fertility policy was devolved from central government to the local level. The aim was to make fertility policy tailorable to local social, economic, and cultural conditions. An early study that analyzes reports from 40 localities in 17 provinces reveals that tremendous diversity already existed, and that "nothing resembled a unified national situation" (Greenhalgh 1986: 499). A study by Gu et al. (2007) provides more comprehensive documentation of the variation in fertility policy at the subnational level. By collecting birth control regulations issued by State Family Planning Commissions (SFPCs) in different provinces, Gu et al. (2007) et al. identify 22 circumstances where couples are eligible to be exempted from the strict one-child policy. These exceptions are further grouped into four categories: gender-based, economic-based, political/ethical/social-based, and entitlement/replacement-based. 3Aside from variations in the written policy, the implementation of fertility policy also varies. A study by Short and Zhai (1998) investigating local fertility policies in 167 communities concludes that the fertility policies vary not only in terms of the strength of the written policy (measured as the percentage of the population who are exempt from the one-child policy) but also in terms of the types of incentive and punishment that are implemented at the local level. Besides the persistent scholarly interest in documenting and analyzing variations in policymaking and implementation procedures, there is also increased speculation as to what extent fertility variation at the subnational level is mainly attributable to the heavy-handed fertility policy. This line of research draws on arguments from classic demographic transition theory and posits that socioeconomic development plays a more important role than fertility policy in driving fertility decline and variation at the subnational level (Birdsall and Jamison 1983;Ping 2000;Poston and Jia 1990;Wang 2008). The process of industrialization and urbanization increases the cost of raising children and women's participation in the labor market, and also facilitates the spread of small-family ideals, ultimately influencing reproductive motivation (Birdsall and Jamson 1983;Tien 1984). For example, using the 2000 census, a study by Cai (2010) compares the county-level fertility rates of two provinces, Jiangsu and Zhejiang, which have similar economic development levels but differently written fertility policies. Cai discovers that although the written policies vary greatly between these two provinces, the county-level fertility rates are still very similar. Cai concludes that even if the fertility decline was initiated by heavy-handed government intervention, the persistently low fertility trend in recent years is primarily driven by socioeconomic factors rather than government intervention. --- Spatial heterogeneity of fertility decline: Theories and new evidence Building upon the classical demographic transition theory (Easterlin and Crimmins 1985;Kirk 1996), existing studies of fertility in China primarily emphasize the role that socioeconomic development has on fertility decline, because it is closely related to the policy debate of whether heavy-handed government intervention is still necessary under the current low fertility regime. The early nonspatial studies highlight rural-urban differences as one major spatial dimension to be considered (Greenhalgh 1988;Kaufman et al. 1989;Lavely and Freedman 1990;Li 1995). However, in demographic literature there is growing interest in directions other than the classical demographic theory of fertility. This is primarily because fertility decline is seen as far from a universal theory, to the extent that "different historical and geographic settings show more diversity than a simple theory of fertility change would predict" (Hirschman 1994: 203). The countervailing argument addresses the role that social interaction and diffusion of knowledge play in fertility behaviors (Axinn and Yabiku 2001;Bongaarts and Watkins 1996;Cleland and Wilson 1987;Hirschman 1994;Montgomery andCasterline 1993, 1996). The central argument is that couples do not only consider their own circumstances when making fertility decisions, but also learn and imitate the fertility behaviors of people close to them. These learning and imitating processes may operate independently of economic considerations. Therefore, the spatial pattern of fertility levels cannot be interpreted as only being in response to underlying socioeconomic structures: It also reflects the spread of innovation, knowledge, norms, and institutional practices (Casterline 2010). Several studies have shed light on the spatial variation of fertility decline in the context of China. For example, Skinner et al. (2000) adopt an explicit spatial approach and propose a hierarchical regional space (HRS) model to study the fertility transition of China in the 1990s. The HRS model takes into consideration two-dimensional spatial structures: core-periphery zones and the rural-urban continuum. By studying fertility decline in the lower Yangzi River region, Skinner et al. (2000) highlight four aspects that contribute to spatial variation in the fertility rate: socioeconomic development levels, family planning policy and practices, changes in traditional family norms, and the spread of sex-selective abortion technologies. These four dimensions covary systematically with the decline of the fertility rate in the lower Yangzi River area through regional space, following a declining trajectory from inner-core cities to farperiphery villages. However, increasing evidence suggests that the spread of ideas, norms, and behaviors may not entirely be in concert with the spatial structures of socioeconomic characteristics. Compared with rapid economic development and institutional change, the changes in culture and norms may be slow. As a result, in certain places distinctive fertility behaviors may differ from general patterns. For example, ethnographic research conducted by Zhang (2007) finds a reversal relationship between fertility rates and economic development levels. In areas where the economic development level is high, wealthy families tend to have more children since they are able to pay the fines, while relatively poor rural families tend to have only one child due to the concern of high costs. The work of Peng (2010) calls attention to the dynamics of birth control policy implementation at the local level, arguing that the collision between pronatalist family norms and antinatalist family planning laws may be particularly salient in some places. He finds that places with strong patrilineal norms (measured by kinship networks and presence of ancestral halls) are particularly resistant to the family planning policy and exhibit high fertility levels. --- Our approach and hypotheses Both theoretical arguments and empirical evidence suggest that understanding fertility decline in China requires explicit recognition of spatial heterogeneity in local characteristics. However, methodologically it is often difficult to describe spatially varying relationships using conventional approaches. The standard regression approaches, such as the Ordinary Least Squares (OLS) models and most of the spatial econometric models (referred to as 'global models' in the literature) obtain only one set of coefficients for all the locations and thus ignore the local and spatial contexts of fertility behavior (Ali, Partridge, and Olfert 2007;Brunsdon, Fotheringham, and Charlton 1996;Chi and Zhu 2008). It can be argued that the varying relationships can be captured by using carefully selected interactions between socioeconomic factors and place-specific dummy variables in a single model. Doing this, however, would require researchers to have intricate knowledge of a specific set of interactions. Other models that deal with local contexts, such as the spatial expansion method and multilevel modeling, require an a priori functional form that results in the nature of the spatial process being modeled as either deterministic or discrete (Fotheringham, Brunsdon, and Charlton 2003). For the purpose of this study we take a blended perspective, considering fertility decline as the result of the structural transformation of socioeconomic characteristics as well as the diffusion of new ideas, technologies, and social norms (e.g., Goldstein and Klüsener 2014;Harrell et al. 2011;Retherford 1985). Specifically, we consider spatial distribution of fertility levels in China as a reflection of the combined forces of economic development, fertility policy implementation and enforcement, and diffusion of new family norms (such as small-family ideals). What distinguishes our analysis from previous studies is that we do not assume any predetermined spatial structures in our model specification (such as rural-urban dichotomy, east-west gradient, or coreperiphery hierarchy), nor do we aim to get more precise global estimates by taking into consideration spatially correlated error terms. Instead, we focus on describing the characteristics of the nonstationarity of fertility behaviors across space, which is often regarded as source of nonrandom error variances that need to be controlled in global models. Using geographically referenced county-level data and applying GWR techniques, we estimate one set of coefficient estimates for each county while taking into consideration the characteristics of its neighboring counties. By so doing, we capture the apparent lack of uniformity in the observed relationship between fertility levels and their determinants and investigate local fertility patterns that are otherwise masked in the global models. We elaborate our hypotheses as follows. To begin with, we expect to find that economic development plays the more complex role in affecting fertility behaviors. Previous literature reveals that the relationship between economic development and fertility decline is not universal, in the sense that the effect is contingent on the microdynamics between pronatalist social norms and antinatalist fertility laws (e.g., Greenhalgh and Winckler 2005;Peng 2010;Zhang 2007). Even in vast rural areas where fertility policies are generally considered to be lenient, the magnitude and type of fertility incentives and regulations that couples are exposed to differ from place to place. To this end, we hypothesize H1a: Although economic development serves as one of the driving forces of fertility decline at the national level, in some places there exist positive relationships between economic development levels and fertility rates. This speaks to the observation that in several economically developed areas, if affluent families desire multiple births they can afford to pay the fines for violating family planning laws and the additional child rearing costs. On the other hand, some studies have found that small-family ideals have begun to emerge in some rural places, such that rural families tend to actively seek only one child, even if they are permitted to have an additional child (Zhang 2007;Greenhalgh and Winckler 2005;Merli and Smith 2002). To this end, we propose H1b: Fertility behaviors vary extensively in rural areas, such that that rural residency is not always an indicator of high fertility levels. We also consider the relationship between female education and fertility levels to be contingent on local economic structures and social/cultural norms. The classic economic theory of fertility decline argues that females with higher education are more likely to participate in the labor market, thus facing higher opportunity costs of childbearing (Becker 1960). However, negative relationships between female education levels and fertility rates are only observed when females have a certain level of autonomy when making childbearing decisions (Jejeebhoy 1995;Jain 1981). In places where son-preference norms are strong, females with higher education have better access to health facilities and maternal care and thus more quickly progress to having a second child (Qian 1997). Therefore, we expect to see H2a: Female education levels are non-negative or insignificantly related to fertility rates in places where traditional family norms are strong. However, we also expect to see some "threshold effects" (Jain 1981), in the sense that certain economic and social thresholds need to be reached before observing the negative effect of women's education on fertility levels. We hypothesize H2b: In places where extreme poverty persists, an increase in female education level is associated with an increase in fertility levels. In terms of cultural/ethnic differences, we hypothesize H3: There exist distinctive fertility patterns in places where ethnic minorities concentrate. For one thing, there are different fertility policies for ethnic groups and the Han Chinese; for another, the diffusion of fertility norms and behaviors is faster in homogenous populations (Attané and Courbage 2000;Childs et al. 2005;Poston, Chang, and Dan 2006). Therefore, we expect to see distinctive fertility patterns in places where ethnic identities are particularly strong. Our last hypothesis considers the spatial clusters of very low fertility rates. There have been many discussions on the emergence of low fertility regimes in the European context. For example, Kohler, Billari, and Ortega (2002) posit that the emergence of a lowest-low fertility (i.e., period fertility lower than 1.3) in Europe is attributable to the combined factors of measurement issues, newly emerged socioeconomic and institutional conditions, and amplified effects of social interaction processes. Although the primary focus of this paper is not to seek explanations for the emergence of spatial clusters of low fertility rates, applying GWR techniques nevertheless enables us to detect distinctive fertility behaviors in low fertility regimes. Therefore, we hypothesize H4: The spatial clusters of low fertility rates are not sufficiently explained by commonly identified socioeconomic, policy, and cultural factors. In other words, we expect to see relatively worse model fit in places where the low fertility rates are clustered. --- Data, measurement, and analytical strategy --- Data and measurement The People's Republic of China has 31 provinces or provincial-equivalent administrative divisions (Figure 1). The nation is conventionally divided into four greater geographic regions that are based on economic development levels: the east coast, the middle, the northeast, and the west, with the east coast being the most developed region and the west being the least developed. The unit of analysis is administrative counties. This study includes 2,869 counties or county-equivalent administrative units in mainland China. 4 The county-level data is primarily drawn from the 2010 population census. The 2010 county-level boundary shapefile is linked to the 2010 population census. The data set and the county boundary shapefile were both obtained from China Geo-Explorer II, which is maintained by the China Data Center (2014) of the University of Michigan. --- Figure 1: People's Republic of China This study uses unadjusted period total fertility rate (TFR) as the measure of county-level fertility. It describes the observed fertility patterns of a synthetic cohort of females who followed the same fertility schedule as those in the 2010 census. The TFR is calculated based on the age-specific fertility rate (ASFR) for women aged 15 to 49 reported in the 2010 census's long form. Compared with the previous five population censuses, the 2010 census is acknowledged as having better quality and more complete enumeration (Wu 2014;Cai 2013). We chose seven covariates as our independent variables, based on previous studies on the determinants of fertility in China. The first two variables are included to take into account the two largest subpopulations that receive nationwide exemptions from the stringent one-child-per-couple restriction. One variable, 'rural Hukou,' is measured as the percentage of the population that has rural residency status. Another variable, 'ethnic minority,' is measured as the percentage of ethnic minorities in the total population at the county level. Aside from the majority Han Chinese, there are 55 ethnic minority groups that are officially recognized by the central government in China. They are identified by ethnic, cultural, and religious criteria (Attané and Courbage 2000). Since the 1970s, national legislation has been consistent in stating that minority couples (or any couple in which one member is an ethnic minority) are not subject to the onechild policy (Attané and Courbage 2000;Gu et al. 2007). We expect 'rural Hukou' and 'ethnic minority' to be positively associated with TFR in our global model. 'Female education' is measured as the percentage of females aged 18 and above who have completed at least nine years of compulsory schooling, equivalent to completing primary and secondary education or higher. Female education has been identified as a strong predictor of fertility levels at both the individual and aggregated level (Jejeebhoy 1995;Martin 1995). Higher female education is expected to be negatively associated with county-level fertility at the national level. 'Net migration' is included in the model to capture the potential migration effect. The net migration rate is calculated as: (1 -Hukou population/census enumeration) <unk>100, following the practice made by Cai (2010). The county level net migration rate thus captures the percentage of population without local Hukou residency status to the total census enumerated population. A positive value of net migration rate indicates migration inflow, while a negative value suggests migration outflow (Cai 2010). We also included the measure 'poor living standard.' The 2010 census asked a list of questions regarding household living conditions, including whether there is a kitchen, tap water, a toilet, or a shower in the house. We calculated the percentage of households that do not have kitchens, tap water, toilets, or showers, for each county. To avoid the issue of multicollinearity, we applied principle component analysis to create a composite measure of poor living standard (Standard Cronbach's Alpha = 0.838) based on the four above-mentioned variables. We also included'sex ratio at birth' (SRB) in the model, measured as the number of boys at age zero divided by the number of girls at age zero. The imbalanced SRB has been a concern in China for the past two decades and can be attributed to three causes: the excessive birth of boys due to sex-selective abortion of female fetuses, the higher survival chances of boys due to infanticide and malnutrition of infant girls (Zeng et al. 1993), and the underreporting of female births in the 2010 census (Goodkind 2016;Shi and Kennedy 2016;Zeng et al. 1993). The imbalanced SRB is considered a result of the practice of son-preference, which is further amplified by the family planning policy. Existing studies have shown that the creation and implementation of family planning policies are heavily dependent on the sex of the surviving children. It is found that in places where there exist strong son-preference norms, couples whose first child is a girl are less likely to receive the one-child certificate, less likely to use contraceptives, and less likely to seek abortion than couples who already have at least one son (Arnold and Liu 1986). Therefore, at the aggregate level, we expect an overall positive relationship between SRB and TFR. The last independent variable we included in the model is 'fertility policy compliance ratio.' It is measured as the lifetime fertility level set by the local State Family Planning Commissions (SFPCs) divided by the observed lifetime fertility level in the 2010 census. The lifetime fertility level set by the local SFPCs, also referred to as 'policy fertility' in some existing literature, is a hypothetical indicator of what the total number of children ever born to a woman would look like in a place if all the households in that place obeyed the local fertility policy. We adopted this indicator at the province level from Table 3 in Gu et al. (2007). 5 Compared with the TFR, which depicts the fertility level in the year 2010, the lifetime fertility level is akin to a stock measure of the fertility levels in the past few decades. 6 A value of fertility policy compliance ratio greater than 1 indicates that the policy fertility is larger than the observed lifetime fertility, suggesting greater compliance with the local fertility policy, while a value of the ratio smaller than 1 indicates greater noncompliance with the local policy. Similar practices that attempted to measure the level of fertility policy compliance can be found in works by Attané (2002). We expect this indicator to be negatively related with TFR; that is, greater compliance with the local policy should be related to lower TFR. --- Analytical strategy We first produced the nonspatial descriptive statistics and conducted an exploratory spatial analysis to detect spatial clusters both visually and statistically. We then applied OLS models to obtain coefficient estimates at the global level. To explore the varying relationships between fertility and socioeconomic predictors, the last step is to apply GWR models. Following Fotheringham, Brunsdon, and Charlton (2003), the basic function of the GWR model is expressed as: = (, ) + <unk> (, ) +(1) where y i is TFR at county i, (, ) denotes the coordinates of the centroid of county i, is the local intercept for county i, and is the local coefficient for predictor n for county i. In GWR models the regression coefficients are estimated for each location independently by applying location-specific weighting schemes; therefore, there are as many 'local' regression models as there are observations (Wheeler and Tiefelsdorf 2005). In matrix form, the vector of local coefficients of is estimated as: =( )(2) where X is the matrix of independent variables, and y is the vector of dependent variables. The estimator in equation ( 2) is a weighted least squares estimator where the weights vary according to the location point of i. There are a variety of weighting schemes available for researchers to choose from (Fotheringham, Brunsdon, and Charlton 2003). We chose the Gaussian weights and their bi-square variations, which are the most commonly used options (Nakaya 2014). Thus in equation ( 2), is an n<unk>n diagonal matrix with the j-th diagonal element equal to 1-( ) <unk> if <unk> and zero otherwise. Noted is the Euclidean distance between location i, where the parameters are estimated, and a specific point in space j at which data is observed (Fotheringham, Brunsdon, and Charlton 2003); where b is the bandwidth size (i.e., the distance between each observation and its neighboring locations specified by the spatial weights). The bandwidth size was determined by an adaptive method (Nakaya 2014). Model comparison between the OLS and GWR models was conducted to illustrate using the Akaike Information Criterion with a correction for finite sample sizes (AICc). One of the advantages of GWR modeling is that researchers can map the local coefficients as well as R 2 in order to better identify spatial heterogeneities. Following this rationale, maps of the local coefficients and local R 2 are presented. The preliminary exploratory spatial analysis and the procedure of mapping the local coefficients were conducted with ArcGIS. The descriptive statistics and the global model were calculated in SAS. The GWR model was conducted in GWR4.0. --- Results --- Descriptive statistics and the OLS regression model Table 1 presents the descriptive statistics of the variables, and documents substantial county-level variation in TFR and socioeconomic factors. For example, although the mean of county-level TFR is 1.24, the lowest TFR is 0.02 in the Culuan district in Heilongjiang province and the highest TFR is 4.72 in the Jinkouhe district in Sichuan province. In terms of the independent variables, 'ethnic minority' on average comprises 16.22% of the total population at the county level. There are 13 counties that have 0% ethnic minority population, while 9 counties have 99% ethnic minority population. The percentage of population with 'rural Hukou' at the county level is 70%. Again, the statistics vary over space from 0% up to 100%. For 'female education,' on average only around 50% of females have at least middle school education. A zero value indicates that no females in a county have higher than middle school education. Table 1 also reveals the skewed sex ratio at birth, as the county-level average SRB is 1.15 males per females, which is higher than the typical range, which usually lies between 1.05 and 1.07 globally (Coale and Banister 1994). The lowest SRB is observed in the Shangganglin district, Heilongjiang province (0.77 males per females), and the largest SRB is observed in Anxi County, Fujian province (1.78 males per females). Lastly, the indicator 'fertility policy compliance ratio' has an average value of 1.15. Recall that this covariate is measured as the ratio of policy fertility to actual lifetime fertility level, where a value greater than 1 indicates that the observed lifetime fertility level in one place does not exceed the threshold set by the fertility planning committee, thus indicating greater compliance with the policy. The last column in Table 1 is the Moran's I statistic, which is a measure of the spatial autocorrelation among the neighboring values (Anselin 1995). The Moran's I statistics for all the variables are relatively high, suggesting strong spatial patterns that exhibit for both dependent and independent variables. The percentile map of the dependent variable, the county-level TFR, is presented in Figure 2a. The fertility rate is higher in the southeast and southwest regions of China and lower in the northeast region and part of the east coast region. To shed further light on the pattern and magnitude of spatial clusters, we explored the significant local spatial clusters/outliers of TFR using local indicators of spatial association (LISA) statistics. LISA is often used as a diagnostic tool for inspecting local spatial heterogeneity in measures of global spatial association (Anselin 1995). Figure 2b reveals that counties with high TFRs tend to cluster on the western boundaries of the Xinjiang and Tibet regions and southern provinces such as Yunnan, Guangxi, Guizhou, and Guangdong. Low TFRs tend to cluster in northeastern counties near the coastal line. Several metro areas, such as Beijing and Shanghai, also exhibit clusters of low TFRs. Aside from the spatial clusters the LISA map also highlights the spatial outliers, which are counties that surround clusters of counties with different values. For example, although high TFR values are concentrated in the southeast regions, there still exist counties that have significantly lower values of TFR compared to their neighboring counties in this region (denoted as 'low-high' outliers). Another place that exhibits low-high outliers in TFR is the Zhujiang River Delta in Guangzhou Province, one of the most economically developed areas in China. By contrast, there are counties in northern provincesnotably Hebei and Ningxia -that have high TFR values and are surrounded by low TFRs in the neighboring counties (denoted as 'high-low' outliers). To briefly sum up, the LISA map shows that the distribution of TFRs varies across space. For example, even among the most economically developed areas the fertility patterns differ between Shanghai and Guangzhou, where the former shows spatial clusters of 'low-low' TFRs and the latter exhibits clusters of 'low-high' outliers. This evidence suggests that a further analysis of the spatial heterogeneity of TFRs is necessary. --- Figure 2: County-level TFR and the local clusters/outliers Table 2 presents the coefficient estimates of the non-spatial global models. Model 1 includes two variables, 'ethnic minority' and 'rural Hukou.' Both variables are statistically significant and have the expected signs. These two covariates explain 23.5% of variation in county-level TFR. The full model is model 2, which includes all the covariates. The full model explains overall 32.2% of variation in the county-level TFR, and the AICc decreased from 2974.7 to 2578.4. In the final model all covariates have the expected signs. For example, aside from the strong positive relationship between ethnic minority and rural Hukou in TFR, model 2 also reveals a negative relationship between 'female education' and TFR. The magnitude of the effect is substantial, as a 10% increase in female education is associated with a 4.3% decrease in TFR.7 Among other covariates, the negative coefficient between 'net migration' and TFR suggests that the places where large population inflows exist are also the places where fertility levels are low, which is expected because migration flows in China are largely from rural to urban places. We found that 'poor living standard' does not exhibit a significant relationship with TFR after controlling for other social and policy-related variables. The global model reveals a positive relationship between observed SRB and TFR, indicating that places with more skewed SRB (more reported male births than female births) have a larger TFR. Lastly, we find a negative relationship between 'fertility policy compliance ratio' and TFR, suggesting that, all else being equal, greater policy compliance is related to lower TFR. We further conducted a multicollinearity test for all the explanatory variables. The variance inflation factors (VIFs) for the explanatory variable parameters are all within the reasonable range, suggesting that multicollinearity is not a big issue for the identified global model. --- Geographically weighted regression results The OLS model shows that the relationship between county-level TFR and socioeconomic factors is consistent with the existing story of socioeconomic development. This section demonstrates how the identified relationships change from one place to another and to what extent these local relationships remain hidden in the global model presented in the previous section. Table 3 presents the estimated coefficients of the GWR model. The last column is the diff-of-criterion test for spatial heterogeneity (Nakaya 2014). The test results suggest that all of the covariates are spatially nonstationary and therefore should all be treated as local covariates. In terms of overall goodness-of-fit, compared to the OLS model the GWR model explains 58.9% of the variation in county-level fertility. The AICc of the GWR model is 1403.7 compared to the 2578.4 of the OLS model, which suggests that the GWR model has a better fit than the global OLS model. Note: Positive value of diff-of-criterion suggests non-spatial variability. AICc = Akaike Information Criterion with a correction for finite sample size. To better present the GWR results, we conducted the following visualization processes. We first created the estimated coefficient maps. The local coefficients are classified by a five-quantile scale and are mapped using multi-hued color schemes. Next, we masked the maps of t-statistics on top of the local coefficient maps so as to only present the local coefficients that are statistically significant at the <unk> = 5% level, following a visualization method proposed by Matthews and Yang (2012). Because the conventional statistical inference procedure for detecting significant local coefficients in GWR models suffers from multiple testing problems (Byrne, Charlton, and Fotheringham 2009;da Silva and Fotheringham 2015), we adopted a correction method proposed by da Silva and Fotheringham (2015) to adjust our inference test results. 8 Our refined maps illustrate the local coefficients that are statistically significant after the adjustment (Figure 3). The local coefficient maps without accounting for significant levels are included in the Appendix. --- Figure 3: The estimated GWR local coefficients at the county level (significant areas only) Note: Local coefficients are displayed in quintile scales. Local inference test results are adjusted to correct multiple testing problems. For a significance level

China has been characterized by persistently low fertility rates since the 1990s. Existing literature has examined the relationship between fertility levels and social, economic, and policy-related determinants. However, the possible spatial variation in these relationships has not been investigated.The purpose of this study is to examine the potential spatially varying relationships between county-level fertility rates and policy and socioeconomic factors in China.Using geocoded 2010 county-level census data, this study adopts the geographically weighted regression (GWR) method to identify place-specific relationships between county-level total fertility rate (TFR) and socioeconomic and policy-related factors.We find that relationships between TFR and widely used social, economic, and policyrelated factors (rural Hukou, ethnic minority, female education, net migration rate, poor living standard, sex ratio at birth, fertility policy compliance ratio) vary spatially in terms of direction, strength, and magnitude. This spatial variation is largely due to differences in local characteristics. The differences between and the complexities of localities cannot be told by a single story of either government intervention or socioeconomic development.

first created the estimated coefficient maps. The local coefficients are classified by a five-quantile scale and are mapped using multi-hued color schemes. Next, we masked the maps of t-statistics on top of the local coefficient maps so as to only present the local coefficients that are statistically significant at the <unk> = 5% level, following a visualization method proposed by Matthews and Yang (2012). Because the conventional statistical inference procedure for detecting significant local coefficients in GWR models suffers from multiple testing problems (Byrne, Charlton, and Fotheringham 2009;da Silva and Fotheringham 2015), we adopted a correction method proposed by da Silva and Fotheringham (2015) to adjust our inference test results. 8 Our refined maps illustrate the local coefficients that are statistically significant after the adjustment (Figure 3). The local coefficient maps without accounting for significant levels are included in the Appendix. --- Figure 3: The estimated GWR local coefficients at the county level (significant areas only) Note: Local coefficients are displayed in quintile scales. Local inference test results are adjusted to correct multiple testing problems. For a significance level of 5%, the proposed correction method compares t-value with <unk>4.273 at the 95% confidence level, instead of the conventional value of <unk>1.96. This is equivalent to saying that our correction uses <unk> = 0.00001602 (calculated based on Equation 8 of da Silva and Fotheringham 2015) for a significance level of 5%. Figure 3a presents spatially varying relationship between 'rural Hukou' and TFR. The result indicates that the positive relationship is only statistically significant in the capital city of Beijing and its surrounding areas. Significant levels aside, a positive relationship between rural Hukou and TFR is found in the vast western region, while the east coast, middle, and northeast regions present a negative relationship between rural Hukou and TFR (see the Appendix). The relationship between 'ethnic minority' and TFR is presented in Figure 3b. Recall that the global model indicates a positive association between the percentage of ethnic minorities and fertility in each county, but the local coefficient map shows that the positive relationship does not hold across space. An investigation of Figure 4b reveals that significant positive clusters are found only on the borders of Gansu and Xijinang provinces, and significant negative clusters are found in southern regions in Tibet. One possible explanation for the different signs of local coefficients that appear in parts of Xijiang and Tibet is that the ethnicities differ in these two places. In Xijiang the majority ethnic groups are Uygurs, Kazaks, and Huis, known as Chinese Muslims. In total, they compose 56.5% of the total population in Xijiang. In Tibet, Tibetans compose 90.5% of the total population (Tibet Autonomous Region Statistic Bureau 2012). It is likely that fertility behaviors vary greatly within ethnic minority populations, especially considering the distinct social, cultural, and religious practices among the different ethnic groups. Figure 3c presents the local coefficients of 'female education.' The map indicates that the expected significant negative relationships are found only in several small places in Sichuan province, suggesting that female education still plays a salient role in reducing the fertility rate in the southwest. However, in northwest regions, most notably in Xinjiang and part of Tibet, the relationship between female education and TFR is positive, suggesting that the higher the percentage of females who have at least middle school education, the greater the county-level TFR. Consistent with several existing empirical studies, our findings suggest that the relationship between female education and fertility is context-specific, with respect to both the level of socioeconomic development and societal gender norms (Jejeebhoy 1995). It is also likely that certain economic and social thresholds need to be reached before the fertility decline commences (Jain 1981). Therefore, our findings suggest that future policy that aims to improve educational levels should target places such as Tibet and Xinjiang, where the socioeconomic development levels are low. The negative relationship between 'net migration' and TFR is partly confirmed in Figure 3d. However, significant local coefficients are only found in counties in Guangdong and Hunan provinces, leaving the majority of regions statistically insignificant. Large rural-to-urban migration flow is a new feature of Chinese population dynamics. One possible explanation for the insignificant relationship that is found in most places might be the census undercount of both fertility levels and migrant populations (Liang and Ma 2004). Despite the fact that the global model implies a positive relationship between 'poor living standard' and TFR, Figure 3e indicates that the local coefficients between poor living standard and TFR roughly follow a northeast-southwest gradient, with northeast regions displaying negative coefficients and southeast regions displaying positive coefficients. The significant negative relationships are clustered in northeastern regions, in particular in Heilongjiang province. Recall that the northeast is where the 'low-low' clusters of TFR are found (Figure 2b): It is possible that in the context of ultra-low fertility the relationship between poor living standard and fertility behavior becomes negative as the cost of childbearing and rearing becomes the most important concern over and above other factors. Figure 3f presents the estimated local relationships between'sex ratio at birth' and TFR. The global model indicates a positive relationship between SRB and TFR, while the local map shows that the relationship between SRB and TFR at the local level is highly variable across space. Specifically, there are significant negative clusters in three places: 1) Tibet, 2) the border between Sichuan and Yunnan provinces, and 3) Hainan province. This indicates that, all else being equal, a more balanced SRB (lower SRB) is associated with a higher TFR. An ancillary analysis of spatial clusters of SRB (available upon request) indicates that among the above-mentioned three areas, 1) and 2) are enclaves of low SRB, while 3) includes clusters of high SRB. Less-skewed SRB in the three areas could be attributed to two factors: weaker son-preference tradition, which is more likely to be observed among ethnic minority populations, and relaxed fertility policies. Specifically, 1) and 2) are places where the majority of the population is non-Han Chinese.9 Although the Han Chinese are the dominant population in 3), the fertility policy is much more relaxed in this region so that the policy fertility is above 2.0 (Gu et al. 2007). Although we can only speculate, the GWR result indicates that the relationship between SRB and TFR is conditional on local culture and socioeconomic context and cannot be summarized by a global model. Figure 3g displays local relationships between 'fertility policy compliance ratio' and TFR. Significant local relationships are found only in Tibet and Xinjiang: in the remaining places the relationship between level of compliance and TFR is statistically insignificant. To further examine the performance of the GWR model, maps of local R 2 and local residuals are presented in Figure 4. The local R 2 map (Figure 4a) shows that the GWR model fits well in most places in China, but with three exceptions. First, Tibet, the southern part of Xinjiang province, and Qinghai province, which have concentrations of ethnic minority groups, including Tibetans and Turkic-speaking populations (Uyghur, Kazakh, Kirgiz). Second, the northeast regions, in particular Heilongjiang and Jilin provinces, where clusters of 'low-low' TFRs are found (Figure 2b). Third, the GWR model does not fit well in central regions, including Hunan, Jiangxi, Hubei, and Henan provinces. --- Discussion and conclusion After almost forty years of economic development and implementation of family planning policies, fertility patterns in China present a mixed picture across space. In some places the variation in TFRs can still be explained within a conventional development framework; in others, clusters of low TFRs might suggest the emergence of new fertility regimes that are distinct from the processes of the first demographic transition. The spatial variation of TFRs might be caused by variation in socioeconomic development and fertility policy and in local cultural contexts and norms. Using geocoded 2010 county-level census data and GWR models, we found that the effect of socioeconomic development and fertility policy factors on TFRs varies spatially in terms of direction, magnitude, and significance levels. Moreover, the varying relationships do not follow predefined provincial administrative boundaries, suggesting spatial variation both between and within provinces. Our four hypotheses are all supported by the GWR results. First, the role that economic development plays varies across space, in particular in rural areas. For example, variables such as 'poor living standard,' measured as households' access to a set of modern household facilities, have different signs in the northeast and northwest areas after controlling for the composition of the rural population. In addition, the significant positive effect of 'rural Hukou' is found only in places close to Beijing, suggesting that there is some unique rural residency characteristic in this place that operates independently of economic and fertility policy factors. Second, the relationship between 'female education' and TFRs is place-specific. The GWR results suggest a significant negative relationship between female education and TFRs in only a few places, such as inland Sichuan province and border counties between Guangxi and Guangdong provinces, while the relationship is positive in northwestern regions. Moreover, although we did find a positive relationship in middle and southern China where traditional family norms are strong (see Appendix), the relationship becomes insignificant after we correct for multicollinearity in the inference test. Third, fertility differences overlap with cultural boundaries. This hypothesis is confirmed in the subsequent GWR models. The local R 2 map (Figure 5a) and residual map (Figure 5b) show that Tibet, the northern part of Xinjiang, and Qinghai have low model performance and non-random distribution of residuals. These results coincide with observations of historical fertility decline in Europe, in the sense that the diffusion of behaviors, technologies, and social norms is faster among people who share the same culture and language (Bongaarts and Watkins 1996). Existing studies of fertility in Tibet and Xijiang have documented distinct social, cultural, and political identities as well as family and childbearing patterns (Attané and Courbage 2000;Childs et al. 2005;Liu, Larsen, and Wyshak 2005). It is possible that these distinctions likewise affect fertility behaviors in these regions. Fourth, the existing socioeconomic and policy variables have low explanatory power in places with significant clusters of low TFRs. The LISA map (Figure 2b) indicates two types of area with spatial clusters of low TFRs: large metro areas (such as greater Beijing and Shanghai) and vast areas in northeast China. The subsequent GWR analysis further shows that our model performs well in metro areas, but not in areas in northeast China. In particular, the model fit is as low as 0.12 in the northeast, indicating that the existing seven predictors (rural Hukou, ethnic minority, female education, net migration, poor living standard, sex ratio at birth, fertility policy compliance ratio) cannot explain the clusters of low TFRs in the northeast. This suggests that there might be new social, cultural, and economic conditions that account for the clusters of low fertility in northeastern areas of China. To this end, our findings speak directly to placespecific and context-specific research on fertility trends in China. The GWR method is used as a "spatial microscope" (Fotheringham, Brunsdon, and Charlton 2003) that detects distinct place-specific relationships that otherwise may be dismissed in global models. That said, this study has two limitations that should be addressed in future research. First, the 2010 county-level TFR suffers from measurement error due to the underreporting of births, which is a well-known issue in the census of China. The issue of measurement error would be of less concern if the errors were spatially homogenous and did not correlate with any independent variables. However, existing literature suggests that the underreporting of births occurs more often in rural areas than in urban areas and that births of girls are more likely to be underreported than births of boys (Goodkind 2011;Li, Zhang, and Feldman 2010;Shi and Kennedy 2016). If this is the case, it is likely that our estimated coefficients of 'rural Hukou' and'sex ratio at birth' are biased downwards. It should also be noted that most existing studies that aim to address the underreporting of births are conducted at the national level, using intercensus backward projection to compare cohorts that are enumerated between censuses (Cai 2013;Goodkind 2011). We used unadjusted data in this study, because in order to correct the underreporting of births at the county level using the same methodology, children's inter-county migration trends need to be taken into account, but the data is not available. Future studies should tackle the issue of birth underreporting at the county level in order to better understand the relationship between TFR and the variables of interest. Second, using only census data limits the study, which does not include direct measures of economic development level such as gross domestic product per capita or average household income. Also, statistics on child mortality are not available at the county level, and this is one predictor that has been found to be highly associated with fertility level (Işik and Pinarcio<unk>lu 2006;Zhang 1990). In addition, this study focused on the spatial heterogeneity of fertility patterns for only a single year; therefore it cannot be used to make inferences concerning the causal relationships of these determinants. To better understand fertility transitions over time and across space, further research is needed that utilizes different data sources for multiple years. --- Acknowledgements We are grateful to Jingjing Wang and Yilang Feng for providing the data. Appreciation is extended to Yong Cai, Wen Fan, Glenn Firebaugh, Leif Jensen, Stephen Matthews, Jennifer Van Hook, the Demographic Research Editor, and three anonymous reviewers for their helpful comments and suggestions on earlier drafts of this paper. The research was supported in part by the National Science Foundation (Award # 1541136) and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (Award # P2C HD041025-16). --- Appendix Estimated GWR local coefficients at the county level (without presenting significant level) Note: Local coefficients are displayed in quintile scales.

China has been characterized by persistently low fertility rates since the 1990s. Existing literature has examined the relationship between fertility levels and social, economic, and policy-related determinants. However, the possible spatial variation in these relationships has not been investigated.The purpose of this study is to examine the potential spatially varying relationships between county-level fertility rates and policy and socioeconomic factors in China.Using geocoded 2010 county-level census data, this study adopts the geographically weighted regression (GWR) method to identify place-specific relationships between county-level total fertility rate (TFR) and socioeconomic and policy-related factors.We find that relationships between TFR and widely used social, economic, and policyrelated factors (rural Hukou, ethnic minority, female education, net migration rate, poor living standard, sex ratio at birth, fertility policy compliance ratio) vary spatially in terms of direction, strength, and magnitude. This spatial variation is largely due to differences in local characteristics. The differences between and the complexities of localities cannot be told by a single story of either government intervention or socioeconomic development.

Spirometry in COVID-19 Times -An Emerging Dilemma Spirometry is useful for the diagnosis, management and monitoring of chronic respiratory conditions in children, especially asthma. As severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can be transmitted via aerosol generation, coughing or sneezing [1], spirometry can pose a risk for transmission of the virus as the procedure requires generation of high minute ventilation and flow, and for the patient to be in close contact with the technician and equipment. We have tried to extrapolate information from adult guidelines on spirometry during the COVID-19 pandemic. As the pandemic evolves over time, prevalence can be classified to be in the pandemic phase, post-peak phase or postpandemic phase, with high, low or controlled community prevalence, respectively. This can be determined by the local health authorities. Level 1 safety recommendations are suggested for those places in the pandemic phase, Level 2 in the post-peak phase, and Level 3 in the post-pandemic phase [2]. Indication for spirometry: During the pandemic phase and postpeak phase, clinicians should restrict referrals for spirometry to those patients who require it urgently or when it is essential for their diagnosis [3]. A pediatrician can teleconsult the patient and determine the need for spirometry, to reduce the number of visits of a child to the hospital. One should; however, not perform spirometry on patients with a clinical suspicion of COVID-19, influenza-like illness (ILI) or severe acute respiratory infections (SARI) [4]. In children who test positive for COVID-19 infection, all pulmonary function tests (PFTs) should be deferred for at least 30 days post-infection, as viral shedding can occur even after 10 days. --- Guidelines for performing spirometry: The following are the Level 1 safety precautions one must follow while performing a spirometry in children during the pandemic phase. Similar precautions are advised for Level 2 in post-peak phase as it might be difficult to determine pre-test probability of infection in children. • Screening: The clinician or technician performing the test, the child and the caregiver, should all be screened prior to entering the PFT room. A proposed triage questionnaire is available in the European Respiratory Society statement [2]. Patients who screen positive should not undergo spirometry. • Infrastructure: Under ideal conditions, negative pressure rooms or HEPA filtration systems with UV germicidal lamps are recommended. However, this may not be available in most centres. Hence, at least a separate enclosed room with adequate ventilation should be designated for performing spirometry [2]. Waiting areas should be re-organized to ensure patients are not in contact with those who are febrile. Thorough cleaning and ventilation of both the room and equipment needs to be performed between each test [5]. The number of air exchanges between procedures need to be VOLUME 57 __ AUGUST 15, 2020 CORRESPONDENCE determined by each facility to ensure removal of 99.0-99.9% of airborne microorganisms calculated as per CDC guidelines [6]. Only one caregiver, who must wear a face mask and follow hand hygiene procedures, should be allowed into the room [5]. • Staff: The person performing the spirometry in the pandemic and post-peak phases should wear full personal protective equipment (PPE) which includes a fit tested N95 mask, eye goggles or face shield, apron and disposable gloves [7]. Strict hand hygiene protocols must be followed by both the operator and the patient. • Equipment: Equipment should be cleaned and disinfected by wiping down all surfaces that the patient comes in contact within a 2-metre radius, using a hospital grade antiviral disinfectant such as 70% isopropyl alcohol (IPA). Recalibration of the equipment after decontamination is suggested [2]. Single use bacterial and viral in-line filters of high specification are required to be used. The ideal filter is one with minimum proven efficiency for high expiratory flow of 600 to 700 L/min [8]. Replace all consumables to single use or disposable ones, wherever possible. Appointments need to be staggered with a gap of 45-60 minutes, taking into consideration the time required for donning and doffing of PPE by the clinician/technician between each patient, post-test cleaning of the room and equipment, and recalibration of the spirometer [5]. All these safety recommendations for performing spirometry must be maintained till the local public health authorities can confirm that the community spread is controlled and the district is in the post-pandemic phase. More specific guidelines for performing lung function tests in children will need to be formulated by global organizations as the pandemic evolves.

other stakeholders like community members, Panchayat Raj Institutions (PRI), Non-Government Organizations (NGO) and other public sector departments as a part of this campaign. Activities from within, like public rallies, marathons, swachhata walks, human chains, street plays/nukkad nataks/folk arts/folkmusic, etc. will act as potent instruments of social advocacy and community participation. It ensures every stakeholder from outside health facility premises and communities are gradually involved for hygiene and infection control and thereby helping health promotion at the grass-root level. World Health Organization estimates that Swach Bharat Abhiyan in India would potentially have a spectacular impact on improving the sanitation of communities and thereby averting disease burden within five years of its launch [4]. The integration and extension of such activities will be another opportunity for healthcare providers to make an impact on health indicators and disease burden. Subsequently, as all stakeholders adopt these initiatives, there will be a visible and viable behavior change of the public at large. To, summarize Beyond hospital boundary will act as a novel, innovative game-changer tool for community participation in sanitation, hygiene, and infection control.

INTRODUCTION Indonesia has a variety of cultural diversity that develops in the lives of its indigenous peoples. This is a form of national wealth which is continuously held and is believed to be down and down as a justified thing. Apart from the various mystical elements that exist, diversity can be used as a study in social studies learning that can be internalized in the character of students. Character training of students includes all fields of study and focuses on teaching and political education. The study of social sciences includes various definitions, data, events, to generalizations that are integrated with the community. Through social studies learning, students are guided to become democratic, responsible, and honorable world citizens. This is an attempt to find alternative solutions to address the increasingly worrying impact of globalization (Safruddin & Ahmad, 2020). The implementation of social studies learning in the globalization era is very important to strengthen the ability to think in the progress of science and technology as the statement by Wilson (1997), technological advances as a result of human culture based on life, community, and the environment. Social studies education plays an important role in passing down culture to continue the future by the nation's aspirations. In connection with tradition, the religious system of Javanese indigenous people has a small but formal small ceremony, not dramatic, and almost contains secrets, called Slametan (sometimes also called Kenduren). Slametan is the Javanese version of the most common religious ceremony in the world, a communal party. Just like in almost all places, the ceremonial activity developed the mystical and social unity of those who participated in it. Slametan can be held to respond to almost all events that want or will be commemorated, redeemed, or sanctified. These events, among them, namely birth, death, marriage, move to residence (house), magic, harvest, having a bad dream, changing the name of someone, opening a factory, sick, asking the spirit of village guards, circumcision and the beginning of a political meeting, and Various things can be the reason for holding Slametan (Geertz, 2013). Banyuwangi has a variety of traditions, and characteristics as ethnic groups found in Banyuwangi, used in Kemiren Village have a variety of traditional cultural forms and traditional ceremonies that are uniquely carried out for generations, namely the tradition of Tumpeng Sewu. The tradition of Tumpeng Sewu is one of the traditional rituals with the presentation of food dishes, namely Tumpeng as a symbol of the tradition they run. Food is an important aspect of human existence therefore food plays an important role as an element of ritual, ceremony, and religious activity (Indiarti, 2015). Tumpeng Sewu comes from the village activities called Slametan that they believe that the activity is carried out as a gratitude to God for the abundance of crops and as a clean ritual of the village so that it is always under His protection. In this regard, the implementation of local cultural values in social studies learning can be studied, namely perennials. Perenialism views education as an important process in the inheritance of cultural values. Cultural values possessed by the community must be transformed in education, so that it is known, accepted, and can be lived by students. Perenialism views that the value born in the past is valuable to be passed on to the younger generation. In social studies education, cultural transformation does not mean indoctrination of the values contained in it, but rather studies logically, critically, and analytically so that students can solve the problems they face in real terms. Social studies education was not in the past, but social studies education also could not ignore the future. Social studies education must accommodate all the needs of students, both inheritance of cultural values, and intellectual development, and prepare students for a better future. This can be done by developing teaching materials in learning. Several previous studies stated that teaching materials based on local values succeeded in inherent cultural values to students (Daryanti, 2021). This is because, revealed that teaching materials are a collection of knowledge that will be consumed by students through fun learning (Iskandarwassid & Sunendar, 2008;Syaharuddin et al., 2020). This shows that students tend to understand the benefits of teaching materials or resources after learning them. Teaching materials can channel signals, can activate the emotions, feelings, and desires of students so that they can inspire students to build a learning process (Filgona et al., 2020;Puspitarini & Hanif, 2019). Thus, the writing of this article aims to explore teaching materials by studying cultural values in the tradition of the Osing tribes as cultural material in social studies learning. --- METHOD The qualitative approach was employed with ethnographic design. It was chosen because it was considered to be able to express a complete and detailed understanding of the background of the phenomenon discussed (Creswell, 2015). While the ethnographic design was chosen because it was considered to be able to explore the behavior and problems faced by certain ethnic groups (Emzir, 2008). Primary data collection is carried out through in-depth interviews with the indigenous people of Osing Banyuwangi. While secondary data collection is carried out through the literature analysis in the order of data collection by documenting, checking various libraries, and analyzing as a source of research and other sources that can be used as supporting the results of primary data collection (Zed, 2004). The results of data collection were then analyzed using interactive data analysis techniques developed by Miles & Huberman (Miles & Huberman, 1992). Data analysis using this model is carried out continuously starting from the collection, reduction, and presentation of data, to withdrawing verification and conclusions. --- RESULT AND DISCUSSION The philosophy of Tumpeng Sewu Tumpeng Sewu is one of the cultures of the Osing tribe. This Osing tribe is one of the typical tribes of the Banyuwangi community. More precisely in the East Banyuwangi area, which is still thick with the peculiarities of the Osing tribe. The language used by the Osing tribe itself is Osing. As for the houses, there are still using the old traditional house. In addition, the Osing community has a livelihood for farmers and tourism. In the Osing tribe, various kinds of cultures are still applied by the surrounding community, one of which is Tumpeng Sewu. In addition to Tumpeng Sewu, there are also other cultures such as Mepe Kasur, Arak-arakan, and so forth. There is also a tourist village, precisely in Kemiren Village, Gelaga District, Banyuwangi. Since the time of Anas Regent, this village often holds a cultural festival. This is one way to preserve existing culture. In addition to the existence of the festival, the community itself also still adheres to and also prioritizes its customs. Therefore, this area is still strong in its culture and customs. In this case, they believe that the things they do will have a good impact, whereas if they do not do or continue the customary culture or existing habits, then bad things can happen to them. This Tumpeng Sewu was held in the month of Dzulhijjah before Eid al -Adha. This event was held simultaneously. And the peak event at the time of sunset. However, before the event began there were various series of events, such as the Village Clean Event and Arak Arakan, then after that, the peak event was Tumpeng Sewu. This event was held in front of each resident's house by holding a mat along the road. Then placed a cone on it. Of all the concoction of the concoction in the cone, there is one type of the Tumpeng Sewu, the Pitik Pecel which is enjoyed together. The procession of the Tumpeng Sewu is divided into three stages, namely preparation, implementation, and also closing. Where in the preparation process is carried out in a meeting preparing the needs and also equipment to support the activity. Then the second stage is the implementation, this first implementation is carried out with a mattress, race, barong wine, and then Slametan or a Tumpeng Sewu event. And the last is closure. The event was closed with cleaning places made by the event. As explained above, this village is a tourist village that often holds a cultural festival. This began in 2016, and since then there was a culture of Tumpeng Sewu. After the Tumpeng Sewu Celebration, not only the village community participated in the activity, but many of the general public participated in enlivening this event. For this Tumpeng Sewu activity, every resident's house was issued a Tumpeng. But if there is one big family then they only issue one cone and enjoyed it together. The event was led by the village head or customary head with prayer together and after that eating together. To prepare a cone one in one house, there is no coercion. And keep doing it well without exception and inviting communities outside the village. This Tumpeng Sewu event is like an ordinary Slametan that was in the ancestors. And the name Tumpeng Sewu only existed in 2015. During the Pandemic period last year, this Tumpeng Sewu activity was still carried out but still with the existing health protocol. Every village must commemorate this event, where this event was held in the month of Dzulhijjah, but in its implementation, it certainly was not together. Several villages have done it and have not yet carried out it. The purpose of this event is to be one of the forms of gratitude for what God has given, this could also be called reducing the Reject of Balak. This tradition is considered special by the local community, this is because the community is still carrying out this tradition for generations and also this tradition is carried out so that residents remain harmonious. And as explained above, this Tumpeng Sewu activity is carried out as a ritual of the village clean village, by their beliefs. This is because the activities are carried out in conjunction with village cleaning activities. As explained above that in addition to cleaning villages, this Tumpeng Sewu has another function, namely to connect friendships between residents. This tradition is carried out based on the Osing community which still believes in mystical things. In this activity, there is a philosophical meaning of equipment in this Tumpeng Sewu event. Like Pitik Pecel which has the meaning of the desire to get good results. Then the tumpeng is meaningful so that the disaster or something that is not good is lost. Sega Golong has that meaning of what is being done is given smoothness. For the last, Jenang Abang Putih has a red meaning as a symbol of the mother and white as a symbol of the father. --- Tumpeng Sewu as a Teaching Material for Social Studies Teaching materials are a guideline for teachers and students to achieve learning goals, as for teachers teaching materials, namely to direct all activities in the learning process and also the substance of competition that will be taught to students. As for students teaching materials function as guidelines in learning and also the substance of competencies that must be learned (Nasution et al., 2021). Therefore, it is important in the learning process to use teaching materials. At this time there are various kinds of teaching materials such as books, articles, and so forth. Therefore, with the presence of teaching materials based on local wisdom, the tradition of the Osing namely Tumpeng Osing is one of the teaching materials that can support social studies learning, especially for the VII class junior high school in the socialization and personality sub-theme in the 2017 curriculum. This teaching material will add to the attractiveness and interest of students to learn and can improve learning. The combination of social studies learning with teaching materials based on local wisdom is very appropriate. This is the purpose of social studies learning itself, namely development, knowledge, and skills (Sariyatun et al., 2021). With this combination, students can analyze daily social activities and this is in line with the ideal concept that learning is not only through teachers and students. However, students become more active in analyzing and also integrating social values around them. Teaching materials based on local wisdom traditions of Osing Tumpeng Sewu is a learning tool that contains subject matter as outlined in print form. This teaching material contains the material for ideas, facts, concepts, principles, and rules regarding the socialization and personality sub-theme in social studies class VII classrooms that are adapted to the tradition of the Osing tribes and the provisions of the existing social studies subject matter. Where in this case the tradition of the Osing tribe cone is a reference for the material used. As we know that in the tradition of Tumpeng Sewu, there are values of socialization and personality in it, as well as cooperation and harmony. Because in the tradition of Tumpeng Sewu, there are activities where the people help each other to carry out this Tumpeng Sewu. In addition, there is harmony in it, proven by this Tumpeng Sewu opening to the public. From this, we can find out that ethnic Osing tribes do not distinguish ethnicity, race, and so forth. This is also a form of unity and unity between citizens. If studied further, this Tumpeng Sewu has many meanings and philosophy that reflects the socialization sub-theme, and therefore the tradition of this Tumpeng Sewu is considered quite relevant if used as teaching material in social studies learning. This study produced teaching materials in the form of textbooks for class VII students with sub-themes regarding socialization and personality. As for the preparation of this teaching material, it is designed in a language that is easy to understand and also not monotonous that is using daily language but does not reduce the correct Indonesian rules. It is intended that the reader is that students are interested and easily understand the material delivered in the book. As for the more detailed specifications regarding the preparation of this teaching material that is there is an introduction, book cover, core part of the material, and also supporting parts such as concept maps, glossary, list of contents summary, and so on. Thus, this teaching material is expected to be accepted by all groups and can help educators and students in the implementation of learning activities, especially in social science lessons. --- CONCLUSION The procession of the Tumpeng Sewu is divided into three stages involving preparation, implementation, and also closing ceremony. The purpose of this event is to be one of the forms of gratitude for what God has given, this could also be called reducing the harmfulness or Tolak Balak. This tradition is considered special by the local community and they have been conducted for generations. The Tumpeng Sewu as a teaching material for social studies lessons, namely the presence of teaching materials based on local wisdom of the Osing Tumpeng Tradition, becomes one of the teaching materials that can support social studies learning, especially for junior high schools in class VII in the socialization and personality sub-theme in the 2017 curriculum. This teaching material will add to the attractiveness and interest of students to learn and can improve learning. The combination of social studies learning with teaching materials based on local wisdom is very appropriate. The purpose of social studies learning includes development, knowledge, and skills. With this combination, students can analyze daily social activities and they not only learn through teachers and students, but they could learn from the traditions and social customs.

This study aims to explore social studies teaching materials based on the Tumpeng Sewu tradition of the Osing tribe. A qualitative approach was used by conducting interviews with informants and strengthen by relevant previous research. The results showed that the culture of Tumpeng Sewu became one of the hereditary cultures of the Banyuwangi people, especially the people in Kemiren Village. This culture is a culinary festival of Tumpeng Sewu in a traditional clean village ritual in Kemiren Village, Banyuwangi. This traditional ritual is carried out by the Osing ethnic group and is held every year by serving food dishes as a form of their traditional tradition. This custom may be great in attracting tourists and improving the economic aspects of the community. The culinary festival is an annual program of the Banyuwangi government which has been held from 2015 to 2019, even though it is hindered by a pandemic, this ritual continues even though the procession is slightly different. The contribution of this research is used as cultural material in social studies learning which refers to literacy studies and interviews with resource persons.

INTRODUCTION In recent years, the growth of online banking has been remarkable, transforming the way individuals and businesses conduct financial transactions. Online banking has become a key player in the realm of business activities, offering unparalleled convenience, accessibility, and a wide range of services [1]. The banking industry has embraced this digital shift, adopting uninterruptible banking services to reduce operating costs and enhance customer experience [2]. Online banking in Nigeria has gained significant momentum, fuelled by the dramatic increase in ecommerce applications and the numerous benefits it offers, including fund transfers, checking account management, and bill payments [3]. However, alongside the immense advantages of online banking, the banking sector in Nigeria faces persistent cybersecurity challenges. Despite substantial investments in securing data, networks, and cyber defines systems, the occurrence of cybersecurity breaches and vulnerabilities is on the rise [4]. The prominence of human factors in contributing to these cybersecurity risks cannot be overlooked. Human behaviour and actions often introduce inconsistencies and errors, posing substantial threats to information assets. Consequently, a comprehensive understanding of the human factors that influence cybersecurity culture is essential for effective risk mitigation [5]. Cybersecurity culture is the collective mindset, attitudes, and behaviours of individuals, organizations, and society as a whole towards ensuring and promoting the security of digital systems and information. It encompasses a shared understanding of the importance of cybersecurity, a commitment to implementing best practices and protocols, and a proactive approach to identifying and mitigating cyber risks. A strong cybersecurity culture fosters a security-conscious environment where cybersecurity is integrated into daily practices and where individuals are vigilant and proactive in protecting against cyber threats [6]. While the significance of human factors in shaping cybersecurity culture has been recognized, there remain notable gaps in the existing literature. Specifically, limited research has been conducted to identify and explore the specific human factors influencing cybersecurity culture among online banking users in Nigeria. The available literature primarily focuses on assessing the state of cybersecurity in Nigeria, with insufficient attention given to cybersecurity culture, standards, interpersonal trust, and social norms. As a result, there is a pressing need to bridge this gap and develop a comprehensive conceptual model that addresses the unique Nigerian context [7]. This study aims to fill the gaps by providing a wellgrounded conceptual model that identifies and examines the key human factors influencing cybersecurity culture among online banking users in Nigeria. This research will contribute to the enhancement of cybersecurity practices, risk mitigation, and the overall cybersecurity culture in the Nigerian online banking sector. The systematic literature review will serve as the foundation for developing a robust conceptual model that encompasses the identified human factors and their interrelationships. By distilling and analysing the literature, this research aims to provide valuable insights into the specific human factors influencing cybersecurity culture. The model will shed light on the complexities and dynamics of these factors, facilitating a comprehensive understanding of their impact on cybersecurity behaviour among online banking users. Also, it is crucial to highlight the limitations of previous milestone works in this field [8][9]. Although previous works have contributed significantly to the understanding of cybersecurity culture and behaviour, their focus on developed countries with distinct demographic, cultural, and infrastructural settings raise questions about the universality of their findings. Moreover, the limited research conducted in Nigeria primarily focused on assessing the state of cybersecurity, rather than delving into the realm of cybersecurity culture. Therefore, this study seeks to rectify these limitations and contribute novel insights specific to the Nigerian context. By addressing the gaps in the existing literature, this research endeavours to provide a comprehensive understanding of the human factors that influence cybersecurity culture among online banking users in Nigeria. The findings will serve as a foundation for designing targeted interventions, policies, and educational programs to promote responsible cybersecurity practices and mitigate cyber threats. Ultimately, the aim is to foster a strong cybersecurity culture that safeguards the interests of online banking users, protects sensitive information, and ensures the sustainable growth of online banking in Nigeria. --- METHODOLOGY The methodology employed for developing the conceptual model involved a systematic review of previous cybersecurity frameworks and research papers relevant to cybersecurity culture. This process aimed to identify and analyze the key factors influencing cybersecurity culture. The gathered literature underwent a qualitative content analysis, whereby relevant documents were identified and classified. This approach allowed for a systematic examination of the variables and constructs proposed in previous research within the cybersecurity culture domain. --- CONCEPTUAL MODEL DEVELOPMEN-T The development of a robust cybersecurity culture model requires consideration of existing frameworks in the field. A thorough review of previous cybersecurity frameworks was conducted, leading to the formulation of the current conceptual model for cybersecurity culture. The primary objective of this comprehensive evaluation was to provide a comprehensive summary and analysis of variables proposed in previous researches within the cybersecurity culture domain, thereby supporting the conceptual model of the present study. Extensive literature searches were conducted across prominent digital databases, including Emerald, AIS, Elsevier Science Direct, ACM, Springer, and Google Scholar, focusing on papers published between 2018 and 2022 and employing keywords such as "Cybersecurity Culture" and "human factors." Through qualitative content analysis, relevant documents were identified and classified. In total, 44 papers specifically addressed Cybersecurity Culture, with 28 papers (representing 64% of the total) aligning with the Cybersecurity Culture framework, in line with the study's objectives. The remaining papers covered a range of topics, including definitions of cybersecurity culture, distinctions between organizational culture and national culture in relation to cybersecurity culture, strategies for developing national and organizational cybersecurity culture, and the goals of cybersecurity culture development. --- Table 1: The Summary of key Human Factors. --- Research Constructs [11] Knowledge, assumptions, norm and value, artifact. [12] Management commitment to information security, security policy and policy enforcement, security Awareness, security training and education, security risk assessment, security compliance, ethical conduct. [13], security, top management, ecurity compliance S., job satisfaction ommunication c [14] Security behaviour, security awareness, social norm, enforcement of information security policy. [15] Security and SETA program, Security policy. ty monitori [16] Organizational level: Assets, continuity, trust, operations, defense, security governance. --- Individual level: Awareness, attitude, behaviour, competency. [17] Organizational culture model: Eighteen studies were retrieved in Table 1. The process used to develop the conceptual model was to extract research in existing information security culture frameworks and models in order to develop an understanding of current information security culture phenomena. For each study, all the proposed constructs were extracted and counted in Table 2. The purpose for counting constructs for each study is to identify top constructs as potential candidates because it is simply impossible to examine every factor that could help conceptualize a security culture. Because of the scope limitation, the current paper will only consider the top constructs where there is strong agreement between academic researchers as to their importance for cybersecurity culture. Table 2 presents top key constructs for that influence cybersecurity culture among online banking users. Cybersecurity In Table 1 above, the review of previous research and the adopted constructs from each study were analysed. The key constructs that influencing cybersecurity culture and led the development of conceptual model are summarized in Table 2. In the section that follows, we lay out the basic framework for modelling cybersecurity culture (see Figure. 1) and hypothesize about the link between cybersecurity cultures and influencing factors. Based on our review of the research, we selected the following top seven key variables as show below: --- Justification for ranking of conceptual factors To determine the ranking of key constructs that influence cybersecurity culture among online banking users, the proposed constructs from each study were counted and analyzed. The purpose of ranking was to identify the top constructs with strong consensus and importance among academic researchers. The ranking of key constructs was based on the frequency of their appearance in the reviewed literature and their recognized significance in influencing cybersecurity culture. The ranking of conceptual factors in the proposed cybersecurity culture framework was determined through a systematic and evidence-based approach. This ranking process aimed to identify the most influential factors based on their frequency in the reviewed literature and their recognized significance in shaping cybersecurity culture among online banking users in Nigeria. Cybersecurity Awareness (Ranked 1): The top ranking of "cybersecurity awareness" is grounded in its consistent and prominent emphasis in the reviewed literature. Multiple studies and academic research papers consistently underscored the critical role of cybersecurity awareness in cultivating a securityconscious mindset among online banking users. It emerged as the most frequently cited and emphasized factor across the selected literature, signifying its central importance in influencing cybersecurity culture. Cybersecurity Policy (Ranked 2): "Cybersecurity policy" secured the second position due to its recognized significance in guiding and enforcing cybersecurity practices. Numerous studies highlighted the importance of well-defined policies in setting the foundation for cybersecurity culture. It was consistently cited and emphasized in the literature as a pivotal factor in promoting security-conscious behavior. Cybersecurity Education (Ranked 3): "Cybersecurity education" claimed the third rank based on its crucial role in equipping online banking users with the knowledge and skills needed to navigate the digital landscape securely. While slightly less frequent than awareness and policy, education was consistently recognized as a fundamental factor in enhancing cybersecurity culture. The ranking reflects the prevalence and importance of each construct as evidenced by the reviewed literature. Notably, "cybersecurity awareness" emerged as the top-ranked factor due to its consistent emphasis and recognition in the field, highlighting its central role in shaping cybersecurity culture. This approach ensures transparency and objectivity in the development of the conceptual model, as the ranking is grounded in a systematic review of existing research rather than subjective opinions. --- Conceptual Model The conceptual model human factors that influencing cybersecurity culture among online banking users in Nigeria. --- DISCUSSION The existing literature review provides and highlighted the key human factors that influence the Cybersecurity Culture among Nigerian internet banking customers. The significance and effectiveness of these human factors on Cybersecurity Culture are vary. From the Table 1 for the summary of previous research on cybersecurity culture the cybersecurity awareness, cybersecurity education and cybersecurity policy are the most construct proposed by the researchers. all of this clearly shows that cybersecurity awareness, education, and policy are the most important factors in identifying the value of cybersecurity, as they work to develop a strategic framework to educate users of the need to follow the cybersecurity policy to avoid any cybersecurity incidents and to enhance the security of users by minimizing the possible cyberthreat. However, the culture of cybersecurity must be strengthened according to strategic plans and scientific methodology, in addition to educating everyone on the necessity and importance of awareness of the concept of cybersecurity culture and exchanging experiences in this regard. Furthermore, the literature review reveals that there is a lack of investigation into what is required of Cybersecurity knowledge that should be incorporated across the organization and individuals to improve Cybersecurity culture. Furthermore, the review also found that very few researchers have addressed the influence of social influence and interpersonal trust on cybersecurity culture. The review also suggested that having an effective cybersecurity culture may potentially contribute to positive cybersecurity behavior, there is a considerable research gap in recognizing each variable and assessing its influences on Cybersecurity culture among organization and individual. In comparing the findings of this study with recent and related published reports, several commonalities and variations can be observed. Firstly, the importance of cybersecurity awareness, education, and policy as influential factors in cybersecurity culture is consistent across multiple studies. These factors are consistently recognized as essential for promoting a security-conscious mindset and behavior among online banking users in Nigeria. However, it is worth noting that while cybersecurity awareness, education, and policy are widely acknowledged, the specific strategies and approaches for implementing them may differ among studies. Some reports emphasize the role of awareness campaigns and training programs, while others focus on the development and enforcement of robust cybersecurity policies. These variations highlight the need for tailored approaches that consider the unique context and characteristics of online banking users in Nigeria. Another notable finding from this study, which aligns with some published reports, is the lack of comprehensive investigation into the required cybersecurity knowledge across organizations and individuals. While the importance of cybersecurity knowledge is acknowledged, there is a research gap in understanding the specific knowledge areas that should be prioritized and incorporated into cybersecurity culture initiatives. Additionally, this study highlights the limited attention given to social influence and interpersonal trust as factors influencing cybersecurity culture. Similarly, some recent reports also identify this gap, emphasizing the need to explore the role of social dynamics and trust relationships in shaping individuals' cybersecurity behaviours and attitudes. --- CONCLUSION To conclude, this study provides valuable insights into the human factors influencing cybersecurity culture among online banking users in Nigeria. Through a comprehensive review of existing literature, several key findings have emerged, highlighting the significance of cybersecurity awareness, education, and policy in fostering a secure online banking environment. The findings emphasize the need for strategic initiatives to enhance cybersecurity culture in Nigeria. By prioritizing cybersecurity awareness campaigns, implementing comprehensive educational programs, and developing robust policies, financial institutions and relevant stakeholders can empower online banking users with the knowledge and skills needed to protect themselves against cyber threats. Furthermore, the study highlights the importance of recognizing and addressing the specific cybersecurity knowledge requirements across organizations and individuals. Efforts should be made to identify and incorporate relevant knowledge areas into training programs and organizational practices, ensuring a comprehensive understanding of cybersecurity principles and practices. The study also emphasizes the role of social influence and interpersonal trust in shaping cybersecurity culture. Recognizing the impact of social dynamics on individual behaviours and attitudes, it is essential to foster a supportive and collaborative environment that encourages responsible cybersecurity practices. Moving forward, it is recommended that further research be conducted to explore the identified research gaps and expand the understanding of cybersecurity culture in the Nigerian online banking context. This includes investigating the effectiveness of different awareness strategies, evaluating the impact of specific cybersecurity knowledge areas, and exploring the dynamics of social influence and trust in shaping cybersecurity behaviours. In conclusion, by implementing the insights gained from this study and adopting a proactive approach to cybersecurity culture, Nigeria's online banking sector can strengthen its resilience against cyber threats and create a safer digital environment for all users.

This study aims to construct a comprehensive conceptual framework that elucidates the critical human factors influencing cybersecurity culture among online banking users in Nigeria. The research methodology is grounded in a meticulous examination of existing literature in the cybersecurity culture domain, serving as the foundation for this framework. The literature review reveals a conspicuous absence of academic research on cybersecurity culture within Nigeria and underscores the importance of comprehending its unique nuances. Key findings from the literature review highlight the prominence of "cybersecurity awareness," "cybersecurity policy," and "cybersecurity education" as influential factors. "Cybersecurity awareness" emerges as the most pivotal factor due to its recurrent emphasis and recognized centrality. "Cybersecurity policy" and "cybersecurity education" secure the second and third positions, respectively, due to their acknowledged significance in cultivating a security-conscious mindset among online banking users. Furthermore, the literature review exposes a research gap concerning the requisite "cybersecurity knowledge" that should permeate organizations and individuals to augment cybersecurity culture. Additionally, it reveals the underexplored influence of "social norms" and "interpersonal trust" in molding cybersecurity culture. This research accentuates the dearth of cybersecurity culture research within Nigeria and underscores the importance of understanding its unique facets. The proposed conceptual framework provides a valuable resource for designing tailored cybersecurity strategies and programs in Nigeria's online banking sector. It advocates for prioritizing cybersecurity awareness, education, and policy, empowering users with the knowledge and skills needed to safeguard themselves against cyber threats. The model also highlights the relevance of recognizing the role played by social dynamics, interpersonal trust, and social norms in shaping cybersecurity behaviours.

INTRODUCTION C linical observation is a method well known to primary care physicians. It is the rare physician who takes a patient's words at face value without also using contextual clues, such as the patient's appearance and behavior, to construct a picture of the patient's health. 1 Indeed, medical education has recently highlighted the need for physicians to be more observant through innovative curricula that teach observational skills by examining art. 2,3 Field observations offer insight into behaviors and the environment 4 and can play an important role in primary care research. A decades-long history of observational research in primary care has demonstrated how contextual factors both in and out of the clinic influence the effectiveness of interventions, as well as clinical outcomes. [5][6][7][8][9] A recent commentary in JAMA Internal Medicine has called for more "ethnographic and field studies" to capitalize on the "value of direct observation," particularly in studies of patient safety. 10(p1024) Unfortunately, most primary care researchers lack a practical approach for including field observations in their studies. Here, we present an overview of unstructured field observations as a qualitative research method for analyzing material surroundings and social interactions, aimed at researchers new to unstructured observations. We then detail a practical approach to collecting and recording observational data through a "3 Cs" template of content, context, and concepts. To demonstrate how this method works in practice, we provide an example of a completed template and discuss the analytical approach used during a study on informed consent for research participation in the primary care setting of Qatar. --- BACKGROUND Observational field research has its roots in the social sciences and is most often associated with participant observation fieldwork in cultural anthropology and sociology, though it has spread to disciplines as diverse as nursing, education, and social work. 11 Although he was not the first to Michael D. Fetters, MD, MPH, MA 1 Ellen B. Rubinstein, PhD, MA 2 undertake fieldwork, the Polish anthropologist Bronislaw Malinowski has been ascribed with popularizing participant observation as a research method. 12 Unlike structured interview sessions, participant observation allows for more flexible interactions with research participants through social events, casual conversations, and semistructured interviews conducted in settings that are a natural part of participants' everyday lives. 13 There is some debate as to how much a researcher should "participate" in daily activities, 12 but it is often enough simply "to be tolerated as an unobtrusive observer." 14(p195) Malinowski's work was influenced by the scientific positivism of the era, but later social scientists launched an antipositivist (interpretivist) critique that shifted the focus of observational research from a search for social facts to an understanding of cultural meanings. 15 We follow an interpretivist approach in believing the purpose of observational research is to arrive at what the anthropologist Clifford Geertz called "thick description." 16 In a famous example, Geertz explains how the same action (a quick blink of the right eye) holds different meanings based on the cultural context in which it occurs. 16 Only in a specific context will a blink be recognized as a wink-a conspiratorial signal to a friend-rather than as a meaningless twitch of the eye. Thick description, then, uses empirical data from multiple sources to contextualize individual behavior and interpret its meaning. A crucial assumption in participant observation is that the researcher is the instrument of data collection. 17 This assumption means all data are filtered through the researcher, through his or her personal characteristics, background, and experiences. 18 The result is a different relationship to objectivity than that found in the natural sciences. Observational research does not seek objectivity through a conceptual separation between researcher and study participants and phenomena. 19 Researchers are not blank slates; instead, they use prior knowledge and experiences as either implicit or explicit bases of comparison to understand what they observe in the field. 20 Personal bias is not seen as a flaw but as part of data collection. 12 Qualitative researchers speak of using bias to explore hunches or ideas, as well as to seek out information contrary to their own views. 21 It is critical before going into the field for researchers to conduct a personal inventory and reflect on their preconceptions, as these will affect their interpretation of events. 21 Researchers must also remain aware of potential differences in status between themselves and their study participants, where higher status may be conferred by socioeconomic, educational, occupational, or other types of privilege, or some combination thereof. 19 Such differences create a certain power dynamic between researchers and participants that can influence data collection. 19 A reflective activity, such as journaling or freewriting, can help define the researcher's lens and help the researcher recognize how he or she may be perceived in the field. 22 --- RATIONALE FOR COLLECTING FIELD OBSERVATIONS Perhaps the most compelling reason for conducting observations is to understand behavior (Table 1). Studies have suggested that nonverbal behavior plays a considerable role in communicating shared cultural meanings. 23 Furthermore, emotions are communicated more through facial expression and vocal tone and inflection than through spoken word. 24,25 Finally, as it pertains to health research, individuals tend to underreport their unhealthy or socially undesirable behaviors 26,27 and overreport the converse. 28 Observational research thus contributes to understanding human behavior in all of these scenarios. Field observations can help researchers understand how the interactions and activities in a given setting inform behaviors and beliefs (Table 1). 29 They contribute to uncovering the broader context of a given scenario, making observational research especially well suited for studying process. 19 In primary care, for example, observational research has deepened understanding of the clinical contexts in which practice transformation to patient-centered medical homes has occurred. [30][31][32][33][34][35][36] Field observations have also helped contextualize the (quantitative) results of randomized controlled trials, by providing insight into why certain randomized controlled trials have succeeded or failed. 9,37 Thus, field observations allow researchers to move beyond selected perceptions and protocols, and better understand the nature of particular activities, who performs them, and the contexts in which they occur. Observational research may not always be appropriate, and under certain circumstances, participants will not tolerate an observer (Table 2). Individuals who engage in socially undesirable or illegal behavior may be reluctant to have an observer present or be at risk personally 12 (although the opposite might also be true 38 ). Sometimes the presence of an observer might change participants' behavior, a phenomenon known as the Hawthorne effect. 39 Research has shown this effect may not be as great as feared, 40 however, and multiple observations over the long term can be a mitigating factor. 29 --- UNSTRUCTURED AND STRUCTURED FIELD OBSERVATIONS A central tenet of observational research is that observations must be recorded to count as data, 12 and it is only through the regular and systematic recording of field observations that researchers create texts for subsequent analysis. 41 Field observations can be structured, unstructured, or a combination of both. Structured observations use a template to record tabulations of specific behaviors that can be measured and analyzed statistically. 42 These observations often focus on highly specific issues and are indicated when the intent is to validly and reliably measure specific behaviors. Unstructured observational data use the researcher's words for thick description of phenomena or events. These words emerge through the researcher's experience in the field. 43 Even though unstructured, the observations are still focused because they address a research question or specific phenomenon. Within that focus, there is great latitude in what the researcher can choose to record (Table 3). To illustrate the potential of field observations, we introduce a study that relied heavily on observational data supplemented by interviews. The first author (M.D.F.) served as co-principal investigator on a multistage, mixed methods parent study in Qatar to develop a health care quality assessment instrument that was adapted for populations whose native languages were Arabic, English, Hindi, and Urdu. 44 A theoretical model, Cultural Construction of Clinical Reality developed by Kleinman et al, 45 guided the research. During the first stage of qualitative data collection, Qatari team members expressed concern about how potential participants would respond to recruitment strategies dictated by US-based institutional review boards. Existing literature had little guidance for recruiting participants in Arabian Gulf countries. We therefore conducted a study to explore how individuals in Qatar responded to in-person recruitment requests. 46 The 3 Cs template emerged from the need for institutional review board approval of data collection instruments and as a training instrument for research assistants. --- THE 3 CS UNSTRUCTURED FIELD OBSERVATIONS TEMPLATE The 3 Cs template provides an easy approach to collecting observational data. A full 3 Cs template (Supplemental Appendix 1, available at http://www.AnnFamMed. org/content/17/6/554/suppl/DC1/) includes the study metadata (details about the project), the research question, and the 3 Cs of context, content, and concepts. --- Field Observation Metadata The study metadata include information such as the project title, the study document type, the observer, the date and time, the location, and a participant description (eg, ID, descriptor if multiple individuals). The metadata from a single observation collected during the Qatar recruitment study are given in Supplemental Appendix 2, available at http://www. AnnFamMed.org/content/17/6/554/suppl/DC1/. Additional information here included the language spoken by the researcher and participant. --- Research Question The research question keeps observations focused on the purpose of the study. Because there is bound to be individual variation in data collection (recalling that the researcher is the instrument of data collection), including the question on the template provides a focal point for recording details. As illustrated in Supplemental Appendix 2, there were both primary and secondary research questions for researchers to address. --- Context, Content, and Concepts The context, content, and concepts are summarized in Table 3 and explained in further detail below. --- Context The first C, context, has a dual meaning: its most immediate referent is the circumstances in which the researcher is making observations (eg, who is doing the observation, where, and when). In a broader sense, context also refers to the researcher's (or team's) prior research experience, and could include knowledge gained from reading the scholarly literature, and additional information about a population, organization, or community-anything that speaks to the social surroundings and recent events that may influence the present interaction. 47 Context can act as a reminder of the researcher's purpose in using observational methods-and may also include information about the researcher's introduction to the field (eg, "invited by Dr A to observe her clinic" or "clinic recruited through practice manager"). Visuals, including sketches or photographs (if possible), may be useful, especially if certain structural features stand out (eg, small vs large waiting room; front staff behind an open desk vs behind a glass partition). This information can enable quick and easy comparisons between multiple locations. --- Content The second C, content, is more focused than context, as it refers directly to what happens during the observation period. (We recommend that researchers record each observation period on a separate template, creating multiple shorter documents, for each episode of observing, rather than in a single long document, for later analysis.) Although observational research is often touted as holistic-that is, covering everything within the 5 senses during a given stretch of time-in truth, researchers continually make decisions about where to point their focus. The research question and the project's theoretical orientation (the research questions, and the first C, context, described previously) are paramount here in guiding the observations. --- Concepts The third C, concepts, is a space for the researcher to connect the minutiae of his or her field observations with the bigger picture, to think back to the research question or hypothesis and compare theory with practice. It can be used to reflect about the process of research, or procedural or ethical questions that arise in the field. It is a space for nascent analysis, where the researcher can try out new ideas based on insights gained in the field. Researchers may take note of emergent patterns or themes (as in grounded theory 48 ), may reflect on their experience in the field, or both. In this way, the third C provides a sort of running commentary on the field observations, which serves both as the foundation for later analysis and as an audit trail that demonstrates how observation, data collection, theory, and analysis are all intertwined. 15 --- USING THE 3 CS APPROACH IN THE FIELD Recording Field Jottings There is never enough time in the field to write down everything one observes. Indeed, recording observations becomes even more difficult when the researcher participates in ongoing activities by, for example, engaging in conversation, directing a lost patient to the exit, or lending a hand to set up for a meeting. Furthermore, most of the time, a field researcher does not bring a computer into the field but rather relies on pen and paper, the latter being less obtrusive and easier to manage (although that has changed with the advent of smartphones and tablets). Generally speaking, therefore, no one writes actual field notes in the field. Rather, they capture field jottings-words, phrases, or drawings to jog their memories after the fact. 41 In the Qatar recruitment study, 46 the research assistants used a printed 3 Cs template for jotting down ideas as they occurred. --- Expanding on Field Jottings After leaving the field site, the observer has the opportunity to construct, from field jottings, a more extensive record of what was observed. 29 It is important to write these expanded field notes as soon as possible after the observation session (hours, not days, later); the more time that passes, the greater chance that the researcher will recall fewer details. A first pass should focus on writing descriptive text focusing on the "what" and "how" while avoiding summative or generalizing statements. The best field notes are rich with details that convey a sense of what it is like to be in the time and space described (ie, thick description). Expanded and edited notes can be found in Supplemental Appendix 2. --- Working With a Research Team and Analyzing the Data Primary care research is typically team based. The 3 Cs approach to field observations can facilitate iterative data collection and analysis. Field notes should be shared among team members both to encourage the entire team's familiarity with the field and to identify salient points for future data collection and pattern recognition. The first few observational sessions should be the most detailed, as this is when the researcher is least familiar with the field. 15 Of course, not all details will ultimately prove relevant, but they can act as a starting point for more refined observations later on. Furthermore, after the researcher begins to recognize what counts as the norm in a given setting, he or she can then more easily identify deviations from that norm. Such surprises in the field can lead to unexpected insights that generate new directions for future research. Field notes can be analyzed like any other qualitative text by, for example, searching for key words and coding for thematic content. Even if they are not formally analyzed, field notes may provide important context for understanding data derived from (timelimited) interviews and surveys. In the Qatar recruitment study, 46 the team conducted iterative analysis of the textual data from field observations and interviews, which led to the creation of a coding scheme in Atlas. ti (ATLAS.ti Scientific Software Development GmbH). The team also integrated numerical data from the demographic instrument and recruitment procedures, to create a narrative format for the results. --- Finishing Field Observations in a Study A final consideration is when to end the observational period or series. One possibility is to have a predetermined end point. In the Qatar recruitment study, 46 field observations continued until the study recruitment goals were met. An alternative is to seek data saturation. Guidelines for determining the saturation point in qualitative research vary, although a recent review calls for saturation to be operationalized in terms of research question(s), theoretical framework, and analytical goals. 49 --- DISCUSSION Field observations are paramount in understanding contextual factors in any research project and have the potential to reveal important insights about the way actors and institutions work in a given place and time. Observational research is inductive and iterative, 15 and its greatest strength is its open-endedness. Because the researcher does not work in a controlled environment or with a standardized checklist or questionnaire, he or she is able to capture any data that do not fit into a priori categories. In such a setting, the research question serves as a guide, not a mandate, and it leaves room to address unexpected occurrences. Although the 3 Cs template provides an excellent entrée for the use of field observations, there are other observational alternatives. Structured observations may be better indicated for inquiry exploring systematically the nature and metrics of phenomena, with integration across multiple observers for the purpose of statistical analysis. Another option is video recording, which yields multiple data sources, such as verbal, paraverbal, and nonverbal responses, and requires specific levels of skills and techniques for analysis. 50 The 3 Cs approach to unstructured field observations can be used when observation is the primary research method 46 or in tandem with another research method, such as qualitative interviews. 51 In the Qatar recruitment study, the observations were conducted more with the intent of being supplemental, but ultimately served as the primary source for a specific publication. 46 The template, which can be submitted for institutional review board approval, provides a straightforward mechanism for recording events and behaviors in almost any project involving human participants. As mixed methods gain increasing popularity in health services research, 52 unstructured field observations can play an important role in contextualizing other types of qualitative and quantitative data, resulting in more rigorous research designs and data collection. --- Context The circumstances (both material and theoretical) under which observations are being conducted, as well as any historical, sociocultural, political, and other information that may directly (or indirectly) influence data collection How do participants respond to the presence of an observer? (Are they excited, anxious, skeptical, wary, etc?) What historical or current events may influence this response?

Most primary care researchers lack a practical approach for including field observations in their studies, even though observations can offer important qualitative insights and provide a mechanism for documenting behaviors, events, and unexpected occurrences. We present an overview of unstructured field observations as a qualitative research method for analyzing material surroundings and social interactions. We then detail a practical approach to collecting and recording observational data through a "3 Cs" template of content, context, and concepts. To demonstrate how this method works in practice, we provide an example of a completed template and discuss the analytical approach used during a study on informed consent for research participation in the primary care setting of Qatar.

Predictors of problematic smartphone use among young adult lesbian, gay and bisexual individuals during the COVID-19 pandemic: a four-year follow-up study Mei-Feng Huang 1, Yu-Ping Chang 2, Wen-Jiun Chou 3,4* and Cheng-Fang Yen 5,6* --- Background Smartphones have become an indispensable tool in modern life. People use smartphones to connect with others, get messages, have fun, learn, and go about their daily lives. However, smartphones have the ability to provide quick fun and close social interaction, which has led to an increasing dependence on smartphones. Individuals who have problematic smartphone use (PSU) experience compulsive smartphone use, tolerance to smartphone use, withdrawal symptoms if smartphones are unavailable, and functional impairment due to PSU [1]. A metaanalysis found that the global pooled prevalence estimate of PSU was 26.99% for PSU [2]. Another meta-analysis on 27 published studies demonstrated that PSU was associated with mental health problems (e.g., anxiety, sleep problems, and depression), physical problems (e.g., musculoskeletal problems), sedative lifestyles, and accidents [3]. Brand et al. proposed a Person-Affect-Cognition-Execution (I-PACE) model to illustrate the development and maintenance of problematic digital devices use [4]. The I-PACE model hypothesizes that problematic digital devices use is the consequence of interactions between predisposing factors (e.g., neurobiological and psychological constitutions), moderators (e.g., coping styles and Internet-related cognitive biases), and mediators (e.g., affective and cognitive responses to situational triggers in combination with reduced executive functioning); conditioning processes may strengthen these associations within an addiction process [4,5]. Tan (2023) also proposed a stimulusorganismresponse-cognitiveadaptivenormative model to examine the drivers of habitual smartphone behavior and PSU [6]. Problematic digital devices use dramatically worsened during the coronavirus disease 2019 (COVID-19) pandemic [2]. People discontinue prepandemic socializing and recreational activities due to the lockdown or restriction of living areas in the COVID-19 pandemic; smartphones become the convenient tool for accessing information, entertainment and interaction with others. A systematic review and meta-analysis on 94 published studies revealed that the prevalence rate of PSU during the COVID-19 pandemic was 30.7%; in the lockdown periods, prevalence of problematic gaming and social media use were higher compared to non-lockdown periods [7]. Schoolchildren with a high level of problematic digital devices use had greater fear of COVID-19 [8] and psychological distress [9,10] compared with those with a low level of problematic digital devices use, especial during COVID-19 school suspension [11]. Therefore, PSU is an emerging health issue during the COVID-19 pandemic that needs to be explored in depth. Lesbian, gay, and bisexual (LGB) individuals are one of the populations at risk for PSU. LGB individuals experience public discrimination and prejudice due to their sexual orientation [12]. LGB individuals may conceal their sexual orientation and restrict their social interaction to avoid sexual stigma and bullying. Compared to the real world, the online world offers them a safer and more private place for obtaining entertainment and social interaction. Social media and dating applications on smartphone also provide the LGB individuals with a quick way to meet LGB friends and find sexual partners [13,14]. However, a study in the United States found that problematic social media use among LGB young adults is associated with depression and low social support [15]. Two studies in Taiwan have also demonstrated that PSU among LGB individuals was associated with depressive and anxiety symptoms [16,17]. A literature review evidenced that problematic Internet use is associated with health issues among the youth minority population [18]. The findings of previous studies highlight the significance of prevention and intervention of PSU in young adult LGB individuals. LGB individuals were disproportionately affected by the COVID-19 pandemic in terms of psychological wellbeing compared with heterosexual individuals [19][20][21]. Disconnection with LGB communities can further worsen LGB individuals' mental health. LGB individuals may rely more on smartphones to obtain social connection and entertainment and thus have a higher risk of PSU compared with heterosexual individuals. Furthermore, PSU can compromise LGB individuals' ability to cope with the predicaments happened during the pandemic. Examining factors that can predict PSU in LGB individuals during the COVID-19 pandemic may provide a reference for developing intervention strategies to reduce the PSU risk. According ecological system theory [22], there may be individual and environmental factors that increase the risk of PSU in LGB individuals. Regarding demographic characteristics, LGB individuals who are male and bisexual have greater PSU than those who are women and gay or lesbian, respectively [16]. Regarding environmental factors, sexual stigma was cross-sectionally associated with PSU severity in LGB individuals [16,17]. However, no prospective study examined the prepandemic individual and environmental predictors of PSU in LGB individuals during the COVID-19 pandemic. In the Taiwanese Study of Sexual Stigma (T-SSS, in 2018 and 2019), the data of PSU, multiple types of sexual stigma (i.e., perceived sexual stigma from family members, sexual orientation microaggression, and internalized sexual stigma), self-identity confusion (i.e., disturbed identity, unconsolidated identity, and lack of identity), emotional problems (i.e., depressive and anxiety symptoms), and perceived family support in 1,000 LGB individuals living in Taiwan were collected [23][24][25][26][27][28][29]. Whether these prepandemic individual and environmental factors can predict the level of PSU in LGB individuals during the COVID-19 pandemic warrants study. This 4-year follow-up study was conducted to investigate the predictive effects of prepandemic individual factors (e.g., demographics, sexual and gender identities, self-identity confusion, anxiety, depression, and PSU) and environmental factors (e.g., sexual stigma and perceived family support) on PSU during the COVID-19 pandemic in young adult LGB individuals. Based on the results of previous cross-sectional studies [16,17]., we hypothesized that greater prepandemic sexual stigma was associated with greater PSU during the pandemic. Moreover, Internet provides young adults who have self-identity confusion with an environment to explore their personal values, beliefs, and goals [30,31]; therefore, we hypothesized that greater prepandemic self-identity confusion was associated with greater PSU during the pandemic. A prospective study found the predictive effect of depression on PSU in Chinese adolescents [32]. A meta-analysis study also demonstrated a positive correlation between PSU and anxiety symptoms [33]. Therefore, we hypothesized that greater prepandemic depressive and anxiety symptoms were associated with greater PSU during the pandemic. Moreover, a prospective study demonstrated the predictive effect of low family support on problematic Internet use in Taiwanese adolescents [34]; therefore, we hypothesized that lower prepandemic family support was associated with greater PSU during the COVID-19 pandemic. --- Methods --- Participants and procedure Both the criteria and methodology used in the T-SSS for recruiting participants at baseline have been described in previous studies [23][24][25][26][27][28][29]. In brief, a cohort of 1,000 individuals (500 men and 500 women) at baseline were recruited through online advertisements on social media platforms-including Facebook, Twitter, and LINE-and a bulletin board system from August 2018 to June 2019. The inclusion criteria were identifying as an LGB individual, being 20 to 30 years of age, and living in Taiwan. The exclusion criterion was having any form of impaired cognition that might interfere with the ability to complete a questionnaire. Four years later (between August 2022 and June 2023), the same 1,000 individuals were contacted by text message and were invited to participate in a follow-up study. Those who responded to this message and agreed to participate received a blank consent form, a research questionnaire, and instructions for how to complete the questionnaire; they were also allowed to contact the research assistant for help if they had any problem understanding the questionnaire. Written informed consent was obtained from all participants. A total of three text messages were sent to the individuals to invite them to participate in the follow-up study, with a 1-month interval between each pair of messages. Those who responded to none of these messages were considered to have been lost to follow-up. This study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital (KMUHIRB-F(I)-20210219). --- Measures The outcome variable was PSU at follow-up. The predicting variables at baseline included prepandemic PSU, demographics, three types of sexual stigma, self-identity confusion, depression, anxiety, and perceived family support. --- Smartphone addiction inventory (SPAI) The 26-item SPAI [1] was used to assess the participants' self-reported severity of PSU in the three months prior to the baseline and follow-up assessments. The participants rated each item on a 4-point scale ranging from 1 (totally disagree) to 4 (totally agree), with a total score ranging from 26 to 104. A higher total score indicated a higher level of PSU. The Cronbach's <unk> coefficient of the SPAI in the present study was 0.96. --- Demographic characteristics We collected data on participants' sex, age, educational level (high school or lower vs. college or higher), sexual orientation (lesbian or gay vs. bisexual), and gender orientation (transgender or not). --- Homosexuality-related stigma scale (HRSS) The 12-item HRSS was used to measure the levels of perceived sexual stigma from family members [35]. Each item was rated on a 4-point Likert scale with endpoints ranging from 1 (strongly disagree) to 4 (strongly agree). A higher total score indicated that the participant perceived a higher level of sexual stigma from family members [356]. The Cronbach's <unk> coefficient for this scale was 0.93 in this study. --- Measure of internalized sexual stigma for lesbians and gay men (TC-MISS-LG) The 17-item traditional Chinese version [29] of the TC-MISS-LG [36] was used to assess each participant's sexuality, identity, and level of social discomfort. Each item was rated on a 5-point Likert scale with endpoints ranging from 1 (strongly disagree) to 5 (strongly agree). A higher score indicated a greater level of internalized sexual stigma. The Cronbach's <unk> coefficient for this scale was 0.76 in this study. --- Sexual orientation microaggression inventory (SOMI) The 19-item traditional Chinese version [25] of the SOMI [37] was used to assess microaggression in the dimensions of anti-gay attitudes and expressions, denial of homosexuality, and societal disapproval over 6 months among LGB individuals. Each item was rated on a 5-point Likert scale with endpoints ranging from 1 (not at all) to 5 (almost every day). A total higher score indicated a higher level of microaggression. The Cronbach's <unk> coefficient for this scale was 0.90 in the present study. --- Self-concept and identity measure (SCIM) The traditional Chinese version [31,38] of the 27-item SCIM was used to assess the level of current self-identity confusion [39,40]. The SCIM assesses three dimensions of self-identity confusion, including disturbed identity, unconsolidated identity, and lack of identity. Items are rated on a 7-point rating scale ranging from 1 (strongly disagree) to 7 (strongly agree). A higher total score indicates a higher tendency for self-identity confusion. Cronbach's alpha of the SCIM was 0.79 in the present study. --- Center for epidemiologic studies depression scale (CES-D) The 20-item Mandarin Chinese version [41] of the CES-D [42] was used to assess the frequency of depressive symptoms within the preceding month. Each item was rated on a 4-point Likert scale with endpoints ranging from 1 (rarely or none of the time) to 4 (most or all of the time). A higher total score indicated more severe depression. The Cronbach's <unk> coefficient for this scale was 0.91 in this study. --- State-trait anxiety inventory (STAI) The Mandarin Chinese version [43] of the STAI [44] consists of 20 items embedded in a single factor of anxiety. All the STAI items are assessed on a 4-point Likert scale, where a score of 1 indicates almost never and a score of 4 almost always. A higher STAI score indicates higher levels of anxiety [44]. The Cronbach's <unk> coefficient for this scale was 0.87 in this study. --- Adaptability, partnership, growth, affection, and resolve (APGAR) index The Chinese version [45] of the APGAR Index [46] was used to assess the participants' perceived support from their families. Each item was rated on a 4-point Likerttype scale with endpoints ranging from 1 (never) to 4 (always). The total scores for the Family APGAR Index range from 5 to 20, with higher total scores indicating higher levels of perceived support from family. In our study, the Cronbach's <unk> values for the Family APGAR Index were 0.94. --- Data analysis All statistical analyses were conducted using SPSS (version 24.0; SPSS, Chicago, IL, USA). We employed descriptive statistics to summarize and analyze the participants' data. The distributions of continuous variables were further tested for skewness and kurtosis to assess their level of departure from a normal distribution, and the results (i.e., absolute values of <unk> 3 for skewness and <unk> 10 for kurtosis) did not reveal any severe deviation [47]. This study detected the associations of prepandemic demographic characteristics, sexual stigma, self-identity confusion, depression, anxiety, perceived family support, and PSU with PSU at follow-up in two stages. In the first stage, the bivariate linear regression analysis involved the entry of only one independent variable at baseline for each time to detect their associations with PSU at followup. In the second stage, factors that were significantly associated with PSU in the bivariate linear regression analysis models were further included into a multivariate linear regression model to identify their associations with PSU at follow-up. Regarding the level of collinearity, the values of variance inflation factor ranged between 1.241 and 3.286; the values of tolerance ranged between 0.304 and 0.806; all values of eigenvalue were larger than 0.01; and the value of condition index was 29.862. The results indicated no problem of collinearity [48]. In order to understand what prepandemic factors were predictive of PSU severity at follow-up in the absence of the influence of prepandemic PSU, we used two regression analyses (one including and another not including prepandemic PSU as an independent variable) to identify the predictors of PSU at follow-up. Stepwise linear regression was performed to select most significant factors in the model. A p value of <unk> 0.05 was regarded as significant. --- Results In total, 673 (67.3%) participants responded to the invitation and agreed participating the follow-up study, 167 (16.7%) responded to the invitation but refused participating in the follow-up study, and 160 (16.0%) did not respond to the invitation. No differences in gender (<unk> 2 = 0.005, p =.946), age (t = 1.890, p =.059), sexual orientation (<unk> 2 = 2.087, p =.149), and gender orientation (<unk> 2 = 15.767, p <unk>.001) were found between those received and did not receive the follow-up survey; however, those who did not receive the follow-up survey were more likely to have an education level of high school or below (<unk> 2 = 15.767, p <unk>.001). Table 1 shows participants' demographics, sexual stigma, self-identity confusion, depressive and anxiety symptoms, perceived family function, and PSU. The gender distribution was relative similar (50.1% were men and 49.9% were women). The participants had a mean age of 24.8 years (standard deviation [SD] = 2.9 years) at baseline, and most of them had a college degree or higher (n = 618; 91.8%). More than half of the participants (n = 373; 55.4%) were gay or lesbian, and 19 participants (2.8%) identified themselves as transgender. The mean perceived sexual stigma from family members on the HRSS was 26.8 (SD = 6.3); the mean internalized sexual stigma on the MISS was 35.6 (SD = 11.5); and the mean sexual orientation microaggression on the SOMI was 42.3 (SD = 11.3). The mean level of disturbed identity on the SCIM was 37.5 (SD = 9.7), unconsolidated identity was 29.4 (SD = 8.7), and lack of identity was 19.2 (SD = 7.5). The mean score of depressive symptoms on the CES-D was 18.9 (SD = 11.3). The mean score of anxiety symptoms on the STAI was 41.2 (SD = 12.7). The mean perceived family function on the APGAR Index was 13.6 (SD = 3.6). The mean PSU on the SPAI was 61.9 (SD = 14.6) at baseline and 58.9 (SD = 17.4) at follow-up. The severity of PSU significantly decreased four years later (paired t = 5.351, p <unk>.001). Table 2 presents the results of bivariate linear regression analysis examining the individual associations of demographic characteristics, sexual stigma, self-identity confusion, depressive and anxiety symptoms, perceived family function, and PSU at baseline with PSU at followup. Three types of sexual stigma, three types of self-identity confusion, depressive and anxiety symptoms, and PSU at baseline were significantly associated PSU at follow-up. Lower perceived family function at baseline was significantly associated PSU at follow-up. Factors that were significantly associated with PSU were further included in a stepwise linear regression model (Table 3). Before the incorporation of PSU at baseline into the analysis model, the results of the model revealed that high levels depressive symptoms (p <unk>.001), disturbed identity (p <unk>.001), and perceived sexual stigma from family members (p =.025) at baseline were significantly associated with PSU at follow-up. After the incorporation of PSU at baseline into the analysis model, the results of the model revealed that high levels PSU (p <unk>.001) and depressive symptoms (p =.002) at baseline were significantly associated with PSU at follow-up. --- Discussion The present study found that the severity of PSU at follow-up significantly decreased compared with the prepandemic severity among LGB individuals. This finding is inconsistent with the findings of other studies examining the changes of digital devices use [2,49]. For example, a 6-month follow-up study that the severity of problematic Internet use among children and adolescents significantly increased during the COVID-19 pandemic [49]. The present study examined the changes of the severity of PSU among young adults during a 4-year period; the participants may have significant growth in mental maturity and increase the ability to self-control smartphone use. However, there was a proportion of LGB individuals having a high severity of PSU measured by SPAI. For example, the total score of the SPAI ranged from 26 to 104, and 15.3% of the participants reported a total score of 78 (three-quarters of the total score) or higher on the SPAI. Moreover, the severity of PSU before the pandemic significantly predicted that of PSU after 4 years, indicating that LGB individuals with high prepandemic PSU tend to maintain their pattern of smartphone use during the COVID-19 pandemic. The result shows the need for early prevention of PSU among LGB individuals. Our findings demonstrated that high levels depressive symptoms, disturbed identity, and perceived sexual stigma from family members at baseline predicted PSU at follow-up. LGB individuals have a higher risk of depressive disorders than do heterosexual individuals [50]. Both sexual minority stress [51] and intraminority LGB community stress [52,53] result in high depressive symptoms in LGB individuals. LGB individuals with depressive symptoms may overuse smartphones to obtain entertainment and online social support for relieving negative emotion. Depression may also compromise LGB individuals' ability to control their smartphone use. Therefore, depressive symptoms can predict the severity of PSU. Because PSU can worsen the individuals' sleep rhythm, physical health, and the ability to cope with stress in the real world, PSU and depression may form a vicious cycle. This study highlights the necessity of early detection and intervention of depressive symptoms for preventing PSU. This study demonstrated that prepandemic high disturbed identity predicted PSU four years later among LGB individuals. The individuals with a maturated selfidentity have clear goals in life, is willing to make commitments, and make efforts to complete developmental tasks [54], whereas the individuals with disturbed identity tend to acquire the thoughts, feelings, and beliefs of others in adulthood [55]. Smartphones provide the individuals with disturbed identity the opportunities to assess the thoughts, feelings, and beliefs of others in the online world; therefore, the individuals with disturbed identity have a high demand for smartphones use and develop PSU gradually. Our findings revealed that high prepandemic perceived sexual stigma from family members predicted PSU at follow-up among LGB individuals. Family is the basic and primitive microsystem in which the individuals are nourished [22]. LGB individuals who perceive sexual stigma from family members have the difficulty in recognizing their sexual orientation and conceal it; they may use smartphones to explore the world outside family, connect with LGB communities, and take a break from family discord. However, overuse of smartphones can lead to PSU. To the best of our knowledge, this study is the first to explore the predictive effects of prepandemic individual and environmental factors on PSU among LGB individuals during the COVID-19 pandemic. Although the severity of PSU was lower at the 4-year follow-up than at baseline, PSU remains a key health concern among LGB individuals and thus warrants attention. The factors significantly associated with PSU should be considered when designing intervention strategies for these individuals. Furthermore, health-care providers should design programs aimed at enabling LGB individuals to develop a mature self-identity and reduce depression. Interventions for the reduction of family and public prejudice toward LGB individuals are also needed. --- Limitations Our study has some limitations. First, because we collected data from a single source, our findings may be subject to shared-method variance. Moreover, because of the self-report measures used in the present study, our findings could be biased by social desirability. Second, our participants were interested in the follow-up survey; thus, our results may not be generalizable to all LGB individuals. Third, several factors were not evaluated at baseline, such as the types of smartphone activities (e.g., Internet gaming, social media, and dating) and impulsivity; therefore, the predictive effects of these factors on PSU remain undetermined. --- Conclusions In this study, higher levels of prepandemic PSU, disturbed identity, depressive symptoms, and perceived sexual stigma from family members were associated with a higher level of PSU at follow-up among LGB individuals during the COVID-19 pandemic. The prevention and cure of PSU are crucial. Health-care providers should design programs to help LGB individuals develop a mature self-identity. Further efforts to modify the family and public attitude and prejudice toward transgender LGB individuals are warranted. Furthermore, mental health problems, such as depressive symptoms, among LGB individuals should be explored as the predictors of PSU. --- Data Availability The data will be available upon reasonable request to the corresponding authors. --- Authors' contributions The work was conceived and planned by MFH and CFY. CFY carried out the analyses. MFH, YPC, WJC and CFY drafted the paper. The authors read and approved the final manuscript. --- Declarations --- Competing interests The authors declare no competing interests. --- Ethics approval and consent to participate The institutional review boards of Kaohsiung Medical University Hospital (approval number: KMUHIRB-F(I)-20210219) approved this study. Informed consent was obtained from all participants prior to the assessment. This questionnaire-survey study did not apply any experiments on humans or the use of human tissue samples. This paper conforms to the Declaration of Helsinki and Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. --- Consent for publication Not applicable. --- Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Background This 4-year follow-up study was conducted to evaluate the predictive effects of prepandemic individual and environmental factors on problematic smartphone use (PSU) among young adult lesbian, gay, and bisexual (LGB) individuals during the COVID-19 pandemic. Methods Data on prepandemic PSU, demographics, sexual stigma (e.g., perceived sexual stigma from family members, internalized sexual stigma, and sexual microaggression), self-identity confusion (e.g., disturbed identity, unconsolidated identity, and lack of identity), anxiety, depression, and family support were collected from 1,000 LGB individuals between August 2018 and June 2019. The participants' PSU was surveyed again after 4 years (between August 2022 and June 2023). The associations of prepandemic individual and environmental factors with PSU at follow-up were analyzed through linear regression.In total, 673 (67.3%) participants completed the follow-up assessment. The severity of PSU significantly decreased after 4 years (p = .001). Before the incorporation of PSU at baseline into the analysis model, the results of the model revealed that high levels depressive symptoms (p < .001), disturbed identity (p < .001), and perceived sexual stigma from family members (p = .025) at baseline were significantly associated with PSU at follow-up. After the incorporation of PSU at baseline into the analysis model, the results of the model revealed that high levels PSU (p < .001) and depressive symptoms (p = .002) at baseline were significantly associated with PSU at follow-up.LGB individuals should be designed considering the predictors identified in our study.

Introduction The shocking statistic that Native Indians use hospital facilities two to two and one half times more than the Canadian national population aver age (Indian Affairs, 1980:20) might suggest that this segment of our society does not share the popular dissatisfaction with Western institutionalized medicine. How wrong such a conclusion would be! Despite Health and Welfare Canada's annual expenditure of approximately one hundred million dollars (Indian Affairs, 1980:119) for the delivery of Indian health care services, Indians continue to reach beyond the professional sector of the medical System for their health care needs; they seek additional therapeutic procedures from other re sources-those of the so-called folk and popular sectors of the health care System. The Canadian medical System can be thought of as a tripartite paradigm composed of the profes sional, the folk and the popular sectors. The professional sector consists of that group of highlytrained experts who administer biomedicine to attack the organic cause ofillness. Indians augment the services of these medical personnel with re sources of the folk sector, most notably the vestiges of traditional Indian beliefs and practices that still exist. These aspects of Native culture attracted CULTURE IV (2), 1984 much attention from early anthropologists and medical personnel who tended to view healing rituals as manipulative conjuring rites that would be superseded by Western medicine, once science had triumphed over superstition. Yet, the efficacy of indigenous folk healers has ensured their survival (e.g., Kew and Kew, 1981). The folk sector of the medical System offers services of other folk healing traditions. Practitioners of everythingfrom Chinese herbalism to iridology are available for consulta tion. However, the most widely used component of the medical System is the popular sector and its selftreatment therapy, including; herbal remedies, pop ular home remedies and commercial patent reme dies. Indian people, like everyone else, rely substantially on their own knowledge of healing procedures to combat illness. Each sector of the medical System attempts to understand and treat illness and therefore develops what Kleinman (1975a;1975b;1978) has called an "explanatory model." The explanatory model addresses five major questions : 1) the etiology or cause of the sickness; 2) the time and mode of the onset of the symptoms; 3) the pathophysiology or nature of the sickness; 4) the expected course and prognosis; and 5) the treatment. Different sectors of the medical System possess distinct explanatory models which may co-exist, complément or compete with one another. As explanatory models are held by practitioners and patients in ail health care sectors, and are socially and culturally determined, they can be formally elicited and used to examine the differential use of health care services among any segment of the population. Through a case study of an elderly Indian man's responses to illness, this paper examines his dif ferential use of medical resources, some of the factors determining his sélection, and the major conflicts that arose in his transactions with the various sectors of the medical System.1 The data used were compiled from interviews and correspondence with the informant over an eight-year period. Emphasis was placed on eliciting the informant's Personal explanation about his illness and how he chose and evaluated treatment. The data are organized chronologically, starting before the informant entered the sick rôle. Unfortunately, it has not been possible to interview the practitioners whom the patient consulted. --- Okanagan-Colville Concepts of Illness HR, born in 1900, is an Okanagan (Interior Salish) man living on an Indian Reserve in southcentral British Columbia. Amongst the Okanagan-Colville, HR is recognized as being culturally conservative and is regarded as one of the last tradition-bearers of his people. HR, like other elderly Okanagan-Colville people, holds to a belief System that classifies illness by etiology-illnesses are caused either naturally or unnaturally. Those disorders believed to be natural or secular include headaches, colds, cuts, toothaches, sores, injuries from accidents and the like. Unnaturally-induced illnesses include: injuries inflicted by animate beings other than humans; diffuse internai illnesses including fever and contagious diseases; afflictions of the mind caused by shamanistic action; spirit-illness; and magical poisoning or sorcery (Ray, 1932:202;Cline, 1938:162-165). Of course, the manifestations of an unnaturally-induced illness can be identical to those of natural origin. Consequently, an ailment is readily designated as unnatural when therapy does not produce immédiate relief. In such cases, a shaman or "Indian doctor" is consulted for a diagnosis. Some Okanagan-Colville people believe that many diseases are brought about by the presence of "small worms" crawling underneath the skin. Other diseases are said to be caused by "poisons," some of which can make the blood weak, requiring the help of tonies to "change the blood." In Okanagan-Colville society today it appears the greatest number of unnatural or prolonged illnesses are thought to be caused by magical "poisoning" or sorcery, known by the Native term ptax. The majority of those still practising ptax are women of post-menopausal âge whose knowledge has been learned in earlier life from mothers and grandmothers. The power of p't'ax allows those women who practice it to alter and manipulate events both for good and evil ends. For example, pta:>ç can be used as a "love charm" to bring two people together, as a "good luck medicine" to make a horse win a race, or as an amulet for protection from ghosts. On the other hand,ptüx is also used to cast spells and to curse people to hâve bad luck or even to die. Ptax is seldom discussed but readily acknowledged amongst the Okanagan-Colville-its mention brings on nervous laughter. It was in the context of this cultural framework that HR perceived his illness and selected treatment. --- Case Study From 1916 to 1942, HR developed numerous minor leg ulcers which he attributed to contact with certain poisonous plants growing in hay fields. HR treated himself with poultices of Canada mint (Mentha arvensis), a natural source of menthol which is both germicidal and antiseptie (Osol and Pratt, 1973:670). Initially, HR did not consider himself to be ill, nor did he recognize any symptoms of disease-ulcers were bothersome but part of his life as a cowboy. When the sores persisted, he augmented his herbal treatments with a home remedy suggested by a local White rancher. But by 1942 the ulcers were affecting his ability to work, so HR treated them with the latex of a poisonous plant, milkweed (Asclepias speciosa), in the belief that a poison was needed to kill the poison in his leg. The treatment proved unsuccessful. In 1942, for the first time in his life, HR sought assistance outside the popular sector of the medical System. A White physician was the most readily available medical resource. The ulcers responded to the physician's treatment and remained healed for nearly thirty years. When the problem sores recurred, however, HR then interpreted the White doctor's therapeutic procedure from thirty years earlier as a causative factor. In describing the physician's earlier treatment, HR recalls that an unidentified "black liquid" was injected into his left hip and then into his right hip. He remembers feeling the liquid travel a short distance down his left leg and stop precisely at the spot where his leg later became ulcerated again. The poison that HR believed to be responsible for the ulcerated leg was now attributed to the black liquid, not to the plants in the hayfields. HR was no longer working actively in the hayfields, and therefore was no longer in contact with the plants which he had believed to be the cause of his leg ulcers. The unfamiliar therapy administered thirty years earlier was outside of the community-held belief System and continued to be misunderstood by HR. Early in 1970, HR sought further medical advice from friends and from other people recognized as knowledgeable in Interior Salish Indian communities, and consulted a known shaman. Because HR classified the ulcers as a naturallycaused disorder, this "Indian doctor" was not requested to make a diagnosis. He did, however, prescribe a herbal treatment, but HR did not receive any symptomatic relief. Later in 1970, HR's concern for effective control of the symptoms made him resort, once again, to a White physician. Pills were prescribed. If the pills would cure him, HR could not understand why he was not given a larger bottle so that he could take them more frequently than was recommended. When the results of this drug therapy were not immediately évident, HR's misunderstanding led to non-compliance and he switched to another prac titioner in another sector of the medical System. HR's quest for a cure led him to a local Chinese cafe-owner. Bearing a note to inform any Cantonesespeaking person of his destination, HR travelled to Vancouver to seek aid from a Chinese practitioner who used biomedicine in his practice. The diagnosis offered by the Chinese healer was in accordance with HR's current theory of the etiology-the healer set about to "kill the poisons that were working together in his body." As well as supplying HR with a larger quantity of pills than the White doctor was willing to dispense, the Chinese healer administered four injections-one in each shoulder and one in each hip. Apparently HR perceived some symptological response, for he complied with the pre scription. Early in 1971, HR once again sought advice from a local White doctor. Varicose veins were now impeding his movement. HR recalls that this condi tion was explained to his satisfaction as being the resuit of "a few poisons working together." The familiar diagnosis and the physical relief that HR obtained as a resuit of the therapy restored HR's faith in Western medicine. At the time HR returned to a White physician two years later, when the problem recurred, he was told he had a circulation problem and was admitted to a hospital for surgery; skin transplants were undertaken from his hip to the ulcers that had developed in his lower leg. This procedure failed, although the swelling subsided once the varicose veins were stripped. With additional treatment, the leg healed eventually and the ulcers did not reappear for several years. The early 1970s marked a profound transition in HR's life. His wife died, leaving him to perform tasks, such as cooking and cleaning, that Indian men normally regard as "women's work." He could no longer manage his ranch, and difficulties he was having concerning his land resulted in litigation with his own relatives. He left his ranch and moved to another house on the other side of the river, but not long after this he was the victim of a freak accident-a rock and mud slide narrowly missed destroying the new house he had moved into. His emotional state at this time was summarized in a letter he wrote to me: Now days Igot no friends, no body would not talk to me, no body likes me, everybody against me, everybody tried to make me mad in some ways... In this dépressive state, HR revised his explanatory model when the ulcers recurred in 1975. The social discord in his life led to a suspicion that his misfortune stemmed from ill will; he believed he was the victim of ptdx (sorcery). He no longer The Quest For A Cure / 23 classified the illness as naturally caused. HR's response to this shift in cognition was then con sistent with his perception of the etiology-he consulted a shaman. HR's visit to the shaman was arranged through an Indian friend who accompanied him on the appointed night. This particular shaman was an Upper Skagit (Coast Salish) man from Washington State, but his curing techniques followed Interior Salish custom, where such healers work for four consecutive nights, beginning their rituals after the last ray of light has faded from the evening sky. In the course of the ritual, which lasted about one hour, the shaman used a basin of water, a fancy handkerchief that he asked HR to sit on, and a blanket that he hung over the back of HR's chair. The techniques used were like those of other shamans whom HR had witnessed, starting with the healer singing his personal "doctoring" songs. Occasionally, the shaman danced. Others who were présent in the room, including HR himself, joined in the singing, but the starting and stopping of the song was determined by the shaman alone. During the pauses he blew forcefully on the back of HR's neck and down the centre of his chest. (This was done, HR explained, to "blow the sickness out of the body".) Then, standing in front of the patient, the shaman moved his hands, one on either side of HR's body but not touching it, down from the head to the feet. Ail the while the shaman made short patting motions to "chase the sickness from the body." Later they had a meal together and the session ended. This was ail in keeping with HR's expecta tions of what the diagnosis would entail. HR describes the first night of doctoring as an "inquest," where the shaman "just looks around and examines the evidence." The resolution of the first night of doctoring came later that night in the shaman's dream. On the second night, the shaman presented his understanding of HR's illness and its treatmenthis personal explanatory model-a culturally-relevant, highly-personal account suggesting social discord in HR's community. He confirmed HR's suspicion that sorcery was the cause of his troubles. Instrumental in this "dirty work," the shaman said, were two women and a male accomplice whose work had begun "quite some time ago." The shaman explained that a man had entered HR's house while he was out and eut small pièces of material from the seams of his clothes. He stored them in a gray sock that had a red mark on it. These bits of cloth were said to hâve been given to the two women who then added some bloodied and pus-coated hair that had been clipped from a cow's injured leg and left on the ground. They took this evil bundle away with them. The shaman informed HR that the women went to a graveyard, where they pushed a long stick into a freshly-covered grave so that the end of the stick reached the gap between the top of the coffin and the side of the grave. They wiggled the stick back and forth to make a passageway into the grave. Next, they inserted HR's specially-treated sock down into the bottom of this passageway. While the women were refilling the hole, they spoke to the sock, saying, "HR will hâve sores like the injured cow. He will suffer like the cow and no doctor, Indian or White, will be able to cure him." Then, the women walked a short distance away, turned back to face the grave, and laughed. This was the cause of HR's ulcerated leg, the shaman announced. The prognosis given as part of this shaman's explanatory model absolved him of any obligation to cure. Yet he spent the third and fourth nights performing ritual therapy. This involved drawing the sickness down the full length of HR's body and capturing it in his hands. The strength of the sickness caused the healer to flail his arms about wildly, as if he were holding a live object. Then he thrust his hands into the basin of water to weaken the sickness and drown it. Now, with his hands still clasped, the shaman walked to an open door and, as if releasing a captured bird, he threw the sickness into the night air. The shaman explained to HR that it would be at least ten days before he noticed any change in his condition, for he was not simply working on HR's symptoms but also trying to withdraw the curse. Nevertheless, no change occurred. In recalling this period of his illness épisode, HR remarks that the therapy was ineffective because the shaman feared the strength of the women's HR does not acknowledge any failing in the healer's abilities. Continuing to seek help, HR resorted to the Shaker Church. [Shakerism is an indigenous West ern North American religious movement whose followers believe that their religion was provided by God to Indian people in their time of need in the 1880s (Barnett 1957).] Shaker ceremonies combine both Christian and aboriginal Indian beliefs. When filled with God's spirit (a state similar to being in contact with one's guardian spirit power), Shakers are believed to prophesy, heal the sick and exorcise evil. HR's friendship with a Shaker follower, in addition to his personal faith in God and also in Indian spirit power, prompted him to attend a Shaker group curing ceremony. Despite dramatic rituals involving the ringing of bells, lighting of candies and "speaking in tongues," the therapy did little to address HR's individual concerns, except in symbolic terms. As the healing ceremony was a social event, it did serve to reunite HR with a community of friends. HR's symptoms worsened over the next few months. But now, having gained an understanding of what he perceived to be the etiology of the illness, he again resorted to treatment from the popular sector of the medical System. He confidently employed his own extensive knowledge of herbal therapy, washing the afflicted area with a décoction made from the tops of young tamarack (Larix occidentalis) trees, and drinking a similar décoction. The wash acted as a counter-irritant because of the natural presence of turpentine (Osol andPratt, 1973:1250). Despite the appropriateness of his home remedy, HR's condition worsened. Again, his recourse was to a professional physician. Following a skin transplant in the summer of 1976, HR had only minor outbreaks of leg ulcers which he treated himself using both Native and commercial medicines. HR's évaluation of the physician's therapy in 1976 was that it was partially effective; the physical symptoms of the disorder were controlled, though the illness was not cured. In a letter to me dated December 1976, HR stated: HR gathering tops of young tamarack (Larix occidentalis). The Doctor says I'm okay but I do not know. The Doctor he doesn't know what was ron [sic] with me but I do. Only thing is nothing I can do about it but only an Indian doctor he help. Western medicine did not alleviate the psychosocial aspects of the illness. HRunderwent little behavior al change and continued to seek out shamans to treat other somatic complaints. His anxiety was still présent. The minor récurrences of his leg ulcers were a reminder that he was still a victim of pTax. In the spring of 1981 the ulcers became large and painful. Once again HR selected a professional physician and once again he was admitted to a hospital for skin transplant surgery. The operation failed; by this time the circulation in his lower leg had become so constricted that a graft was not possible. He was discharged, well supplied with ointments and bandages for self-treatment at home. HR's simultaneous use of healers from the professional, the folk and the popular sectors of the medical System began in mid-1981. At this time his faith in ail healers was tempered by a practical skepticism. Although he believed that professional physicians were most efficient at controlling his symptoms, his most recent hospital treatment had not been successful. And he believed that the shamans whom he had hired did not possess guardian spirit powers strong enough to remove the curse which he perceived to be the cause of his problems. Also, he realized that there had been little change in his symptoms following the Indian healer's therapy. HR continued applying the ointments prescribed by a local White physician. He supplemented this therapy with attendance at a curing ceremony held at a relative's home, with a Blackfoot shaman from Montana as host. The presence and the supportive rôle of HR's relatives during the ceremony helped to meet his needs, with the resuit that he felt better. His évaluation of the shaman's capabilities for reducing his physical complaints was equally posi tive. While he believes that the herbal medicine given to him by this Blackfoot doctor may hâve eventually healed his leg, HR was not left with what he considered enough medicine to complété the treatment. By the spring of 1982, HR's leg ulcers had again deteriorated. At this time a White woman, belonging to a local church, was organizing groups to visit a faith healer in the Philippines. Stories of the healer's abilities in expelling cancerous organs and in other feats resulting in miraculous cures enticed HR to sign up for the trip. However, when the date of departure was delayed several times and a number of HR's friends asked him to reconsider his The Quest For A Cure / 25 decision, he concurred that there were too many indications that he should not go. Continuing his quest for a cure, HR again turned to self-treatment. He prepared poultices from the peeled and pounded roots of chocolaté tips (Lomatium dissectum), a plant considered so poisonous by the Okanagan-Colville that it was used in former times to make an infusion for killing fish in streams. But it is also well known as an effective medicine for treating cuts, sores, boils and dandruff (Turner, Bouchard and Kennedy, 1980:66-68). The herbal treatments were supplemented with visits to a local White physician. When HR was hospitalized for six weeks in June 1982 he certainly did not expect a culturally-meaningful explanation of his disorder from the Western medical personnel. He also did not expect that their therapy would cure his illness, although he did believe that they could bring his physical symptoms under control. He commented on this in an interview with me following his departure from the hospital: --- Whiteman doctors can help it, but they can 't cure it. Itfools them. It looks like it gets better, but it doesn 't, it cornes out again. Whiteman doctors can't cure it because they don't understand what caused it. The explanatory models held by the hospital personnel and HR were quite dissimilar. In fact, the hospital routine was antithetical to HR's personal view of appropriate therapy. HR was assigned to a large hospital room with five other men, none of whom was Indian. The elderly patient occupying the next bed had difficulties with his bowels, which meant that the nursing staff often had to attend to his needs. HR was very much aware of the smell of feces permeating his room. But the window could not be opened because of another patient's condi tion, so there was no fresh air. In addition, the filled bedpans were apparently left in the adjoining washroom, where HR believed the hospital staff "were drying the dirt so that they could examine it." Weeks passed and the smell continued. Finally, HR left a note on one of the filled bedpans in the washroom, requesting the staff to remove the cause of the "stink." What was perceived as merely unpleasant to other patients was very offensive indeed to HR, whose culture views ail human wastes as potential "poisons" that are severely detrimental too healing and, worse still, can be used in ptax. There was no acknowledgement of HR's request. HR's dépressive feelings began to hâve the biological concomitants of lack of appetite and insomnia. He spent longer and longer periods of time withdrawn in the smoking lounge (where the Windows could be opened) until past curfew, when he had to be ordered to go to bed. But from HR's cultural perspective, the time when he was ordered to bed was precisely the time when healing vis believed to occur-late at night. (In Okanagan-Colville society, many medicines are taken in secrecy, either late at night or early in the morning, and often in association with steambathing to cleanse and ritually purify the body.) Because ofthe medical staffs lack ofknowledge of HR's explanatory model, they were unable to respond with culturally-appropriate therapy. For example, the leg ulcers were washed daily with a diluted solution of hydrogen peroxide, even though HR urged them to use it full strength. His request stemmed from the culturally-conditioned assumption that only the visual evidence of the peroxide's foaming action would indicate that the sickness was being drawn from his leg. This concretized symbolism is a common feature of shamanic performances. When HR asked to be moved to an empty room that he knew existed on the same floor of the hospital, his request was neither discussed nor acted upon. He then took his complaint to the head nurse. Three days later, when no decision had been made, HR announced that he was leaving the hospital. His doctor was called in to speak with him but HR explained that he would "soon be dead" if he stayed because "the hospital is too dirty for an old Indian." Unfortunately, the physician-patient communica tion came too late. Disappointed and distrustful, HR went home, vowing never to return again. While in the hospital, HR had received an Indian visitor who told him about another Blackfoot shaman in Montana. Stories of this shaman's ability to cure Indian people's disorders that Western medicine had "given up on" only served to reinforce HR's decision to leave the hospital. In September 1982, HR travelled to Browning, Montana for the Blackfoot shaman's healing ceremony. Certain advance préparations were necessary -HR was instructed to purchase spécial foods to be eaten at the midnight feast. Also, he was told to bring certain coloured ribbons that the healer later fashioned into long sashes. HR maintained that he was a victim ofptax, so this Blackfoot shaman was not asked to give a diagnosis. Instead, he proceeded directly to the mechanics of exorcism, brushing his hands up and down HR's body while singing his personal doctoring songs. Those présent joined in the singing and danced where they stood. The ceremony took place in a mountain cabin and the actual curing ritual was performed in total darkness, according to Blackfoot custom. Before the second and final night of the curing ceremony, HR was asked for some gas money so that the shaman could visit a particular mountain where he would gather a certain root for HR's ulcerated leg. The root, unfamiliar to HR, was to be pounded and used as a poultice. But after HR had applied the root for several days, his leg became swollen and painful and he discontinued the procedure. A ribbon sash made for him by the Blackfoot shaman was hung above the headboard of HR's bed, where, he was told, he should "look every day to see if the ribbons hâve moved." The shaman told him his guardian spirit power would, through the power imbued in the sash, protect HRfrom evilly-disposed spirit powers of other shamans. (This sash is still hanging above HR's pillow, with a rosary and a picture of Christ.) The Blackfoot shaman cautioned HR that three additional visits would be required before he would be cured, yet HR is hésitant to résumé the sessions and risk the chance of adverse reactions to the unidentified herbal poultices. Not long after returning from Montana, HR pursued still further treatment. In October, 1982, on the recommendation of a friend, an anthropologist with whom he had collaborated, HR visited a Chinese herbalist in Vancouver. The herbalist, an HR being treated by Chinese herbalist, November 1982. old man of 91, had been trained in China by master herbalists. In his basement he kept a large supply of various préparations he had imported from his homeland. This herbalist showed HR "before and after" photographs of leg ulcers he had treated successfully. Visits to this herbalist were to be made at twoday intervals for six weeks. The regular treatment procedure included washing the ulcers and applying a powder, followed by a fresh herbal poultice. The Chinese herbalist recommended abstinence from certain foods believed to be critical to the healing process, and dietary modifications that HR strictly complied with while undergoing the therapy. A friendship developed between HR and the herbal ist and his wife, with HR often delaying his departure to sit and smoke with the couple. But three weeks after the treatment had begun, the elderly Chinese herbalist died. A full six weeks' supply of poultices had been prepared, so the widow was able to continue the treatments, as she had assisted her husband on numerous occasions. Feeling better and perceiving some alleviation of his symptoms, HR returned home to the interior of British Columbia in No vember, 1982. Upon his return, HR began a program of selftreatment. Friends and relatives suggested several different home remedies, including mega-vitamin therapy, drinking tonies made from Aloe vera and parsley, and applying Aloe vera to the ulcers and covering them with parsley poultices. HR also drank a total of 31 quarts of a tamarack décoction he made at home. The tamarack décoction, HR explains, "heals the sores from the inside of the body." During this period, HR also hired a Colville shaman from Washington State whose healing abilities had only recently been acknowledged. This man's doctoring was similar to other shamans HR had called upon. This shaman left HR with two large handkerchiefs which were tied together to form a loop that was placed over his head and under one arm. These he was to continue wearing for one month's time, after which he was to put them underneath his bed pillow. HR followed this shaman's advice, despite being dissatisfied with the explanation given by him for the cause of the problem-he suggested that HR was the victim of the pth% of two people, one of whom he could see and the other of whom he could not see. The shaman informed HR that the hair from the tail of an ornamental China horse,that HR has hanging on his wall, had been used by an old woman to put a curse on him. The old woman was someone HR knew, lived on the west side of the river not too far away, and would now be The Quest For A Cure / 27 ill herself. Although HR was very much aware that an old woman living nearby would probably like to make him suffer (because of a land dispute in which he was the victor), he dismissed the explanation on the basis that this particular woman was healthy. Also, the shaman was unable to "see" the incident that had been described by the Upper Skagit shaman in 1975. The Col ville shaman encouraged HR to continue with whatever home remedies he was using and cautioned him not to expect any immédiate change in his condition for about eight weeks. Ten weeks passed before HR asked the Colville shaman to return. Again, he performed his exorcism ceremony. He told HR to call him in another four weeks, but when the ulcerated leg worsened, HR decided to seek other courses of treatment. HR's self-treatment program now consisted of using large quantities of hydrogen peroxide which he poured liberally over the sores. This he purchased at the local pharmacy. HR's purchases of hydrogen peroxide grew so large that the pharmacist began to ask what he was using it for. When HR showed the pharmacist his badly-ulcerated leg, he suggested immediately that HR consult the town's new physician, a young female general practitioner. In the following few weeks HR's condition deteriorated rapidly. He was found bedridden, in pain and without medicine of any description when his anthropologist friend subsequently visited him. He was too ill to visit the new female physician and too distrustful to go to a hospital, so the doctor went to see him at his home. She initiated therapy involving white sugar poultices (see Knutson et al., 1981, for a report on this treatment of chronic leg ulcers), a form of treatment similar to the poultice treatments which HR himself had been using. And most importantly, she instructed Indian health workers, women from HR's own community and speakers of HR's native language, how to apply these poultices. Other women from his community were hired to visit HR daily to préparé him a nutritious meal and tidy his house. By the summer of 1983, the results of this therapy were apparent, but still, HR felt the treatment was incomplète. In an interview I had with him at that time concerning the White doctor's treatments, HR stated: Even ifshegets them better-if they get better Ican begood enough to go someplace. I willgo to an Indian doctor. Even ifshegets them better. Sheget them better, but they'll corne back again. In the fall of 1983, however, HR's established routine was interrupted when the town doctor announced she was leaving, his Indian health worker went on vacation, and the substitute worker applied the poultices in a slightly different manner. HR was very distraught; he augmented the sugar treatments with his own home remedies. He obtained the services of yet another shaman, a Coast Salish man from the Lummi tribe in Washington State. As a resuit of this man's therapy, HR again felt better. But the condition of his leg ulcers did not improve. --- Discussion At a time when Western medical personnel are lamenting their inability to treat certain Native disorders, there is a need to evaluate how and why Indian people like HR are using such diverse health care alternatives. Indeed, the most striking feature of HR's "illness trajectory" is its pluralism. He has selected and combined therapy from the popular, the folk and the professional sectors of the Canadian medical System, switching from one resource to another. HR's pathway through medical options should not be viewed as naive, spéculative, or irrational, but as a pragmatic search for symptom relief in a milieu in which he maintains control. HR's case study demonstrates how his explanatory model underwent alteration as he received new information, negotiated with friends, and utilized alternative resources. But does this illness trajec tory, as illustrated in figure 1, suggest a pattern of health care use? What was the process of his interaction with the various health care sectors? What were the major déterminants in HR's healthseeking behaviour? A number of models hâve been proposed to examine the "processual" approach to illness-the stages typically passed through by a person who believes himself or herself

This paper examines the use of health care alterna tives by a culturally-conservative Canadian Indian man who suffers from statis ulcers, a disorder he has attributed to both natural and unnatural origins. A case study is presented that utilizes a conceptual framework from medical anthropology. This case study illustrâtes the cultural and social déterminants of health seeking; the perceived etiology, degree ofimpairment, and the efficacy of the treatment and its cultural relevance are ail found to be significant factors in the sélection of therapeutic re sources. Ritualized performances of Native shamans are found to be an intégral part of the healing process.En présentant cette étude de cas, nous suivons un cadre conceptuel propre à l'anthropologie médicale. Sont mis à jour les déterminants culturels et sociaux intervenant dans la cure; l'étiologie retenue, le degré de maladie, l'efficacité du traitement choisi et les interférences culturelles appa raissent comme autant de facteurs significatifs lors du processus de sélection des thérapies. Les rituels des cha mans indiens font aussi partie de ce processus de guérison.

ability to treat certain Native disorders, there is a need to evaluate how and why Indian people like HR are using such diverse health care alternatives. Indeed, the most striking feature of HR's "illness trajectory" is its pluralism. He has selected and combined therapy from the popular, the folk and the professional sectors of the Canadian medical System, switching from one resource to another. HR's pathway through medical options should not be viewed as naive, spéculative, or irrational, but as a pragmatic search for symptom relief in a milieu in which he maintains control. HR's case study demonstrates how his explanatory model underwent alteration as he received new information, negotiated with friends, and utilized alternative resources. But does this illness trajec tory, as illustrated in figure 1, suggest a pattern of health care use? What was the process of his interaction with the various health care sectors? What were the major déterminants in HR's healthseeking behaviour? A number of models hâve been proposed to examine the "processual" approach to illness-the stages typically passed through by a person who believes himself or herself to be ill. One such model formulated by Chrisman (1977) distinguishes five éléments of the health-seeking process: symptom définition, illness-related shifts in behaviour, lay consultation and referral, treatment actions, and adhérence. Chrisman (1977:353) suggests that "the ability to relate conceptual éléments with [illness] chronology should increase our capacity to explicitly link sociocultural factors such as health beliefs... to behaviors during sickness." HR's case study shows that the presence of symptoms may not necessarily be sufficient to precipitate help-seeking beyond the popular sector of the medical System. It appears that HR has been af<unk>licted with the same circulatory disorder, varying in its severity, since 1916. Primary treatment was, and continued to be throughout his illness épisode, became a critical déterminant in selecting alternative resources from that point onwards. The question of "why" he was af<unk>licted was now as important as "how" the illness was manifested. The disorder was now given meaning within his Native community, allowing HR to call upon those who shared his health beliefs and practices for suggestions about treatment and recommendations of healers. Thus, he first chose practitioners who shared his primary etiological assumption. But when the problem persisted, his explanatory model fluctuated enough to allow his etiological explanation to become multi-factorial, encouraging an amplified use of resources. He consulted a greater number of people for advice and acquired a greater number of options from which to choose appropriate therapy. Once HR perceived his illness to be of unnatural origin, the nature of his transactions with practitioners altered. He no longer expected a cure to be administered by any one healer. He also moved from the rôle of "patient" to the rôle of "patron," selecting spécifie services from spécifie practition ers. He understood his problem to be an "Indian illness" for which the efficacy of biomedicine was limited. While recognizing that Western medicine is proficient at temporary symptom control but neglects the psychosocial needs of a patient, HR considered the therapeutic procedures of shamans to be especially capable of fulfilling the latter. HR's pattern of using professional physicians and Western medical resources suggests that the severity of the problem was his major déterminant in selecting that sector of the medical System. His adhérence to the prescribed régime was greatest when there existed some commonality between the explanatory models held by HR and his practition ers, when there was immédiate visible improvement in his condition, and when there was personal support from members of his community. However, once the symptoms were under control, HR sought amelioration of what he perceived to be the cause of the illness by attending Indian healing ceremonies. His expectations regarding this form of therapy and the practitioner's beliefs and function were always satisfied as his illness was given meaning within his own cultural framework. Inasmuch as these cere monies were social events focusing on HR, he generally experienced significant behavioural changes following the treatment-quite simply, he felt better. He was, therefore, reluctant to deny the efficacy of the herbal remedies prescribed by the shamans. Instead, he rationalized the lack of symptom response while using such remedies by stating that he was not given enough of them, that he broke a cultural taboo while using them, or that Native cures require a long period of time before the results can be evaluated. This contrasts sharply with his view of Western drug therapy for which he has higher expectations, including quicker results. On one occasion, when HR perceived no immédiate symptom response while undergoing drug therapy, he stopped treatment and switched to an alternative medical resource. On another occasion he stopped treatment because he perceived it to be antithetical to his expectations of appropriate therapy and, consequently, potentially harmful. The greatest conflicts arose when there were major discrepancies between the explanatory models of HR and the practitioners. Foremost was the discrepancy in expectations of HR and of the medical personnel during his 1982 hospital stay. Apparently no attempt was made by the attending nurses or HR's physician to gain any understanding of the cultural influences governing HR's response to his illness. Yet both HR and the hospital personnel viewed the nature of his disorder, its cause, and its treatment and eventual outcome, quite differently. Thus, a problem in communica tion ultimately resulted in poor patient care, and HR left the hospital with a renewed need for the services of a shaman. --- Conclusion HR is a fiercely independent individual but he is not unique amongst Indian people in his multiple use of treatment alternatives. It is generally agreed that the demands of Canada's Indian population are not being met by Western medicine. A recent éditorial (1982) by John Last, an Ottawa physician and editor of the Canadian Journal of Public Health, laments, "we must confess our failures to diagnose and treat the disorders that exist here [among Indian people]." Yet despite this, only a few medical projects in Canada hâve attempted to integrate the available medical resources into a syncretic System offering more culturally-relevant health care to Indian people. Here in British Columbia, mental health research by Drs. Jilek, Jilek-Aall and Todd demonstrates the effectiveness of some Native therapeutic procedures (for example, in the treatment of anomie dépréssion) and has prompted these medical people to co-operate closely with the few remaining Coast Salish Indian healers (Jilek, 1980:130;cf. Jilek and Todd, 1974; and Jilek and Jilek-Aall, 1978). Unfortunately, such a sensitive concern with Indian patients' psychosocial needs, as is taken by Jilek and associâtes, has not been fortheoming from the larger body of general medical practitioners. Indeed, the Jilek approach has been criticized as being "anti-western, pro-nativist and anti-positivist" (Hippler, 1980:192). The extent to which Indian people like HR are using alternative medical resources, and the know ledge and logic that are operative in their sélection, await further clarification. It is perhaps timely for Western medical per sonnel and anthropologists alike to address the issue of discrepancy between patients and practitioners and to focus upon the actual transactions between the two. Only when we gain an understanding of the cultural dimensions of health care within our pluralistic medical System will people like HR be assured of better health care. In the meantime, HR's quest for a cure con tinues. --- NOTES 1. The data included in this paper were elicited while conducting Salishan ethnographie research under the auspices of the British Columbia Indian Language Project, funded by the Social Sciences and Humanities Research Council of Canada. I wish to acknowledge the assistance of Randy Bouchard, Heather Clarke, Dr. Wolfgang Jilek, Dr. Louise Jilek-Aall, Dr. Arthur Klein man, Dr. Michael M'Gonigle, Dr. Peter Stephenson, Dr. Wayne Suttles, Dr. Norman Todd, Dr. Nancy Turner, and Dr. Wendy Wickwire for their valuable criticisms of earlier drafts of this paper. Spécial thanks go to HR, whose quest continues. I alone, of course, am responsible for remaining deficiencies. --- Apple's (1960) thesis that it is the social disruption caused by symptoms that triggers the health-seeking process. In HR's case, cultural factors had a distinct influence on symptom définition, the first step of the health-seeking process. Initially, HR assigned his problem to the category of natural illness. The most appropriate therapeutic procedure for a disorder of this etiology is self-medication, which HR tried. When this approach was not successful, HR resorted to a local White physician, not to a shaman whose therapy for a natural illness would be too closely aligned to self-treatment. The options that HR considered appropriate to combat the "poisons" which he believed had invaded his body, consisted most often of biomedicine available through White physicians, Chinese healers and a nearby hospital. These were chosen after consulting with friends (some of whom had comparable problems in the past) to détermine the severity of the disorder and the necessary action to be taken. The health-seeking process was most dynamic following HR's re-assessment of his symptoms to the category of unnatural illness. Again, cultural and social factors were instrumental in his percep tion of the disorder, which recurred during a time when he felt surrounded by social discord. Pîax (sorcery) was suspected immediately. HR sought confirmation not from his community of friends but from a shaman whose responsibility it was to identify the person(s) whose evil intentions were causing HR's misfortune. HR's explanatory model then reflected a rétrospective assessment of the years' events as causal agents for his illness. His ré évaluation of the symptoms and their meaning The Quest For A Cure / 29

This paper examines the use of health care alterna tives by a culturally-conservative Canadian Indian man who suffers from statis ulcers, a disorder he has attributed to both natural and unnatural origins. A case study is presented that utilizes a conceptual framework from medical anthropology. This case study illustrâtes the cultural and social déterminants of health seeking; the perceived etiology, degree ofimpairment, and the efficacy of the treatment and its cultural relevance are ail found to be significant factors in the sélection of therapeutic re sources. Ritualized performances of Native shamans are found to be an intégral part of the healing process.En présentant cette étude de cas, nous suivons un cadre conceptuel propre à l'anthropologie médicale. Sont mis à jour les déterminants culturels et sociaux intervenant dans la cure; l'étiologie retenue, le degré de maladie, l'efficacité du traitement choisi et les interférences culturelles appa raissent comme autant de facteurs significatifs lors du processus de sélection des thérapies. Les rituels des cha mans indiens font aussi partie de ce processus de guérison.

INTRODUCTION The increase in life expectancy and the aging of the population in Brazil and in the world have occurred in an accelerated fashion, which makes it important to reflect on the conditions under which the elderly are aging. Aging is a complex, intersectoral process that encompasses discussions on health, education, social assistance, social security, and housing, including the social and economic aspects that affect the quality of life of the aging population. In order that this intersectorality is put into practice, there must be some communication among public policies, and these should aim at the same interests: social protection and coping with social inequalities (1). Brazil has been trying to prepare itself to respond to the growing demands of its aging population, through actions and policies implemented with the purpose of guaranteeing the elderly rights to health and quality of life in the aging process. At present, the predominance of chronic non-degenerative diseases is highlighted, and they are a major cause of death, representing a great health problem (2). In this context, promoting aging with quality of life becomes a challenge that requires strategies capable of modifying behavior towards healthy aging. In the city of Rio de Janeiro, a program of activities called Academia Carioca da Sa<unk>de (Health Academy of Rio de Janeiro) has been conducted since 2009 with the objective of promoting health and preventing diseases and their complications, ensuring access for the population to practices promoting physical, mental, and social well-being. This program has proved to be a significant resource for the construction of a more active society with a healthier lifestyle and a better quality of life. The phenomenon of quality of life in the health area has been conceptualized in two different ways: quality of life in a generic sense and in relation to health. The first presents a wider meaning, used in sociological studies, which are concerned with the cultural and social mobilization that this phenomenon imposes on society, and does not refer to it only as a process of illness or injury (3). Regarding the different approaches, concepts, and evaluations, the scientific literature presents the discussion of several authors about this phenomenon (4). Quality of life is approached through discussions of representations, according to a polysemic notion, which includes different senses, meanings, knowledge, and experience. Considering this theoretical reference, the relation of the context in which people are inserted to their standard of behavior is worth mentioning. Thus, the understanding of the construction of the knowledge of the elderly included in the program Academia Carioca on living with quality, and the links of this construction with the practices adopted for the benefit of health, constitute an instrument capable of subsidizing interventions that promote active and successful aging, meeting the problem outlined. --- OBJECTIVE To identify the social representations of older adults regarding quality of life, and to analyze the practices of care adopted to promote it. --- METHOD --- Ethical aspects The study was approved by the Research Ethics Committee of the Nursing School Anna Nery/Health Care Institute S<unk>o Francisco de Assis. The participants signed the Free and Informed Consent Term and had their identities kept confidential by using identification codes: male (M) or female (F), followed by the interview number and age. --- Theoretical-methodological framework The procedural approach of the Theory of Social Representations (TRS) was applied to explore the content that delineates the knowledge of the elderly about the object in question. Social representation is responsible for guiding behaviors and reshaping the elements of the environment in which they occur (5). Therefore, the application of the TRS in the present study aims to identify how the representation of quality of life affects the attitudes of the elderly, and in what way they can influence this in the context of the group of Academia Carioca. --- Study type This was a qualitative, exploratory, and descriptive study, with the goal of understanding the meanings attributed by the people to the researched object in its production context. --- Methodological procedures --- Field Research was conducted from December 2015 to June 2016, in an Academia Carioca de Sa<unk>de at the Family Health Strategy Clinic located in the city of Rio de Janeiro. It is a space for health promotion, development of community, and multisectoral action, centered on the practice of physical activities. It has the participation of healthcare and physical education professionals who work as partners. This choice was made because it is a place of socialization among the elderly, where they establish conversations and acquire knowledge about health issues; therefore it is useful for the construction of social representations. --- Data source The source was a qualitative, non-probabilistic, random, and convenience sample. Inclusion criteria were: people aged 60 years and over; of both genders; with verbal communication and cognition preserved, who were participants for at least 30 days. Exclusion criteria were: irregular frequency of activities and/or absence on days of data collection. Participants were 30 elderly people, 25 women and five men, defined based on a preliminary analysis of the data and configuration of the empirical framework of the research. The sufficiency to answer the guiding questions and to reach the objectives determined the cessation of participant gathering (6). --- Data collection and organization First, an exploratory observation and familiarization with the environment for later gathering of participants was performed for two months. In the second phase, an instrument was applied Social representations of older adults regarding quality of life Ferreira MCG, et al. individually to collect the social and demographic data of the elderly. In the third phase, in-depth interviews (average duration of 60 minutes) were conducted with open questions about the daily life and activities performed by the participants in the spaces of the Academia de Sa<unk>de, maintaining the participants' privacy. The testimonies were transcribed using Alceste 2012 software standards. --- Data analysis Data analysis occurred after the interview processing in the Alceste software, which divided the text into elementary context units (uces), reducing the roots of the main words, and distributing the lexicons, separating them into frequencies and associations according to content similarity or non-similarity. The software captures groupings and performs lexical classifications, and the researcher is responsible for tracing the meanings of the formed classes, naming them and performing an analysis of the content based on the distribution and association of the lexicons. The class to be analyzed deals with the multidimensionality of the concept of quality of life, which emerged in lexical Class 1. Triangulation occurred in order to contemplate the analysis of the descending hierarchical classification, ascending hierarchical classification, and content of the units of the lexical class. This triangulation of data analysis applied in the studies of social representations is primordial for the deepening and validity of the results (7). --- RESULTS Of the 30 participants in the study, 25 were female (83.3%) and five were male (16.7%). The age ranged from 60 to 82 years, according to the inclusion criterion in the sample, with a predominance of individuals between 60 and 65 years old (53.3%). About 16 elderly people (53.3%) were married, 20% were divorced, 20% widowed, and 2% single. Most of them (90%) lived with relatives, and only three (10%) lived alone. Of the total number of participants, 18 (60%) were retired and 12 (40%) were still working. As for the data processed by the Alceste software, the corpus submitted to the software analysis consisted of 30 units of the initial context (uci), which were distributed into 1,391 units of elementary context (uce), formed by 4,447 words or distinct words. The software reduced the vocabulary to its roots, originating 859 analyzable words (nouns, adjectives, verbs) and 266 supplementary forms (articles, pronouns). The processing resulted in the construction of five classes, which addressed: quality of life; daily life and leisure; and therapy applied in primary care and its consequences for health. To meet the objective of this article, we will specifically analyze Class 1 on the quality of life. The multidimensionality of the concept of quality of life emerged in lexical Class 1 of the production of data processed by the Alceste software. This class consists of 436 uces, the equivalent of 46% of the uces used, and about 146 words analyzed, as shown in Figure 1. In the descending hierarchical classification, most of the incidental terms refer to the psychosocial, subjective, and multidimensional aspects of the researched object (quality of life). Class 1 originated from the questions related to what the elderly think about their quality of life, through which means they construct their conceptions, knowledge, and how they act in relation to these factors; that is, it refers to the image they construct around the subject in question. In the lexical analysis ascending hierarchical classification (AHC), shown in Figure 2, two groups stand out. The first bring together the words "money" and "want," which refer to the terms "conditions" and "I had," and both are connected to the words "I consider" and "happy." This, in turn, is linked to the second group of the terms "physical activity" and "should," which establishes relations with the words "food", "to eat", "good," and "health." These relationships express the social determinants of health and suggest the correlation with money and its importance for supportive care such as food. Meanings highlighted were those that relate quality of life to social determinants and in what way they affect health. The uces in this class involve economic, cultural, and social situations related to housing, food, education, income, and employment conditions, which are states that influence health. The social determinants of health are present in many points in this class, especially in terms of aspects that involve financial and social difficulties, which permeate the representations and allow the emergence of existing inequalities and inequities in the population studied. The associated lexicons and their meanings in the uces demonstrate a protagonism and a proactive movement of the elderly towards actions that take them to quality of life in the face of what they consider to be its promoter. This understanding can be identified in the verbs used in the construction of the thought expressed in the uces, such as "demand", "to do," and "to search." Social representations of older adults regarding quality of life Ferreira MCG, et al. In the ascending lexical analysis it is observed that the lexicons "to try" and "to search" link to the verb "to look for," which expresses the proximity and frequency with which they are used together. They are linked to the lexicons "socialize with" and "way," which may mean that the elderly seek to live together in a more harmonious way. This connection between the terms in this class emphasizes the importance of good judgment for the search of social life, using the term "fundamental." I've heard of quality of life. Quality of life is a person who tends to exercise, seeks to make friends, not to be alone, feeds him/herself well. Good food; do not overdrink. The person having some time for resting. (F1, 62 years) This proactive movement is focused on elements that constitute a representation of quality of life, based on the gregarious social question, good nutrition, rest, and physical activities. The latter integrate the immediate context of the participants, because the research was performed in an environment that is part of the daily life of the elderly. The word "food" and its complementary forms were among the main categories mentioned by the participants in the research, with the occurrence of the typologies of food and forms of food related to socialization ("eating out"), and to the economic and financial condition for acquiring what they consider "good food." The elderly emphasized the importance of food quality and healthy eating habits for maintaining health and achieving a satisfactory quality of life. According to the ascending analysis, the word "food" frequently appears related to the term "eating," and both relate to the word "power." They establish relationships with the terms "good" and "health" that allow us to infer that attitudes toward food and eating are valued as necessary for good health. The terms "good" and "health" are also related to the words "I should" and "physical activity," but not directly to "food." In this sense, physical activity is also interpreted as an attitude that contributes to health. The importance of being in harmony with their families, children, and grandchildren is evident in the uces. They also emphasize the relevance of living with friends, and the fact that they are inserted in, and feel belonging, to a social group. It's absurd. This way we cannot have a good life. Quality of life has to do with having health as well. Without health, the person is always sad. To live well at home in harmony. Our home first, our children, our family, our friendships. To be prepared for old age. (F17, 61 years) --- I think that a person who has quality of life can do what he/ she wants, walk, travel, eat well. Find friends at any time of the day without even worrying about how you will pay the bill. (F8, 60 years) The social, political, and economic dimensions in the uces were also striking when the participants were questioned about quality of life. In this sense, ideas and meanings related to money, purchasing power, food, product acquisition, housing, and leisure emerged as factors dependent on economic power. According to the ascending lexical analysis, the terms "money" and "I want" are related to the terms "I had" and "conditions," and both relate to the terms "consider" and "happy." These relationships demonstrate that money is seen as something that enables better conditions (housing, leisure, purchasing power, food), which are important for living well. Because dietary products are much more expensive, whole grains are also good for health, everything is very expensive. Then I would buy a lot of fruit so I could eat whole food. But The issue of purchasing power appears in the lexicon composing the uces through explanations of the ability of the elderly to acquire goods, products, and services. This constitutive financial element representative of quality of life is also evident in the conditions of access of these individuals to physical activities conducted by professionals, which represent costs. In the ascending lexical analysis, it is observed that the words "I should" and "physical activity" are related to the terms "money" and "I want," which shows the relation established by the elderly between economic power and the practice of physical activities. --- Because I would not be able to find another place. Perhaps, of course, I would have to pay; I would have to have money. And I don't have it. In a way, this helps me a lot. I like it. The teacher is cheerful, playful, says something to try to boost our self-esteem. (F20, 61 years) The service offered, and the environment of Academia Carioca integrates the representations of quality of life, serving as a means to fulfill the daily life of those who experience the transition from active/productive life as workers to retirement. --- The environment influences a lot; here we live well with other people, keep good sense. We can't stay still; you feel depressed if you stop. It's important to be on the move. If you have always moved, worked, when you retire you have to get something to do. You can't stop; you get sick. (M18, 68 years) And as I'm retired, I have to have the peace of mind that a retiree deserves, in a place where I can safely do my things, being healthy and able to have my leisure at ease. (M12, 70 years) In the construction of the image of the object quality of life, the elements that constitute the discourse of the reified universe of thought are conveyed by health professionals and the media. In this respect, they present a reified speech of accountability of the other represented by the expression "helps us to do." --- I hear on television, in that TV program called "Bem Estar"; I really enjoy watching it. Here we also listen to it, in the gym. There are also books that talk about quality of life. They say that we have to maintain a healthy life, eat varied dishes, practice physical activities and not stop the mind. (F4, 60 years) These uces represent a dimension based on the social image of active aging influenced by health professionals and the media. It is noted that participation in Academia Carioca favors reflection on the benefits of positive attitudes towards health, and regarding the importance of participating in social groups so that the individuals can live longer and with more quality. Many things are within the program of quality of life, important things, mainly for us, who are older. It is important that we have positive influences, and that's why quality of life is good. They say that the person has an even longer life, because with quality of life the person lives longer and well. (M18, 68 years) It helps in the way of thinking, living, moving, making friends. All this changes our lives and makes people have quality of life, be happy, not be sad. (F6, 82 years) --- DISCUSSION The concept of quality of life can be considered multidimensional according to a general or holistic approach. Studies deal with the conceptual aspects and participants' difficulties in discussing their approach (2)(3). Considering this multidimensionality, it is well known that the social determinants of health influence people's quality of life, especially the elderly's. The reference to the health social determinants by the participants in the uces demonstrates the anchoring process of social representation, so that the dimensions that permeate the representational field of these elderly people about quality of life affect them and are meaningful when they relate them to social determinants. Anchoring is the process of cognitive integration that assigns meaning to the object. The subject turns to what is familiar to him, setting the new idea within a preexisting thought (8). It is observed in the uces that the knowledge and ideas used to construct common sense about food and eating practices have aspects originating from the reified universe of knowledge. Thus, the participants incorporated information discussed in their group, and transformed this knowledge into meaningful knowledge for their daily life, thus justifying good eating practices in benefit of their quality of life. Knowledge about appropriate food practices and healthy foods is adjusted to a biomedical, therapeutic, prescriptive, and identity discourse, typical of the reified universe. At certain moments, this speech co-occurs with the recognition of the benefits of physical exercise for the participants' quality of life. The verbs "look for" and "seek" and their variations, "I look for", "searching for," and "I search" reflect actions that the participants accomplish or idealize to be appropriate for attaining quality of life. The noun "person" had an impact on the set of Class 1 lexicons, relating to the verbs cited. These associations reveal a proactive attitude towards quality of life, evidencing an initiative to practice physical activities and satisfaction in participating in social groups. The presence of these verbs reflects the symbolic dimension of representation, because it has to do with the dissemination of knowledge by the reified universe on the themes of physical activity and food, articulated to the promotion of quality of life, through discussions with the health professionals of Academia Carioca and information transmitted in the media, for example, on TV shows. These discussions disseminate a discourse of the accountability of the other, which generates a coherence between the thought of the elderly when they investigate and their action. In this sense, the verbs they use indicate that they take responsibility for the action, incorporating it into their daily lives, and understanding how beneficial it is to their health. The predominance of terms such as "doing exercises", "not standing still," and "doing activities" also demonstrates this awareness of the importance of practicing physical activities, and the denial of sedentary attitudes. The subjects construct ideas about the object and act according to the influences of the environment to which they belong which, in the case of this research, is evidenced by attitudes that reveal the importance of an active and healthy aging for a successful aging. The social representation allows us to evaluate how the social life intervenes in the construction of the subject, and how this subject affects the environment in which he lives. The incorporation into everyday life of knowledge provided by the social environment only occurs if it has meaning or affects the person in any way. In this sense, knowledge about the importance of physical activities and good food, disseminated in media such as television programs and during the activities at Academia Carioca contains messages that inspire the elderly, making them incorporate them into their daily lives. This demonstrates the importance of the affective dimension to the representational field. These data are congruent with other studies that indicate the practice of physical activities as an element that favors the psychophysical health and quality of life of individuals in the aging process (9). A cross-sectional study carried out in Finland, Poland, and Spain evaluated the determinants of the quality of life of elderly people, and showed that the practice of physical activity and the establishment of bonds through the insertion of elderly people into social groups are some of the factors that contribute to the increase in quality of life of the aging population (10). A study carried out in Brazil showed that the elderly did not isolate themselves at home; they sought active insertion into groups in the community, establishing new bonds and friendships (11). The search for social interaction was also perceived by the participants as a relevant point for quality of life; thus loneliness is translated as negative for well-being. Among the motivations for participation in social groups are the search for improvement in health and quality of life as well as an escape from loneliness, aspects that evidence the important role of social insertion of the elderly for psychic health (12). The importance of support from family and social relationships for a healthy aging is also highlighted (13). A study conducted in Turkey shows that high levels of social support and the practice of activities of daily living contribute to a better quality of life and lower rates of depression (14). These results reiterate how independence and social participation positively affect the quality of life of the elderly. On the other hand, fragile health and the incapacities imposed by aging tend to weaken the social interaction and community participation of the elderly, with risks of social isolation and limitations in lifestyle. Therefore, the insertion of these individuals into the free social groups offered by SUS brings benefits to the interaction, as well as favoring changes in lifestyle and maintenance of successful aging (15). The strong influence of the current economic conditions experienced by the Brazilian population is also observed in the uces. With the economic and financial crisis, metaphors and words that express dissatisfaction and emotional, social, and economic instability have become evident. As we age, the issue of social security becomes present in both the social and economic spheres. With the increase in longevity and better health conditions, the number of elderly people who will benefit from social security increases, which demands greater government spending. Studies on the financial condition of the elderly also relate purchasing power to quality of life, as it affects access to goods and services, for example, quality food and leisure (16)(17). In the results of this research, Class 1 lexicons and their uces show that money allows quality eating and practice of leisure activities that, in their turn, relate to aspects of social satisfaction. This affects attitudes regarding the compliance with advice from health professionals to purchase and consume foods that require better economic conditions. The need for money for the practice of physical activities as highlighted by the elderly has links with the overvaluation in the media and in society of the cult of the body. Awareness of the benefits of exercise has permeated the people's initiative and increased the search for gyms, which influences the demand of the market. In a study that discussed aspects of the deconstruction of the image of the elderly due to media influence, the dissemination of information that increasingly encourages the adoption of new habits of life and maintenance of a healthy body among the elderly was observed (18). The imagistic dimension of active aging is disseminated by the media and also by health professionals. The social image of active aging is brought by the media as synonymous with longevity and health and reaches the individual image, which contributes to the search for the Academia. The recognition of the relevance of the Academia Carioca by the research participants goes beyond the scope of physical health, and reaches the psychological sphere, due to the benefits to mental health. The results show the building of bonds in the context of the group, and the importance of being inserted into a space that intensifies social life and contact with other people that can help the participants to reach psychic wellbeing and quality of life. In a study carried out in a co-living group in the city of Mogeiro in the state of Para<unk>ba, the biopsychosocial contribution of the living spaces was reaffirmed, because after the insertion in the group, the individuals began to show more living satisfaction, and felt more socially valued. The group facilitates conviviality and social relations, as well as facilitating the expression of feelings (19). The actions carried out in the environment of the socialization group of the Program Academia Carioca de Sa<unk>de go Social representations of older adults regarding quality of life Ferreira MCG, et al. beyond the sphere of the practice of physical activities, becoming a mechanism of health promotion through necessary and important information for the subjects inserted into it. Thus, the reified universe of knowledge is present in the representations of participants when using technical knowledge about health care. This occurs through testimonies about healthy practices and the necessary care that they carry out with the purpose of preventing or mitigating the consequences of problems related to non-communicable chronic diseases. These are apprehended in the context of the actions carried out in the group. The diffusion of strategies and practices that stimulate healthy habits is an alternative allowing for an old age with fewer chronic diseases or functional impairments. However, the economic and social conditions of an individual also directly interfere with the quality of their food and the possibility of carrying out activities, factors that may make the promotion of healthy living habits an even more challenging task for health professionals. It should be noted that Academia Carioca was planned and structured to be a space for sharing knowledge and information about health care, a place of production of meanings and transformation of attitudes. The elderly who participate in it are aware of the importance of being inserted into groups in which they can socialize and carry out activities to benefit their quality of life. In a study that evaluated aspects of the health-related quality of life of the elderly in situations of social vulnerability, social adversity, lifestyle, health conditions, and functional aspects of social relations stood out as important markers for understanding this phenomenon (20). --- Limitations of the study Conducting this study in only one field, without possibilities to significantly increase the number of participants, did not allow an analysis by gender and by age group segments. --- Contributions for nursing and public policy Getting to know the knowledge and practices of the elderly in terms of quality of life allows nursing professionals to plan and implement health-promotion care that is more consistent with the context of the elderly, considering their lives and experiences. In addition, it reaffirms the importance and contribution of Academias de Sa<unk>de that are part of the Family Health Clinics, because they are spaces where gerontological nursing has a great opportunity to work, especially the Academia Carioca Program, for the development of better health conditions of the elderly, extending the conditions of multidisciplinary work in health care. --- FINAL CONSIDERATIONS The social representations of the elderly with regard to quality of life are built on the social determinants of health and on the discourse of social health production, which demonstrates a global understanding of the concept of quality of life. Such representations denote the articulation of knowledge and care practices, with valuation of physical activities and proactive attitudes towards aging. They also provide care for the prevention or mitigation of chronic noncommunicable diseases. The elderly have communicated their knowledge about what, in the light of their representations, they consider quality of life, and they are able to recognize it in everyday situations, through examples. However, images are identified in their speeches that describe actions and conditions that refer to an understanding of the ideal or imagined quality of life in a discourse about the future, but not the one currently lived by them. The characterization of practices promoting quality of life evidenced attitudes related to healthy eating habits, daily physical activities, social participation, life and interaction with friends in social groups, the accomplishment of daily tasks with independence and autonomy, support and family contact, the importance of sleep and rest, as well as leisure activities in the company of friends and relatives. These were the main actions of the elderly in favor of their quality of life. In view of the above, Academia Carioca is part of the daily life of the elderly, promotes their resocialization through dialogue, and allows the construction of knowledge and actions to incorporate healthy attitudes in their daily lives. It is, therefore, a pulsating space of circulation of knowledge on issues that matter to health, which contributes to social insertion and care for the elderly.

Objective: to identify the social representations of older adults regarding quality of life, and to analyze the care practices adopted to promote it. Method: qualitative, exploratory, descriptive research, applying the Theory of Social Representations. Thirty older people from a Health Academy of Rio de Janeiro participated in the study. The software Alceste was used, and lexical analysis of data was performed. Results: social representations of quality of life are based on the social determinants of health; they evidence knowledge and practices of care by valuing physical activities. The practices promoting quality of life comprise healthy eating habits, daily physical exercise, social participation, interaction and socialization, accomplishment of leisure activities and daily tasks with independence and autonomy, and support and family contact. Final considerations: the elderly have a global understanding of the concept of quality of life, coordinate knowledge built in daily life and knowledge coming from the technical-professional fi eld, which evidences the multidimensionality of the concept. Descriptors: Quality of Life; Aged; Motor Activity; Health Promotion; Nursing. RESUMO Objetivo: identifi car as representações sociais de idosos sobre qualidade de vida e analisar as práticas de cuidado por eles adotadas para promovê-la. Método: pesquisa qualitativa, exploratória, descritiva, com aplicação da Teoria das Representações Sociais. Participaram 30 idosos de uma Academia Carioca de Saúde. Utilizou-se o software Alceste e realizou-se análise lexical dos dados. Resultados: As representações sociais de qualidade de vida sustentam-se nos determinantes sociais de saúde, evidenciam saberes e práticas de cuidado, com valorização de atividades físicas. As práticas promotoras de qualidade de vida congregam hábitos alimentares saudáveis, exercícios físicos diários, participação social, convívio e interação, realização de atividades de lazer e tarefas cotidianas com independência e autonomia, apoio e contato familiar. Considerações fi nais: Os idosos têm uma compreensão global do conceito de qualidade de vida, articulam saberes construídos no cotidiano e advindos do campo técnico-profi ssional, o que evidencia a multidimensionalidade do conceito. Descritores: Qualidade de Vida; Idoso; Atividade Motora; Promoção da Saúde; Enfermagem. RESUMEN Objetivo: identifi car las representaciones sociales de ancianos sobre calidad de vida y analizar las prácticas de cuidado por ellos adoptadas para promoverlas. Método: investigación cualitativa, exploratoria, descriptiva, con aplicación de Teoría de las Representaciones Sociales. Participaron 30 ancianos de una Academia Carioca de Salud. Se utilizó software Alceste y se efectuó análisis lexical de datos. Resultados: las representaciones sociales de calidad de vida se sustentan en los determinantes sociales de salud, evidencian saberes y prácticas de cuidado, valorizando la actividad física. Las prácticas promotoras de calidad de vida incluyen hábitos alimentarios saludables, ejercitación física diaria, participación social, convivencia e interacción, realización de actividades recreativas y tareas cotidianas con independencia y autonomía; apoyo y contacto familiar. Consideraciones fi nales: los ancianos poseen una compresión global del concepto de calidad de vida, articulan saberes construidos en el día a día, devenidos del área técnico-profesional, evidenciando la multidimensionalidad del concepto.

Background While most research on common or rare disorders has in the past been conducted in European and East Asian populations, there is increasing interest in conducting such studies in sub-Saharan African populations. There is a growing recognition that health disparity research will be improved by including populations with genetically diverse background. This means we need to include more sub-Saharan Africans in genetic research for common and rare-disorders, especially as this region is known to harbor the greatest amount of genetic variation within our species [1]. Within the United States, minority recruitment for genetic/family studies has been recognized as crucial to understand genetic variation and to enable generalization of research findings. Pharmacogenomic studies, for example, investigate the efficacy of pharmacologic agents in populations with certain genetic characteristics [2][3][4]. Genetic research in African populations is needed to study the genetic predisposition of Africans to certain diseases. Currently there are projects under way to investigate proposed rheumatoid arthritis loci in African populations [5], to investigate susceptibility of Nigerians to risk loci for chronic kidney disease [6] or to study the effect of certain promoter variants for modulation of serologic autoimmunity to SLE [7]. The Yoruba ethnic group of Nigeria has been sampled in previous genetic studies. The Yorubas were part of the international HapMap studies that led to the sequencing of the Human Genome [8] and to create a map of human genome sequence variations [9]. Samples from the Yoruba ethnic group were also included to collectively identify >3 million common DNA variants, primarily SNPs [10]. SNP data and their linkage disequilibrium patterns provide basic information for genome-wide association studies in African or admixed populations [8,[10][11][12][13]. However, to our knowledge there are no genetic studies that use the Nigerian or Yoruba population as their primary target. Our study is in the process of identifying the genetic basis for keloid formation, a wound healing disorder, with the Yoruba ethnic group as the primary target population. Keloids are scars that grow over the margin of the original injury to the dermis (Figure 1). This often disfiguring, painful, and stigmatizing scar can occur after a minor cut or trauma to healthy skin. Keloids occur mostly in dark pigmented populations [14,15] including the Yoruba people of Nigeria. There is no good and efficient treatment for keloids [16,17]. It has been generally accepted that there is a genetic basis for keloids [18][19][20][21]. However, the mode of inheritance is still unsettled. Some studies suggest that the keloid inheritance pattern is autosomal dominant [21,22] while a study of 34 Yoruba family pedigrees (from Nigeria) supported autosomal recessive inheritance [23]. The Yoruba awareness of keloids predates the first publication on keloids in Western literature [24,25]. According to Omo-Dare, Yorubas for long have referred to keloids in their art and literature [26]. Indeed, one of the 140 chapters in the Ifa literary corpus is dedicated to keloids; with artifacts/sculptures and folklore reflecting keloids. The Yorubas, with their local customs of facial markings and earlobe perforations, recorded their familiarity with keloids 10 centuries before Alibert and Retz [27]. For modern-day genetic research, it is often necessary to recruit thousands of participants, especially for association studies. In the instance of case-control studies, only a single affected participant or a single unaffected control individual per family is recruited. Other strategies involve the recruitment of trios (father, mother, and affected offspring) or sib-pairs. Recruitment for those studies may be relatively straight-forward, although not trivial. For linkage studies, however, large families must be identified where the disorder of interest is transmitted. Challenges arise when many informative family members need to be recruited to give enough power to a pedigree for independent identification of a disease gene locus. Developing strategies to present the research to potential study participants in a manner that enhances voluntary participation of family members from all generations becomes essential, and can be a challenge. Suggestions to enhance recruitment for genetic/family studies with emphasis on minority recruitment in the US or recruitment in Africa have been addressed in the literature [28][29][30][31]. However, literature on recruitment strategies or recruitment experience for genetic/family studies in the Yoruba population is limited. This article documents our recruitment experience for a genetic linkage study in the Yoruba population in Nigeria and describes the challenges we faced as well as methods used to reduce these challenges. --- Setting: Nigeria Nigeria is a country in West Africa with an estimated population of 170,000,000 (census 2010), which is composed of 250 ethnic groups including Hausa and Fulani (29%), Yoruba (21%), Ibo (18%), Ijaw (10%), Kanuri (4%), Ibibio (3.5%), and Tiv (2.5%). Nigeria was a port for the slave trade and obtained its independence in 1960 from Great Britain. The official language is English but the three major local languages are Hausa, Yoruba, and Ibo, along with more than 200 dialects. The Yoruba homelands are in the western part of Nigeria, the Igbo reside in the east, while the Hausa and Fulani live in the north. --- Lessons learned This study was carried out with careful consideration of cultural issues in the host country. We followed recommendations to enhance recruitment in such low resource-limited settings [29,31]. Here we share our experience on several issues that impacted our recruitment for consideration in future studies. --- Methods --- Recruitment sites Participant recruitment was carried out in Ibadan (Oyo State) and Osogbo (Osun State) as well as in surrounding towns and villages. Ibadan is the capital of Oyo state with a population of 1.3 million, the third-largest metropolitan area in Nigeria after Lagos and Kano. Ibadan is located in south western Nigeria with the principal inhabitants being ethnic Yoruba. There are several smaller towns and villages around Ibadan where recruitment of study participants took place including Ogbomoso, Egbeda, Oyo, Iseyin and others. Osogbo, the second recruitment site for this study is the capital of Osun State with a population of approximately 160,000 (by 2006 census). Most of the inhabitants are Yorubas. Recruitment around Osogbo extended to surrounding towns, including Ile-Ife, Modakeke, Ikirun, Ilesa, Ede, and Ikire, among others. The population structure and culture in Oyo and Osun States are very similar. --- Study preparations and interaction between US and Nigerian research groups Potential study sites and plastic surgeons who were interested in a genetic study were identified through personal contact with a health care professional in Ibadan. The US researchers worked with the local researchers from the inception of this project. Ethics approval from an IRB in the US and ethics committees in Nigeria was obtained for the initial pilot study prior to the US researchers' first visit to Nigeria. During the first visit to Osogbo, which was paid for by institutional funds and by the University of Connecticut Health Center (UCHC) General Clinical Research Center (GCRC), the US scientists met with Nigerian plastic and general surgeons and chief medical directors of different hospitals as well as with provosts of colleges of health sciences to describe the goals and objectives of the research project. Discussions followed on the prevalence of keloids in their clinics and the surrounding communities, as well as on how to engage the community leaders and how to dispel fears and concerns related to research amongst the keloid patient population. Plans were developed on how to address cultural beliefs, and which methods would be most appropriate to promote the research study. Also discussed was the feasibility of hiring staff through local hospitals, research and ethics training for staff, plans for transporting samples from surrounding towns, data collection and the transfer of samples to the US. Additional activities included FWA registration of the Nigerian institution in Osogbo and human subjects training (CITI) for those investigators who did not already have that certification. As a team, we outlined a feasible roadmap for the pilot study with the goal of obtaining NIH funding (R01) for largescale recruitment. Communication between the Nigerian teams and the research group from the US involved emails and telephone contacts between regular visits to Nigeria. Consenting and privacy (HIPAA) documents were reconciled regularly to be acceptable to the ethics committees and IRB on both sides. As NIH funding was obtained, subcontracts between institutions were developed and an additional recruitment site in Ibadan was instituted. Investigators reviewed recruitment processes collaboratively, planned and implemented recruitment in surrounding towns and villages and updated staff training to the Nigerian research teams during regular visits throughout the recruitment period. While in Nigeria, the US team took part in recruitment efforts, meetings with community leaders, educational forums and media presentations. --- Study design The primary goal of this cohort study was to recruit families where keloids are inherited in an effort to identify the genetic basis for keloid scar formation. The focus of recruitment was on large multi-generation families. Keloids were assumed inherited when several members of a family were affected. Probands who did not report any additional affected family member were classified as sporadic or singular cases. Data discussed here were obtained between August 2005 and February 2013. Ethics approval was obtained from the Institutional Review Board of the University of Connecticut (IRB#03-007) and the ethics committees of the Ladoke Akintola University of Technology (LAUTECH), College of Health Sciences, Osogbo and the College of Medicine, University of Ibadan/University College Hospital (UI/UCH), Ibadan. All oversight committees had obtained Federal Wide Assurance (FWA). Recruitment of study participants was performed by trained recruitment teams under the supervision of plastic surgeons at LAUTECH and the (UI/ UCH). Recruitment was supported by NIH-funded subcontracts with both institutions. Each recruitment team consisted of trained nurses, research assistants and two surgeons to direct the recruitment efforts as co-investigators and to diagnose participants. Training of staff included human subjects training as well as training to differentiate between keloids and other wound healing disorders (e.g., hypertrophic scars). All team members and surgeons were Yorubas who lived in the community and were employed at the respective teaching hospitals. The US research team included a Yoruba nurse research associate who was the professional and cultural liaison between the research teams from the US and Nigeria and the Nigerian community. The American research team travelled to Nigeria 1-2 times a year to provide support and technical assistance to the Nigerian teams. --- Recruitment strategies Probands were recruited either in the hospitals at LAUTECH in Osogbo and UCH in Ibadan or in the community. Outside the hospital settings, participants were recruited in private and public clinics, private homes or by invitation in churches and mosques. --- Patient advisors We utilized a modified community-based participatory research approach to engage keloid patients and the community at large. A number of keloid patients who were recruited in the early phases of the study served as patient advisors. The research teams educated them about the goals of the project and the involvement of research participants. Patient advisors assisted by passing information about the project to individuals with keloids in their communities, and instructed them to contact the recruitment team if they were interested in participating. While patient advisors were generally able to answer many questions of participants, they did not participate in the actual recruitment. --- Community leaders To initiate recruitment in a new town, the groups visited the king or his chiefs and ward leaders, who hold important leadership roles within the community, to explain the purpose of the research and the need for cooperation of the population. At those meetings the research team could answer questions from the leaders and/or the community. Posters and handbills (flyers) were distributed during the visits. --- Health directors Medical and nursing directors of local health centers and private hospitals were approached with the request to identify study participants which would then be recruited by the Nigerian research staff. Posters and handbills with pictures of keloids were made available in the clinics to introduce the research project to patients. --- Churches/mosques We solicited the assistance of pastors and imams to share the research study with their congregations. The religious leaders gave us permission to explain the study during a special announcement period. Research teams stayed after the service to answer questions and set up appointments for recruitment. --- Advertisement Handbills and posters were distributed in the communities, specifically in market places, private and public health clinics, schools, churches, pharmacies and clinical laboratories. The posters and handbills in English and Yoruba language contained information about the study, contact information and stipend. --- Media The research team made radio and TV presentations in English and Yoruba language describing the purpose of the study, eligibility criteria and contact information. In addition, radio and television announcements were run to advertise the study and to dispel some myths about keloids. --- Affected/enrolled participants with keloids Often times, enrolled participants were asked to spread word about upcoming recruitment visits in their community. As a result, additional keloid patients usually came to the recruitment site to join in the study. --- Participant recruitment Recruitment and enrollment of participants took place in the plastic surgery clinics in Ibadan or Osogbo hospitals and in the community (participant's home, work place and other chosen site). Diagnosis of keloids was confirmed by the clinicians heading the recruitment teams. The consenting process involved explaining the study to potential participants. Prospective participants were given copies of the consent form and related documents to read or have them read to them. Participants were given sufficient time for review and to ask questions. They had the option to participate the same day or at a later day if they would like to discuss further with their family. Usually, married women liked to discuss with their husbands before agreeing to participate. After consent was obtained, the recruiter and the research participant filled a questionnaire to document the participant's personal information, keloid history and past medical history. Minors between ages 7 and 12 signed assent forms and in addition, the recruiter obtained parental consent. Older minors signed the consent forms together with a parent. Keloid site(s) and sizes of lesions were documented in a questionnaire (body map) and photographs of keloids and other scars were taken. Participants consented to have images of their keloids (Figure 1) used in scientific articles. Research assistants constructed pedigrees when probands reported other affected family members. Probands were asked to inform other family members about the study so they could contact the research team if interested. Family recruitment often took place in their homes which required the research team to travel to their towns or villages. We also recruited large cohorts of patients with keloids without family history (or with no other family members available for the study) and control individuals without keloids. Eligibility criteria for participation as unaffected control individuals were the absence of keloids or suspicious scars and that no member of their families had keloids, suspicious scars or skin diseases. Only one control individual per unaffected family was recruited. Recruitment settings for controls were the same as for the keloid case recruitment. Contact information of all participants was collected in case clarifications were required during the course of the study. Participants were given a sum of approximately $10 (USD) (Naira 1,500.00) as compensation for transportation and inconvenience. Venous blood samples (Vacutainer, Becton Dickinson, USA) or saliva samples (OraGene saliva kits; DNA Genotek, Ontario, Canada) were collected and transported to the US for processing at the University of Connecticut Health Center. Each participant was given a unique identification number. --- Analysis Pedigrees were entered into Progeny pedigree drawing software (Progeny Software, LLC, Indianapolis, IN, USA) and were updated whenever new information about a family was obtained. Statistical analyses were performed in Excel and SPSS. Challenges observed in the process of recruitment and issues that could affect genotype analysis were documented. --- Results --- Administrative challenges Before the actual study could commence a number of administrative steps were necessary to establish agreements between institutions, to train research staff and to ensure that ethics approvals were in place so that the project would accord with the latest standards of human subject protection. The first recruitment site for our study, LAUTECH Consenting documents had to be crafted to suffice all human subjects protection elements and oversight requirements for UCHC protocols. Problems for the Nigerian ethics committee arose when paragraphs were included that are only relevant to the US but may be confusing to participants in Nigeria. Examples of where study participants can get confused are references to: a) Health Insurance Privacy Portability Act (HIPAA) and health insurance providers, because health insurance is unavailable to most Nigerians; b) Requirements of a specific US State, which do not apply to Nigeria; c) US oversight committees that may gain access to consenting materials (Protected Health Information (PHI) in case of project audits. Consents and HIPAA forms need to be carefully crafted to consider the environment and culture of research participants [31]. Recruitment materials needed to be approved by ethics committees/IRBs of both institutions. Training of all research staff in ethics and research compliance was needed before any approval could occur. Training was performed by the US staff during extended visits. --- List of administrative challenges --- Inexperience of investigators and institutions in international collaborations Raising funds for pilot study to demonstrate feasibility of study to funding agencies Registration of Nigerian recruitment sites for Federal Wide Assurance (FWA) Adapting wording of consent/HIPAA forms to local requirements Reconciling IRB requirements of US and Nigerian institutions Human subjects training of new recruitment teams Implementation of subcontracts with different expectations of financial administrations Maintaining ongoing communication between teams to optimize recruitment and for trouble shooting Funds transfer due to differences in banking systems Significant start-up funds were needed to initiate this collaboration as funding agencies (e.g., NIH) will not allow spending funds abroad until all compliance issues are fulfilled and the foreign site is registered with the funding agency. However, in order to start such a genetic collaboration from scratch, travel costs, resources for salaries in the host country as well as computers, internet connection, cameras for documentation, recruitment documents and operating funds had to be funded in advance. Most institutions in low resource countries cannot afford to finance salary expenses upfront and be reimbursed later. Whenever banking systems are different, even the transfer of funds in itself can pose an administrative challenge. Since Principal Investigators at UCHC and LAUTECH had no prior experience with the administrative and regulatory challenges of such a collaborative project and because such a large international genetic study was new to administrations at both sides, our project suffered significant delays. The University of Ibadan, which was later added as a second study site, had experience with international collaborations as it had participated in previous research activities including the HapMap project. As a result of the experience gained with establishing the first recruitment site at Osogbo, it was easier, although not trivial, to add the University of Ibadan as the second recruitment site. --- Recruitment challenges There were several other challenges to successful recruitment (see List of recruitment challenges section). In most families there were some members who lived far away, often located throughout Nigeria and even abroad. Those who lived within the country but in other states did not return to their home towns every year, which prevented timely and complete recruitment of many large families. --- List of recruitment challenges Obtaining referrals from clinics and community centers without payment Unreliable transportation system and road infrastructure Participants are less motivated to participate in research without direct (health) benefits Participants reside in remote locations Participants lack funds to maintain mobile phones Participants do not show for appointments Difficulty to understand research project and purpose Inaccuracies in family and clinical histories Lack of birth records for older adults Open family system which does not distinguish between biological siblings and adopted or socially related siblings Polygamy with families living far apart Hesitation to donate sample because of beliefs in "voodoo" and "juju" Paternity issues Furthermore, even though most people had mobile phones, communication was problematic at times as most probands did not have contact to all of the family members. Frequently, the research group had to travel long distances for recruitment or to confirm pedigree structures, especially in polygamous pedigrees where members from one branch of the family had insufficient knowledge of other parts of the family. On the other hand, we found that children from one wife sometimes lived with families of other wives and were fully accepted as children of the family. Information about "brothers" or "sisters" had to be confirmed by several sources since in many cases, families did not distinguish between "sibling" or "half-sibling". Also, in this open family system, other relatives living with families were often reported as immediate family members and were therefore wrongly recruited. For travel within the state, the recruitment teams had to rely on public transportation. Public transportation was plagued by unreliable connections; hence it was frequently time consuming to commute to recruitment sites. Poor road conditions made travel even more challenging, thus reducing staff time for actual recruitment. Many of the participants (older generations) did not have formal education and found it difficult to understand the research concept with no immediate benefit for their keloids, which they often viewed as a stigmatizing condition. They were more interested in a study that treats or cures keloids and not particularly eager to participate in a study that will take years before yielding meaningful results. Some individuals only wanted to participate if they could receive free treatment. We explored access to keloid patients in Ibadan and Osogbo or in neighboring towns through local hospitals, community clinics and private clinics. This approach often failed as clinicians, especially in private clinics, were reluctant or not willing to inform keloid patients about the study for fear that patients could be diverted from them. Some health care providers (including some hospital staff at host institutions) expected payment for patient referrals to the study, commonly referred to as "head money". We considered payments for referrals as unethical. To reimburse any clinic for actual time and efforts spent on recruitment, we would have needed a formal contractual agreement. Establishing a formal contractual agreement would have been a major administrative effort, necessitating involvement of the funding agency and institutions managing the funds. Accurate pedigrees and clinical histories were sometimes difficult to obtain for several reasons. Some participants, especially those without formal education, had no records of their birth dates and hence were guessing their age, which was at times far off when we compared the parental age to that of the children, who did have birth certificates and some level of education. This problem was more common with women who did not attend school and had no need to obtain official birth records. We were therefore aware that we may have received incorrect information regarding age or relationships despite direct questions. In some instances, we found discrepancies with genotyping results and had to re-visit families for clarification. We had to exclude several individuals in key families for our linkage study because participants could no longer be reached or discrepancies could not be fully resolved. Other issues included friends posing as family members, or that individuals were addressed by different names by different members of the family, which is not uncommon in the Yoruba culture. These cultural idiosyncrasies made it necessary to re-evaluate pedigrees whenever new participants were recruited. For some families it took 5 years to recruit all informative family members who were interested in participating. There were circumstances where remuneration became a major factor for recruitment. While most study participants were happy with the amount of compensation, some participants tried to negotiate for higher compensation. Some individuals believed they should be paid more because they had more keloids or because they had a longer way to travel to a study site. Others wanted extra payment before they would introduce the research team to other family members. Recruitment teams sometimes had to provide prepaid phone cards to probands who were unable to call other family members because of their financial situation. Occasionally, there were extended delays in communication when family members could not be reached because they lacked funds to purchase phone cards. Some private clinics or probands indicated that they were willing to work with the research team if we let them distribute the study stipend to referred participants. This would have allowed them to split the remuneration after subject enrollment. The study also faced challenges from religious beliefs and cultural practices. The Yorubas have strong beliefs in voodoo or juju, which vary within religious groups. Some were convinced that their saliva and/or blood would be used for evil rituals. As a result, families withdrew from the study and the recruitment teams had to destroy some samples that were already collected. We strongly believe that deeply ingrained cultural beliefs also played a role when research teams had difficulties to re-contact study participants. In such circumstances, participants would find excuses or make appointments with the recruiters and not show up rather than frankly telling the research team that they were afraid of voodoo or juju or do not wish to participate for other reasons. This led to further delays in the recruitment progress and sometimes was a dead end for the recruitment of a family that would have been informative for the genetic study. During sample analysis, we discovered inconsistencies with paternity and false allocations of children from different wives. We also encountered friends or distant relatives posing as close family members in at least 5 out of more than 100 families. Recruitment in 3 large families from the Osogbo group and 10 families from the Ibadan group were delayed significantly because family members relocated to different parts of Nigeria or to foreign countries. Three Osogbo-based families withdrew from the study or could not be recontacted for unknown reasons while 9 families withdrew or could not be recontacted by the Ibadan recruitment center. The fear of voodoo or juju prohibited the recruitment of a larger number of family members in three families in Osogbo while 5 individuals raised the issue of juju among prospective participants in Ibadan. Most other family members could usually be recruited. Total recruitment exceeded 4200 study participants over a 7-year period (Table 1). Sample collection in a number of families was incomplete for various reasons, either because family members refused to participate for fear of rituals or for unknown reasons or because of family members living in other parts of the country. Participants who had no family history or where no other family member was available for recruitment were included in a case cohort for a future genome wide association study (GWAS). We were able to recruit approximately 1,900 individuals with keloids during the recruitment period. In preparation for a GWAS, we also recruited more than 2,000 unaffected control individuals. Our initial goal, however, was to recruit mostly large families with inherited keloid susceptibility. We identified and recruited from more than 100 families with more than 550 participants. For 79 families, we were able to establish complete 3-generation pedigrees. Of those families, only 27 had more than 10 participating family members and 9 had more than 20 members participating in the study. --- Discussion Scientific literature offers several recommendations on ethical, legal and cultural considerations for research in low resource settings and for recruitment of minority populations. Experiences with community engagement and informed consent for genetic studies in Africans [28] and for obtaining DNA samples from four populations including the Yorubas from Ibadan (Nigeria) have also been reported [29]. However, there is no report on issues that should be considered when recruiting entire families in rural and urban settings with genetic research as the main goal. Successful recruitment of diverse populations such as the Yorubas for genetic studies is important for advancement in biomedical research. The Yoruba ethnic group was chosen for this linkage analysis approach because we were expecting that large families with keloids could be recruited. Due to the expected locus heterogeneity for causative variants leading to keloids [32][33][34], recruitment from the Yoruba population seemed ideal for finding significant linkage because of relatively low admixture [35,36] compared to many other populations. To attain acceptance of a genetic study within multiple generations of a family, it is important to appreciate social, religious, and cultural beliefs that may impact participation. The Yoruba ethnic group that occupies mostly the western part of Nigeria highly values their cultural practices and religious beliefs. Although most Yorubas are Christians or Muslims, they retain their practice of worshiping different gods such as Sango (god of thunder), Obatala also known as Orisa-nla (the shaper, former, maker), Yemoja (god of the river) and Ogun (god of iron). All these beliefs create perceptions about the possible causes for a disease and often determine the approach to treatment. Traditional healers still have high influence on people's decisions. Part of their health beliefs is that illnesses may originate from one or a combination of three forces; the magical practices (witches or sorcerers), natural events (environmental, hereditary) and supernatural occurrence (destiny or the individual's double spirits) [37][38][39]. It has also been documented that many Nigerians approach treatment of diseases by utilizing Western medicine in combination with treatments offered by traditional healers and "spiritual" centers [37]. For example, one study reported that some Yorubas identified environmental factors such as diet and use of unorthodox traditional herbs as the cause for cancer [38]. The interest in experimenting with different treatments could explain why some of the participants in our keloid study reported treatment of their keloids with a combination of traditional herbs and Western medicine and why they were willing to contribute to science to find better treatment or cure. However, stigmatizing and/or painful diseases such as cleft lip and palate, cancer or keloids are by some perceived to be in part due to a "punishment" for evil deeds or result from retaliation by a jealous foe [37,38,40]. As a result, many Nigerians are weary of participating in studies that have the potential of "affecting" them or other family members, especially if they donate blood, saliva or hair samples. Recruitment was also affected by challenges that had already been described in other cultural environments [30,41]. Such issues included economic factors, lack of trust, different cultural background of researchers and study participants, or issues with time and transportation needed to reach remotely located family members. Meeting regulatory and administrative requirements (IRB, ethics committees and FWA) as well as staff training was less of a problem compared to recruitment challenges. Research staff took the Collaborative Institutional Training Initiative (CITI) training for good clinical practice before they engaged in recruitment. An online education module for ethics training on research based on the Nigerian Code for Health Research Ethics has recently been developed and validated. Future studies will require research staff to take this module that encompasses the Nigerian cultural and social principles [42]. Our experience with this study confirmed the need to allocate extra time for all stages of the study. Extra time was needed to educate the public and prospective research participants about the study and to make up for delays due to travel logistics when recruiting relatives residing outside a proband's home town or when family members were not available for recruitment in a timely manner. Additional challenges arose during the rainy season when treacherous road conditions led to extended delays. Unreliable public transport slowed down recruitment throughout the study period. While purchasing a car dedicated to the study might have been the most economical way to travel to recruitment sites, administrative barriers and uncertain liability issues prevented us from doing so. Working with an institution with established research infrastructure versus a site without any research infrastructure or experience with foreign collaboration should be taken into consideration when planning a study as this will likely impact timely study initiation, staff recruitment, and research coordination. There is recognition that capacity building in African countries must be addressed for successful integration of human genetic studies [43]. Nigerians and specifically the Yorubas are generally not research-na<unk>ve. Participation of Yorubas or any underserved population in genetic studies can be enhanced by community involvement from inception of the study as well as by regular feedback [30], especially when the study design involves participation of family members from several generations. Our study plan included the assistance from probands or other individuals with keloids and contact with community leaders, kings or chiefs where applicable. Interaction with members of the oldest generation or children of old family members may be difficult when the culture of scientists and recruiters is different from that of the participants. To address these and other issues, our study involved research staff from the local community who were born and raised in the Yoruba culture. Visits to kings, chiefs, churches and mosques were useful to convince the community that blood and saliva samples would not be used for voodoo or juju. It was important and reassuring to be accompanied by research participants from their community at such meetings to demonstrate that nothing bad had happened to them since recruitment. It was also important to ensure that clinic or hospital administrators were aware of this study so that they could then encourage the population to participate. We were allowed to recruit in local palaces, churches and health centers/hospitals. Additionally, the research staff was available for recruiting during special holidays when relatives would return home for family celebrations. Cultural sensitivity during study coordination was ensured by a member of the US team that was born and raised in the Yoruba community. Genetic analysis identified non-blood relatives in some families. Some pedigrees had to be updated several times as additional information became available. While scientists require accurate information about a participant's position in a family it is more important for the traditional Yoruba family where individuals fit within a family than how they are biologically related. When obtaining family histories, we found that "adopted" distant relatives were sometimes identified as children or parents. We also encountered situations where friends of the family were reported as members of the family. These situations required that research staff approached a certain member of the family with clear questions to obtain accurate information before establishing a pedigree. During DNA analysis we still identified participants who did not fit in the family structure. The need for confidentiality in these situations was discussed among researchers and participants were either removed from the pedigree or in rare instances the family was eliminated from analysis. With increasing education, the community may become more aware of inheritance and biological relationships within a family and better understand the significance of genetic family studies. Until then, investigators should be prepared to address these factors in their study plan. Even though remuneration is always a difficult topic for recruitment in low resource settings, it is still an incentive for participants who otherwise may not be able to travel to a recruitment site or miss some work/earning to volunteer for a study "without immediate benefit" to them. Other studies also reported that some participants expected more payment for their contribution [29], while others preferred to have free treatment/cure in lieu of money [44]. Investigators should be aware of the potential for abuse by opportunists that may impact recruitment. Our research protocol, for example, required that all study participants be treated and compensated equally. In a review of Project SuGar, the author identified some dynamics that negatively affected recruitment of African-American families into genetic studies and offered suggestions to enhance recruitment [30]. Impacting factors included economic barriers, core values including cultural, spiritual and social factors, as well as health practices of the community. Previous studies have demonstrated that the majority of Nigerians would participate in research activities. Several approaches have been recommended or used to engage, recruit and retain Nigerians for genetic studies [45]. Such methods include the use of community advisory councils, incentives, local partners/champions, focus groups, awareness through public forums and less cumbersome wording during the informed consent process. Our study employed many of these suggestions. Knowledge of genetic variations and their roles in disease pathology and for treatment outcome is significant in the quest to reduce health disparities. While there seems to be a recent increase in genetic studies conducted in African countries, the majority is usually in collaboration with and led by Western scientists. African countries must continue to be involved and take a leadership role in genetic studies so their people can benefit from the advances of the genomic and the post-genomic era. One could argue that the Yoruba people

Background: More involvement of sub-Saharan African countries in biomedical studies, specifically in genetic research, is needed to advance individualized medicine that will benefit non-European populations. Missing infrastructure, cultural and religious beliefs as well as lack of understanding of research benefits can pose a challenge to recruitment. Here we describe recruitment efforts for a large genetic study requiring three-generation pedigrees within the Yoruba homelands of Nigeria. The aim of the study was to identify genes responsible for keloids, a wound healing disorder. We also discuss ethical and logistical considerations that we encountered in preparation for this research endeavor. Methods: Protocols for this bi-national intercultural study were approved by the Institutional Review Board (IRB) in the US and the ethics committees of the Nigerian institutions for consideration of cultural differences. Principles of community based participatory research were employed throughout the recruitment process. Keloid patients (patient advisors), community leaders, kings/chiefs and medical directors were engaged to assist the research teams with recruitment strategies. Community meetings, church forums, and media outlets (study flyers, radio and TV announcements) were utilized to promote the study in Nigeria. Recruitment of research participants was conducted by trained staff from the local communities. Pedigree structures were re-analyzed on a regular basis as new family members were recruited and recruitment challenges were documented. Results: Total recruitment surpassed 4200 study participants over a 7-year period including 79 families with complete three-generation pedigrees. In 9 families more than 20 family members participated, however, in 5 of these families, we encountered issues with pedigree structure as members from different branches presented inconsistent family histories. These issues were due to the traditional open family structure amongst the Yoruba and by beliefs in voodoo or in juju. In addition, family members living in other parts of the country or abroad complicated timely and complete family recruitment. Conclusions: Organizational, logistics and ethics challenges can be overcome by additional administrative efforts, good communication, community involvement and education of staff members. However, recruitment challenges due to infrastructural shortcomings or cultural and religious beliefs can lead to significant delays, which may negatively affect study time lines and expectations of funding agencies.

and compensated equally. In a review of Project SuGar, the author identified some dynamics that negatively affected recruitment of African-American families into genetic studies and offered suggestions to enhance recruitment [30]. Impacting factors included economic barriers, core values including cultural, spiritual and social factors, as well as health practices of the community. Previous studies have demonstrated that the majority of Nigerians would participate in research activities. Several approaches have been recommended or used to engage, recruit and retain Nigerians for genetic studies [45]. Such methods include the use of community advisory councils, incentives, local partners/champions, focus groups, awareness through public forums and less cumbersome wording during the informed consent process. Our study employed many of these suggestions. Knowledge of genetic variations and their roles in disease pathology and for treatment outcome is significant in the quest to reduce health disparities. While there seems to be a recent increase in genetic studies conducted in African countries, the majority is usually in collaboration with and led by Western scientists. African countries must continue to be involved and take a leadership role in genetic studies so their people can benefit from the advances of the genomic and the post-genomic era. One could argue that the Yoruba people still have to offer a lot to science. Nigerians have been called upon to embrace the opportunity of improving their people's quality of life with a timely response to technology gaps in genomics [46]. Initiatives such as the MalariaGen [47] and the Human Heredity and Health in Africa (H3Africa) project, among others, have begun to address this concern (H3Africa: Human heredity and Health in Africa, http://www.h3africa.org/). --- Conclusions We believe that our recruitment experience amongst the Yorubas in Nigeria will be useful to other researchers who plan similar large genetic study studies in Africa or in other low resource settings. We have identified administrative and cultural differences as well as challenges due to insufficient infrastructure that can lead to delays in successful recruitment. As other genetic multigenerational studies are undertaken in low resource settings, investigators will recognize that a combination of methods that address the issues outlined in this article can enhance the success of their studies. --- Competing interests The authors declare that they have no competing interests. Authors' contributions PBO, VO and EJR contributed to the conceptual development of the drafting of the manuscript. PBO, VO, SA, SOF, OMO and EJR contributed data for the manuscript and were involved in critical reading. All authors approved the final manuscript.

Background: More involvement of sub-Saharan African countries in biomedical studies, specifically in genetic research, is needed to advance individualized medicine that will benefit non-European populations. Missing infrastructure, cultural and religious beliefs as well as lack of understanding of research benefits can pose a challenge to recruitment. Here we describe recruitment efforts for a large genetic study requiring three-generation pedigrees within the Yoruba homelands of Nigeria. The aim of the study was to identify genes responsible for keloids, a wound healing disorder. We also discuss ethical and logistical considerations that we encountered in preparation for this research endeavor. Methods: Protocols for this bi-national intercultural study were approved by the Institutional Review Board (IRB) in the US and the ethics committees of the Nigerian institutions for consideration of cultural differences. Principles of community based participatory research were employed throughout the recruitment process. Keloid patients (patient advisors), community leaders, kings/chiefs and medical directors were engaged to assist the research teams with recruitment strategies. Community meetings, church forums, and media outlets (study flyers, radio and TV announcements) were utilized to promote the study in Nigeria. Recruitment of research participants was conducted by trained staff from the local communities. Pedigree structures were re-analyzed on a regular basis as new family members were recruited and recruitment challenges were documented. Results: Total recruitment surpassed 4200 study participants over a 7-year period including 79 families with complete three-generation pedigrees. In 9 families more than 20 family members participated, however, in 5 of these families, we encountered issues with pedigree structure as members from different branches presented inconsistent family histories. These issues were due to the traditional open family structure amongst the Yoruba and by beliefs in voodoo or in juju. In addition, family members living in other parts of the country or abroad complicated timely and complete family recruitment. Conclusions: Organizational, logistics and ethics challenges can be overcome by additional administrative efforts, good communication, community involvement and education of staff members. However, recruitment challenges due to infrastructural shortcomings or cultural and religious beliefs can lead to significant delays, which may negatively affect study time lines and expectations of funding agencies.

Introduction The coronavirus disease 2019 (COVID-19) pandemic continues to burden health systems and communities globally, with millions of cases and deaths [1]. Because of the significant and continued impact of COVID-19, vaccines have been developed at a rapid and unprecedented pace to control the spread of the virus, and prevent hospitalisations and deaths [2]. Many vaccines have been shown to be safe and effective with high-income countries having vaccinated more than half of their population [3]. Despite the availability of these vaccines, countries are faced with various challenges including vaccine hesitancy and anti-vaccination sentiments, limited global supply, and inefficient vaccine deployment [4,5]. These issues in vaccine uptake, together with declining community acceptance of other public health interventions, will mean a delayed recovery and prolonged pandemic [6]. The World Health Organization (WHO) in 2019 identified vaccine hesitancy or the reluctance to vaccinate as one of the top ten threats to global health despite evidence of the important role of vaccines in improving population health outcomes [7]. Together with weak primary health care and other health challenges, countries especially low-and middle-income countries (LMICs) will struggle to meet the demands of the communities within their health system capacity. With the pandemic, countries are further burdened with many health systems overwhelmed throughout its course. The Philippines presently faces these challenges: vaccine hesitancy and increasing anti-vaccination sentiments, a weak primary health care system with efforts to strengthen it through the recently implemented Universal Health Care Law, and an overwhelmed health system because of the demands of COVID-19 and other public health problems [8][9][10][11][12][13]. These challenges are further compounded by a global shortage of vaccine supply with inequitable vaccine distributions [14]. Historically, the Philippines was one of the countries with generally high vaccine confidence rates [15]. Following the dengue vaccine controversy in 2017 however, confidence levels have dramatically dropped and have impacted succeeding vaccination efforts including the COVID-19 vaccination campaign [9,12,[15][16][17]. Dengvaxia, the world's first commercially available dengue vaccine developed by Sanofi Pasteur, was introduced as part of a national school-based immunization programme despite the lack of empirical data on the risks associated with administration of the vaccine among those not previously infected with dengue or seronegative children [9,12,[15][16][17]. By the time reports were released that the vaccine may cause more severe disease among seronegatives, the Philippines had already inoculated more than 800,000 Filipino school-age children [9]. This was highly politicised, and damaged trust in vaccines and the health sector [9,12,[15][16][17]. As a result, immunisation rates dropped and the country saw outbreaks of previously controlled vaccine-preventable diseases such as measles and polio [18,19]. In addition to vaccine hesitancy, the Philippine health system is not prepared for additional health care demands. As early as the first phase of the pandemic, critical care capacity was overwhelmed with the influx of patients in hospitals [10,11]. As of 16 September 2021, the Philippines ranks third among countries with the highest number of newly confirmed cases per one million population [1,20]. Globally, 42.9% of the world population have received one dose of a COVID-19 vaccine, with much lower rates in LMICs like the Philippines [20,21]. Only 55% of Filipinos have expressed willingness to be vaccinated against COVID-19, and as of 16 September 2021, only 30% of the population have been fully vaccinated [21,22]. To end this pandemic, it is critical to implement all possible public health interventions and strategies from face masks, physical distancing, to getting vaccinated [4,23]. However, there is a need to recognise that the adoption of all these interventions is influenced by individual risk perceptions, and these perceptions are shaped by various sources of information and experiences [24]. Additionally, there are interpersonal and structural factors that influence health decisions of individuals. Recognising the multiple dimensions in which behaviours and decisions occur, theories and models have been proposed to explain how individuals make decisions on their health based on factors that change over time and context [25,26]. The social ecological model provides a useful framework for investigating health behaviours and decisions by recognising that a multiplicity of factors interacts to influence health of individuals [26]. These include individual factors representing biological or behavioural characteristics, interpersonal factors representing networks and social capital operating within a defined boundary, and structural factors that include health systems and are mediated through laws and policies [26]. Published studies on vaccination that utilised this model reported that vaccine intentions and attitudes operate along multiple dimensions, with a series of events influencing decisions related to vaccination [17,27,28]. Improving adherence to interventions and vaccination rates therefore requires a better understanding of the different reasons behind vaccine mistrust and not just determining their individual beliefs, knowledge, and levels of trust [17,27,28]. A recently published scoping review supports the use of the social ecological model in understanding attitudes towards COVID-19 vaccination [29]. The review showed that influencing factors are embedded within the social ecological model and that multilevel interventions are needed to improve uptake of vaccines [29]. This scoping review of 50 articles had representation from various countries, but did not include data from the Philippines. We address this gap by exploring the vaccination narratives and challenges experienced and observed by Filipinos during the early COVID-19 vaccination period. We used qualitative data from a mixed-methods study conducted from June to August 2021 that aimed to understand how people in the Philippines view COVID-19 and what influences their behaviours. With these findings, we hope to provide insights to possible avenues of future research and directions for improving COVID-19 vaccine uptake and reach. --- Material and methods --- Design and setting We conducted an online survey among adults ages 18 and older in the Philippines (n = 1,599) from June to August 2021. A subsample participated in the semi-structured interviews (n = 35) with representation from the general population and health workforce from July to August 2021. Data from the interviews informed the findings of this paper. --- Participants and recruitment We aimed to interview participants from different regions in the Philippines, various age groups, socio-economic classes, and vaccination status and attitudes. This allowed us to ensure maximum variation sampling, which aims to capture as many population contexts as possible. We contacted a total of 115 individuals through the information they provided (i.e., mobile number, phone number, e-mail). Out of the 115, 35 participants completed the interviews. The remaining 80 either refused or could not be contacted after a maximum of three attempts. We classified participants according to their vaccination priority group based on the COVID-19 Vaccination Program's prioritisation framework [30]. Those in the first priority group (A1) were frontline workers in health facilities; other priority groups (A2 to C) comprised and represented the general population (Table 1). --- Data collection We conducted the interviews in Filipino, Cebuano, and/or English via online platforms such as Zoom or via phone call. The interview guide included questions about their views on COVID-19, vaccines, and their risk perceptions and behaviours. We recruited interview participants until saturation was reached (i.e., no new information was being obtained from the interviews) [31]. The interviews lasted between 60 to 90 minutes with a token amounting to USD 6 provided to each participant. All participants consented to the interview being recorded. --- Data analysis The interviews were digitally recorded, transcribed verbatim, and translated from Filipino or Cebuano to English. The research team are native and/or fluent speakers of the three languages, and checked for linguistic and conceptual equivalence in the translated documents. We de-identified all participants and assigned pseudonyms. We analysed the data using inductive content analysis focusing on the experiences and views towards vaccination [32]. Our analysis was guided by principles of grounded theory. Transcripts of the interviews were read to identify themes and two investigators (AMLA, VCFP) independently coded the interviews according to emergent themes in Microsoft Excel [33]. We used coding language that was close to the participants' terms and phrases to ensure that we were co-constructing accurate categories reflective of their responses [34]. The codes were reviewed, and areas of disagreement were resolved between the two investigators. Themes from the interviews were further explored through discussions with the other members of the team. We considered reflexivity throughout data collection and analysis, acknowledging that our preconceptions and experiences about vaccination as public health practitioners and health professionals may influence the way we analyse and interpret data. Our use of the grounded theory allowed us to explore the experiences of our participants and our own shared experiences, and avoided being limited by how we view COVID-19 vaccination [35]. To highlight the complex reasons for delaying and/or refusing COVID-19 vaccination, we embedded our findings within the social ecological model with three broad themes: individual factors (attitudes, beliefs, knowledge, behaviours), interpersonal factors (relationships and social networks), and structural factors (health systems and service delivery; media; and policies, regulations, and laws at the local, national, and global level) [26] (Fig 1). The quotes presented in this paper are either in the original English or translated from Filipino or Cebuano. --- Patients and public involvement The public were not directly involved in the design, recruitment, conduct, reporting, or dissemination plans of this research. Their only involvement was as research participants. --- Ethics statement This study was approved by the University Research Ethics Office of Ateneo de Manila University (Study No. SMPH CORISK 2021). All participants were informed about the aims and objectives of the study by including the written consent form in the email correspondence. Prior the interview, the research team thoroughly explained the study to them and provided them the opportunity to ask questions they may have. Written digital consent was taken from study participants before the interview. --- Results We interviewed 35 participants with representation from different vaccination priority groups working in various parts of the country. Our participants also had different educational backgrounds, employment status, and vaccination attitude (Table 2). There was an almost equal proportion of females and males (females: 19; males: 16) with a median age of 38 years old (range: 21 to 74 years old) in the overall study population. Participant views on the barriers to COVID-19 vaccination are presented below, organised using the three tiers of the social ecological model. Individual barriers include perceptions; attitudes; and beliefs about the science, about vaccines, about the health system and government. Interpersonal barriers are the networks and social capital that influence health beliefs and decisions. Vaccine procurement, supply, and logistics, together with media-and policy-related issues, comprise the structural barriers. Where there are differences between the general population and health workers, these are highlighted in the text. --- Individual barriers Vaccine brand hesitancy and brand preferences. Vaccine brand hesitancy or delay in getting the vaccine due to brand preferences was a common theme among the participants. The country's first administered vaccine was Sinovac-CoronaVac, which is manufactured by a Chinese biopharmaceutical company. This was given to health workers despite lack of published data on effectiveness at the time and initial announcements that these were not recommended for high-risk individuals (Quote I1, Table 3). In addition to concerns about the effectiveness of the vaccine, participants also read and heard information on how this vaccine was made. They believed this specific vaccine was using the same virus to 'immunise' an individual's system, which may have unintended effects (Quote I2, Table 3). Other participants cited that this specific brand was not recognised by other countries, and therefore wanted and waited for other vaccines. Meanwhile, others refused to receive mRNA vaccines due to beliefs about its safety and effectiveness. Negative experiences with the health system as source of vaccine hesitancy and anti-vaccination sentiments. The participants cited negative experiences in the past, whether these happened recently or decades ago, as causes of their negative attitude towards vaccines. Three participants who identified themselves as COVID-19 'anti-vaxxers' or those opposed to vaccines, had different sources of anti-vaccination sentiments. These three participants belong to different priority groups. One belongs to the A1 or frontliner group and is working as a Barangay Health/Emergency Response Team (BHERT) member who responds to COVID-19 related health care needs in the community. The second is a retired professional (A2 or senior citizen group) while the third is an environmental protection officer who oversees implementation of public health standards in the community (B2 or other government workers). These participants experienced an undesired event related to vaccines and/or medical care from four years to more than three decades prior the pandemic (Quotes I3-I5, Table 3). Except for one antivaxxer, no other health worker reported negative experiences that caused mistrust in the COVID-19 vaccines and vaccination campaign. Vaccines are viewed as unsafe and deadly. Perceptions on risk of getting infection with and dying from the virus varied among the participants. However, for those who were opposed to the vaccines, their fear of the COVID-19 vaccine and its effects was greater than their fear of the virus and outcomes (Quote I6, Table 3). This fear and their view of vaccines being unsafe and deadly resulted to vaccine refusals or delays. According to them, the deaths observed after --- I1 Vaccine brand hesitancy and brand preferences I had concerns with Sinovac. I read about the studies published about vaccines and Sinovac initially did not publish their results. For me, I did not want Sinovac. [...] There was even a time when the Department of Health said, "Sinovac is not for health workers". I still think some vaccines are better than others (A1, 51-60 years old, male, Laguna). --- I2 Vaccine brand hesitancy and brand preferences I had myself registered, then I did not go, then I registered, then did not go. Then when I went through the third registration, they [local government unit] asked in their website, "Why did you not show up the last time you were scheduled"? The choices for refusal were: conflict in schedule, choice of vaccine, and three other reasons. I chose choice of vaccine. After that, they scheduled me to another vaccination site to get my vaccine. When I went, I was still not in any way sure that it was Pfizer. But I knew it would not be Sinovac. So I was open to whatever vaccine it will be, except for Sinovac. So knowing that it was not Sinovac convinced me to go. Well, regardless if it was AstraZeneca or other vaccines, I was okay with it as long as it was not Sinovac. Because I heard that they used the same virus, they cased it in the vaccine, and it was like you will have the virus in your body? So the negativity as well as the news about it. That was really a major reason that I wouldn't go for Sinovac (A3, 41-50 years old, female, National Capital Region). --- I3 Negative experiences with the health system as source of vaccine hesitancy and anti-vaccination sentiments I self- --- I4 Negative experiences with the health system as source of vaccine hesitancy and anti-vaccination sentiments I was working in a school and was looking after a child whose mother was busy. The child was very smart, but was diagnosed with autism. When he had check-ups, I would accompany him and observe. Then I researched about autism, and found that it is caused by vaccines. No wonder there are many children with autism; it is associated with vaccines (A2, 61-70 years old, female, Camarines Norte). (Continued ) administration of the vaccine are caused by the vaccine; however, medical doctors and hospitals report the death as being caused by underlying conditions such as comorbidities (Quotes I7-I8, Table 3). Some participants also believed the circulating theory that the life span of those who are vaccinated is shortened and they only have two to three years to live: "you are healthy but because of the vaccine, you suddenly die". In addition to the belief that vaccines cause death or shorten an individual's life span, participants also had doubts about the COVID-19 vaccines particularly the mRNA vaccines that use a relatively new technology (Quote I9, Table 3). These concerns about the safety profile of vaccines either caused delays in vaccine acceptance and uptake or refusals. The reverse was reported among most of the health workers and other participants who viewed vaccines positively. They believed that the vaccine protects them from severe illness, hospitalisation, and death, and that vaccines only have minimal risk. Vaccines are viewed as unnecessary and insufficient to prevent disease. Vaccines were viewed as unnecessary by some participants, especially those in older age groups who are not allowed to go out (Quote I10, Table 3). Those in lower priority groups felt that others needed the vaccine more than them. Younger participants shared that they were COVID-19 survivors even without the vaccine; but those at high risk especially the elderly and persons with comorbidities will need the vaccine to protect them (Quote I11, Table 3). The participants also viewed vaccines as insufficient-they expected that getting vaccinated means no longer needing other public health interventions but were disappointed to learn that vaccines are only one part of the solution. Participants therefore questioned the need for the vaccines given the information they have read and/or watched about still being at risk of getting infected despite being vaccinated (Quote I12, Table 3). The lack of clarity in the role of the vaccines has negatively influenced people's decisions on getting the vaccine. Skepticism towards vaccine incentives. Vaccine incentives in the country, such as promotions and offers for those vaccinated, created skepticism among some of the participants. These incentives 'bothered' participants and raised questions about the role of vaccines and the intentions of the government. As a result, these incentives 'disincentivised' participants from getting the vaccine as participants felt being forced to take it (Quote I13, Table 3). Use of vaccines not fully approved by the Food and Drug Administration (FDA). Participants viewed decisions to vaccinate individuals as 'rash' and expressed concerns about vaccines not yet being fully approved by the Food and Drug Administration (FDA). Some also shared concerns about the rapid development of vaccines compared to other vaccines that took decades to develop (Quote I14, Table 3). Participants felt that they were being experimented on using an unproven vaccine, relating this with the dengue vaccine controversy (Quote I15, Table 3). This caused delay or refusal in getting the vaccines when it was offered to them. Low health literacy and lack of critical skills to evaluate health information. Health literacy or how people acquire, evaluate, and apply health information to inform their decisions, including getting the vaccine, is an important but underestimated tool to combat misinformation. Participants shared that Filipinos seemed to know a lot about vaccines, but only superficially. They shared that those among low-resource communities and older population groups were especially vulnerable to misinformation (Quote I16, Table 3). This lack of awareness and critical skills to evaluate information, together with the rapid spread of misinformation, influences people's decisions to get their first dose, to return to their second and get fully vaccinated (Quote I17, Table 3). There were also several participants who shared that they were confused with the contradictory information they were reading and hearing (Quotes I18, Table 3). Religious beliefs do not support vaccines. 'Antichrist'-this was how one participant described the vaccines against COVID-19. Another participant shared concerns about the vaccines and how they would replace antibodies created by God (Quote I19, Table 3). She mentioned that these vaccines have active chemicals that are causing unintended side effects and deaths. --- Interpersonal barriers Family influence and opposition to vaccines. Participants recognised the influence of their family on their health decisions including getting vaccinated. One participant who was opposed to COVID-19 vaccines shared that everyone in their family was unvaccinated because they believed her (A1, 51-60 years old, female, Misamis Oriental). Similarly, a mother who had a negative experience related to the dengue vaccine that was administered to her child, refused to have herself and her family vaccinated against COVID-19 (B2, 41-50 years old, female, National Capital Region). Misinformation spread by networks. Rumours and misinformation about COVID-19 vaccines are easily spread by networks, whether by word of mouth or through social media. A participant said her "eyes have been opened only now because of YouTube" (A2, 61-70 years old, female, Camarines Norte). Participants believed that this affected vaccine uptake, especially among individuals who do not have the opportunity to receive accurate information from official sources including the Department of Health (Quote IC1, Table 4). Perceived conflicts of interest of health professionals. Participants viewed key figures in the response to the pandemic as having conflicts of interests. This perception of having 'hidden agenda' created mistrust in the information provided health professionals, health organisations, and other figures and institutions. These conflicts of interest, whether financial or nonfinancial, subject evidence and data to bias especially if there are undesired adverse effects to the treatment or vaccine (Quote IC2, Table 4). --- Structural barriers: Health systems and service delivery Inadequate supply of vaccines. Observations of participants regarding supply of vaccines varied according to location and membership to the vaccine priority groups. Participants, especially those from cities and provinces outside of metropolitan areas, reported that the supply of vaccines was insufficient to meet the demands and needs of the communities (Quote S-HS1, Table 5). However, even within highly urbanised areas, participants shared that there were those who did not get their second doses on time because no vaccines arrived (Quote --- S-HS6 Perceived inefficiencies of the vaccination system "Nadidismaya" [or disappointed] because on the day of vaccination, there is a two-hour seminar about COVID and vaccines [...] There were people leaving the vaccination site because they found the two-hour seminar long, and they were afraid of crowding in one area (A2, 61-70 years old, male, Nueva Vizcaya). --- S-HS7 Perceived inefficiencies of the vaccination system I believe there's this glitch in the registration system of the city. There's a bug in the system that significantly slows down the vaccine rollout (B1, 21-30 years old, male, National Capital Region). --- S-HS8 Perceived inefficiencies of the vaccination system The system is not centralised. For example, you were already vaccinated in one site. But you are also in the list of another site. I think there should be a feedback mechanism, "I am vaccinated already, you may remove me from the list". So that others waiting can get the slot (A4, 31-40 years old, male, Rizal). --- S-HS9 View that the vaccination system is inflexible and excludes vulnerable and marginalised populations I got my vaccine ahead of my parents. Because I would always wait for posts from the local government on Facebook. At the time they announced that registration was open, I registered myself immediately. But my parents, they are not inclined to technology so they would just wait for guidance (C, 21-30 years old, male, Cebu). (Continued ) --- S-HS14 Health professionals seen as amplifiers of misinformation In addition to Ivermectin, vaccines are another debate within the medical community. I even have a classmate [in medical school] who is an anti-vaxxer. I said, "Let's wait" [for the evidence]. Because others were already fighting (A1, 51-60 years old, male, Laguna). --- S-HS15 Health professionals seen as amplifiers of misinformation There are actually doctors who are anti-vaxxers. There is this specific doctor who had a talk with a public radio station. Previously, she would not give vaccines, according to my classmates [in medical school] working with her because she is doing private practice. She had a pregnant patient who was referred to her. She would convince the mother not to have the child vaccinated (A1, 21-30 years old, male, Iloilo). (Continued ) S-HS2, Table 5). Health workers found that vaccines for them were easily accessible, however those in other groups had to wait longer before getting the vaccine (Quote S-HS3, Table 5). Perceived inefficiencies of the vaccination system. Participants highlighted issues with the system including the slow rollout of vaccines, long waiting time, inefficient registration systems, and lack of a centralised system. Participants mentioned getting frustrated with the speed at which vaccines are being distributed and administered in the country (Quote S-HS4, Table 5). Participants also mentioned issues with the waiting process to get a slot after registration and the waiting time at the day of the vaccination, with some being asked to stay at vaccination sites for two hours to watch a seminar on COVID-19 and vaccines (Quotes S-HS5-6, Table 5). There was perceived risk of exposure, which could be lessened if the process was faster and more efficient. There were also glitches in the online registration systems used by local governments that caused additional delays in getting people vaccinated (Quote S-HS7, Table 5). Local governments are responsible for the distribution and administration of vaccines among their constituents, and individuals may register with various local governments depending on their place of residence or work. This lack of a centralised system makes it difficult to track who have already been vaccinated and where they have been vaccinated such that those who are still waiting for a slot are unable to secure one (Quote S-HS8, Table 5). View that the vaccination system is inflexible and excludes vulnerable and marginalised populations. The current vaccination system of some local governments is viewed as inflexible that excludes vulnerable and marginalised populations. There are individuals who lack access to technology and digital platforms. Especially in rural areas and among the elderly, their exclusion due to access issues is further compounded by their low digital health literacy. These individuals are then unable to register online and get the vaccine (Quote S-HS9, Table 5). While registration is online, even those in older age groups who are part of highly prioritised groups because of their susceptibility to the virus are required to go to the vaccination centre (Quote S-HS10, Table 5). Similarly, those belonging to marginalised groups and communities also encounter considerable challenges to getting the vaccine (Quote S-HS11, Table 5). Logistical challenges. A participant recognised that there are also logistical constraints in the distribution of vaccines, in addition to problems with supply. The COVID-19 vaccines have different temperature requirements with some requiring special distribution systems (S-HS12, Table 5). These logistical challenges influence the distribution of vaccine brands to areas that have the capability to store them and affect decisions to delay getting the vaccine especially among those who prefer other brands (S-HS13, Table 5). Health professionals seen as amplifiers of misinformation. Misinformation on vaccines and treatment were not only observed within families and social networks, but also within the medical community reported by participants who are health professionals themselves. There have been debates about Ivermectin as treatment for COVID-19, as well as vaccines, which have created factions within the group (S-HS14, Table 5). Some of these health professionals who are anti-vaxxers or opposed to vaccines publicly share their views in media and in their practice (S-HS15, Table 5). Because of the stature and credibility of health professionals, their views, whether backed by science or not, get amplified in the media and communities. Pandemic response deemed as ineffective affects trust in health institutions. The response and messaging of health organisations, together with other key figures and institutions in the country, were viewed by participants as ineffective (S-HS16, Table 5). As a result, there is declining trust in these organisations with participants doubting information provided, such that Filipinos no longer take the pandemic seriously (S-HS17, S-HS18, Table 5). In turn, participants turn to other sources of information that they think are more credible and trustworthy. --- Structural barriers: Media and policies Traditional and digital media accelerating the infodemic. Information on the virus and vaccines are easily and effectively amplified by the media. With the infodemic (portmanteau of information and epidemic) or the exponential production of information whether scientifically accurate or not, traditional media and digital media become drivers of (mis)information or fear towards vaccines (Quotes S-MP1-S-MP2, Table 6). Information that participants were receiving from these sources influenced their health beliefs and vaccine decisions (Quote S-MP3, Table 6). Perceived poor policy implementation and lack of evidence-based policies contributing to loss of confidence in vaccines and health institutions. The government developed the Philippine "National Deployment and Vaccination Plan for COVID-19 Vaccines" that identifies population groups to be prioritised ensure vaccine equity accounting for different risks and needs [36]. This plan also stated that only vaccines granted with emergency use authorisation (EUA) or certificate of product registration (CPR) by the Philippine FDA will be purchased by the government. However, this was reported by participants to be poorly implemented with others using connections also known as 'palakasan' system to get the vaccine ahead of those in the priority list (Quote S-MP4, Table 6). Even within the government, the Presidential Security Group were given vaccines even without EUA and/or CPR registration from the FDA (Quote S-MP5, Table 6). In addition, the government purchased vaccines that did not publish their results, and reportedly had lower efficacy rates but more expensive (Quote S-MP6, Table 6). As a result, participants felt that the government was'settling for S-MP4 Perceived poor policy implementation and lack of evidence-based policies contributing to loss of confidence in vaccines and health institutions We cannot avoid it-that those who have connections get the vaccine first. And then they [government] promised us that we frontliners and our family members [extended priority list] will be prioritised for the vaccines. But this is not true. I was not prioritised and this was the same experience for my co-workers (A1, 31-40 years old, female, Pampanga). S-MP5 Perceived poor policy implementation and lack of evidence-based policies contributing to loss of confidence in vaccines and health institutions The FDA approved Sinopharm [for compassionate use] even if there were no published trials yet. The vaccination of the Presidential Security Group was illegal. Why were they vaccinated illegally? There are many issues with the government, which make people question the vaccines. So there were many people who hesitated getting the vaccine, and they lost confidence in the available vaccines. We, health workers, had difficulties persuading or convincing people to get vaccinated (A1, 21-30 years old, male, Albay). (Continued ) less' and that Filipinos deserved better (A4, 21-30 years old, female, National Capital Region). These issues contributed to declining confidence in vaccines and health institutions, with Filipinos questioning the safety of such vaccines and the implementation of these prioritisation frameworks. --- S-MP6 National and local political issues. Past and current political issues contributed to refusals to specific vaccine brands. Together with reports of how the virus emerged from Wuhan, China, these triggered skepticism towards vaccines manufactured in their country. Participants mentioned the dispute of the Philippines and China regarding contested territory at the West Philippine Sea (South China Sea) as a reason for not preferring and/or refusing vaccines from their country, even when donations of Sinovac from China were the first vaccines to be available (Quote S-MP7, Table 6). This dispute also influenced how participants thought about the origins of the virus and why other countries developed their own vaccines (Quote S-MP8, Table 6). Locally, participants viewed politics to have influence on which cities or provinces receive preferred vaccine brands. They mentioned that these 'favored hospitals and provinces' were prioritised, which was perceived as unfair and causing further delays in the vaccination rollout (Quote S-MP9, Table 6). --- Discussion One of the most effective public health strategies, vaccination, has been the focus of false and inaccurate information with rapidly declining rates of acceptance. [37]. In the Philippines, vaccine confidence plummeted after the Dengue vaccine controversy [9,12,[15][16][17]. While antivaccination views and vaccine hesitancy are not yet the main barrier to vaccination in the Philippines which still struggles with vaccine access and distribution, lessons from other countries indicate that these equally and urgently need to be addressed in addition to other challenges [38]. Our study supports the findings of other published research that report a host of individual, interpersonal, and structural barriers that work individually or collectively against vaccination uptake and reach [29]. Therefore, there is a need for a holistic approach to promote COVID-19 vaccination that not only addresses barriers at the individual level, but also at the interpersonal and structural levels [38,39]. Individual perceptions, beliefs, and experiences play a major role on the decision to vaccinate. These are shaped by exposure to (mis)information spread by networks, by key health figures and institutions, and through the media [40][41][42][43]. Misinformation regarding vaccines have been present since vaccines were first developed [44][45][46], but the advent of social media made its propagation much easier [43,45,47]. Unique to the Philippine context is vaccine brand hesitancy, specifically towards Chinese manufactured vaccines and mRNA vaccines. This is caused in part by lack of transparency and scientific information, and spread through networks and the media. Further aggravating the issue is how some people attempt to correct misconceptions in a way that alienates people instead of addressing misinformation. People involved in vaccine promotion activities, especially primary care providers, may need to be trained on how to engage with vocal vaccine deniers and promote vaccination. The World Health Organization document outlining how to respond to vaccine misinformation would be an important resource in such an endeavour [48]. Celebrities and social media influencers may also play a role in promoting vaccination [41], but it is essential that they disclose conflicts of interest to develop trust with their audience. The media also needs to be trained on how to present news regarding adverse effects following immunsation, and regarding COVID-19 in general, so as not to create unnecessary panic and dissuade people from getting vaccinated. A study reported that there may be a need to use first-person, people-centred narratives to prevent 'psychic numbing' and give faces to numbers [49]. In all these, it is vital to engage with the public, especially those who are vaccine hesitant, in order to promote vaccination using language that is inclusive and applicable to their context [48]. The health system and one's interactions with it also contribute to one's decision to get vaccinated. As in this study, trust in the health system has been found to be a major factor in getting COVID-19 vaccine [41,50]. The Philippine government has instituted several health system confidence-building policies. The recent COVID-19 Vaccination Program Act stipulates the provision of free COVID-19 vaccines to all Filipinos and the establishment of an indemnification fund for people who could possibly develop adverse effects following immunisation [51]. Perceptions of 'palakasan' (i.e., use of political connections), stemming from instances during the course

Effective and safe COVID-19 vaccines have been developed at a rapid and unprecedented pace to control the spread of the virus, and prevent hospitalisations and deaths. However, COVID-19 vaccine uptake is challenged by vaccine hesitancy and anti-vaccination sentiments, a global shortage of vaccine supply, and inequitable vaccine distribution especially among low-and middle-income countries including the Philippines. In this paper, we explored vaccination narratives and challenges experienced and observed by Filipinos during the early vaccination period. We interviewed 35 individuals from a subsample of 1,599 survey respondents 18 years and older in the Philippines. The interviews were conducted in Filipino, Cebuano, and/or English via online platforms such as Zoom or via phone call. All interviews were recorded, transcribed verbatim, translated, and analysed using inductive content analysis. To highlight the complex reasons for delaying and/or refusing COVID-19 vaccines, we embedded our findings within the social ecological model. Our analysis showed that individual perceptions play a major role in the decision to vaccinate. Such perceptions are shaped by exposure to (mis)information amplified by the media, the community, and the health system. Social networks may either positively or negatively impact vaccination uptake, depending on their views on vaccines. Political issues contribute to vaccine brand hesitancy, resulting in vaccination delays and refusals. Perceptions about the inefficiency and inflexibility of the system also create additional barriers to the vaccine rollout in the country, especially among vulnerable and marginalised groups. Recognising and addressing concerns at all levels are needed to improve COVID-19 vaccination uptake and reach. Strengthening health literacy is a critical tool to combat misinformation that undermines vaccine confidence. Vaccination systems must also consider the needs of marginalised and vulnerable groups to ensure their access to vaccines. In all these efforts to improve vaccine uptake, governments will need to engage with communities to 'co-create' solutions.

-19 in general, so as not to create unnecessary panic and dissuade people from getting vaccinated. A study reported that there may be a need to use first-person, people-centred narratives to prevent 'psychic numbing' and give faces to numbers [49]. In all these, it is vital to engage with the public, especially those who are vaccine hesitant, in order to promote vaccination using language that is inclusive and applicable to their context [48]. The health system and one's interactions with it also contribute to one's decision to get vaccinated. As in this study, trust in the health system has been found to be a major factor in getting COVID-19 vaccine [41,50]. The Philippine government has instituted several health system confidence-building policies. The recent COVID-19 Vaccination Program Act stipulates the provision of free COVID-19 vaccines to all Filipinos and the establishment of an indemnification fund for people who could possibly develop adverse effects following immunisation [51]. Perceptions of 'palakasan' (i.e., use of political connections), stemming from instances during the course of the pandemic where powerful individuals seem to be above the law [52], contribute to vaccine hesitancy and poor uptake of vaccines. These negative impacts are further compounded by the highly politicised Dengvaxia controversy where individuals, especially parents of school-age children, felt that health institutions and governments were experimenting on them [9,12] with our participants relating the COVID-19 vaccine 'experiment' with the dengue vaccine. In addition, inadequate supply, logistical challenges, and perceptions about the inefficiency and inflexibility of the system negatively impact vaccination rates in the country. As of 16 September 2021, only 3 in 10 Filipinos received one dose with significant differences between population groups: almost all frontline and health workers have been vaccinated while only 2 in 5 elderly Filipinos received their first dose [21]. Those in the third priority group have higher rates than the elderly population group, which were offered the vaccines earlier. Apart from individual reasons, marginalised and vulnerable groups such as the elderly have reported not being able to get their vaccine due to lack of home vaccination services and guidance in using online registration systems. The system will need to consider needs of all population groups to improve vaccination uptake. In all these, trust in the health system needs to be maintained, while disregarding regulations and policies in place can erode trust in the vaccination process. In the Philippines, the national government has the responsibility to procure, allocate, and distribute the vaccines to the different provinces and municipalities, but it is the local government that is responsible for last-mile transport and actual inoculation. This results in wide variations in client registration and procedures between different localities. This underlines the need to identify best practices in vaccine rollout systems to implement a system that is efficient and inclusive to ensure that access to technology and mobility will not be barriers to vaccination. There are a number of limitations that need to be considered when interpreting our findings. First, we were not able to have representation from the A5 priority group (indigent population). While we initially were able to get a participant from this group based on the survey response, we later found during the interview that this individual belonged to a different vaccination priority classification. This may point to issues with online data collection where researchers are unable to reach individuals from low-resource households. Second, there may be social desirability bias because we were unable to ensure if the respondent had other people with them that may have caused a change in their responses. Additionally, we did not disclose any political affiliations and interests, but participants may have been cautious in mentioning negative experiences related to vaccination. Participants may also have chosen more positive responses considering our background as health researchers. However, we emphasised that they will remain anonymous and their data treated with utmost confidentiality. Lastly, factors influencing COVID-19 vaccination uptake is context-specific, and this paper does not aim to represent all situations and circumstances. --- Conclusion Challenges to COVID-19 vaccination may be individual, interpersonal, and/or structural, which interact to influence decisions. Individual perceptions play a major role in the decision to vaccinate, and such perceptions are shaped by exposure to (mis)information amplified by the media, the community, and the health system. In the Philippines, vaccine brand hesitancy and misinformation are prevalent due to their rapid spread through social media and sensationalism in traditional media. Information on the effectiveness of safety of vaccines regardless of brand needs to be communicated to the public to increase COVID-19 vaccine confidence. At the interpersonal level, exposure to networks and health workers who are opposed to vaccines heightens public skepticism of vaccination. Structural barriers including political issues and poor implementation further contribute to vaccine refusals. The ongoing infodemic and anti-vaccination sentiments operating at all three levels (individual, interpersonal, structural) require empowering individuals to evaluate health information, and therefore health literacy becomes a critical tool to combat misinformation. Families and peers also need to be involved in these discussions as they influence vaccine uptake. Individuals engaged in vaccine promotion activities may need to be retrained on how to engage with vocal vaccine deniers in public. Given the involvement of traditional media, trainings on public health and science communication may be helpful in reporting vaccination-related news. Public figures need to disclose conflicts of interests and be transparent to the public, laying out the risks and benefits of vaccines. Laws should be well-implemented and equally implemented regardless of socioeconomic class or social position to encourage trust in the health care system and in vaccination initiatives. There is also a need to study best practices in vaccine rollout to implement systems that are efficient and inclusive so that we can vaccinate as many people against COVID-19 as quickly and as inclusively as possible: provide technological support particularly among older populations and allow flexible options for receiving the vaccine such as home vaccination. Given resource limitations, the vaccination rollout could also be improved by increasing the role of the private sector in the rollout and administration of the vaccine. The government and health organisations will need to connect with individuals, communities, and other institutions, including those who are against vaccines or hesitant towards vaccines, to co-create effective and sustainable solutions. --- All data relevant to the study are included in the article. --- Author Contributions Conceptualization: Arianna Maever L. Amit, Veincent Christian F. Pepito, Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit. Data curation: Arianna Maever L. Amit, Veincent Christian F. Pepito. Formal analysis: Arianna Maever L. Amit, Veincent Christian F. Pepito. Funding acquisition: Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit. Investigation: Arianna Maever L. Amit, Veincent Christian F. Pepito, Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit. Methodology: Arianna Maever L. Amit, Veincent Christian F. Pepito, Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit. Project administration: Arianna Maever L. Amit, Lourdes Sumpaico-Tanchanco. Resources: Lourdes Sumpaico-Tanchanco. Supervision: Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit. Validation: Veincent Christian F. Pepito, Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit. Visualization: Arianna Maever L. Amit. Writing -original draft: Arianna Maever L. Amit, Veincent Christian F. Pepito. Writing -review & editing: Arianna Maever L. Amit, Veincent Christian F. Pepito, Lourdes Sumpaico-Tanchanco, Manuel M. Dayrit.

Effective and safe COVID-19 vaccines have been developed at a rapid and unprecedented pace to control the spread of the virus, and prevent hospitalisations and deaths. However, COVID-19 vaccine uptake is challenged by vaccine hesitancy and anti-vaccination sentiments, a global shortage of vaccine supply, and inequitable vaccine distribution especially among low-and middle-income countries including the Philippines. In this paper, we explored vaccination narratives and challenges experienced and observed by Filipinos during the early vaccination period. We interviewed 35 individuals from a subsample of 1,599 survey respondents 18 years and older in the Philippines. The interviews were conducted in Filipino, Cebuano, and/or English via online platforms such as Zoom or via phone call. All interviews were recorded, transcribed verbatim, translated, and analysed using inductive content analysis. To highlight the complex reasons for delaying and/or refusing COVID-19 vaccines, we embedded our findings within the social ecological model. Our analysis showed that individual perceptions play a major role in the decision to vaccinate. Such perceptions are shaped by exposure to (mis)information amplified by the media, the community, and the health system. Social networks may either positively or negatively impact vaccination uptake, depending on their views on vaccines. Political issues contribute to vaccine brand hesitancy, resulting in vaccination delays and refusals. Perceptions about the inefficiency and inflexibility of the system also create additional barriers to the vaccine rollout in the country, especially among vulnerable and marginalised groups. Recognising and addressing concerns at all levels are needed to improve COVID-19 vaccination uptake and reach. Strengthening health literacy is a critical tool to combat misinformation that undermines vaccine confidence. Vaccination systems must also consider the needs of marginalised and vulnerable groups to ensure their access to vaccines. In all these efforts to improve vaccine uptake, governments will need to engage with communities to 'co-create' solutions.

Introduction Historically, many gender variant individuals have lived in a chronic state of conflict between self-understanding and physical being, with a continual misalignment between others' perceptions of them and their internal self-perception of gender. Only recently have professionals from mental health and medical realms come together to provide services to youth and, hopefully, some validation. As with other newly evolving fields of study, initial interventions were applied without the benefit of much research or precedent for guidance, and at times in an atmosphere of professional division (see Drescher & Byne, 2012, for a summary of continued controversies). The Gender Management Services-Disorders of Sexual Development Program (GeMS-DSD) evolved due to the dearth of available services for two distinct populations: a) youth with Disorders of Sexual Development (DSD) and b) gender variant youth. DSD refer to biological conditions in which anatomic sexual development is atypical (Houk, Hughes, Ahmed, & Lee, 2006) whereas gender variance refers to gender expression and/or identity inconsistent with prevailing societal expectations and norms (Kulick, 1999). The term transgender typically refers to those individuals for whom genotype and phenotype are mismatched. Therefore, biologically male children may self-identify as female and vice versa, or youth may not fit neatly into either category. This paper will focus on the gender variant group served by GeMS-DSD. We highlight clinical challenges, and provide clinical vignettes to illuminate the psychologist's critical role. Please refer to the online supplemental materials for further description of terms relevant to gender, sex and sexuality, and a summary of suggested psychosocial evaluation recommendations. The development of the GeMS-DSD Program was made possible because the initiative of an endocrinologist with prior expertise treating transgender adults, and a strong passion to assist gender variant youth without access to care. As with any novel program, a vision and a sense of possibility are essential aspects of effective action. With a strong belief in the need for such a program in a multidisciplinary hospital setting, the GeMS-DSD service was developed, partially dependent upon the persuasive abilities of the founding physicians, but also within the structure of an institution that encouraged care for underserved youth and with clinic directors and hospital administrators who fostered innovation. The GeMS-DSD program became the first multidisciplinary mental health and medical program housed in a pediatric academic center in North America to serve youth with DSD or gender variance, and has forged a path for the development of other clinics in the United States. Many mental health professionals, medical students, pediatric house officers, endocrine fellows, and staff endocrinologists have participated in our program. --- Program Development The development of GeMS-DSD was a shared effort, requiring extensive multidisciplinary collaboration. Consultation was sought from urology, endocrinology, medical ethics, genetics, neonatology, gynecology, psychology, and hospital administration. When the program opened, it was co-directed by a pediatric urologist with expertise treating children with DSD and a pediatric endocrinologist, working in tandem with a psychologist to provide evaluations and services for gender variant youth and their families. The remainder of the discussion will focus on the gender variant group in the GeMS program, with an emphasis on the crucial role of psychologists within this multidisciplinary team. In order to develop our mental health protocols, our hospital supported the GeMS psychologist receiving training in Amsterdam from Peggy Cohen-Kettenis, PhD and her team, pioneers in assessing and treating transgender youth. The purpose of the trip was to learn and adapt the Dutch protocol for use in the United States. The Amsterdam group opened the first specialized gender identity clinic for children and adolescents in 1987 (deVries & Cohen-Kettenis, 2012) and have published numerous studies based on their protocol and interventions (e.g., Delemarre-van de Waal & Cohen-Kettenis, 2006;deVries, Steensma, Doreleijers, & Cohen-Kettenis, 2011;Wallien & Cohen-Kettenis, 2008;deVries & Cohen-Kettenis, 2012). During the training trip, the GeMS psychologist and endocrinologist participated in the first international Adolescent Gender Identity Research Group Meeting. Psychological measures were selected collaboratively for clinics to use in the evaluation of transgender youth, based on shared experience with this population, while each clinic adapted and added measures as needed for individual sites. When opened, the GeMS clinic was flooded with inquiries from families, not only from the local region, but also from across the nation and internationally. Notably, before the GeMS program existed, the demand for services was largely invisible. In addition, children and families struggled to identify resources (many of which were predominantly non-existent) without the aid of trained professionals, while sometimes coping with significant and multifaceted psychosocial challenges. These could include a range of issues such as managing family responses, including anxieties and discord related to atypical gender expressions and/or disclosures of children; managing peer, school and other social circumstances in contexts that were often less than accepting; and managing mental health issues. Numerous articles have been published outlining similar multifaceted issues gender nonconforming children and families may face (e.g., Dreger, 2009;Ehrensaft, 2007;Malpas, 2011;Menvielle, 2012). In response to the increasing volume of cases a social worker joined the team to conduct pre-screening telephone intakes, aid families in finding resources, and to help develop written clinic protocols in collaboration with the psychologist. --- Clinic Practice The GeMS program, based on the model of care first developed and shaped in Amsterdam, continues to be adapted over time in response to new developments in the field and service demands. Our protocol relies on existing guidelines and standards for working with transgender individuals developed by various disciplines. For example, the World Professional Association for Transgender Health (WPATH) Standards of Care (Coleman et al., 2011), the Endocrine Society Guidelines (Hembree et al., 2009), the Report of the American Psychological Association (APA) Task Force on Gender Identity and Gender Variance (2009; http://www.apa.org/pubs/info/reports/gender-identity.aspx), and the American Counseling Association Competencies for Counseling with Transgendered Clients (2010) each offer valuable recommendations for working with the transgender population. Generally, these guidelines and standards are similar in that they all recommend supporting transgender individuals in their affirmed gender identity, which often includes assisting in medical interventions that will help make the individual's body congruent with their affirmed gender. The APA Task Force report (APA, 2009) states support for the "efficacy, benefit, and medical necessity of gender-transition treatments for appropriately evaluated individuals..." (p.67), a statement consistent with the goals of the GeMS team. Nevertheless, many of these guidelines do not focus on issues specific to transgender youth. The Society for Adolescent Health and Medicine (2013) has issued recommendations for promoting the health and well-being of lesbian, gay, bisexual and transgender adolescents, and the American Academy of Child and Adolescent Psychiatry (2012) has published practice parameters addressing gay, lesbian, bisexual, gender nonconforming and gender discordant children and adolescents. The APA also published a helpful and accessible pamphlet regarding gender identity and gender expression, with some information about transgender youth (http://www.apa.org/topics/sexuality/transgender.pdf). They note that "it may be helpful to consult with mental health and medical professionals familiar with gender issues in children" (p. 3), while also emphasizing that "identifying as transgender does not constitute a mental disorder" (p. 3) and that "it is not helpful to force the child to act in a more gender-conforming way" (p. 3). This position is aligned with our gender affirming approach to care (see Hidalgo et al., 2013 for an elaboration of a gender affirming model) which views gender variations as part of an expected diversity, and not pathology. Mental health challenges may emerge related to cultural and social responses to a child or co-exist with gender non-conformity. Consistent with much literature (e.g., Hidalgo et al., 2013;Steensma, McGuire, Kreukels, Beelman & Cohen-Kettenis, 2013;Wallien & Cohen-Kettenis, 2008) we view gender as sometimes fluid over time, recognizing that not all gender non-conforming children fit neatly into male or female identities, and that gender identity (internal sense of self) and gender expression (outward expression of gender) may modify over time. Members of the GeMS team have played a role in the development of standards and guidelines, including as a member of the active APA Task Force to develop guidelines for psychological practice with transgender and gender non-conforming clients. As time has elapsed, and our clinical expertise has developed, we have advanced to a more flexible, individualized approach to care than was utilized at the clinic's inception, which may evolve further with increasing research to inform best practices. Within our current model we continue to prioritize evaluation and treatment, mental health and readiness for medical treatment, but allow for a variable structure and account for the unique circumstances of the youth and family. Therefore, the model set forth below is adaptable, serving as a guide for care as opposed to an inelastic protocol. Clinical discretion and family needs are prioritized, as deemed appropriate by the psychologist working within a multidisciplinary team. In addition, as the field evolves, our future practices may vary from those delineated. However, we anticipate that our fundamental approach will endure, and can be described as the intertwining of mental health and medical expertise, each informing the other to best assist families and youth. --- Intake The initial telephone intake, conducted by a GeMS clinical social worker, includes gathering a substantial amount of information and allows the parent and/or guardian the opportunity to tell their story to a knowledgeable professional, often for the first time. The information includes reasons for concerns about gender variance, current crises, and developmental, medical, and mental health history. Other services include support, psycho-education, explanation of protocols, outside referrals and scheduling a clinic appointment when appropriate. We believe that it is imperative for a qualified and experienced clinician to be the first point of clinical contact to set the roadmap for future care, and to act as an identified trusted individual to whom the family can turn. The intake frequently plants the seeds of hope, providing relief for families who have been enduring the stress of a situation for which they have had little preparation, often within a context of isolation. A description of the patient population presenting in GeMS through the year 2010 indicated that the mean age at intake was approximately 14, with a slight preponderance of genotypic female to male patients, many of whom (approximately 44%) presented with a significant psychiatric history (Spack et al, 2012). It is important to note that the earliest we medically treat children is when puberty has just begun, medically defined as Tanner Stage 2 (Marshall andTanner, 1969, 1970). A youth's chronological age is less relevant than their biological development and a cognitive level necessary to adequately assent to treatment. However, we do not accept new patients for treatment older than eighteen. In the case of younger children who are not yet approaching puberty, guidance is often sought for gender related challenges, in which case we provide psycho-education, and offer referrals for families to receive supportive mental health counseling. These services may assist the youth in clarifying their gender identity, and help youth and families navigate the many anticipated and unanticipated issues they may confront, including whether or not to initiate a social transition (presenting in social settings as the affirmed gender). Children may experience anxiety and depression, often secondary to the social and familial ramifications of their gender questioning and/or atypical presentation, and a mental health professional with relevant expertise can be tremendously helpful. When a child is seeking services closer to puberty, our current model typically recommends three to six months of psychotherapy. For some children who feel a compelling sense of urgency in light of impending physiological changes, this recommendation may be modified, especially when complicating factors are absent and the child is well supported. This aspect of the model reflects our recognition that many youth and/or parents seeking services in our clinic are in the early stages of gender exploration and consideration of medical intervention options, and need a safe forum in which to learn more about the issues involved, and treatment available. Further, we have found psychotherapy exceedingly helpful for treating co-occurring mental health issues and for exploring the child and/or adolescents' thought processes, family functioning, strengths and support systems. In addition, psychotherapy enables a deeper exploration of the child's Gender Dysphoria (GD), the range of gender expression and gender identity questioning, and whether the subjective experience fits more into a model of binary identity (e.g., male/female) versus a fluidity of gender and gender nonconformity. Mental health intervention can also support problemsolving regarding the medical and social challenges that lie ahead. It helps facilitate discussion between families and other support systems (schools, extended family, religious/ sectarian community affiliates) as next steps are contemplated. Many authors also have noted the importance of mental health services (e.g., Bernal & Coolhart, 2011;Menvielle, 2012;Turek, 2011). Drescher & Byne (2012) emphasize that "the majority of adolescent persisters do well when they receive family and professional support for early interventions" (p. 504). Therefore, GeMS patients are asked to continue working with their outside mental health provider during the course of medical treatment in our clinic. One of the purposes of the puberty blocking medical intervention (described below) is to buy time for the adolescent to continue exploring gender identity issues without the added stress of a puberty that is inconsistent with their self-identity. In our view, it is often unrealistic to expect an adolescent to sort through the myriad of issues related to gender variance without the help of a professional. Many of the challenges adolescents face regard the reactions of others to their gender identity and/or expression, but can also include gender-related questioning and confusion (see Cohen-Kettenis, Steensma & de Vries, 2011, for an interesting discussion of psychological interventions for adolescents with GD). --- Psychological Evaluation The goals of evaluation, conducted by a licensed psychologist, are to further understand the child and family's needs, and to inform medical treatment interventions. Before initiating the evaluation, we typically request a letter from the child's outside community therapist composed with the aid of a guide we provide. The therapist is asked to address their understanding of the patient's gender identity history, including length of time the patient has had gender questioning feelings, how long he/she has been living in the role of a different gender (if at all), and how persistent his/her identification with a different gender has been, if ever, over the course of time. The letter includes the therapist's impression of the patient's supports, the therapist's perception of other mental health issues or developmental concerns, and finally, the therapist's perception of benefits/drawbacks related to medical intervention. Assuming that the therapist's letter is generally supportive of medical intervention, following review by our mental health clinicians, we move forward with an on-site psychological evaluation. This evaluation consists of extensive interviews of youth and families, and measures of anxiety, depression, self-concept, behavioral and social functioning, autism spectrum disorder (ASD), and gender identity. With consent, outreach is often made to collateral informants, and we review relevant documents (e.g., neuropsychological evaluations), as appropriate. In the clinical interview, we address what the youth and parents hope to accomplish from the evaluation, family and developmental history, school and academic history, mental health and medical history, substance use, and trauma history. We gather an extensive gender history including the youth's subjective experience of gender across time, gender presentation, gender role expression, and sexual orientation. Considerable attention is paid to factors that make these cases more complicated, such as patients presenting with features of ASD, severe psychiatric concerns (e.g., suicidality, self-harming behaviors, psychosis, violence and aggression, and history of abuse/trauma), and/or complicated family factors (e.g., divorced parents, unsupportive family members). We assess support structures and strengths, familial attitudes about non-traditional gender roles and sexual orientation preferences, religious, cultural and ethnic background, and additional individual and family stressors. The youth's age at first signs of GD or disclosures is always noted; families may be caught off guard when their children first disclose gender questioning close to adolescence or after the onset of puberty, and often the evaluations of these youth and families are particularly complex. Consistent with psychological evaluations in general, the rationale for numerous measures and methods of information gathering is to obtain the most authentic and comprehensive clinical picture possible. This is particularly critical, given that the results and clinical formulation play the primary role in deciding whether to move forward with a potentially life-changing medical intervention for the adolescent. We synthesize and interpret the information obtained, and use the evaluation as a way to understand the youth and family's state of mind, ambivalences, and overt and covert pressures. We also want to ensure that, to the extent possible, a youth's cultural and social environment will support their chosen gender identity and provide a safety net as they move forward. A full clinical report is written that integrates the information, and provides a formulation and recommendations. The team psychologist then meets with the family to review this information. Medical interventions that often follow are either in the form of puberty blockers, and/or cross-sex hormone therapy, described below. As noted above, continuing psychotherapy for youth is typically recommended by our protocol. At times we recommend family treatment and/or support groups to help with the family's adjustment to their child's transition. The GeMS team then remains in contact with community providers as clinical care dictates. In addition, youth treated in our program return for regular clinic visits, meeting with both mental health and medical team members, in order to provide continuity of care and further assist adolescents and family members as needed. --- Medical Intervention Medical intervention with transgender youth in GeMS occurs under the auspices of a subdivision within the Endocrine Department. In brief, as alluded to above, with children who have recently begun puberty, puberty-blocking hormones are often prescribed. These are administered in the form of subcutaneous implants in the upper arm, which last two to three years, or monthly injections. These treatments are not routinely covered by health insurance in the United States and may range in cost from $120 to over $1,000 per month. Other medical services, laboratory tests, and sometimes cross-sex hormones may be covered by insurance. In the absence of pubertal blockers, biological males with affirmed female identities may experience significant growth, permanent facial hair and vocal changes, and intolerable erections. A voice that has deepened cannot be raised through hormone therapy, and requires difficult and expensive speech therapy, in order to affect a higher voice. Similarly, without such intervention, biological females who identify as male may experience menstruation and breast development; the latter can only be modified through surgery. Nevertheless, an adolescent who has initiated puberty blockers can decide to terminate the intervention and allow physiological changes to occur as they would have, had the medical intervention never been initiated. Only with an older adolescent, typically around age sixteen, are irreversible interventions initiated, and only after psychotherapy and a careful psychological evaluation has taken place. In this way, we try to ensure that an adolescent is not ambivalent, and that these interventions are well thought through and understood without coercion from others, and with full consent. When these conditions are met, an adolescent may be placed on cross-sex hormones (estrogen for genetic males and testosterone for genetic females), to facilitate a more complete transition into that individual's affirmed gender. When natal puberty has been previously blocked, the cross-sex hormones are even more effective in rendering a more gender consonant, "typical" presentation. For male-to-female (MTF) patients, treating with pubertal suppression in early puberty followed by estrogen in later adolescence causes enhanced breast development, vocal quality consistent with the affirmed gender, no development of a protruding larynx or "Adam's Apple", absence of male-typical facial or body hair, and diminished masculinization of the body frame and facial bones. For femaleto-male (FTM) patients, pubertal suppression in early puberty followed by treatment with testosterone later in adolescence leads to development of facial and body hair, deepening of the voice, masculinization of the body frame and facial bones, no need for mastectomies, and no menarche (see Delmarre-van de Waal &Cohen-Kettenis, 2006 andShumer &Spack, 2013 for further information). A common scenario is for GeMS to recommend puberty blockers, when the youth and/or the parent may feel that it would be best to start cross-sex hormone therapy instead. The delay of puberty, rather than the immediate onset of the puberty of choice (utilizing cross-sex hormones) is sometimes difficult for the youth or family to accept. This is an area where we currently have little research to guide us, and the decision of whether to block puberty, or instead move forward with an affirmed gender (i.e., cross-sex hormones) must be weighed carefully. Aside from the irreversible nature of cross-sex hormone initiation, this intervention has significant ramifications for fertility, while puberty blockers do not (Lazar, L, Meyerovitch,, de Vries,, Phillip & Lebanthal, 2014). Anecdotally, we have found that the GeMS evaluation has been invaluable by providing information to guide subsequent psychosocial and medical decision-making. In general, adolescence is marked by a search for identity and personal transformation, and at times impetuous decision-making. Given the implications of social transition and medical intervention, coupled with the developmental challenges of identity consolidation, we feel the need to progress with care and forethought, to ensure that all interventions proceed safely, to minimize medical and psychosocial contraindications or complications, and to make sure it is the appropriate timeframe for intervention. We also want to ensure that the child/adolescent who may be gender variant does not feel compelled to choose a gender (male/female), when in actuality they may not fit into a typically recognized gender identity. Nevertheless, these considerations always need to be balanced by the very real physiological ticking clock, especially for the younger child on the verge of a puberty that they deeply want to avoid. --- Challenges and Dilemmas of Psychosocial Practice --- Child and Family Expectations When confronted by a gender variant child, a parent may be caught very much off guard, with no ability to rehearse the best response to such an unanticipated circumstance. In addition, for a parent, it may feel like a loss of the daughter or son to whom they became so bonded. Moreover, some families are aware of their child's GD in early childhood while others are surprised to learn about it when their child is in their teens. Both instances carry particular emotional impact for families. Many parents are resilient and loving in the face of these challenges, but may experience an understandable drive for rapid certainty and solutions. We have also encountered parents who are resistant to accepting this diagnostic picture, and believe their child's gender variance is a phase, or a manifestation of some other psychological issue that can be resolved, thus resolving the gender variance. Unfortunately, the problems and issues that often exist for gender variant children and their families are nuanced and indeterminate, and the resolutions may evolve through a time consuming process without a known end. This can add to the stress and consequent pressure to "solve" the issues (see Bernal & Coolhart, 2012, Dreger, 2009, Menvielle, 2012and Turek, 2011 for further discussion of family issues). It can be particularly challenging when two parents or guardians with legal custody are in dissent about how to proceed, especially in contentious divorce situations when communication is minimal or hostile, yet medical consensus needs to be reached. Typically, our program requires consent of both parents before medical treatment can go forward and mental health and/or medical clinicians may need to be proactive in trying to resolve disputes with sensitivity. --- Psychosocial Considerations Any number of psychological, social and cultural factors can impinge upon youth and their family, and influence decision-making, expectations and emotional reactions. The Report of the APA Task Force on Gender Identity and Gender Variance (2009) summarizes some of these factors, including general behavior problems, peer related problems and other mental health issues. Below we outline some of the common issues we have encountered in our work. Not infrequently, children and adolescents are involved in meaningful activities, which will be likely impacted by a gender transition. Prominent among these are youth sports teams, which are typically grouped by gender. Adolescents are often loath to lose these areas of gratification, along with the opportunity for social bonding. Other hobbies and interests that are often impacted include dancing, theatre, cheer leading and sleep-away camp, and children and families may be unable to forecast how they will weather these transitions. Therefore, a child may face the dilemma of losing the opportunity to sustain an ability or talent they value in order to live in a gender they embrace. A youth's environment and culture is essential to consider when evaluating treatment options. Ideally, the family and community should provide every child safety, love and solace, and the support a gender questioning child and/or adolescent needs (as any youth does) to thrive into a healthy maturity. However, such youth often struggle for acceptance within their families and communities. We know from prior research (Dean, et al., 2000;Fitzpatrick, Jones, & Schmidt, 2005;Gibson & Catlin, 2011;Grossman & D'Augelli, 2007;Hass, et. al., 2010;Spack et al., 2012) that many children with GD become deeply anxious and depressed, and resort to suicide attempts. Others are at risk of leaving home and living a life with high costs and risks, including of exploitation, abuse, and as victims of violence, while obtaining hormones illicitly without the oversight of a qualified medical professional. Even when families and children seek professional service and care, external factors beyond their control can impede access. Many geographic areas still lack basic services for children with GD, and traveling for access to medical care is not always an option for families living within modest means. Furthermore, schools and religious institutions vary in level of comfort dealing with transgender children, and may not have the understanding or training to navigate the complexities of their transgender student or member's needs. Learning to deal with social issues such as bullying and isolation, and practical issues such as bathroom and locker use, requires open and honest dialogue with experts familiar with gender issues; not all communities are able or willing to avail themselves to this kind of discussion. One positive outgrowth of the Internet and widespread coverage of transgender issues is mainstream access to information about gender variance and dysphoria. Families can become much less isolated by accessing on-line social networks and organizations such as Parents, Families, and Friends of Lesbians and Gays (PFLAG), even when there is not a chapter in their vicinity. However, the increased availability of differing professional standards and practices can sometimes also confuse families, who may specifically seek out professionals who seem open to providing services desired by the patient or parents, even if they are inconsistent with typical practice standards. This could result in circumventing the input of mental health professionals, or providing irreversible intervention for a young or ambivalent child. --- Mental health Sadly, we know that transgender youth are at risk for anxiety, depression, self-harm, suicidal ideation, psychiatric hospitalizations, homelessness, exploitation, and abuse (Dean, et al., 2000;Fitzpatrick, Jones, & Schmidt, 2005;Gibson & Catlin, 2011;Grossman & D'Augelli, 2007;Hass, et. al., 2010Grossman & D'Augelli, 2007;Spack et al., 2012). In addition, the spectrum of issues that can present in any child or adolescent can present in gender variant youth, including history of trauma, oppositional defiant disorder/conduct disorder, and learning disabilities. These youth may do poorly in school, and/or have difficulty with socializing, and negotiating the normal developmental challenges of adolescence. Optimally, a pubescent child and adolescent should be stable, safe, and supported in advance of receiving medical interventions such as puberty blockers or cross-sex hormones. Yet, for many, medical intervention is an antidote for some of their mental health problems. This poses a dilemma for the clinician, who may be averse to going forward with medical intervention, but feel compelled to do so in case that is the critical step needed to jump start a child's recovery. Such intervention should only take place once the crisis of active suicidal ideation, behavior and/or self-harm has receded, and following a full psychosocial evaluation if it had not taken place already, as well as with close monitoring to ensure that the child is safe and that the dangers continue to remit. Delays can be particularly difficult and contribute to a child's distress because of the limited physiological time frame. At the very least, psychological services should help to ensure adequate support systems before any medical intervention occurs, and puberty blockers can buy time and allow for a child to make thoughtful decisions about his or her gender. Finally, there appears to be a higher than expected incidence of co-occurring GD with ASDs based on clinical experience as well as research, although more empirical study needs to be completed (e.g. deVries, Noens, Cohen-Kettenis, van Berckelaer-Onnes, & Doreleijers, 2010;Drescher, 2012;Spack et al, 2012). Very often adolescents on the autism spectrum know they are different from peers, but have only recently identified gender identity as a factor contributing to this divergence. Sometimes they and their families believe that a gender transition will solve all problems, and/or latch on to gender as the sole reason they are unlike their peers. Similarly, parents may believe that the GD is a manifestation of the ASD, and resist treatment. Parents of youth on the autism spectrum may be concerned that their child's intense focus on gender is a fleeting concern, particularly if their child has a history of transitory preoccupations. When children with an ASD are evaluated, it is often more difficult to discern the degree of gender variance given the relatively concrete and binary thought processes and communication patterns that typify this population. A child with an ASD already has challenges in social realms and is faced with an additional unique and complex set of social circumstances. A comprehensive evaluation should help sort through these issues and it may be necessary to move forward cautiously. However, it is our opinion that treatment not be withheld indefinitely as these youth experience the same biological time constraints characteristic of all pubescent individuals, and therefore need to receive optimally timed interventions to the extent possible. --- Service Gaps and Evolution of Practice Watching clinical services grow is rewarding, especially when they translate into more contented and peaceful lives for youth and their families. Nevertheless, evidence-based practices are aspirational when a new field emerges with no guiding clinical precedent. Controversies among providers in the mental health and medical fields are abundant. Drescher & Byne (2012) and Stein (2012) provide excellent discussions of issues of consensus versus continued controversies. These include differing assumptions regarding whether early intervention with gender variant youth can encourage desistance, and whether that is an appropriate practice. Other areas of debate include the age at which children (or adolescents) should be encouraged or permitted to socially transition; whether cross-sex hormones and surgery should be offered to youth, and if so, at what age; whether parental consent be required for these medical interventions; and whether mental health involvement be required, including psychological evaluation, prior to each stage of medical intervention. These issues are complex and providers in the field continue to be at odds in their efforts to work in the best interest of the youth they serve. Addressing each of these controversies goes beyond the scope of this paper; however, the GeMS team continues to stay abreast of these issues and actively participates in ongoing discussion and research (see Schwartz, D., 2012;Ehrensaft, D., Minter, S.P., 2012;Zucker, K.J., Wood, H., Singh, D., & Bradley, S.J., 2012;and Shwartz, D., 2012 for discussions of some of the issues and differing viewpoints). An important priority going forward is to develop research to enhance our understanding of what typifies this population of children, and their developmental course and patterns, and to examine the long-term outcomes of treatment. The field needs to better comprehend which children are most likely to have a life-long and persistent identification with a different gender than the one they were assigned versus those who cease to self-identify as transgender over the course of time. Although some information is available (e.g., American Psychiatric Association, 2013;Steensma, McGuire, Kreukels, Beekman, & Cohen-Kettenis, 2013;Zucker, Wood, Singh, & Bradley, 2012) much more research in this area is needed. Other high priority areas for systematic examination include the effects and side effects of various medical interventions, especially given that they are initiated with youth who may be on a lifetime course of hormone treatment, and psychosocial outcomes for youth who receive medical intervention during adolescence. Finally, we can only report on children with access to services; youth may not have access because of geography and lack of availability, lack of financial means, and/or because of social structures that do not support them. As noted earlier, these children are at risk to be exploited, to be runaways, street youth and sex workers, and to self-medicate and self-harm. Prevention and outreach, to shelter at-risk youth from damaging and avoidable traumas, and to improve access to mental health services, should be one of the highest priorities for health care providers. --- Clinical Case-Composites The following represent composites, not actual cases, to serve as examples of how GeMS has addressed common clinical scenarios Case Scenario # 1: Early Puberty Referral Information: M. is a 10 year old Black natal female who identifies as male. He and his parents came to the clinic stating a desire to initiate puberty blockers to avoid feminizing. History: Although only 10, M's pediatrician had put his pubertal development as Tanner Stage 2 (pubertal), and he was developing breasts. He had been living as a boy at school and elsewhere for two years, and was quite concerned that his pubertal changes might alert others to his natal gender, and was very also very assertive about his desire to avoid the onset of menstruation. He had been in therapy for two years, and was also being treated by a psychiatrist for

Historically, many gender variant individuals have lived in a chronic state of conflict between selfunderstanding and physical being, one in which there was a continual misalignment between others' perceptions of them and their internal self-perception of gender. Only recently have professionals from mental health and medical realms come together to provide services to these youth. This paper describes an innovative program: the first mental health and medical multidisciplinary clinic housed in a pediatric academic center in North America to serve the needs of gender variant youth. We describe our model of care, focusing on the psychologist's role within a multidisciplinary team and the mental health needs of the youth and families assisted. We highlight clinical challenges and provide practice clinical vignettes to illuminate the psychologist's critical role.

sex workers, and to self-medicate and self-harm. Prevention and outreach, to shelter at-risk youth from damaging and avoidable traumas, and to improve access to mental health services, should be one of the highest priorities for health care providers. --- Clinical Case-Composites The following represent composites, not actual cases, to serve as examples of how GeMS has addressed common clinical scenarios Case Scenario # 1: Early Puberty Referral Information: M. is a 10 year old Black natal female who identifies as male. He and his parents came to the clinic stating a desire to initiate puberty blockers to avoid feminizing. History: Although only 10, M's pediatrician had put his pubertal development as Tanner Stage 2 (pubertal), and he was developing breasts. He had been living as a boy at school and elsewhere for two years, and was quite concerned that his pubertal changes might alert others to his natal gender, and was very also very assertive about his desire to avoid the onset of menstruation. He had been in therapy for two years, and was also being treated by a psychiatrist for anxiety symptoms. His therapist had written a letter in support of M living in his affirmed gender. --- Psychological Evaluation: The formal psychological evaluation indicated that M had a longstanding identification as male, which emerged in his early preschool years, as well as ongoing GD, which predominantly took the form of anxiety. His anxiety diminished, according to him and his family, as well as his therapist and psychiatrist, as he transitioned socially and began to live and be treated as a male at home and at school. Information from school revealed that he was viewed as normal and high functioning in all areas. As an example of a response to gender-related questions, M stated that he was not a transgender boy, but just a regular boy. M did report significant anxieties related to social situations, as well as to bathing and bathroom situations. M resides with two biological parents who were both supportive and in accord with pursuing medical treatment, although they reported that it initially had been difficult for them to accept his social transition. Recommendations: Given his long-standing history of GD, positive adjustment at school, the consistency of data obtained from the his psychiatrist, psychologist, both parents and himself, the GeMS team recommended puberty blockers as well as continued psychological treatment to help diminish his anxiety and problem-solve social situations as they may arise. Continuing follow-up with the GeMS psychologist indicated that his anxiety diminished as his impending puberty was forestalled, with strong acceptance for his affirmed gender from his family and others. --- Case Scenario # 2: Parent: Adolescent Conflict Referral Information: E. is a 17 year old Hispanic natal male who came to the clinic with her parents, who immigrated to the United States soon after E's birth. E was hoping to be able to be treated with puberty blockers and female hormones, while her parents were unified in believing that psychotherapy could resolve her GD, and were hoping to have this confirmed by a psychological evaluation. History: E's parents were invested in her remaining male, partially due to the elevation of male status in their traditional culture. Reportedly, E. had been interested in receiving care for her gender dysphoria for several years prior to the current appointment, to avoid the onset of pubertal changes she was already experiencing. However, her parents had been resistant. She had been in therapy with a psychologist for many years, and her therapist was instrumental in helping to persuade her parents to bring her to the clinic. --- Psychological Evaluation: The evaluation revealed that E had identified as female since the age of 5, including using female pronouns, attempting to wear female underwear, playing with traditionally female toys, and identifying with female characters during pretend play. At present, E wore female clothing and had grown her hair, but appeared androgynous due to a deep voice and some light facial hair. She was generally assumed to be male at school and elsewhere, although her closest friends used her female name and pronouns at her request. The psychological evaluation revealed a strong cross-sex identification as female, and mild depression. Recommendation: Puberty blockers were recommended, with possible cross-sex hormones in about six months. The psychologist spent considerable time with E's parents and with E, reviewing the results of the evaluation, and the basis for the recommendations. E's parents were distressed during discussion to learn that there was some urgency to proceed quickly, believing incorrectly that medical intervention could reverse pubertal changes. The treatment recommendations also included family therapy, to facilitate positive communication within the family and provide support and psycho-education for E's parents. We also recommended a continuation of psychotherapy for E., to help her adjust to personal and social changes, provide support, and to help her cope with family discord. E. continues to be seen by the psychologist in our clinic for consultation, and is adjusting well to the initiation of hormone treatment. --- Case # 3. Ambivalence and Mental Health Complexity Referral Information: L. is a 16 year old White European American natal female who presents as male, and has chosen a male name and male pronouns. He has been in therapy since the age of 8, and was initially evaluated and put on pubertal blockers in our clinic at age 13. His mother called the clinic requesting that L. be considered for cross-sex hormones. L. was not seen for a full evaluation as he is an ongoing patient in our service, but for a screening related to his mother's request that cross-sex hormonal treatment be initiated. History: L. was adopted at the age of 1, and his early history is not known. He has been diagnosed with depression, anxiety, and Conduct Disorder. He has a history of self-harm related to depression, academic pressure, and of being bullied in school. His social, academic, and emotional functioning tends to be poor, and he is emotionally and behaviorally dysregulated, with periods of rage at school and at home, and some known drug use. He was recently suspended at school for cheating and for provoking physical altercations. His mother believes that cross-sex hormones would alleviate his distress and dysregulation, Psychological Screening: L's therapist, when contacted with the family's consent, indicated that L. appears ambivalent about his affirmed gender, and therefore did not believe that cross-sex hormones should be initiated. Other aspects of our evaluation also suggested ambivalence on L's part. Although he ultimately agreed with his mother that he should start testosterone, he began the evaluation by suggesting it was "too early" to start them. In addition L. reported that he binds his breasts on occasion (1 <unk> per week) to present convincingly as male, but mostly does not, and that he has been involved in an ongoing heterosexual romantic relationship as a male. He stated that this relationship has been very gratifying, and indicated concern about losing his girlfriend when he started testosterone. Although he stated that he wants to be viewed as male, L also stated that he did not look forward to the changes that testosterone would cause. Recommendations: Given that L was initially resistant to the initiation of cross-sex hormones, and that his mother initiated the consultation, along with L's ambivalence about the changes that testosterone would precipitate, cross-sex hormones were not recommended at this juncture. Instead, we recommended that L. continue to sort out his desires in his therapy relationship, while also addressing some of his other concerning behavioral and mental health issues. We also recommended family therapy, as it appeared that parental anxieties and pressures may have been impacting L's choices. We agreed to consult with L and his family again in 3 to 6 months. --- Case #4.: Autistic Spectrum Disorder Referral Information: B. is a 12 year old White European American natal male, Tanner stage 1, who has been increasingly presenting as female for approximately six months to one year. She and her parents presented in our clinic seeking an evaluation and recommendations for treatment. History: B. was diagnosed with high functioning ASD at the age of 7, after experiencing social difficulties for several years. Although intellectually bright, B. has not done well in school. B. spends much of her spare time on the computer, investigating various subjects and reporting the details to her parents. Her parents worry about her poor academic progress and her socialization, and she has been in treatment since her initial diagnosis. B. disclosed that she was a girl to her therapist and her parents 6 months earlier, after increasing depression and suicidal feelings. --- Psychological Evaluation: The evaluation revealed that B. strongly identified as female. B. stated that this feeling had begun within the past year at the start of the school year. Her parents indicated that they would support her if she were truly transgender, but expressed concern that B. may be unhappy socially and using a transgender diagnosis as a means to attempt to resolve her social isolation, and as a result of self-hatred. They also expressed concern that B.'s identification as female is a passing phase, similar to other passing phases/ obsessions she experienced throughout her life, rather than an enduring identification, and that B had limited understanding of the impact of changing genders. B.'s therapist was unsure of whether B. should be treated with hormones yet, expressing similar concerns to her parents. School reports indicated that B. was sometimes taunted by peers, apathetic about schoolwork, often inattentive, and increasingly isolated. All data consistently indicated depression and anxiety. Recommendations: Because of the complexities of B.'s situation, including a relatively recent identification as female, and limited social understanding, we recommended continued psychotherapy and monitoring of her GD, with treatment addressing her depression and anxiety, without immediate medical intervention. We also recommended that her therapist consult with her school to problem-solve solutions to isolation and bullying, and interventions to increase gratifying activities for B. outside the home. We recommended a psychiatric consultation for possible psychopharmacological intervention as well, and a return visit in 3 months to monitor B's progress and her gender identification in light of the new interventions. --- Supplementary Material Refer to Web version on PubMed Central for supplementary material.

Historically, many gender variant individuals have lived in a chronic state of conflict between selfunderstanding and physical being, one in which there was a continual misalignment between others' perceptions of them and their internal self-perception of gender. Only recently have professionals from mental health and medical realms come together to provide services to these youth. This paper describes an innovative program: the first mental health and medical multidisciplinary clinic housed in a pediatric academic center in North America to serve the needs of gender variant youth. We describe our model of care, focusing on the psychologist's role within a multidisciplinary team and the mental health needs of the youth and families assisted. We highlight clinical challenges and provide practice clinical vignettes to illuminate the psychologist's critical role.