Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Sociocultural determinants of HIV vulnerabilities among ethnic and religious minority men who have sex… Pan, Stephen Warren 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_february_pan_stephen.pdf [ 3.37MB ]
Metadata
JSON: 24-1.0221671.json
JSON-LD: 24-1.0221671-ld.json
RDF/XML (Pretty): 24-1.0221671-rdf.xml
RDF/JSON: 24-1.0221671-rdf.json
Turtle: 24-1.0221671-turtle.txt
N-Triples: 24-1.0221671-rdf-ntriples.txt
Original Record: 24-1.0221671-source.json
Full Text
24-1.0221671-fulltext.txt
Citation
24-1.0221671.ris

Full Text

 SOCIOCULTURAL DETERMINANTS OF HIV VULNERABILITIES AMONG ETHNIC AND RELIGIOUS MINORITY MEN WHO HAVE SEX WITH MEN IN NORTH CHINA by Stephen Warren Pan M.S.P.H, The University of Alabama at Birmingham, 2007B.A. (Government), The University of Texas at Austin, 2005 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Population and Public Health) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) December 2015 © Stephen Warren Pan, 2015 ii  Abstract Background: In China, 29% of new HIV cases are among men who have sex with men (MSM). Sociocultural factors may be influencing the HIV vulnerabilities of certain ethnic and religious minority MSM, but little such research exists in China. This dissertation seeks to understand how ethnicity and religion are shaping HIV vulnerabilities and testing among MSM in China.  Methods: Data were collected in three phases in Beijing and Tianjin, China from January 2013 – April 2014. Participants were self-identified MSM over 18 years old. In phase one, in-depth qualitative interviews were conducted with MSM of various ethnic and religious affiliations (n=41). Interview transcripts were analysed using an interpretive thematic approach. In phase two, findings from phase one informed the design of a structured questionnaire, which was pilot tested in two feedback sessions (n=8). In phase three, the structured questionnaire was used in a clinic-based cross-sectional survey (n=400). Statistical tests were conducted to assess if HIV infection, vulnerability, and testing were correlated with: ethnic and religious affiliation, acculturation, religiosity, and ethnic and religious discrimination. Inter-ethnic and inter-religious sexual partnerships were also assessed. Homonegativity was also assessed as a possible mediating variable between: (a) ethnicity and religion and (b) HIV-related outcomes.    Results: Ethnic and religious affiliations profoundly influenced individuals’ social reference groups. Religious social norms, codes of sexual morality, and beliefs in supernatural interventions appear to be inhibiting same-sex sexual behaviours among religious MSM, especially Muslims. Buddhists and ethnic minorities of South China had higher odds of various HIV vulnerabilities. Religiosity appears to have both positive and deleterious influences on HIV vulnerability. Sexual minority religious organisations can provide unique social support. Buddhists, Muslims, and ethnic Hui were more likely to report sex with their own religious and ethnic peers. Homonegativity did not account for any associations between ethnicity and religion and HIV-related outcomes.  Conclusion: This dissertation is the first study to explicitly explore the intersectionality of ethnicity, religion, sexuality, and HIV vulnerability in China. Findings highlight the ethnic and religious diversity among MSM in China and its salience for HIV prevention. Public health models of conceptualizing ethnocultural difference in China should be re-examined.  iii  Preface This statement is to confirm that the work presented in this dissertation was conceived, conducted, analysed, and written by Stephen Warren Pan (S.W.P.). S.W.P. conceptualized the study design, collected significant parts of the data, and performed the analysis of the data. With the guidance of supervisor Dr. Patricia M. Spittal, and committee members Drs. Martin T. Schechter, and Richard M. Carpiano, S.W.P. established the research objectives and hypotheses, designed the data collection instruments, conducted all of the analyses, and wrote each chapter. Drs. Yuhua Ruan, Yiming Shao, Zheng Zhang, and Dongliang Li also provided important guidance and support during the data collection phases of the project. The qualitative data presented in Chapter three were gathered by S.W.P. The quantitative data presented in Chapters four through six were gathered by S.W.P and study project staff affiliated with the Chinese Centers for Disease Control and Prevention and the Beijing Centers for Disease Control and Prevention. The study protocol for this research project received ethics approval from the National Center for AIDS/STD Control and Prevention, Chinese Centers for Disease Control and Prevention (Project number: X120717232, Co-PIs: S.W.P., Yuhua Ruan, and Yiming Shao), and the Providence Health Care Research Institute, University of British Columbia Research (REB certificate number: H12-00975, PI: Patricia M. Spittal). All non-original images in the dissertation have been reprinted with permission from the copyright holder, or do not require copyright permission.  iv  Table of Contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iii Table of Contents ......................................................................................................................... iv List of Tables ..................................................................................................................................x List of Figures ............................................................................................................................... xi Glossary ....................................................................................................................................... xii List of Abbreviations ................................................................................................................. xiii Acknowledgements .................................................................................................................... xiv Dedication .....................................................................................................................................xv Chapter 1: Introduction, background, & objectives ..................................................................1 1.1 Introduction ..................................................................................................................... 2 1.2 Intersectionality & HIV vulnerability ............................................................................. 3 1.3 Background ..................................................................................................................... 5 1.3.1 HIV/AIDS in mainland China: Epidemic and social transitions ............................ 6 1.3.1.1 Blood-borne transmission phase, 1980s to 1990s ............................................... 7 1.3.1.2 Sexual transmission phase, 2000s to present ...................................................... 8 1.3.2 Men who have sex with men in China .................................................................. 11 1.3.2.1 MSM life in China at the end of the twentieth century: precursor to the epidemic ......................................................................................................... 11 1.3.2.2 MSM since the 2000s: Dynamic sub-cultures amidst mainstream heteronormativity ........................................................................................... 12 1.3.2.3 HIV and MSM in China .................................................................................... 17 1.3.3 Ethnic affiliation and HIV .................................................................................... 22 1.3.3.1 Conceptualizing ethnic affiliation in mainland China ...................................... 23 1.3.3.2 Six ethnic groups in the PRC ............................................................................ 26 1.3.3.2.1 Han ................................................................................................................ 26 1.3.3.2.2 Manchus ........................................................................................................ 27 1.3.3.2.3 Mongols ........................................................................................................ 27 1.3.3.2.4 Uyghurs ......................................................................................................... 28 1.3.3.2.5 Hui ................................................................................................................ 29 1.3.3.2.6 Zhuang .......................................................................................................... 30 1.3.3.3 Ethnic affiliation and HIV in the PRC .............................................................. 31 1.3.4 Acculturation and HIV .......................................................................................... 32 1.3.4.1.1 Acculturation within the PRC ....................................................................... 34 1.3.4.1.2 Lack of HIV-related acculturation research in mainland China ................... 36 1.3.5 Religion and HIV .................................................................................................. 36 v  1.3.5.1 Conceptualizing religious affiliation in the PRC .............................................. 36 1.3.5.2 Religious social organisation and belief systems in the PRC ........................... 39 1.3.5.3 Religiosity in the PRC ...................................................................................... 40 1.3.5.4 Religious affiliation, religiosity, and sexual HIV vulnerability ........................ 41 1.3.5.5 Lack of information on religion and HIV vulnerabilities in China .................. 42 1.3.6 Ethnocultural discrimination ................................................................................. 43 1.3.6.1 Discrimination and HIV vulnerability .............................................................. 43 1.3.6.2 Ethnic and religious discrimination in the PRC................................................ 44 1.3.7 Ethnocultural minority MSM, HIV vulnerability, and HIV testing...................... 45 1.3.7.1 Ethnoracial affiliation & HIV epidemiology .................................................... 46 1.3.7.2 Acculturation & implications for HIV vulnerability and testing among MSM 47 1.3.7.3 Religion & HIV vulnerability and testing ......................................................... 48 1.3.7.4 Ethnocultural discrimination & HIV vulnerability ........................................... 49 1.3.8 Ethnocultural minority MSM and HIV vulnerability in the PRC ......................... 51 1.3.8.1 Ethnic minority MSM in the PRC .................................................................... 51 1.3.8.2 Religious minority MSM in the PRC ............................................................... 52 1.3.8.3 Ethnicity, religion, and HIV vulnerability among MSM in China ................... 53 1.4 Dissertation objectives and organisation ...................................................................... 55 Chapter 2: Methodology..............................................................................................................57 2.1 Conceptual model ......................................................................................................... 58 2.2 “MSM” terminology ..................................................................................................... 58 2.3 Study setting ................................................................................................................. 59 2.3.1 Beijing, the nation’s capital .................................................................................. 60 2.3.2 Tianjin, an overshadowed city of 14 million ........................................................ 60 2.4 Qualitative in-depth interviews ..................................................................................... 61 2.4.1 Maximum variation sampling ............................................................................... 61 2.4.2 Recruitment & enrollment .................................................................................... 63 2.4.3 Qualitative data collection .................................................................................... 65 2.4.4 Transcription ......................................................................................................... 66 2.4.5 Data analyses ........................................................................................................ 66 2.4.6 The candidate as the qualitative research instrument ........................................... 68 2.5 Questionnaire design ..................................................................................................... 69 2.5.1 Designing the survey data collection instrument .................................................. 69 2.5.2 Focus group pilot testing of the questionnaire ...................................................... 70 2.6 Quantitative survey ....................................................................................................... 71 2.6.2 HIV diagnostics .................................................................................................... 73 2.6.3 Data management.................................................................................................. 73 2.6.4 Measures ............................................................................................................... 73 2.6.4.1 Dependent variables .......................................................................................... 73 2.6.4.2 Independent variables ....................................................................................... 74 2.6.4.2.1 Ethnic affiliation & ethnic identity affirmation ............................................ 74 2.6.4.2.2 Religious affiliation & religiosity ................................................................. 75 2.6.4.3 Control variables ............................................................................................... 76 2.6.4.4 Mediator variables ............................................................................................ 76 vi  2.6.5 Quantitative analyses ............................................................................................ 76 2.6.5.1.1 Testing for correlations between independent and dependent variables ...... 77 2.6.5.1.2 Power calculation .......................................................................................... 77 2.6.5.1.3 Testing for mediation .................................................................................... 78 2.7 Ethical reviews and considerations ............................................................................... 79 Chapter 3: Narratives of intersecting ethnic, religious, and sexual identities: implications for HIV vulnerability and testing ...............................................................................................80 3.1 Introduction ................................................................................................................... 80 3.1.1 Ethnicity in the PRC ............................................................................................. 81 3.1.1.1 Han .................................................................................................................... 81 3.1.1.2 Manchus ............................................................................................................ 81 3.1.1.3 Mongols ............................................................................................................ 82 3.1.1.4 Uyghurs ............................................................................................................. 82 3.1.1.5 Hui .................................................................................................................... 83 3.1.1.6 Zhuang .............................................................................................................. 84 3.1.2 Religion in the PRC .............................................................................................. 84 3.1.3 Ethnicity, religion, and HIV among MSM ........................................................... 86 3.1.4 Ethnic and religious MSM in China ..................................................................... 88 3.2 Objectives and rationale ................................................................................................ 89 3.3 Methods ........................................................................................................................ 89 3.3.1 Study site: Beijing ................................................................................................. 89 3.3.2 Recruitment & enrollment .................................................................................... 90 3.3.3 Data collection ...................................................................................................... 91 3.3.4 Transcription ......................................................................................................... 91 3.3.5 Data analyses ........................................................................................................ 92 3.3.6 The researcher as the qualitative research instrument .......................................... 93 3.4 Results ........................................................................................................................... 94 3.4.1 Sociocultural worldview ....................................................................................... 94 3.4.1.1 Ethnic social identities ...................................................................................... 94 3.4.1.1.1 Devalued, yet undeniable ethnic identities ................................................... 94 3.4.1.1.2 Evolving attitudes towards ethnic identity .................................................... 96 3.4.1.2 Religious worldview ......................................................................................... 97 3.4.1.2.1 Sense of religious social identity .................................................................. 97 3.4.1.2.2 Omnipotent higher power ............................................................................. 99 3.4.1.2.3 Cosmic moral order .................................................................................... 100 3.4.2 Negotiating ethnicity, religion, and sexuality ..................................................... 101 3.4.2.1 Reconciling religious identity and sexual behaviours .................................... 101 3.4.2.1.1 Perceived incompatibilities of religion and sexuality ................................. 102 3.4.2.1.2 Identity integration ...................................................................................... 104 3.4.2.1.3 Religion in context: mainstream heteronormativity and homonegativity .. 107 3.4.2.2 Religion and sexual behaviours ...................................................................... 108 3.4.2.2.1 Defining the boundaries of a moral sex life ................................................ 108 3.4.2.2.2 Sexual inhibitions ....................................................................................... 110 3.4.3 Ethnocultural bias ............................................................................................... 113 vii  3.4.3.1 Cultural devaluation in heteronormative society ............................................ 113 3.4.3.2 Ethnocultural status & sex partner selection ................................................... 115 3.4.3.3 Uyghur men and social segregation ................................................................ 117 3.4.4 Discussion ........................................................................................................... 117 3.4.4.1 Ethnocultural affiliations, social norms, and sense of social identity ............. 118 3.4.4.2 Cosmic moral order, supernatural punishment, & sexual behaviour .............. 119 3.4.4.3 Acculturation and ethnocultural bias .............................................................. 122 3.4.4.4 Limitations ...................................................................................................... 123 3.4.4.5 Conclusion ...................................................................................................... 124 Chapter 4: Ethnicity, acculturation, and ethnic discrimination among men who have sex with men in North China: Implications for HIV vulnerability and testing..........................125 4.1 Introduction ................................................................................................................. 125 4.1.1 Ethnic affiliation & HIV vulnerabilities ............................................................. 125 4.1.2 Acculturation: implications for HIV vulnerability and testing ........................... 127 4.1.3 Discrimination and HIV ...................................................................................... 130 4.2 Objectives and rationale .............................................................................................. 133 4.3 Methods ...................................................................................................................... 133 4.3.1 Study setting: Beijing & Tianjin ......................................................................... 133 4.3.2 Four ethnic groups of North China ..................................................................... 134 4.3.3 Ethical reviews .................................................................................................... 137 4.3.4 Participant enrollment and survey data collection .............................................. 137 4.3.5 HIV diagnostics .................................................................................................. 138 4.3.6 Data management................................................................................................ 138 4.3.7 Measures ............................................................................................................. 139 4.3.7.1 Dependent variables ........................................................................................ 139 4.3.7.2 Independent variables ..................................................................................... 139 4.3.7.3 Control variables ............................................................................................. 140 4.3.8 Data analyses ...................................................................................................... 140 4.4 Results ......................................................................................................................... 141 4.4.1 Descriptive findings ............................................................................................ 141 4.4.2 Hypothesis 1: Association between ethnic affiliation and HIV vulnerabilities & testing ............................................................................................................... 150 4.4.3 Hypothesis 2: Association between participant ethnic affiliation and Ethnicity of sex partner ........................................................................................................ 151 4.4.4 Hypotheses 3-4: Association between ethnic identity affirmation and HIV vulnerabilities & testing; Association between ethnic identity affirmation × ethnic affiliation and HIV vulnerabilities & testing ........................................ 154 4.4.5 Hypothesis 5: Association between mandarin proficiency and HIV vulnerabilities & testing ........................................................................................................... 155 4.4.6 Hypothesis 6: Association between ethnic discrimination and HIV vulnerabilities & testing ........................................................................................................... 156 4.5 Discussion ................................................................................................................... 158 4.5.1 HIV vulnerabilities vary substantially by ethnic affiliation ................................ 158 4.5.2 Ethnic sexual mixing patterns do not occur at random ....................................... 160 viii  4.5.3 Ethnic identity affirmation associated with less sex with men ........................... 160 4.5.4 Language proficiency and ethnic discrimination each/respectively have no significant associations with HIV vulnerability or testing............................... 161 4.5.5 Limitations .......................................................................................................... 162 4.5.6 Re-conceptualizing indicators of ethnocultural difference ................................. 163 Chapter 5: Religious affiliation, religiosity, and religious discrimination among men who have sex with men in North China ...........................................................................................164 5.1 Introduction ................................................................................................................. 164 5.1.1 HIV & men who have sex with men in mainland China .................................... 164 5.1.2 Religious affiliation and religiosity in the People’s Republic of China ............. 165 5.1.3 Religious affiliation, religiosity, and religious discrimination: implications for HIV vulnerability ............................................................................................. 168 5.1.4 HIV, religion, and men who have sex with men ................................................. 170 5.2 Objectives and rationale .............................................................................................. 172 5.3 Methods ...................................................................................................................... 173 5.3.1 Study design and recruitment.............................................................................. 173 5.3.2 HIV diagnostics .................................................................................................. 174 5.3.3 Measures ............................................................................................................. 174 5.3.3.1 Dependent variables ........................................................................................ 174 5.3.3.2 Independent variables ..................................................................................... 175 5.3.3.3 Control variables ............................................................................................. 176 5.3.4 Data analyses ...................................................................................................... 176 5.4 Results ......................................................................................................................... 177 5.4.1 Descriptive findings ............................................................................................ 177 5.4.2 Hypothesis 1: Association between religious affiliation and HIV vulnerabilities & testing ............................................................................................................... 185 5.4.3 Hypothesis 2: Association between participant religious affiliation and sex partner religious affiliation ........................................................................................... 185 5.4.4 Hypothesis 3: Association between religiosity and HIV vulnerabilities & testing ......................................................................................................................... 188 5.4.5 Hypothesis 4: Association between religiosity × Religious affiliation and HIV vulnerabilities & testing ................................................................................... 190 5.4.6 Hypothesis 5: Association between religious discrimination and HIV vulnerabilities & testing ................................................................................... 194 5.5 Discussion ................................................................................................................... 194 5.5.1 HIV vulnerabilities vary significantly by religious affiliation ............................ 195 5.5.2 Religious sexual mixing patterns do not occur at random .................................. 196 5.5.3 Religiosity may have both protective and deleterious impacts on HIV vulnerability ..................................................................................................... 197 5.5.4 Directionality and magnitude of associations between religiosity and HIV vulnerabilities vary by religious affiliation ...................................................... 198 5.5.5 No evidence that religious discrimination increases HIV vulnerability or reduces testing practices................................................................................................ 198 5.5.6 Limitations .......................................................................................................... 199 ix  Chapter 6: Religion, ethnicity, homonegativity, and HIV vulnerabilities: a mediation analysis ........................................................................................................................................201 6.1 Introduction ................................................................................................................. 201 6.2 Methods ...................................................................................................................... 206 6.2.1 Study design and recruitment.............................................................................. 206 6.2.2 HIV diagnostics .................................................................................................. 207 6.2.3 Measures ............................................................................................................. 207 6.2.3.1 Dependent variables ........................................................................................ 207 6.2.3.2 Independent variables ..................................................................................... 207 6.2.3.3 Control variables ............................................................................................. 209 6.2.3.4 Putative mediator variables ............................................................................. 209 6.2.4 Data analyses ...................................................................................................... 210 6.3 Results ......................................................................................................................... 211 6.4 Discussion ................................................................................................................... 217 6.4.1 Limitations .......................................................................................................... 218 6.4.2 Conclusions ......................................................................................................... 219 Chapter 7: Recommendations & conclusion ...........................................................................220 7.1 Summary of study findings ......................................................................................... 220 7.2 Dissertation strengths .................................................................................................. 223 7.3 Unique contributions ................................................................................................... 224 7.4 Implications for research and policy ........................................................................... 226 7.5 Limitations .................................................................................................................. 228 7.6 Conclusion .................................................................................................................. 229 Bibliography ...............................................................................................................................230 Appendices ..................................................................................................................................256 Appendix A: Official ethnic minority groups of the People’s Republic of China (2010) ...... 256 Appendix B: Participant consent form .................................................................................... 257 Appendix C: Qualitative in-depth interview guide ................................................................. 261    x  List of Tables  Table 1.1: Select characteristics of six ethnic groups in the PRC ................................................ 31 Table 2.1: In-depth interview participant characteristics (n=40) .................................................. 63 Table 2.2 : Sociodemographics of focus group participants (n=8) ............................................... 71 Table 4.1: Survey participant characteristics, by ethnic affiliation ............................................ 144 Table 4.2: Survey participant characteristics, by ethnic identity affirmation ............................. 146 Table 4.3: Survey participant characteristics, by language use & ethnic discrimination ........... 149 Table 4.4: Ethnic affiliation and dependent variable correlations .............................................. 152 Table 4.5: Ethnic identity affirmation and dependent variable correlations ............................... 155 Table 4.6: Language proficiency, ethnic discrimination, and dependent variable correlations . 157 Table 5.1: Survey participant characteristics, by religious affiliation ........................................ 179 Table 5.2: Survey participant characteristics, by religiosity ....................................................... 181 Table 5.3: Survey participant characteristics, by religious discrimination ................................. 183 Table 5.4: Religious affiliation and dependent variable correlations ......................................... 186 Table 5.5: Religiosity and dependent variable correlations ........................................................ 189 Table 5.6: Religiosity and dependent variable correlations, by religious affiliation .................. 193 Table 5.7: Religious discrimination and dependent variable correlations .................................. 194 Table 6.1: Survey participant characteristics, by perceived external & internalized homonegativity ............................................................................................................... 212 Table 6.2: Mediation effects of perceived external homonegativity .......................................... 215 Table 6.3: Mediation effects of internalized homonegativity ..................................................... 216  xi  List of Figures  Figure 1.1: Map of China ................................................................................................................ 1 Figure 1.2: Distribution of select ethnic minority “homelands” in China .................................... 24 Figure 1.3: Dominant religions of China by county ..................................................................... 38 Figure 1.4: Proportion of Chinese citizens who feel that religion is rather important or very important in their life, by administrative region ............................................................... 41 Figure 2.1: Timeline of dissertation data collection ..................................................................... 58 Figure 2.2: Dissertation conceptual model with fictionalized sexual networks ........................... 58 Figure 2.3: Study project sites ...................................................................................................... 59 Figure 2.4: In-depth interviews by mode of recruitment (n=40) .................................................. 64 Figure 2.5: Conceptual model for mediation ................................................................................ 78 Figure 5.1: Predicted probabilities of unprotected anal intercourse by organizational religiosity and religious affiliation ................................................................................................... 190 Figure 5.2: Predicted probabilities of anal sex by private religiosity and religious affiliation ... 191 Figure 5.3: Predicted probabilities of circumcision by intrinsic religiosity and religious affiliation......................................................................................................................................... 192    xii  Glossary  Areligious For this dissertation, refers to an individual without any religious affiliation, while acknowledging that an individual with no religious affiliation may still engage in religious practices.  Heritage culture The traditional culture of an ethnic group and its constituent members.   Homosexual Social label for individuals with same-sex attraction.  Men who have sex with men Individuals who identify as a man and have anal or oral sex with other men. For this dissertation, sex may have occurred recently or in the distant past.   Tongzhi Social label commonly used in mainland China that refers to sexual minorities. Includes men and women with same sex attraction. Literally means “same purpose” or “same will” in English. Also means “comrade”.    xiii  List of Abbreviations ART Anti-retroviral therapy CBO Community based organisation CDC Centers for Disease Control and Prevention DUREL Duke University Religion Index HIV Human Immunodeficiency Virus MEIM Multigroup Ethnic Identity Measure MSM Men who have sex with men NGO Non-governmental organisation PRC People’s Republic of China STD Sexually transmitted disease UAI Unprotected anal intercourse UNAIDS Joint United Nations Programme on HIV/AIDS VCT Voluntary counseling and testing     xiv  Acknowledgements Sole authorship of this dissertation belies the hundreds of individuals whose inconspicuous support has buoyed me throughout my PhD program. To the intrepid men who so graciously agreed to participate in this project, thank you for sharing your lived experiences to make this work a reality. To my supervisor, Dr. Patricia Spittal, I will forever be indebted to you for your unflagging encouragement and mentorship as I navigated my career and the vicissitudes of life. You challenged me to break free of my comfort zones, and for that, I have become a more versatile investigator. My committee members, Drs. Richard Carpiano and Martin Schechter, have been well-springs of wisdom at every juncture during my doctoral training, and I am profoundly grateful for the time they so magnanimously invested in me. To my field supervisor, Dr. Yuhua Ruan, my public health work in China would never have been possible if not for your decision to mentor a wide-eyed 24-year old Master’s student in 2006. 谢谢. Guidance from Drs. Charlyn Black, Bonnie Henry, and Ying McNab were also of great benefit during the planning phases of the dissertation. Special thanks to Drs. Yiming Shao, Zheng Zhang, and Dongliang Li for providing critical in-kind support and input, and to Drs. Gary Poole, Leanne Currie, and Frank Wong for serving on my PhD examination committee and strengthening this dissertation.    I am immeasurably appreciative of the friends who so diligently assisted with key project activities in China: Yunhua Zhou, Yue Zhang, Yang Han, Xiaohui Li, Chunhui Li, Zheya Sun, Yunnan Xu, Xuebing Leng, Wei Kan, Lu Yin, Zhenhai Zhou, Xiao Qi, Xiao Ke, Xiao Fan, Hai Yue, and my brother, Daniel Goan. Many sage colleagues and friends also found time to offer insightful feedback as the project unfolded – such persons include Drs. Ernest Volinn, Yiliang Zhu, Joan Kaufman, Joseph Tucker, Neil Schmid, Gordon Shen, Kumi Smith, Han-zhu Qian, Sakura Christmas, Jenny Hsi, Dai Kojima, Mitchell Luo, Nora Kleinman, Diane Gu, Rod Knight, Alexis Crabtree, and Katherine Muldoon. Credit is due to the three extraordinary friends who meticulously commented on an earlier version of the dissertation and always made time for me: Willa Dong, Casey Miller, and David Shallcross. My enduring gratitude goes to each of you. Scores of other individuals also brought me much inspiration, camaraderie, and levity during my PhD program. In particular, Drs. Matthew Weins, Andrew Tu, Margo Pearce, Putu Duff, Angeli Rawat, Gareth Mercer, Andrea Goldson, Hongbin Zhang, Ben Brisbois, Guopeng Fu, Joyce Lee, Ute Carkner, Ben Wang, Beth Hensler, Alex Liautaud, Chris Laugen, Joseph Puyat, Alden Blair, Anton Friedman, Richa Sharma, Kate Jongbloed, David Zamar, Auntie Liu, Ling Guan, Katsushi Nakano, Aunt Jenny, Auntie Kim, and my sister, Maria Pan. Many others merit explicit acknowledgement, but in the interest of personal privacy, are left unnamed.   Funding for this dissertation project was provided by the US Fulbright program, the University of British Columbia, the Liu Institute for Global Issues, and the Chaoyang District and Chinese Centers for Disease Control and Prevention. The Canadian Institutes of Health Research also sponsored presentation of dissertation findings at multiple academic conferences.   To my father, Alfred Pan, your interminable dedication to your children and vocation has taught me what it means to be a professional, a father, and a husband. Mom, I miss you. To my sons, Taisei and Taiyo, you have already enriched my life in ways I never could have imagined. To my inimitable wife, Eri Sasaki: 頭が下がる.  Lastly, I give my utmost gratitude to God.  xv  Dedication     To Eri-chan   1   Chapter 1: Introduction, background, & objectives Figure 1.1: Map of China  (Central Intelligence Agency [cartographer], 2011)  2   1.1 Introduction Today, it is estimated that over 780,000 people in China are currently infected with HIV, equivalent to approximately 0.05% of the general population (L. Zhang, Chow, Jing, et al., 2013). The burden of HIV, however, is unevenly borne by certain key subpopulations. Men who have sex with men (MSM) represent only 0.3% of China’s total population (UNDP, 2012), but accounted for 29.4% of all new HIV infections nationwide in 2011(Ministry of Health of the People’s Republic of China, 2012). Moreover, despite committed HIV-prevention campaigns targeting MSM, HIV continues to spread inexorably among the MSM population in China (Ministry of Health of the People’s Republic of China, 2012). Between 2004 and 2014, the national HIV prevalence among MSM in China increased from approximately 1% to 8% (National Health and Family Planning Commission of The People’s Republic of China, 2015). It is important to recognise, however, that MSM in China are an ethnoculturally diverse subgroup with diverse HIV prevention needs and circumstances. Ethnic and religious minority groups in China have heterogeneous social networks and cultural norms that invariably shape the contexts within which HIV transmission occurs (Aral, Adimora, & Fenton, 2008; Friedman, Cooper, & Osborne, 2009). International research has demonstrated that the intersection of ethnicity, religion, and sexual minority identities can have profound effects on vulnerability to HIV and testing (Garofalo et al., 2014; C. S. Han et al., 2014; Maulsby et al., 2014; Mor, Grayeb, Beany, & Grotto, 2013; Santos et al., 2013; Sheehy et al., 2013; Toleran et al., 2013; Vu, Choi, & Do, 2011), but little is known about the HIV vulnerabilities of ethnic and religious minority MSM in China. In response, this dissertation seeks deeper understanding about how ethnicity and religion is influencing HIV vulnerabilities and HIV testing among MSM in two cities in North China.   3   In exploring the research question, this dissertation contributes novel and actionable knowledge that can: (a) inform future HIV prevention planning in China, (b) stimulate more nuanced, multi-dimensional understandings of ethnocultural and sexual identities in contemporary China, and (c) provide a Chinese perspective to the burgeoning literature concerning ethnocultural determinants of HIV vulnerability and health in general.  The remainder of this introductory chapter continues with a background overview of: “intersectionality” as a research framework, concepts of HIV vulnerability, the HIV epidemic in China, MSM in contemporary China, and the HIV implications of ethnicity, religion, and ethnocultural discrimination. Following this review of social, behavioural, and public health literature relevant to the dissertation, the chapter concludes by describing the objectives and organisation of the dissertation.   1.2 Intersectionality & HIV vulnerability The individual embodiment of multiple intersecting identities and social positions is a universal feature of human society. Statuses such as gender, sexual orientation, religious affiliation, and ethnicity overlap to mark unique social conditions and experiences distinct from any of its constituent identities in isolation. Expressions of each type of co-occurring social status are continually being shaped by each other, as well as the broader cultural contexts within which they are simultaneously embedded. Historically, however, public health research practices in general have tended to depict certain social identity categories as being mutually exclusive. For example, the influential regulatory document entitled, US National Institutes of Health Policy and Guidelines on the Inclusion of Women and Minorities as Subjects in Clinical Research, has elicited sharp criticism due to its implicit presentation of gender and ethnoracial  4   minorities as mutually exclusive categories, contrary to the lived human experiences of ethnoracial minority women (Bowleg, 2012).  In response to growing concern that individual health effects tied to certain social statuses are dramatically modified by other embodied social statuses, an emerging paradigm has begun to gain momentum within the public health literature: intersectionality. Tracing its roots back to black feminist studies in the US (Crenshaw, 1991), intersectionality has been described as a “theoretical framework for understanding how multiple social identities such as race, gender, sexual orientation, SES (socioeconomic status), and disability intersect at the micro level of individual experience to reflect interlocking systems of privilege and oppression (i.e., racism, sexism, heterosexism, classism) at the macro social-structural level” (Bowleg, 2012). Technically, all individuals possess multiple social identities (e.g., gender, nationality, ethnicity), but intersectionality focuses upon the overlap of individual identities and social positions associated with heightened social marginalization. Given the potential combinations of co-occurring social identities, intersectionality offers promise for new insights as a framework with which to approach health disparities research, including that of HIV vulnerability.  The terms “vulnerable” and “vulnerability” are commonly invoked within the field of HIV prevention research, but are nevertheless subject to myriad conceptual interpretations given their inherent ambiguity and complexity. Due partly to HIV “risk group” labeling conventions (N. Glick, Crystal, & Lewellen, 1994) and definitions of “vulnerable groups” as outlined in the 1964 World Medical Association Declaration of Helsinki (World Medical Association, 1964), there exists a strong propensity to conceptualize HIV vulnerability as an immutable attribute dichotomously applied to social groups and all their constituent members. While recognising the realities of epidemiological disparities between social groups that have and have not been  5   deemed an HIV “risk group”, this dissertation approaches HIV vulnerability as a dynamic and situationally-based concept measured on a continuum, rather than binary classification, and one which will naturally vary accordingly to within-group heterogeneities. Thus, social groups and their affiliated individuals are not “vulnerable” to HIV infection per se. For the present dissertation, “HIV vulnerability” is understood as a temporal condition created by interacting individual- and environmental-level factors that jointly elevate an individual’s susceptibility to HIV infection (Delor & Hubert, 2000). The intersectionality of gender, sexuality, ethnicity, and religion is of significant consequence for individual HIV vulnerabilities and population-level HIV disparities (Watkins-Hayes, 2014). Each social group represents a specific configuration of organisational structures, cultural norms, material resources, senses of identity, and prestige – all of which have implications for individual-level HIV vulnerability and HIV testing patterns (Adimora & Schoenbach, 2005; Kenyon & Zondo, 2011), and whose effects are simultaneously attenuated, enhanced, or modified by the other social groups to which individuals belong. Generally, the preponderance of HIV-related intersectionality research has centered on minority ethnoracial women and minority ethnoracial sexual minorities in North America (Watkins-Hayes, 2014). However, such research is severely lacking in mainland China, where HIV has claimed the lives of no fewer than 136,000 persons as of 2013 (National Health and Family Planning Commission of The People’s Republic of China, 2014), many of whom were undoubtedly among the most marginalized of Chinese society in more ways than one.  1.3 Background  At the end of 2011, it is estimated that over 780,000 people in the People’s Republic of China (PRC) were living with HIV, representing approximately 0.058% of the general  6   population (Ministry of Health of the People’s Republic of China, 2012). Though the HIV epidemic in China is now unlikely to become that of a generalized epidemic (defined as HIV prevalence >1% among pregnant women in antenatal care, (UNAIDS, 2011)), such figures mask the grossly disproportionate suffering of particular subgroups and geographic regions over the past thirty years. The following sections chronicle the epidemiological evolution of HIV/AIDS in China and provide an overview of ethnicity, religion, and MSM social life within contemporary Chinese society. 1.3.1 HIV/AIDS in mainland China: Epidemic and social transitions In 1985, AIDS was identified for the first time in the PRC (World Health Organization, 1986). Four years later, the first domestic cluster of HIV/AIDS cases was reported in Yunnan province (Jia et al., 2010), an ethnically diverse province in Southwest China bordering Myanmar and Vietnam. Since then, the HIV epidemic has spread inexorably to all 31 provinces, infected millions of individuals, and claimed well over a hundred thousand lives in China (Ministry of Health of the People’s Republic of China, 2012). However, the rapid proliferation of HIV throughout China over the past thirty years is attributed to far more than a novel pathogen’s evolutionary efficiency at viral replication. To be sure, China’s new socioeconomic liberalization policies in the late 1970s fostered a social environment increasingly conducive to the transmission of communicable diseases such as HIV.      The PRC was founded upon the social and economic tenets of Marxist-Leninism-Mao Zedong thought on October 1, 1949 (China, 1982). In pre-reform China (1949-1976), market capitalist principles were eschewed in favour of a planned economy, and as a consequence, domestic travel was fairly restrictive. Geographically constrained by a stringent household registration system (hukou) that precluded them from migrating or even obtaining food rations in  7   other parts of the country, the vast majority of Chinese citizens tended to live, work, and die within close proximity of their birthplace (X. Wu & Treiman, 2004). Living standards were relatively rudimentary compared to industrialized nations at the time, but virtually guaranteed employment and income protection by the state ensured a modicum of lifetime economic stability for most Chinese citizens (Leung, 1998, p. 618). Compounded by aggressive social campaigns to eliminate sexually transmitted diseases, limited mobility and a planned economy effectively helped reduce STI incidence to virtually zero (M. S. Cohen, Henderson, & Aiello, 1996; Dikötter, 1997; S. Pan, 2006). Likewise, mass anti-narcotics campaigns and implementation of austere legal repercussions for drug trafficking (e.g., execution) significantly cut the consumption and transport of illicit narcotics within a few short years after the PRC was founded (Lowinger, 1977; L. Lu, Fang, & Wang, 2008). However, in 1978, the Chinese central government began shifting its policies towards that of a free market economy and what would later be termed “socialism with Chinese characteristics” (China, 1982). As the effects of socioeconomic reform wore on over time, China’s government began systematically drawing down its commitments of lifetime economic security for all, and in turn, Chinese citizens were increasingly afforded new privileges of unencumbered domestic travel and accumulation of private wealth (Liang, 2001; X. Wu, 2001). It is at the dawn of this post-reform China that HIV made its indelible entry into China in two successive epidemiologic phases: the blood-borne transmission phase (from the 1980s to 1990s) and the sexual transmission phase (from the 2000s to the present). 1.3.1.1 Blood-borne transmission phase, 1980s to 1990s  During the early phases of the epidemic in the 1980s, the preponderance of HIV infections were attributed to injection drug use (IDU) in Southwest China, and were largely  8   sustained by opiates originating in a region of Southeast Asia commonly known as the “Golden Triangle” (Z. Wu, Rou, & Cui, 2004). Newly adopted economic policies in China accelerated narcotics consumption and illicit trade with Southeast Asia, and by 2000, IDU transmissions accounted for 67% of reported HIV incidence, with the vast majority of HIV cases being reported from five border provinces with large ethnic minority populations: Yunnan, Guangxi, Sichuan, Guangdong, and Xinjiang (The UN Theme Group on HIV/AIDS in China, 2002).    By the early 1990s, market-oriented economic reforms were well underway, and in the process, unprecedented pressures, as well as opportunities, arose for Chinese citizens to augment their income (Liang, 2001). In large rural swathes of central China, selling plasma quickly caught on as a popular means of easily supplementing personal income (The UN Theme Group on HIV/AIDS in China, 2002; Z. Wu, Liu, & Detels, 1995). Compared to whole blood extraction, selling plasma carried lower risks of acute anemia, thus enabling an individual to sell their plasma at more frequent intervals (The UN Theme Group on HIV/AIDS in China, 2002). Unfortunately, underdeveloped regulatory oversight mechanisms at the time were not commensurate with the vast proliferation of plasma collection enterprises, and unhygienic clinical practices directly exposed tens of thousands of plasma donors to equipment contaminated with HIV (Dou et al., 2010; Z. H. Qian, Vermund, & Wang, 2005; Z. Wu et al., 1995). Eventually, over 35,000 Chinese citizens contracted HIV via iatrogenic plasma donation (Dou et al., 2010), representing 10.7% of all HIV cases in China by 2005 (Ministry of Health of the People’s Republic of China, UNAIDS, & WHO, 2006). 1.3.1.2 Sexual transmission phase, 2000s to present  Due to pragmatic implementation of harm reduction measures (H.-Z. Qian, Schumacher, Chen, & Ruan, 2006; Sullivan & Wu, 2007), drug treatment programs (Yin et al., 2010), and  9   stricter regulation of plasma collection facilities (Henan Province Health Department, 2006; The National People’s Congress of the People's Republic of China, 1997), HIV incidence attributed to IDU and plasma donation declined, and the dominant modes of HIV transmission in China began rapidly shifting from blood-borne transmission to sexual transmission in the 2000s. Between 2006 and 2014, annual reported HIV incidence attributed to blood-borne transmission (IDU or blood/plasma donation) dropped from 51.1% to below 7.0%, while incidence attributed to sexual transmission (heterosexual or homosexual) rose from 33.1% to 92.2% (National Health and Family Planning Commission of The People’s Republic of China, 2015).  In addition to interventions related to blood-borne HIV transmission, underlying forces for the transition to a sexually-driven epidemic trace back to socioeconomic reforms of the late 1970s and the structural transformations in Chinese society that followed. First, the momentous shift towards a market-based economy helped commodify sex and incentivize individuals to enter the lucrative sex work trade, particularly those with low education or vocational skills (Sutherland & Hsu, 2012b, pp. 33–4). Unlike the pre-reform era which was characterized by stringent bans on sex work (M. S. Cohen et al., 1996), by the mid-2000s, female commercial sex workers in China had numbered between 2.8 and 4.5 million (F. Lu et al., 2006). Second, the easing of restrictions on domestic mobility enabled tens of millions of Chinese citizens to freely migrate from rural to urban areas, as well as between cities (Liang, 2001). As a result, migrant and sojourning Chinese citizens gained access to a significantly larger pool of potential sex partners with whom they could discreetly have relations with unbeknownst to members of their existing social or sexual network. Rural-to-urban migrants have consistently reported greater tendencies to engage in risky sexual behaviours such as unprotected sex compared to non-migrant counterparts (Xiaoming Li et al., 2007; Sudhinaraset, Astone, & Blum, 2012; X. Yang,  10   Derlega, & Luo, 2007). Nationwide, migrants accounted for 18.2% of all reported HIV cases in 2011, up from 10.2% in 2008 (Q. Qin et al., 2013). Among those who migrate unaccompanied by members of their established social network, riskier sexual behaviours also appear to be mediated by psychological distress attributed to social isolation in the urban environment (X. Yang et al., 2007). Compared to the general population, HIV prevalence is significantly higher among rural-to-urban migrants in the city (0.38% vs. 0.06%), as well as rural-to-urban migrants who returned back to their point of origin (0.18% vs. 0.06%) (L. Zhang, Chow, Jahn, Kraemer, & Wilson, 2013). Finally, fundamental changes in the interpretive meanings of sex and its perceived purpose appear to have increased social tolerance for a greater diversity of sexual practices (E. Y. Zhang, 2011). For example, in the six brief years between nationally representative surveys conducted in 2000 and 2006, the proportion of individuals who felt that laws criminalizing prostitution were excessively harsh increased from 13.9% to 26.7%, and the proportion of men who were indifferent to a prospective bride’s virginity increased from 5.9% to 22.4% (S. Pan, 2008). Such findings suggest that Chinese individuals may now be less deterred from engaging in sexual behaviours that previously elicited greater social stigma. In sharp contrast to the pre-reform Chinese government’s moral education and anti-prostitution campaigns that only sanctioned sex as a means to generate progeny within a monogamous heterosexual marriage, large-scale national surveys unequivocally indicate that sex in China over the past thirty years has increasingly become an activity of pleasure conducted outside of marriage (D. Liu, Ng, Zhou, & Haeberle, 1997; S. Pan, 2006; Parish, Laumann, & Mojola, 2007). According to the China Health and Family Life Survey, the proportion of those born between 1965 and 1980 who reported premarital sex with a commercial sex worker was over eight times higher than those born between 1945 and 1964 (8% vs. 1%), and married men born  11   in the 1960s were over twice as likely to have had a concurrent non-spousal sexual relationship than their counterparts born in the 1950s (21% vs. 8%) (Parish et al., 2007).  1.3.2 Men who have sex with men in China It would be difficult to overstate the magnitude of influence that China’s open reforms have exerted on the sexual minority community’s development. Analogous to its impact on sexual attitudes and practices within mainstream China, policies of socioeconomic liberalization fostered social and physical environments within which new sexual minority identities and communities could be formed. Also known as tongzhi in Chinese, sexual minorities in China, however, have not developed tongzhi communities without serious constraints.  Social stigma against homosexuality, or “homonegativity”, continues to prevail at nearly every level of Chinese society ranging from the central government to the family household, and poses a serious threat to HIV prevention and overall health among the estimated 2-10 million men who have sex with men (MSM) in China (F. Lu et al., 2006). More information about tongzhi terminology and identity can be found in the glossary and Chou (2001). 1.3.2.1 MSM life in China at the end of the twentieth century: precursor to the epidemic Reforms beginning in the early 1980s liberalized sexual attitudes and facilitated greater movement within China, but the taboo nature of male-to-male sex and lack of private spaces continued to severely inhibit discussions of  same-sex behaviours and dissuade a large proportion of men from acting upon their same-sex desires throughout the 1980s and much of the 1990s (W.-S. Chou, 2000, pp. 102, 105–6; Kong, 2011, pp. 153–4; Y. Li & Wang, 1992, p. 114). Loss of employment, demotion, damaged social standing, police harassment, criminal punishment, and electro-shock “therapy” were just a few of the possible consequences of inadvertently disclosing one’s same-sex sexual behaviours (W.-S. Chou, 2000, pp. 110–2; Y. Li, 2006, pp. 86– 12   94; Wan, 2001). “Homosexuality” was officially classified as a mental disorder by the Chinese Psychiatric Association (W.-S. Chou, 2000, p. 111), and same-sex sexual activities were potentially subject to criminal prosecution under the ambiguously defined offense of “hooliganism” (liu mang) (Bullough & Ruan, 1993, p. 47; W.-S. Chou, 2000, p. 108; Y. Li, 2006, p. 83). Such sentiments were also reflected in the general population as well. In 1989, a nationwide survey of college students in China demonstrated that 71% of respondents would advise their a friend to see a doctor if they self-identified as a “homosexual” (D. Liu et al., 1997, p. 193).  Among those intrepid enough to risk incurring such devastating repercussions, options for meeting other MSM were relatively scarce. Parks, bathhouses, and public restrooms served as essentially the only practical venues for men to seek out MSM  (Bullough & Ruan, 1993, p. 48,50; Y. Li & Wang, 1992, p. 159), thus constraining the scope and connectivity of MSM sexual networks. Moreover, for many men with same-sex attraction who had never been exposed to concepts of homosexuality and anal sex, notions of having sex with another man remained utterly foreign and beyond the scope of their imagination; as late as 1989, fully 35% of rural married residents had never even heard of homosexuality  (D. Liu et al., 1997, p. 288). Given the stigma and practical barriers to same-sex sexual relationships during the end of the twentieth century, it is perhaps unsurprising that between 1985 and 2005, only a small percentage of detected HIV cases were attributed to MSM transmission (Ministry of Health of the People’s Republic of China, 2012).  1.3.2.2 MSM since the 2000s: Dynamic sub-cultures amidst mainstream heteronormativity Towards the end of the 1990s, a profusion of same-sex cultural expression was clearly beginning to emerge from China’s increasingly vibrant tongzhi community, particularly in  13   popular migrant destination cities (Kong, 2011, p. 164; Rofel, 2007, p. 86). Recent estimates show that migrants represent 75.6% of the MSM in Beijing, 78.0% in Shanghai, and 88.2% in Shenzhen (J. T. Lau, Chow, Li, & Zhang, 2014). By the late 1990s, legitimate sexual minority bars were well established in major urban centers (W.-S. Chou, 2000, pp. 129–31; Rofel, 2007, p. 86). Moreover, MSM began circulating periodicals concerning same-sex issues throughout the country without interference (W.-S. Chou, 2000, p. 133), and numerous sexual minority advocacy organisations began operating nationwide programs for greater social acceptance of tongzhi and same-sex sexual behaviours (Moreno-tabarez et al., 2014). During this time, sexual attitudes in China had already begun to shift, and mainstream media also began to take notice of the tongzhi sub-culture. From pop culture talk shows to academic sexologists, MSM had entered the public discourse and at minimum appeared to be eliciting greater curiosity, if not less persecution from the general public (Kong, 2011, p. 155). Furthermore, over ten years of social reform and opening up had softened official discrimination of MSM, and by 1997, homosexual behaviour was no longer legally classified as a form of “hooliganism”; four years later, in 2001 homosexuality was delisted as a mental illness (UNDP & USAID, 2014). Regarding homosexuality, the government’s de facto approach has been characterized as the triple-no policy: no approval, no disapproval, and no promotion (Hongjie Liu, Liu, Cai, Rhodes, & Hong, 2009). With little doubt, much of the sexual minority community’s exponential growth beginning in the late 1990s could not have occurred without the transformative advent of the Internet and new modes of telecommunications (W.-S. Chou, 2000, pp. 134–5; Ho, 2010, pp. 104–7; Miège, 2009). Contrary to the preceding decades in which research about same-sex attraction and meeting MSM were logistically far more cumbersome, the Internet suddenly  14   offered unprecedented opportunities for men with same-sex attraction to discreetly learn about alternative sexual perspectives and engage in online discussions about the tongzhi community, as well as efficiently meet MSM from throughout the country. By 2004, there were already approximately 360 male tongzhi websites operating in China (Tong Ge, 2004, p. 190), and as Internet penetration expands in China, it is expected that more men with same-sex attraction will develop awareness about MSM-associated identities and social ties with MSM (Ho, 2010, p. 105; Miège, 2009). Naturally, social ties formed between MSM online sometimes led to sexual relationships (Q. Li et al., 2011; D. Zhang et al., 2007), thereby increasing the scope and connectivity of MSM sexual networks. In 2014, China’s leading mobile dating app for MSM, “blued”, had over 15 million users, making it the largest MSM dating app in the world (Larson, 2014). Based on social network studies that demonstrate how concurrent sexual partnerships catalyze the speed and scope at which sexually transmitted infections propagate in a population (Doherty, Padian, Marlow, & Aral, 2005; Doherty, Shiboski, Ellen, Adimora, & Padian, 2006; Potterat et al., 1999), it is not unreasonable to conclude that the Internet has indirectly accelerated the spread of HIV/STIs among MSM in China. On the other hand, the Internet and mobile technologies have also enabled HIV prevention messaging and interventions to reach millions of men with same-sex attraction (J. Li & Chen, 2012), particularly those living in rural areas devoid of MSM advocacy groups’ off-line messaging (Gao et al., 2011). Internet-based interventions appear to be a promising, if not necessary, platform for future scale-up HIV prevention messaging and testing among younger MSM in China (Chen, Wang, Xu, & Chen, 2011; Z. Wu et al., 2013; Zou et al., 2013) as mobile phones and growing wireless Internet coverage continues to rapidly expand. By the end of 2013, the China Internet Network Information Center estimated that China had approximately 618 million distinct Internet users  15   that year, representing 45% of  the national population (China Internet Network Information Center, 2014). The confluence of mobility, the Internet, and a market-based economy has also created space for a booming market of male commercial sex. Bluntly called “money boys” or “M.B.” within the sexual minority community, male sex workers in China are generally younger, more likely to identify as bisexual, and more likely to be non-resident migrants when compared with the overall MSM population (E. P. F. Chow, Iu, Fu, Wilson, & Zhang, 2012). For rural young men with limited education or vocational skills, sex work offers a lucrative opportunity to earn income that would be otherwise unimaginable. In 2006, a study of over 500 male sex workers in China concluded that on average, a typical sex worker could expect to receive 200-500 CNY (~25-63 USD as of June 30, 2006) for one anal sex session with a client, while male sex workers in high demand could receive 800-1000 CNY (~100-125 USD as of June 30, 2006) (Tong Ge, 2007). As a point of reference, the average monthly income in China in 2006 was 980 CNY for urban residents (~123 USD as of June 30, 2006), and just 299 yuan for rural residents (~37 USD as of June 30, 2006) (National Bureau of Statistics of China, 2007). Although male sex workers have greater numbers of sex partners than the overall MSM population, they are also more likely to use condoms; a meta-analysis from 2012 showed that HIV prevalence among male sex workers was comparable to that of the overall MSM population (E. P. F. Chow et al., 2012).     China’s sexual minority communities have developed rapidly and gained greater visibility in popular media, but to be sure, Chinese society has by no means become a bastion of sexual acceptance free of homonegativity. Despite revisions to the criminal code and professional psychiatric manuals, stigmatization of MSM among the general public persists. In 2013, a nationally representative survey by the Pew Research Center showed that 61% of  16   Chinese citizens still regarded homosexuality to be morally unacceptable, compared with only 13% who found homosexuality to be acceptable (17% of respondents did not consider homosexuality to be a moral issue) (2014). Although instances of overt enacted homonegativity may not occur frequently due to the inherently hidden nature of sexual-orientation, feelings of homonegative shame and perceived external homonegativity remain common among even urban MSM in China (Feng, Wu, & Detels, 2010; J. X. Liu & Choi, 2006; Miège, 2009; Steward, Miège, & Choi, 2013; Zhongxin, Farrer, & Choi, 2006). As a result, many MSM refrain from openly disclosing MSM activities and substantial numbers of men with same-sex erotic attraction attempt to modify their sexual-orientation, at times with the assistance of paid service providers (“‘Cures’ for homosexuality,” 2014; Zheng, Wu, Pang, Xiao, & Li, 2008).  Further complicating the challenges of homonegativity have been the issues of filial piety and gender roles, as MSM in China are often confronted with deeply engrained cultural expectations to enter conventional marriages and procreate on behalf of their parents, family lineage, and country (Kong, 2011, p. 160; Koo et al., 2014; Y. Li & Wang, 1992; Miège, 2009; Miller, 2013, pp. 162, 336; Steward et al., 2013). According to the immeasurably influential Chinese philosopher Confucius, “There are three types of filial impiety, having no descendants is the gravest” (bu xiao you san, wu hou wei da) (Lui, 2009). Hence, issues of homonegativity notwithstanding, MSM in China must often make excruciating personal decisions about how to reconcile personal same-sex sexual desires with conventional reproductive duties that they are expected to fulfill as men in China. Some have opted to avoid siring children indefinitely and risk incurring the displeasure of their family members, but a significant number of men have compromised by discreetly maintaining sexual relationships with other men after establishing a conventionally heterosexual family. According to a nation-wide survey of just under 50,000  17   MSM, 20.3% of MSM in China are currently married  (Z. Wu et al., 2013), while another multi-city study of 2,046 MSM demonstrated that only 30.6% of respondents did not intend to enter a conventional heterosexual marriage (B. Zhang et al., 2008). Preference for long-term, female-centered contraception methods such as the intrauterine device (39.6%) and female sterilization (33%) instead of condoms (4.3%) in China (Population Reference Bureau, 2008) also makes it difficult for married MSM to justify using male condoms without arousing suspicions of marital infidelity or MSM activities. Given the propensity for many married MSM to concurrently engage in unprotected vaginal sex with their wives (Koo et al., 2014), there is also growing concern about HIV vulnerabilities of the estimated 0.5 to 1.1 million women married to MSM in China (E. P. F. Chow, 2013).  1.3.2.3 HIV and MSM in China By all accounts, the HIV epidemic among MSM in China is large and rapidly spreading. In 2011, MSM are estimated to have accounted for 29.4% of all 48,000 new HIV infections nationwide (Ministry of Health of the People’s Republic of China, 2012) but only represented 0.3% of the country’s population (UNDP, 2012). Between 2001 and 2010, the rate of MSM transmission increased from 0.39 to 0.98 per 100 person-years, while HIV prevalence had risen from 1.77% to 5.98% (L. Zhang, Chow, Jing, et al., 2013) during the same time frame. In the city of Chongqing in Southwest China, HIV prevalence among MSM had already surpassed 10% by 2007 (Xiao et al., 2009).  The expeditious spread of HIV among MSM in China can be attributed to both social and biological factors. Sexual networks, homonegativity, and changing social norms alone cannot explain the alarming rate of HIV transmission among China’s MSM. Due primarily to the biological fragility of the rectum’s tissue lining, HIV is approximately 18 times more likely to be  18   transmitted by receptive anal sex than vaginal sex (Boily et al., 2009), a risk that extends equally to both males and females alike (Baggaley, White, & Boily, 2010; Beyrer et al., 2012; Boily et al., 2009). Thus, due to the efficiency of HIV transmission via unprotected anal intercourse (UAI) and the fact that many MSM are sexually versatile (i.e., practice both insertive and receptive anal sex), HIV is biologically able to proliferate far faster among MSM than non-MSM populations.  According to mathematical epidemic modelling, worldwide HIV incidence among MSM would be 50-98% lower if the per-act transmission rates of receptive anal sex were comparable to that of vaginal sex (Beyrer et al., 2012). Compelled by this confluence of intractable social and biological factors driving the HIV epidemic among MSM, in 2010, the directors of the Chinese Center for Disease Control and Prevention (CDC) and National Center for AIDS/STD Control and Prevention stated that: “Among the new HIV/AIDS challenges emerging in China, the expanding epidemic among MSM is undoubtedly the gravest of these new challenges regarding transmission of HIV” (Z. Wu & Wang, 2010).   One of the most disconcerting features of the HIV epidemic is that approximately 61 – 87% of HIV-infected MSM in China are estimated to be undiagnosed (E. P. F. Chow et al., 2014). Compounded by the fact that only 38% of MSM in China have received an HIV test in the previous 12 months (Zou, Hu, Xin, & Beck, 2012), due partly to homonegativity and HIV-associated stigma that deter MSM from accessing HIV testing and prevention services (Xuefeng Li et al., 2012; Santos et al., 2013; Y. Song et al., 2011; Wei et al., 2014; Z. Wu, Sun, Sullivan, & Detels, 2006), it follows that the majority of HIV-infected MSM in China are unable or unwilling to regularly access HIV testing services, and thus more prone to unwittingly transmit the virus via sexual intercourse. In response, the Chinese government and foreign donors have devoted considerable resources towards the nation-wide scale up of HIV testing among MSM   19   (E. L. Fan, 2014; Jacobs, 2009), an initiative which is subsumed under China CDC’s broader “Treatment as Prevention” HIV/AIDS prevention strategy (Woo, 2013). According to the treatment as prevention concept, anti-retroviral therapy (ART) regimens are used to minimize HIV viral loads for the maximal number of eligible HIV-positive persons in order to reduce the probability and rate of HIV-serodiscordant transmission at the dyadic and population level, respectively (J. Cohen, 2011). Fully subsidized ART regimens are available for all HIV-positive Chinese citizens eligible for treatment (F. Zhang et al., 2007), although research suggests that unregistered rural-to-urban migrants face significantly greater obstacles to ART access compared to those with official urban residency status (Todrys & Amon, 2009).     Ultimately, the success of HIV “treatment as prevention” strategies among MSM will rely heavily upon identification of undiagnosed HIV cases by testing. However, there exists several serious challenges to scale-up of HIV testing among MSM in China. First, intransigent homonegativity and stigma against AIDS continues to dissuade many MSM from uptake of HIV prevention and testing services; healthcare workers in hospitals and health departments are often perceived to be untrustworthy, liable to breach codes of confidentiality, and prejudiced against MSM (Feng et al., 2010; Y. Song et al., 2011; Wei et al., 2014; W. C. Wong, Zhang, Wu, Kong, & Ling, 2006) and people living with HIV (L. Li et al., 2013; UNAIDS, 2009; Z. Wu, Sun, Sullivan, Detels, & Rico, 2010; Y. Yang, Zhang, Chan, & Reidpath, 2005). In a sample of MSM in Beijing, 47% reported that they had not tested for HIV because of fears that their same-sex sexual activities might be made known to others (K.-H. Choi, Lui, Guo, Han, & Mandel, 2006). Second, large proportions of MSM in China reportedly do not test for HIV because they do not perceive themselves to be at serious risk of infection (E. P. Chow et al., 2013). Qualitative studies also indicate that many believe testing would connote unfaithfulness to a sexual partner  20   who expects monogamy (Wei et al., 2014). Third, misunderstandings about ARVs and HIV infection have misled many MSM to question the utility of HIV testing (Y. Song et al., 2011); in Beijing, 61.3% of previously undiagnosed HIV-positive MSM reported that the prohibitively expensive cost of HIV treatment was a reason that dissuaded them from testing for HIV (Xuefeng Li et al., 2011), despite the fact that ARVs are free for all HIV-positive Chinese citizens (F. Zhang et al., 2007).  Although the current HIV prevention discourse has taken a decided shift towards biomedical strategies such as “treatment as prevention” (“HIV treatment as prevention--it works.,” 2011; International Council of AIDS Service Organizations, 2009; Lancaster, Nguyen, Lesko, & Powers, 2013) and curative therapies (Deeks et al., 2012; International AIDS Society, 2012), condoms remain a critical and effective component in preventing sexual transmission of HIV. In a national survey of almost 50,000 MSM throughout China, odds of HIV infection among respondents who did not consistently use condoms with male partners during sex were 50% higher than those who did (Z. Wu et al., 2013). That said, condom usage varies substantially by partner type. Generally, consistent condom use among MSM in China has been lowest for “regular” sexual partnerships (19.9%), as opposed to non-commercial casual sexual partnerships (30.4%) (E. P. F. Chow, Wilson, & Zhang, 2011), partly due to beliefs that condoms signify unfaithfulness and would be unnecessary between monogamous sex partners (H. Li, Lau, Holroyd, & Yi, 2010; W. C. Wong et al., 2006). Diminished physical sensation, perceived low risk, and attenuated psychological pleasure are also commonly reported reasons MSM in China elect not using condoms (H. Li et al., 2010; W. C. Wong et al., 2006). Research within and outside China has also found consistent statistical associations between UAI and various forms of homonegativity among MSM, including internalized and perceived external homonegativity  21   (Andrinopoulos et al., 2014; K. Choi, Hudes, & Steward, 2008; Neilands, Steward, & Choi, 2008; Newcomb & Mustanski, 2011; Ross, Kajubi, Mandel, McFarland, & Raymond, 2013; Ross, Berg, et al., 2013). Internalized homonegativity refers to the internalization of heteronormative prejudices against sexual minorities by sexual minorities (Williamson, 2000), while perceived external homonegativity alludes to the expectation of stigmatization and discrimination based on sexual minority status (Meyer, 1995). Strong internalized homonegativity is linked to less frequent HIV testing (Pyun et al., 2014; Santos et al., 2013), and may be a function of reduced socialization with other MSM. Aversion to homosexuality may deter MSM from socializing with other MSM who may otherwise be important sources of information about HIV knowledge, prevention skills, and testing services (Huebner, Davis, Nemeroff, & Aiken, 2002). Aside from transfer of information, popular opinion leaders and peers within MSM social networks can establish positive norms surrounding safer sex and HIV testing (Kelly et al., 1991; Hongjie Liu et al., 2009). In the absence of such social ties and normative pressures, MSM may be more likely to engage in riskier sexual behaviours and less likely to test for HIV. MSM who harbor strong internalized homonegativity may also have less interest in developing steady sexual relationships with other men, and thereby may have more non-committal sexual partners (Stokes & Peterson, 1998). Prevalence of internalized and perceived external homonegativity appears to remain high among MSM in China. In a sample of 477 MSM in Shanghai, 89% had heard that homosexuals were not normal (Neilands et al., 2008) and 9% had seriously considered undergoing clinical treatment to alter their sexual orientation (UNDP & USAID, 2014).   In light of these challenges, MSM-affiliated community-based organisations (CBOs) have been contracted by local health departments throughout the country in an effort to  22   significantly boost HIV testing and condom promotion rates among MSM (E. L. Fan, 2014; Miller, 2013, p. 210; Z. Wu et al., 2010). Unlike CBOs, government health departments are rarely able to establish the antecedent community trust required to successfully recruit large numbers of MSM for HIV testing  (E. L. Fan, 2014; Miller, 2013, p. 239; UNDP and WHO, 2008). As of 2011, the United Nations Development Program estimates approximately 300 MSM-oriented CBOs and non-governmental organisations in China (2012).  However, the ability of CBOs to successfully build ties with MSM has been challenged by the diversity of assumed sexual identities and distinct sub-groups under the broad MSM rubric (Chapman, Cai, Hillier, & Estcourt, 2009; H. H. Li, Holroyd, & Lau, 2010; Tong Ge, 2005, p. 551). For example, HIV prevention strategies targeting enterprising transgender commercial sex workers may not be appropriate for the married men who discreetly maintain steady sexual relationships with other men. Researchers have begun to appreciate the complexity of sexual identities, gender role expectations, and sex work experiences among China’s MSM, and are now appealing for greater segmentation of China’s MSM HIV prevention strategies (Feng et al., 2010; J. T. Lau et al., 2014). Nonetheless, two major types of social identity have hitherto received little to no consideration in the voluminous MSM HIV prevention literature in China: ethnic and religious affiliation.  1.3.3 Ethnic affiliation and HIV Ethnic affiliations are socially constructed identities delineated by a political consensus that particular groups of people share some distinct degree of common history, practices, and/or values (Eriksen, 2010, pp. 15–7). Ethnic affiliations are often also a powerful predictor of individual vulnerability to HIV. Invariably, each potential incident of human-to-human HIV transmission is situated within particular sociocultural, historical, and geographical contexts  23   which indirectly affect the probabilities of transmission (Aral et al., 2008; Friedman et al., 2009). Since the beginnings of the global epidemic, such structural elements often engender skewed distributions of HIV disease and vulnerabilities at the societal level, whereby marginalized ethnic minorities endure disproportionate suffering due to HIV (Piot, Bartos, Ghys, Walker, & Schwartländer, 2001). The following sections briefly describe the evolution of ethnic affiliation and HIV in the PRC. 1.3.3.1 Conceptualizing ethnic affiliation in mainland China Shortly after the PRC was founded in 1949, the Chinese central government created a new national ethnic classification system which consolidated hundreds of self-identified ethnic minority groups based loosely on whether or not groups shared a “common territory, common language, common economic mode of production, and common psychology or culture” (Fei, 1981, p. 60; Mullaney, 2011, pp. 3, 11). Upheld for over 60 years with few modifications, the existing ethnic classification system in China has become well-established within bureaucratic systems, and powerfully shapes how Chinese citizens conceptualize ethnic affiliation and their senses of ethnic identity (Mackerras, 1994, p. 144; Mullaney, 2011, pp. 134–6). Figure 1.2 illustrates the geographic distribution of six official ethnic minority group “homelands” in China today.   24   Figure 1.2: Distribution of select ethnic minority “homelands” in China      (Adapted from Mackerras, 1995, p. xiii; Y. Zhou & Bloch, 2009) Currently, the PRC recognises exactly 56 official ethnic groups within China that range in population from several thousand (e.g., the Lhobo) to over 1 billion (ethnic Han majority) (National Bureau of Statistics of China, 2013). Collectively, China’s 55 ethnic minority groups total about 111 million persons (~9% of the national population) and are characterized by highly heterogeneous histories, religions, socioeconomic statuses, languages, diet, acculturation, and geographic “homelands”  (National Bureau of Statistics of China, 2013; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b). Most ethnic minority groups  25   have their own distinct spoken dialect, but the growing influence of Mandarin Chinese (the dominant and official spoken language in China) is rapidly marginalizing minority dialects in contemporary China (Shih, 2002a; M. Zhou, 2000a). Depending on location, some ethnic minority groups are also able to exercise semi-autonomous self-governance from the township to provincial level, whereby local ethnic minority communities essentially nominate and elect their own governing leaders (Wang, 2002). Unlike most polities within China, these special ethnic minority administrative areas also enjoy policies explicitly designed to protect cultural practices such as ethnic minority languages  (United Nations’ Committee on the Elimination of Racial Discrimination, 2009b).     Each Chinese citizen has one and only one official ethnic affiliation which is determined at birth according to the official ethnic status of their mother and/or father. Parents with discordant ethnicities are free to assign either the father or mother’s ethnic affiliation to the child at birth, although the child is able to change his/her ethnic affiliation to that of the other parent once they reach 18 years of age. No changes in official ethnic affiliation are permitted after age 20 (Mullaney, 2011, p. 123). Hence, official ethnic classification in China is notably dissimilar from that of North America, where self-reported and open-ended ethnic categories can change over time (Kaplan & Bennett, 2003). Generally, all officially recognised ethnic minorities in China enjoy special privileges such as exemptions from the one-child policy, relaxed testing criteria for admissions to high school and college, and subsidized school tuition from the government (Hesketh & Zhu, 1997; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b).      26   1.3.3.2 Six ethnic groups in the PRC Though the constitution of the PRC explicitly states that all ethnic groups in China are equal (China, 1982), the large number of ethnic minority groups and their differing population sizes have elevated the influence and national profile of certain groups over others. According to the 2010 national census, only 18 of the 55 ethnic minority groups had a population over one million, and 19 ethnic minority groups had populations below 10,000 (National Bureau of Statistics of China, 2013). The following sections provide a succinct and selective overview of six ethnic groups who hold strong presence within China’s national consciousness and who play prominent roles in this dissertation. 1.3.3.2.1 Han In terms of population and political clout, Han are the dominant ethnic group in mainland China and are closely associated with notions of “Chinese” culture and identity (Gladney, 1998, p. 11). Ethnic Han are indigenous to much of modern day China, save for several provinces that currently share land borders with other countries (e.g., Tibet, Xinjiang, Inner Mongolia, Heilongjiang, etc.). Socioeconomic development for ethnic Han in China has been inseparable from China’s social and economic reforms over the past 35 years, and though abject poverty remains, material standards of living have risen dramatically among ethnic Han overall (Sutherland & Hsu, 2012b, pp. 25–6). Confucian ideals of family lineage and gender roles retain strong influence within ethnic Han society, despite efforts by the government to eliminate Confucian influences during the first 30 years of the PRC (Adamczyk & Cheng, 2015; Steward et al., 2013). Buddhism and atheism are arguably the most popular official religious orientations among ethnic Han, but Christianity is also rapidly becoming adopted throughout urban and rural regions alike (“Cracks in the atheist edifice,” 2014; Pew Research Center, 2012). To be sure,  27   ethnic Han are far from a culturally monolithic social group (Gladney, 1998, p. 11). Regional distinctions such as dialect (e.g., Cantonese in the South) and mode of agricultural production (e.g., wheat in the North, rice in the South) have fostered unique cultural differences that have been scientifically tested and widely acknowledged in Chinese society (Talhelm et al., 2014).   1.3.3.2.2 Manchus Progenitors of present day ethnic Manchus originated from Northeast China and established China’s Qing dynasty (1644-1911) after the ethnic Han rulers of the Ming dynasty had been overthrown (Spence, 2013, p. 33). Ironically, the new governing Manchus began to adopt cultural practices of their ethnic Han subjects, and the Manchu language was eventually displaced by Chinese (Crossley, 1997, p. 127; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b). Similarly, Manchu shamanistic religious practices gave way in favour of Tibetan Buddhism during their centuries of rule (Dede, n.d.; Goossaert & Palmer, 2011, p. 32). Today, among the ten million-plus ethnic Manchus in China, there exist fewer than 50 native speakers of the Manchu language (Dede, n.d.). With generations of heavy intermarriage between Han and Manchus, ethnic Manchus of contemporary China have become culturally indistinguishable from Northern ethnic Han in all but name (Mackerras, 2003, p. 16).  1.3.3.2.3 Mongols Almost four hundred years prior to the founding of the Qing dynasty, ethnic Mongols established China’s Yuan dynasty (1271-1368) (Hanson, 2000, p. 331). As the Mongol empire expanded from present day Mongolia to what is now Tibet, ruling Mongols began learning and adopting teachings of Tibetan Buddhism, as opposed to Chinese variants of Buddhism (Morgan, 2007, pp. 109–10). Compared to the Manchus, cultural assimilation to Han culture has been historically more limited for the Mongols. As of 1994, 80% of ethnic Mongols could still speak  28   their heritage language (Mongolian), compared to less than 1% of ethnic Manchus (M. Zhou, 2003). However, heritage language and religion is fading among ethnic Mongols of contemporary China. Practice of Tibetan Buddhism has become relatively weak within contemporary Mongol culture in China (Goossaert & Palmer, 2011, p. 369), and the dominance of Mandarin Chinese in the economy is rapidly marginalizing the use of Mongolian, as well as other minority languages (Bulag, 2003, pp. 234–5). Yet, despite weakening Mongol cultural practices, certain ethnocultural stereotypes persist. Mongols’ tradition of nomadic pastoralism has distinguished them from most other ethnicities of China, and contributes to their reputation within China as a free-spirited people of the plains, analogous to evocative imagery of the American cowboy (Dede, n.d.). Relations between ethnic Mongols and Han throughout periods of the 20th century were marked by tensions concerning issues of political autonomy and national identity (Bulag, 2003, p. 229), but ethnic Mongol nationalism had largely subsided by the 21st century (Goossaert, 2011, p. 369). In contrast, while ethnocultural politics of discontent may have abated among other ethnic minority groups, relations between Uyghurs and ethnic Han remain heavily strained by mutual distrust.      1.3.3.2.4 Uyghurs Of China’s 55 ethnic minority groups, the Uyghurs of Xinjiang province are among the most culturally distinct and least assimilated into Han society (Mackerras, 2003, p. 16). In addition to their Central Asian features and Turkic-based language system which continues as the mother tongue of most today (Dwyer, 2005), the Uyghurs boast a long Islamic tradition tracing back to at least the 16th century (Gladney, 2004, p. 213). In reality, many ethnic Uyghurs may not be practicing Muslims who abide by religious rituals such as calls to prayer, but Muslim traditions are inextricable from Uyghur culture and sense of ethnic identity within China.  29   Unfortunately, recent concerns with violent Islamic extremism have prompted officials in Xinjiang province to issue restrictions on religious cultural practices such as fasting during Ramadan (“Beijing bans Xinjiang officials and schools from participating in Ramadan,” 2014). Controversial social policies such as this have fueled Uyghur discontent and exacerbated growing interethnic tensions over the past decade. In 2009, weeks of interethnic rioting in the provincial capital of Xinjiang eventually left no fewer than 197 dead and 1600 injured (“Innocent civilians make up 156 in Urumqi riot death toll,” 2009). Currently, this scale and intensity of interethnic strife appears unique to ethnic Uyghurs, and not simply a function of cultural discord with Islam in general (Bovingdon, 2010, p. 3). Relations with ethnic Han have remained relatively free of conflict among China’s most populous Muslim ethnic group: the Hui.    1.3.3.2.5 Hui The Hui are unique as an ethnic group in China because they lack their own distinct heritage language and primarily use the local language, most often Chinese (Gladney, 1998, p. 37). Essentially, the Hui ethnic classification was created based on lineage to Persian, Arab, Turkish, and Mongol migrants who settled throughout China prior to the 14th century and who retained their Islamic faith, but adopted local Chinese cultural practices such as attire and language (Gladney, 1998, p. 49). Due largely to their eclectic histories and geographic dispersal throughout most of the PRC, ethnic Hui of different regions engage in dissimilar cultural practices and subscribe to dissimilar notions of what defines Hui ethnicity. For example, ethnic Hui in North China closely link their ethnic status with being Muslim, while some Hui in Southeast China primarily associate ethnic status with lineage (Gladney, 1998, pp. 166–7). Consequently, conventional Islamic practices such as male circumcision vary widely among the Hui, and are generally less prevalent when compared with ethnic Uyghurs (Mackerras, 1995, p.  30   117). Nonetheless, in spite of their differences, ethnic Hui in China readily endorse the Hui ethnic classification and identify with other Hui as ethnic peers (Gladney, 1998, p. 167). Aside from one riot in central China between ethnic Hui and Han in 2004 (Kahn, 2004), relations between ethnic Hui and Han in the 21st century have for the most part been amicable and devoid of interethnic violence (Crane, 2014).    1.3.3.2.6 Zhuang Another prominent and culturally diverse ethnic group is the Zhuang, reportedly indigenous to Southwest China’s Guangxi and Yunnan provinces (Kaup, 2000, p. 27). Unlike the aforementioned ethnicities, the Zhuang lack a unified religion or language. Each Zhuang village pays homage to local deities which may or may not be recognised by ethnic Zhuang in neighbouring villages (Kaup, 2000, pp. 43–4), but Chinese Buddhism has also taken root among the Zhuang (Wickeri & Tam, 2011, p. 55). Distinct Zhuang language dialects also contribute to social divisions among the Zhuang, whereby Mandarin will often be the language of choice between Zhuang who speak different dialects (Kaup, 2000, p. 176). Hence, despite being China’s most populous ethnic minority group, the Zhuang possess relatively weak ethnic solidarity. Social development and material standards of living among the Zhuang generally lag behind ethnic Han – a trend that is common among many marginalized ethnic minorities in China (Kaup, 2000, pp. 150–2) – partly due to differing rates of urbanization (Guo, 2013, pp. 4–5) and uneven economic policies. For centuries, Zhuang, Uyghur, and Mongols alike have taken umbrage at exploitative economic policies that favour Han populations at the expense of their own ethnic groups, at times resulting in violent conflict (Kaup, 2000, pp. 166–7). Resentment against Han of course varies widely, as some urban ethnic Zhuang highly assimilated to Han-society perceive no discrepancies between ethnic Han and Zhuang (Kaup, 2000, pp. 176–7).   31    Table 1.1: Select characteristics of six ethnic groups in the PRC Ethnic group Population (2010) Traditional heritage Language  Proportion who spoke heritage language (1994) Proportion who spoke  Chinese language (1994) Traditional religious association Total per capita annual income among urban residents (2005; CNY) Proportion who live in cities and towns (2005) Han 1,225,930,000 Chinese n/a 100% Buddhism, Daoism, & Chinese folk religion 11,080 44% Manchu 10,387,958 Manchu <1% >99% Buddhism, Daoism, & Chinese folk religion 9,871 53% Mongol 5,981,840 Mongolian 80% 57% Tibetan Buddhism 9,654 48% Uyghur 10,069,346 Uyghur >99% <5% Islam 8,805 37% Hui 10,586,087 Chinese  n/a 100% Islam 9,718 40% Zhuang 16,926,381 Zhuang  97% 57% Shamanism and Animism 10,248 34%  (Adapted from Guo, 2013, pp. 4–5, 126–7; National Bureau of Statistics of China, 2013; M. Zhou, 2000b, 2003)  1.3.3.3 Ethnic affiliation and HIV in the PRC During the earlier phases of the HIV epidemic in China, ethnic minorities were highly overrepresented in HIV incident cases. By the year 2000, ethnic minorities only comprised 8% of the total population, but accounted for 36% of all reported HIV cases (Jing & Huan, 2010). Since the turn of the century, Han-ethnic minority HIV disparities have become less pronounced as the epidemic has spread beyond ethnic minority social circles and HIV surveillance has improved, but the toll of HIV has remained strikingly uneven between different ethnicities. Highly assimilated ethnic minorities such as Manchus have socioeconomic statuses and cultural practices/values comparable with that of majority Han (Mackerras, 2003) and there exists no evidence to suggest that they significantly differ from ethnic Han in terms of HIV burden. However, many less assimilated ethnic minority groups in Southern and Western China have  32   been grossly overrepresented in HIV burdens. Within some ethnic minority communities in Southwest China, HIV prevalence in the general population has risen as high as 11% (Dong et al., 2014), compared to 0.058% nationwide (Ministry of Health of the People’s Republic of China, 2012). Among female sex workers in Yunnan province, HIV prevalence among ethnic Zhuang  has been significantly higher than their Han counterparts (3.1% vs 1.5%) (Jia et al., 2010). Regarding people who inject drugs in Xinjiang province, HIV prevalence among ethnic Uyghurs has been shockingly higher than that of their Han counterparts (41% vs. 14%) (Y. Zhang et al., 2007). One possible factor for such interethnic HIV disparities is ethnic acculturation. 1.3.4 Acculturation and HIV At the individual level, acculturation can be understood as “the changes that take place as a result of contact with culturally dissimilar people, groups, and social influences” (Schwartz, Unger, Zamboanga, & Szapocznik, 2010). Specifically, HIV-related acculturation research has identified cultural practice (e.g., language usage) and cultural identity (e.g., ethnic identity affirmation) as two distinct acculturation domains with distinct effects on HIV vulnerability (Schwartz et al., 2011, 2014). For example, among Hispanic adolescent males in the US, Hispanic cultural practices was associated with fewer oral sex partners, but ethnic identity affirmation was actually associated with more partners (Schwartz et al., 2014). Moreover, cultural practices and identity do not necessarily correlate. It is entirely plausible that an individual can exhibit weak heritage cultural practices, but simultaneously have strong affirmation of their heritage ethnic identity. Such findings suggest that combining cultural identity and practices into a single measure may attenuate qualitatively unique measures of effect.  33   Language is a popular measurement of cultural practice in the public health sciences (Abraído-Lanza, Armbrister, Flórez, & Aguirre, 2006), and has strong potential to impact ethnic minorities’ HIV-related knowledge, attitudes, and practices. For example, poor proficiency in the dominant culture’s language can constrain “the extent to which people can access health information, communicate their health problems and questions to healthcare providers, understand health instructions, and participate in interventions” (Unger & Schwartz, 2012, p. 353). Studies have shown that Hispanics in the US who preferred using Spanish over English were less likely to utilize HIV-related health services than those who preferred English (Kinsler et al., 2009; Wohl, Tejero, & Frye, 2009). Ethnic identity affirmation refers to the degree one values and feels attached to the ethnic group(s) which they claim membership (Schwartz et al., 2010). Studies with heterosexual ethnic minority groups in the US and South Africa have indicated that ethnic and racial identity affirmation may inhibit riskier sexual behaviours such as unprotected sex (Beadnell et al., 2003; Nyembezi et al., 2014; Siddiqui, 2005), earlier sexual debut (Jarrett, 2011), and sexual concurrency (Oparanozie, Sales, DiClemente, & Braxton, 2012). Although the mechanisms of association remain inconclusive, evidence suggests that ethnic identity affirmation may not directly promote healthier behaviours, but rather acts as a buffer against negative stress by enhancing coping skills, self-esteem, and sense of belonging (Jarrett, 2011; Smith & Silva, 2011). The relationship between ethnic affirmation and HIV vulnerabilities, however, can vary significantly by gender, sexual orientation, ethnic group, social context, and historical experiences. To date, few if any studies have examined the HIV-related implications of language or ethnic identity affirmation in China.       34   1.3.4.1.1 Acculturation within the PRC  Within most urban settings in mainland China, Han represent the dominant ethnic group with whom most ethnic minorities and ethnic minority groups negotiate their degree of acculturation. Chinese language proficiency is arguably one of the most conspicuous examples of Han ethnocultural practice, and in fact, the term “Han language” (han yu) is commonly used interchangeably with “Chinese” (zhong wen). Naturally, Chinese serves as the native language for ethnic minorities who lack their own distinct heritage language (e.g., ethnic Hui) or whose heritage language has become socially obsolete (e.g., ethnic Manchus). Ethnic minorities raised in majority Han social environments also typically acquire strong Chinese literacy. By contrast, mastery of Chinese can be considerably more challenging for ethnic minorities who have limited Chinese language exposure in daily life. Chinese proficiency has tended to be stronger among ethnic minority Hui and Manchus, but lower among ethnic Uyghurs, Zhuang, and Mongols (M. Zhou, 2000b). Another important acculturation measure of ethnic Han practice is the nature and extent to which social ties develop between an ethnic minority person or ethnic minority group and the Han majority. Given the ubiquity of ethnic Han within Chinese society and institutions of influence (e.g., health departments and social advocacy organisations), interpersonal connections with ethnic Han have immense consequences for personal social support and access to resources within China. Factors influencing interethnic social ties are complex and can range from local sociodemographics (e.g., the ratio of Han to ethnic minority) to personal ethnic biases, but certain ethnic minority groups have emerged as being more prone to social segregation along ethnic lines. Due to strained interethnic relations, social distance between ethnic Uyghurs and Han is arguably the most pronounced of any interethnic relationship within China. One survey in  35   a Western Chinese city found that although 79% of the population was ethnically Han, 19% of Uyghur respondents reported not having a single ethnic Han friend (Yee, 2003).  Heritage ethnic cultural practices and senses of ethnic identity vary widely both between and within ethnicities. Regarding preservation of cultural practices, certain ethnic minorities have been more successful than others at maintenance of heritage language use, as previously mentioned. However, due to China’s national education policies and transition to a market-based economy, the general trend across all ethnicities has been declining usage of heritage languages with each successive generation (M. Zhou, 2003). Migration, interethnic marriages, and pro-assimilation social policies have also made it more difficult for some ethnic minorities to preserve heritage cultural values such as religious faith. For example, Islamic belief is commonly cited as a defining characteristic of ethnic minority Uyghurs and Hui, but historical small-scale migration to Han-dominated regions has often led to loss of Muslim religious beliefs and cultural practices such as taboos surrounding consumption of pork (Gladney, 1998; Shih, 2002b). But despite the lack of Islamic faith, such communities remain classified and strongly identify as ethnic Muslim minorities. Interestingly, a growing sense of ethnic consciousness has been developing among many – if not most – of China’s ethnic minority groups, not just those of an Islamic religious persuasion. In a recent multi-site survey of ethnic minority adolescents in far West China, ethnic Hui, Mongol, and Uyghur groups each reported a stronger sense of closeness to their ethnic group than ethnic Han (Tang & He, 2010). Similarly, another survey among adults in West China demonstrated that 91% of Uyghurs reported being proud of their ethnicity, compared to just 67% of ethnic Han respondents (Yee, 2003).    36   1.3.4.1.2 Lack of HIV-related acculturation research in mainland China Despite myriad ethnic minorities and languages, HIV-related acculturation research in mainland China remains sparse. Few, if any, HIV-related studies from mainland China which have explicitly sought to measure effects of ethnic acculturation in either English or Chinese. On April 9, 2015, a Chinese language search for “(acculturation AND HIV)” (wenhua shiying, aizibing bingdu) on wanfangdata.com.cn yielded just one article qualitative study of HIV in an ethnic minority village in Northwest China (N. Zhang, 2008). Nonetheless, some evidence suggests that acculturation is highly pertinent to HIV vulnerability among certain ethnic minorities in China. Up until the late 2000s in Xinjiang province, official dissemination of HIV prevention information was typically only conducted in Chinese. Naturally, such health promotion interventions would have been ineffective at augmenting HIV prevention knowledge, attitudes, and practices for the many marginalized Uyghurs who lacked sufficient Chinese comprehension (A. Hayes, 2012).       1.3.5 Religion and HIV 1.3.5.1 Conceptualizing religious affiliation in the PRC Historically, most denizens of dynastic China engaged in fluid Daoist, Confucian, and Buddhist religious practices with little if any notions about mutually exclusive religions and discrete religious identities. Such syncretistic practices are sometimes referred to as “popular Chinese religion” (minjian xinyang) (Goossaert, 2011, pp. 22–3). However, beginning with its introduction from the West in the early 20th century and formally institutionalized by the Chinese government’s Religious Affairs Bureau in the 1950s (Goossaert, 2011, pp. 152–4), concepts of religion (zongjiao) and religious faith (zongjiao xinyang) have gained increasing traction within the consciousness of Chinese society over the past century (Goossaert, 2011; Stark & Liu, 2011).  37    The Marxist-Leninist principles upon which the Chinese Communist Party was founded are unequivocal in its commitment to atheism. According to the Chinese Communist Party’s Official Policy Statement on Religion, “We Communists are atheists and must unremittingly propagate atheism” (Chinese Communist Party, 1982). Nonetheless, leaders within the Communist Party recognised that sweeping attempts to abolish religion were impractical and counterproductive to their efforts at consolidating political support during the formative years of the PRC. Hence, in 1954, the Communist Party established the Religious Affairs Bureau to oversee policies and activities of what would become the five officially state-sanctioned religions: Islam, Buddhism, Protestantism, Daoism (also known as Taoism), and Catholicism (Goossaert & Palmer, 2011, p. 153). Public expressions of religion were later severely repressed during the Cultural Revolution in the 1960s and 1970s (F. Yang, 2012, pp. 72–3), but since the early 1980s, Chinese citizens have essentially been free to affiliate with and openly practice “normal religions” as defined by the state, so long as activities are conducted within officially state-sanctioned religious apparatuses and are not seen as provoking social unrest (Laliberté, 2011; F. Yang, 2012).   38   In conjunction with China’s broader socioeconomic reforms of the late 1970s, these religious freedoms have helped spur exponential growth in religious affiliations among Chinese citizens. Between 1982 and 2009, the number of Protestant clergy rose from ~5,900 to ~37,000; during this same time period, the number of Islamic clergy doubled (~20,000 to ~40,000), while the number of Buddhist monks/nuns multiplied by over seven-fold (~27,000 to ~200,000) (F. Yang, 2012). Definitive population estimates remain elusive, but official reports from the Chinese government approximate 16 million Protestants (1.2% of the national population), 21 million Muslims (1.6%), and 5.3 million Catholics (0.4%) (United Nations’ Committee on the Elimination of Racial Discrimination, 2009b) in China as of 2009, though a survey commissioned by the Pew Research Center estimated as many as 68 million Christians (5.1%) and 244 million Buddhists (18.2%) (2012). Figure 1.3, which illustrates the dominant religions at the county-level, clearly indicates the extensive spread of Christianity in China, as well as the tendency for members of a common religious affiliation to cluster geographically. Reprinted from [Why are China’s religions so hot?], In Sina.com.cn, n.d., Retrieved April 15, 2015, from http://history.sina.com.cn/cul/zl/2014-07-04/142294628.shtml. © 2015. The Center for Religion and Chinese Society at Purdue University. Reprinted with permission.  Figure 1.3: Dominant religions of China by county  39   1.3.5.2 Religious social organisation and belief systems in the PRC In addition to serving as a form of social identity, personal religious affiliation is often a marker for specific religious organisational structures and beliefs. Generally, religions in China can be crudely classified into two categories: churched religions and non-churched religions (Stark, Hamberg, & Miller, 2005). Churched religions are characterized by congregations of members who subscribe to explicit institutionally sanctioned religious creeds. In China, Islam, Catholicism, and Protestantism constitute the largest churched religions, and each maintains its own religious codes of moral comportment. Given that members are typically expected to believe specific religious creeds and partake in social religious activities, churched religions naturally have considerable capacity to instill a strong sense of religious identity and exert social pressures among their congregants. In effect, the religious community to which an individual claims affiliation with serves as an important social reference group (Bock, Beeghley, & Mixon, 1983). In contrast, non-churched religions impose far fewer expectations on its members. Buddhism and Daoism are the main unchurched religions officially recognised by the Chinese state, and neither requires its members to acknowledge specific religious creeds or partake in religious social activities. Buddhist and Daoist venues (e.g., temples and shrines) primarily cater to individualized religious practices and are thus less well suited for supporting religious communities and religious social norms (Eric Y Liu, 2011, p. 145).  Besides differences in social organisations, religious affiliations in China fundamentally differ in their views of the supernatural. Christianity and Islam both espouse beliefs in an omnipotent higher being consciously concerned with the moral rectitude of humanity (i.e., the Lord or Allah), while Buddhism and Daoism are religions primarily founded upon beliefs in a “natural order” to the universe driven by karma and “the way,” respectively. Essentially, the core  40   principle of karma holds that “deeds by all forms of sentient beings create life circles and consistently influence past, present, and future experiences” (Eric Y Liu, 2011, p. 144). Daoism can be understood as “the paramount force behind the natural order that keeps the universe ordered and balanced” (Eric Y Liu, 2011, p. 144). As neither Daoism nor Buddhism demand exclusive religious recognition of its members, the distinction between self-proclaimed Daoists and Buddhists in China are often blurred, with worldviews and practices being routinely and unconsciously blended (L. Fan & Whitehead, 2011, p. 21). 1.3.5.3 Religiosity in the PRC While individual religious affiliation is defined by membership or non-membership in a religious social group, religiosity is an inherently multi-dimensional construct that can be conceptualized as “a level of adherence to, participation in, influence of, or identification with a set of [religious] beliefs” (Shaw & El-Bassel, 2014). Within the sexual HIV prevention literature, religiosity has been commonly operationalized as religious faithfulness (i.e., importance of religion in one’s life), frequency of participation in religious activities, influence of religion during childhood/adolescence, influence of religion on behaviour, and self-appraised degree of being religious (Shaw & El-Bassel, 2014). As with any religiously pluralistic society, manifestations of religiosity in China vary by religious affiliation. For example, frequency of participation in religious activities might include sutra chanting for Buddhists, Bible studies for Christians, or Friday prayers for Muslims.  In recent decades, religiosity in China has increased dramatically. As a proxy national metric of participation in religious activities, the number of Buddhist/Daoist temples increased from ~14,557 in 1995 to ~23,000 in 2009; within that same time frame, the number of Christian churches and meeting points rose from ~41,377 to ~64,000 (F. Yang, 2012). Due to both shifting  41   concepts of religion and greater religious faithfulness, Chinese citizens increasingly view religion as an important element of their lives (Figure 1.4).  Figure 1.4: Proportion of Chinese citizens who feel that religion is rather important or very important in their life, by administrative region  (Adapted from World Values Survey Association, n.d.) 1.3.5.4 Religious affiliation, religiosity, and sexual HIV vulnerability Religious affiliation and religiosity wield strong potential to influence individual and group-level sexual HIV vulnerabilities via myriad pathways (Shaw & El-Bassel, 2014). Religious affiliations delineate boundaries of distinct organisational social structures and cultural norms between given religious groups, and have been shown to correlate with population HIV prevalence, individual HIV infection, and riskier sexual behaviours (Shaw & El-Bassel, 2014). Lower HIV prevalence and HIV risk among Muslim communities and Muslim individuals have respectively emerged as the most consistent finding among international research concerning HIV and religious affiliation (P. B. Gray, 2004; Nattrass, 2009; Obermeyer, 2006; Shaw & El-00.10.20.30.40.51990 2001 2012Jiangsu provinceLiaoning provinceBeijing municipalityHubei provinceShanghai municipalityGuizhou provinceFujian provinceShaanxi province 42   Bassel, 2014), while effects associated with Christian, Hindu, and other sub-denominational affiliations have been more varied by setting and study (Shaw & El-Bassel, 2014). Few epidemiological studies have explicitly examined the relationship between Buddhist religious affiliation and HIV vulnerabilities, and no consistent trends have yet emerged (Adamczyk & Hayes, 2012; Visser, Smith, Richters, & Rissel, 2007).  Religiosity has generally predicted lower likelihood of HIV infection and riskier sexual behaviours, but the magnitude of statistical effects vary by how religiosity is operationalized (Shaw & El-Bassel, 2014). Mechanisms of association between religiosity and sexual HIV vulnerability are not yet fully understood, but religious rituals and codes of moral conduct clearly play important roles. For example, research has indicated that lower HIV prevalence among Muslim communities may be partly attributed to firmly established social norms proscribing sexual concurrency and consumption of alcohol, and requiring the practice of male circumcision (Adamczyk & Hayes, 2012; Badri, 2009; P. B. Gray, 2004; Obermeyer, 2006), all factors which have been associated with HIV vulnerability (Bailey et al., 2007; Baliunas, Rehm, Irving, & Shuper, 2010; R. H. Gray et al., 2007; Kalichman, Simbayi, Kaufman, Cain, & Jooste, 2007; Woolf & Maisto, 2009).   1.3.5.5 Lack of information on religion and HIV vulnerabilities in China In light of the growth of religion in China and international research linking religion to HIV, there is strong reason to believe that religion is profoundly shaping the HIV epidemic within China. However, literature concerning the influence of religious affiliation or religiosity on HIV vulnerability in China remains underdeveloped and limited to incidental findings of a few studies (Albrektsson, Alm, Tan, & Andersson, 2009; Deng, Li, Sringernyuang, & Zhang, 2007; Hyde, 2007). One ethnographic study of a rural Theravada Buddhist community in  43   Southwest China suggested that beliefs in reincarnation may indirectly discourage behavioural change among people who inject drugs. In theory, individuals who believed in reincarnation were less motivated to modify behaviours deemed as religiously immoral (e.g., injecting drugs) because they believed that adopting more virtuous behaviours could be postponed until their next Earthly reincarnation (Deng et al., 2007). In addition to religion and religiosity, another prominent lacuna in HIV prevention research in China is the relationship between ethnocultural discrimination and HIV vulnerabilities.  1.3.6 Ethnocultural discrimination 1.3.6.1 Discrimination and HIV vulnerability Discrimination has been defined as a “behavioural manifestation of a negative attitude, judgement, or unfair treatment toward members of a group” (Pascoe & Richman, 2009), and can occur at the interpersonal and/or structural levels (Link & Phelan, 2001). Interpersonal discrimination includes personally-mediated acts of devaluation and exclusion, while structural discrimination involves social norms and institutionalized practices which undermine the welfare of stigmatized groups, with or without interpersonal discrimination (Angermeyer, Matschinger, Link, & Schomerus, 2014). Acts of interpersonal and structural discrimination can be committed intentionally or unintentionally, and may or may not be perceived by the putatively discriminated individual (Jones, 2001).  Current research indicates that discrimination based on race/ethnicity, religion, and sexual orientation has negative impacts on a broad spectrum of health indicators including substance use, psychological distress, and HIV sexual vulnerability (Krieger, 2014; Rippy & Newman, 2006). Compelling empirical evidence suggests four possible mechanisms linking ethnocultural or sexual orientation discrimination and HIV vulnerability. First, instances of  44   perceived discrimination can trigger psychological distress which prompts negative coping strategies such as unprotected sex with casual partners (Díaz, Ayala, & Bein, 2004; Earnshaw, Bogart, Dovidio, & Williams, 2013).  Second, discrimination can constrain access to important HIV prevention information and services. For example, discriminatory language policies can impede access to HIV prevention literature among linguistic minority groups (Wohl et al., 2009), while perceptions of discrimination can dissuade marginalized ethnocultural minorities from establishing contact with health service institutions associated with the dominant social group (Earnshaw et al., 2013).  Third, ethnocultural discrimination can create interpersonal power imbalances that disadvantage discriminated individuals leading up to and during sexual situations, whereby safer sex becomes harder to negotiate and practice (C. S. Han et al., 2014; C. Han, 2008).  Fourth, ethnic/racial discrimination when selecting sex partners can create ethnoculturally segregated social and sexual networks, and constrain one’s ability to partner with members outside their ethnocultural group. If HIV prevalence varies between such ethnoculturally segregated sexual networks, then members of disparate ethnocultural groups may have different risks of exposure to HIV, independent of individual sexual behaviours (Maulsby et al., 2014; B. Mustanski, Birkett, Kuhns, Latkin, & Muth, 2014).  1.3.6.2 Ethnic and religious discrimination in the PRC Discrimination against ethnocultural minority groups in China has been reported by numerous sources and is of concern to both the United Nations and the Chinese government (United Nations’ Committee on the Elimination of Racial Discrimination, 2009a, 2009b). For example, at the institutional level, certain practices such as religious fasting have been formally prohibited with the intention of preserving social stability (“Beijing bans Xinjiang officials and  45   schools from participating in Ramadan,” 2014), and promotion of standardized Mandarin Chinese has frequently marginalized minority languages (M. Zhou, 2003, p. 28). At the interpersonal level, experiences of ethnocultural discrimination in China varies significantly between social groups and has ranged from ethnoculturally-based adolescent bullying (S. W. Pan & Spittal, 2013) to large-scale interethnic violence (“Innocent civilians make up 156 in Urumqi riot death toll,” 2009). Ethnocentric Han national narratives have at times cast ethnic minorities as less modern (Gladney, 2004, pp. 60–1) and ethnic stereotypes indeed inhibit interethnic trust (Dautcher, 2009, p. 63; S. Zhang et al., 2013). However, aside from several prominent ethnic groups (e.g., Uyghurs) and exceptional incidents (Kahn, 2004), relations between Han and most ethnic minorities in 21st century China have been relatively free of violent strife.  In recent years, several ethnographies have critically examined the sociopolitical determinants of health among deprived ethnic minority groups in Southwest China (Hyde, 2007; S. Liu, 2011; Lyttleton, Deng, & Zhang, 2011), but few if any studies have specifically addressed the health implications of ethnocultural discrimination. One exception is a study which showed that adolescents in urban Chinese cities were more likely to experience depression, suicidal ideation, and interpersonal violence if they perceived being bullied because of their ethnicity or religion (S. W. Pan & Spittal, 2013). Few if any quantitative studies have examined ethnocultural discrimination and HIV vulnerability in China, but the aforementioned literature strongly suggests a decidedly positive relationship.    1.3.7 Ethnocultural minority MSM, HIV vulnerability, and HIV testing Currently, the literature concerning ethnocultural determinants of MSM HIV vulnerability and testing remains sparse and overrepresented by studies with non-white minority Americans and Judeo-Christian traditions. Most ethnic minority and religious MSM HIV-related  46   studies have overwhelmingly relied upon African-, Asian-, Latino-American, and Christian sub-populations. Aside from one study report addressing ethnic Uyghur MSM in Xinjiang province (Kamaliti, 2011),  such research remains virtually non-existent in mainland China. Nonetheless, international empirical studies strongly suggest that ethnocultural dynamics are influencing MSM HIV vulnerability within other settings that, like mainland China, maintain a pluralistic culture. In general, ethnic minority and religious MSM often encounter distinct HIV challenges and vulnerabilities when compared to their social counterparts. Guided by the dissertation’s aforementioned objectives, this chapter briefly describes the extant literature concerning ethnocultural determinants of HIV vulnerability among MSM.  1.3.7.1 Ethnoracial affiliation & HIV epidemiology One of the important observations from the small but growing body of minority ethnoracial MSM HIV research is that neither religious nor ethnic minority status per se is associated with greater HIV risk. The breadth of diversity between ethnoracial groups and cultural contexts simply precludes any such universal theories. For example, in the US, African Americans represent 35% of incident HIV infections among MSM (Maulsby et al., 2014), but only ~14% of the total US population (Rastogi, Johnson, Hoeffel, & Drewery, Jr., 2011). On the other hand, minority Asian American MSM have consistently had lower HIV prevalence and incidence than MSM of all other US racial groups (Wei et al., 2011). Similarly divergent trends have also been observed between minority “Blacks” and “Asians” in the United Kingdom (Elford et al., 2012). In Malaysia, one cross-sectional study found that HIV infection and UAI were both significantly more prevalent among ethnic majority Malay MSM than ethnic minority Chinese MSM (5.3% vs. 1.8% and 49.4% vs. 39.8%, respectively) (Kanter et al., 2011).   47   1.3.7.2 Acculturation & implications for HIV vulnerability and testing among MSM In addition to ethnoracial affiliation, ethnic practices and ethnic identity affirmation have gained increasing recognition as critical influences on HIV vulnerabilities and testing among MSM.     Heritage language use (where applicable) has served as a popular metric of heritage ethnoracial practice among HIV-related acculturation research among MSM and non-MSM alike. One study with Asian MSM in the US found that the odds of HIV testing were almost three times lower for those who did not speak English at home (Toleran et al., 2013). In theory, language barriers impede access to HIV testing services and diffusion of safer sex innovations among those who are less proficient in the dominant-group language (Unger & Schwartz, 2012). Empirical research with Latino MSM in the US has linked greater use of Spanish with UAI among non-main sex partners (Donnell et al., 2002), though no association was detected between Spanish/English language preference and HIV serodiscordant UAI (Mizuno, Borkowf, Ayala, Carballo-Diéguez, & Millett, 2013).  Ethnic identity affirmation can be defined as “the extent to which one is attached to one’s racial/ethnic group and views that group positively” (Schwartz et al., 2011). Along with any sense of identity derived from one’s sexual behaviours, ethnic identity affirmation among ethnic minority MSM has important implications for individual self-worth and mental health, which in turn directly impact one’s resolve and capacity to adopt HIV preventative behaviours (Chae & Yoshikawa, 2008; Nemoto et al., 2003; Vu et al., 2011). Despite stronger heteronormative and homonegative values among certain ethnocultural groups, evidence suggests that ethnic identity affirmation among MSM is inversely associated with various HIV sexual vulnerabilities. Among MSM in the US, greater affirmation of Latino ethnic identity was associated with less UAI  48   (Donnell et al., 2002), and pride in Asian/Pacific Islander ethnic identity was correlated with more frequent HIV testing (Vu et al., 2011).  Notably, frequent HIV testing was even more likely when Asian/Pacific Islander MSM in the US reported strong ethnocultural identity in combination with strong sexual identity (Vu et al., 2011).  1.3.7.3 Religion & HIV vulnerability and testing For many men, religion is a fundamental framework around which their lives and world perspectives are organised, and which often intertwines with ethnic affiliation. Due to heteronormative and austere religious codes of sexual morality, MSM of certain religious faiths (e.g., Islam or Christianity) may be more likely to encounter and internalize stigmatization of their sexual orientation and sexual behaviours (Keogh, n.d.; Kugle & Chiddy, 2009; Siker, 2007; Smallwood, 2013; Wagner et al., 2012; Wilson, Wittlin, Muñoz-Laboy, & Parker, 2011). Homonegative religious sexual norms in particular might inhibit same-sex sexual behaviours, but may also limit information and access to health services such as HIV testing and counseling by socially isolating MSM of faith from other MSM (Andrinopoulos et al., 2014; Santos et al., 2013). Currently though, epidemiological evidence documenting HIV vulnerabilities among MSM by religious affiliation remains scarce and inconsistent. Results of the few such studies indicate that Muslim MSM in West Africa appear less likely to test for HIV and more likely to have sex with females when compared with their Christian counterparts (Lorente et al., 2012; Sheehy et al., 2013). In Israel, Muslim MSM appear to have lower HIV prevalence than Jewish MSM (Mor et al., 2013). However, no significant differences in HIV vulnerabilities were observed between religious groups among Malaysian MSM (Muslim, Hindu, Buddhist, Christian, no religion, and other religion) (Kanter et al., 2011) or between Christian denominations among US MSM (Garofalo et al., 2014).    49   Compared to religious affiliation, effects of religiosity on HIV vulnerabilities have been less studied, but results have been more consistent. Profound differences in doctrine, practices, and norms present challenges for generalizing about religiosity across religious affiliations, but religious faithfulness and participation in religious activities have been linked to less risky sexual behaviours among MSM and transgender women in Chicago (Dowshen et al., 2011; Garofalo et al., 2014). Unfortunately, mechanisms of association between religiosity and reduced HIV-related sexual vulnerabilities remain poorly understood. In theory, greater religious faithfulness and participation in religious activities may reduce HIV-related sexual vulnerabilities by promoting social support and personal coping skills (Garofalo et al., 2014), but religious social stigmatization of homosexual and extramarital sex may also be compelling religiously faithful and religiously active MSM to suppress same-sex behaviours in accordance with their beliefs about sexual morality. The relation between internalized homonegativity and HIV sexual vulnerabilities remain in dispute (Newcomb & Mustanski, 2011; Ross, Berg, et al., 2013), but associations between religiosity and internalized homonegativity among American MSM have been relatively robust (Ross, Rosser, Neumaier, & Positive Connections Team, 2008; Smallwood, 2013). 1.3.7.4 Ethnocultural discrimination & HIV vulnerability Ethnocultural discrimination manifests interactively at multiple social levels (e.g., inter-personal and structural) and has the potential to exacerbate HIV vulnerability among MSM by modifying individual behaviours and sexual networks, as well as the social contexts within which they take place. Perceived ethnoracial discrimination can undermine psychological well-being and promote negative coping behaviours such as substance use, riskier sexual behaviours, or avoidance of health services providers (Paradies, 2006; Williams & Mohammed, 2009).  50   Among ethnoracial minority MSM in the US, perceived racism alone and in combination with perceived external homonegativity has consistently been linked to UAI (Ayala, Bingham, Kim, Wheeler, & Millett, 2012; C. S. Han et al., 2014; Huebner et al., 2013; Mizuno et al., 2012). But even without cognizant acknowledgement, ethnoracialized social structures and notions of sexual desirability among MSM can lead to interracial sexual power imbalances that disproportionately place certain marginalized ethnocultural minorities at elevated vulnerability to HIV infection (Arnold, Rebchook, & Kegeles, 2014; C. Han, 2008; McAdams-Mahmoud et al., 2014; Ro, Ayala, Paul, & Choi, 2013). For example, at the interpersonal level, internalized and enacted racism against Asian-American MSM can adversely influence safer sex practices with white MSM. According to Han (2008), the confluence of wide-spread discrimination against Asian-Americans by white and Asian-American MSM has encouraged many Asian-American MSM to take on riskier sexual behaviours such as receptive anal sex in order to increase their favourability with the fewer number of white MSM willing to partner with Asian-Americans. The consequences of racial discrimination are even more pronounced between white and African-American MSM. Driven in-part by de facto racial residential segregation (H. F. Raymond et al., 2014) and racial bias against African-American MSM (H Fisher Raymond & McFarland, 2009), African-American MSM sexual networks have become more racially homogenous, interconnected, and segregated (Maulsby et al., 2014; B. Mustanski et al., 2014). As a consequence of racialized sexual network patterns and racially distinct background prevalence, African-American MSM may face greater risk for HIV infection, independent of any personal behaviours. In short, the current literature demonstrates that sweeping generalizations cannot be made about the HIV implications of ethnic minority affiliation or acculturation among  51   MSM, though religiosity appears to be associated with greater homonegativity and reduced HIV vulnerabilities.  1.3.8 Ethnocultural minority MSM and HIV vulnerability in the PRC 1.3.8.1 Ethnic minority MSM in the PRC  Ethnic and religious minority MSM remain a neglected and poorly understood sub-group within Chinese society. Despite burgeoning social science research on sexual and ethnocultural minorities in China, both populations have usually been conceptualized with little consideration of the other; studies of male sexual minorities in China have rarely accounted for ethnocultural minority experiences, and vice versa. Nonetheless, several exceptional qualitative studies offer insight into the range of embodied experiences of ethnic and religious minority MSM in China. Qualitative interviews with ethnic minority MSM suggest that intensity of homonegativity varies widely by ethnicity in China. Mutual masturbation and anal sex among adolescent boys has been reported as a common and socially acceptable practice within rural ethnic Zhuang communities (Tong Ge, 2005, p. 129), while ethnic Muslim MSM perceive same-sex sexual activities to be highly taboo within their ethnocultural communities (Kamaliti, 2011; Tong Ge, 2005, pp. 127, 130–1). Aside from potentially greater homonegativity within their ethnocultural community, some ethnic minority MSM also encounter ethnic segregation and discrimination from other MSM (Kamaliti, 2011; Tong Ge, 2005, p. 130). According to an interview with a Uyghur MSM, ethnic tensions transcend sexual minority communities and have led to ethnically distinct social and sexual networks:    52   The larger cities in Xinjiang also have gay ‘spots’. If you go there, you’ll notice that the Uyghur gays will stand at one end and the Han gays will stand at the other end without any interaction in-between …. In Xinjiang, Han gays don’t respect Uyghur gays because they feel Uyghurs are dirty and uncivilized …. If you get fucked by a Han, your body is soiled, ethnic pride is shamed … it’s like you’ve betrayed your own ethnicity. (Tong Ge, 2005, p. 130)  1.3.8.2 Religious minority MSM in the PRC Oftentimes, interethnic tensions and attitudes towards MSM are intertwined with issues related to religion. Generally, extant research indicates that Christian- and Muslim-affiliated MSM in China are prone to greater internalized and perceived external homonegativity (Chan & Huang, 2014; Kamaliti, 2011; Tong Ge, 2005, pp. 130–1). In addition to MSM stigmatization in mainstream Chinese society, many Christian and Muslim MSM in China must also contend with the religious-based stigmatization of same-sex sexual behaviours within their religious social reference groups. As elsewhere in the world, mainline Christian and Muslim institutions in China view same-sex sexual behaviours as inherently immoral acts which violate religious codes of conduct (Chan & Huang, 2014; Kamaliti, 2011). That being said, some Christian MSM in China have begun to challenge their religion’s conventional view of same-sex behaviours by creating informal sexual minority religious organisations which encourage alternative perspectives on religious codes of sexual morality. Though few in number, these sexual minority Christian social organisations provide important space for Christian MSM to mutually reciprocate a type of peer social support that would be otherwise unavailable in mainstream Christian or tongzhi communities (Chan & Huang, 2014). By contrast, heteronormative Islamic codes of sexual morality have remained largely uncontested in China’s public sphere. Moreover, few if any, Muslim religious organisation in China specifically serve sexual minority populations.      53   All major Buddhist sects discourage acts of “sexual misconduct” in a generic sense (Corless, 1998, p. 253; Mitchell, 2008, p. 121; Sweet, 2007, p. 76), but conflicting interpretations of what constitutes “sexual misconduct” has led to differing contemporary Buddhist perspectives on the morality of same-sex sexual behaviours in China. One prominent Chinese Buddhist master has averred that homosexuality is neither inherently right or wrong (Hsing Yun, 2001), while another felt that homosexuality “planted the seeds which lead to rebirth in the lower realms of existence” (Corless, 1998, p. 255). Thus, Buddhism within China today currently lacks a dominant religious view on the moral implications of same-sex sexual behaviour.   But regardless of how Buddhists define “sexual misconduct,” sex between Buddhist men in practice was highly prevalent and well-tolerated throughout much of Dynastic China (Hinsch, 1990, p. 97). As a continuation of this legacy, MSM in China today appear to encounter relatively limited religious-based sexual stigmatization from Buddhist populations when compared with Christian or Muslim communities. Indeed, ethnographic interviews indicate that Buddhism has become an increasingly popular religion among MSM in China (Miller, 2013, pp. 343–5; Tong Ge, 2005, pp. 131–2).  1.3.8.3 Ethnicity, religion, and HIV vulnerability among MSM in China The intersections of ethnic, religious, and sexual minority statuses carry potentially profound implications for HIV vulnerability among men in China. For example, ethnic minority MSM who primarily have sex with ethnic peers may have greater risk of exposure to HIV if there is higher background prevalence within their overall ethnic community. Or, religious codes of sexual morality may at times be inhibiting religious minority MSM in China from seeking out male sex partners. Or, as a consequence of perceived ethnoreligious discrimination, ethnocultural minority MSM may be less inclined to contact and access HIV prevention services provided by  54   ethnic Han CBOs, whether or not the organisation is operated by peer sexual minorities. However, given the lack of understanding about ethnocultural minority MSM in China, propositions surrounding their HIV vulnerabilities remain as unresolved conjecture.  Deeper understanding about HIV vulnerability among ethnic and religious minority MSM in China has remained hampered by two major barriers. First, there is a severe paucity of studies concerning ethnic or religious minority MSM in China. The only such quantitative study I am aware of indicated that many ethnic Uyghur MSM in Xinjiang province continue to engage in unprotected anal and vaginal sex (70% and 47%, respectively), despite being well informed about HIV/AIDS (J. Zhang, Li, & Li, 2010). I am unaware of any epidemiological study with religious minority MSM in China. Second, conventions of crudely dichotomizing ethnic status into “Han majority” and “other” have obfuscated any meaningful interpretations of ethnicity as a sociocultural construct. Presumably for the sake of statistical power, epidemiological HIV surveys in China typically aggregate all 55 ethnic minority groups into a single category. As a result, significant associations for specific ethnic groups become masked and comparisons between studies left untenable because the ethnicities of participants classified under the pan-ethnic minority rubric can drastically change from study to study. For example, one Beijing study in 2008 indicated that HIV was spreading 4.7 times significantly faster among ethnic minority MSM (S. Li et al., 2011), while another Beijing study in 2009 did not detect any significant disparity between majority Han and ethnic minority MSM (D. Li et al., 2012). In Yunnan province, HIV incidence rates have been 5.7 times higher among ethnic minority MSM (vs majority Han), but again, without disaggregated data by ethnicity, the interpretive and applicative utility of such findings becomes disputable (Xu et al., 2013).  55   1.4 Dissertation objectives and organisation In light of the described literature and lacuna surrounding religion, ethnicity, and HIV among MSM in China, this dissertation uses qualitative in-depth interviews and quantitative survey data to meet four key study objectives: Objective 1: To explore processes by which ethnicity and religion may be influencing HIV vulnerabilities and testing among ethnocultural minority MSM  Objective 2: To assess how ethnic affiliation, language preference, ethnic identity affirmation, and ethnic discrimination are potentially impacting HIV vulnerabilities, sexual networks, and HIV testing  Objective 3: To assess how religious affiliation and religiosity are potentially influencing HIV vulnerabilities, sexual networks, and HIV testing  Objective 4: To assess sexual minority stigma as a potential mediator of associations between religion and ethnicity with HIV vulnerabilities  Guided by these objectives, this dissertation is organised into seven chapters. Chapter two introduces the dissertation study setting, methods, and broader multi-disciplinary framework. Chapter three presents analyses of qualitative interviews that, consistent with objective one, aim to depict how notions of ethnicity and religion shape the worldviews, sexual behaviours, social structures, and lived experiences of ethnocultural MSM in North China. Chapter four focuses on objective two and uses primary survey data to assess how individual MSM HIV vulnerability and testing patterns may be impacted by ethnic affiliation, language preference, ethnic identity affirmation, and ethnic discrimination. Chapter five corresponds with objective three and uses data from the same survey to assess how individual MSM HIV vulnerability and testing patterns may be impacted by religious affiliation, religiosity, and religious discrimination. Chapter six addresses objective four and assesses the extent to which internalized and perceived sexual minority stigma mediates associations between (a) religion and ethnicity, and (b) HIV  56   vulnerabilities. As chapters three through six were written as manuscripts for submission to refereed journals, each chapter contains a background and methods section which reiterates the content presented in chapters one and two. Lastly, chapter seven integrates results from chapters three through six and discusses the study findings within the broader international and cross-disciplinary context. Recommendations and implications for future study are presented accordingly.   57   Chapter 2: Methodology This multi-disciplinary dissertation project was conceptualized in the tradition of intersectionality (Bowleg, 2012; Crenshaw, 1991). Results were based upon primary data responses from self-identified MSM who were living in Beijing and Tianjin, China at some time between February 2013 and April 2014. Given the dearth of information about ethnocultural minority MSM in China, a three-phase exploratory study design (Figure 2.1) was determined to be the most appropriate for developing and testing emergent hypotheses relevant to the dissertation’s research question (Clark & Creswell, 2010). In phase one, I conducted in-depth qualitative interviews with 41 self-identified MSM at private locations throughout Beijing from February to December 2013. In phase two, results from phase one were used to inform the design of a structured questionnaire, which was pilot tested in two feedback sessions (n=8). In phase three, the structured questionnaire was used in a clinic-based cross-sectional survey which included blood-based HIV testing (n=400). To enhance capacity to “construct meaningful propositions” salient to the dissertation objectives, methodological triangulation (Mathison, 1988) was completed by complementing findings from semi-structured qualitative interviews with results from the cross-sectional epidemiological survey, and vice versa. I was personally responsible for conceptualizing the dissertation study design, conducting all qualitative interviews, supervising all project activities (e.g., participant recruitment, transcription, data entry and cleaning, etc.), and performing all data analyses. Key project partners included government institutions (The Chinese CDC, Chaoyang District CDC) and three tongzhi advocacy non-governmental organisations (organisation names suppressed for privacy). This chapter details the dissertation project’s conceptual model, study setting, and methods concerning data collection, management, and analyses.  58   Figure 2.1: Timeline of dissertation data collection  2.1 Conceptual model Through the lens of intersectionality, the current dissertation examines the “embodied experiences” (Starks & Trinidad, 2007) of religious and ethnic minority MSM in North China, and how such multi-dimensional identities influence behavioural, interpersonal, and social group dynamics pertinent to HIV vulnerability. While acknowledging the diversity within each social category, the intersectionality conceptual model of this dissertation (Figure 2.2) depicts the overlap and sexual connectivity of three behavioural/social groups: MSM, ethnic minority, and religious minority. Enveloping the Venn diagram is non-religious, ethnic majority Han, heteronormative mainstream Chinese society. 2.2 “MSM” terminology In the 1990s, the behavioural-based term “men who have sex with men” emerged from HIV prevention discourses in an effort to more accurately describe men thought to have greater HIV vulnerabilities due to engagement in anal sex with other men, irrespective of their sexual social identities (Young & Meyer, 2005). Labels based upon sexual orientation (e.g., “gay”, “tongzhi”, “homosexual”, “hijras”, “queer”, “two-spirited”, or “bisexual”) reflect distinct social identities that can have implications for HIV vulnerability, but not all men who claim such Figure 2.2: Dissertation conceptual model with fictionalized sexual networks  59   identities necessarily have sex with other men, and therefore would not necessarily experience greater HIV vulnerability. Moreover, invariably some MSM will not identify with any type of sexual minority label. Thus, despite the term’s eschewal of sociocultural context, I deferred to official UNAIDS practices (UNAIDS, 2011), and decided that the term “MSM” most appropriately represents the population of interest given the aims and scope of the dissertation. 2.3 Study setting The PRC is currently home to over 1.3 billion people (The World Bank, 2015) and has the second largest national economy in the world (The World Bank, 2014). Since the late 1970s, the PRC has made significant progress in all human development indictors (e.g., life expectancy, years of education, purchasing power parity) (United Nations Development Programme, 2014), although wealth and industrialization has been disproportionately concentrated along the coastal provinces and in urban economic hubs. Due in large part to decades of steady rural-to-urban migration, 52% of China’s population had urbanized by the year 2012 (The World Bank & Development Research Center of the State Council P. R. China, 2014).  All data collection activities were conducted in Beijing and Tianjin, two adjacent “mega” cities in North China. Beijing and Tianjin were chosen as study sites because of the strong and established working relationship between project partners and the local MSM population. In addition, the ethnocultural diversity of Beijing and Tianjin’s populations was conducive to drawing comparisons between multiple ethnic and religious minority groups.  (Central Intelligence Agency [cartographer], 2011)  Figure 2.3: Study project sites  60   2.3.1 Beijing, the nation’s capital Beijing ranks among China’s most prosperous and prestigious cities. In 2013, Beijing’s gross regional product (GRP) per capita was 2.2 times higher than the national average (~$15,000 USD vs. ~$7,000 USD), and post-secondary educated residents as a proportion of the total regional population was almost quadruple the national rate (32% vs. 9%) (National Bureau of Statistics of China, 2014). Its robust economy and cosmopolitan social scene draw millions of migrants from across the country and around the world; in 2010, 37% of Beijing’s 20 million inhabitants were migrant residents (C. Huang, 2014). Although all 55 ethnic minority groups of China are represented among Beijing’s population, 96% of Beijing residents are ethnically Han (Beijing Statistical Information Net, 2011), and standard Mandarin is the dominant dialect in most personal and professional spheres. Manchus, Hui, and Mongols constitute the vast majority of Beijing’s ethnic minority population (83%) (Beijing Statistical Information Net, 2011). Hundreds of Buddhist, Christian, and Muslim religious venues are located throughout the city, though most residents do not claim any specific religious affiliation (China.com.cn, 2008). Beijing’s sexual minority community is among the most vibrant and socially active in China, with an estimated MSM population size between 115,731 and 239,258 persons (Guiying et al., 2014). Since 2009, HIV prevalence estimates among Beijing MSM have ranged from 2.5% to 8.0% (Yunhua Zhou et al., 2014).   2.3.2 Tianjin, an overshadowed city of 14 million  Tianjin is a relatively wealthy Chinese municipality that attracts millions of domestic migrants and plays a vital role in the North China economic zone; of its 14 million inhabitants, 25% are migrants (C. Huang, 2014). However, Tianjin does not command nearly as much wealth or prestige as neighbouring Beijing. In 2013, Tianjin’s per capita disposable income and  61   percentage of post-secondary educated residents were 35% and 44% lower than that of Beijing, respectively (National Bureau of Statistics of China, 2014). The vast majority of residents in Tianjin are ethnically Han (97%) and nonreligious (china.com.cn, 2009; National Bureau of Statistics of China, 2014), while ethnic Muslim Hui represent the bulk of Tianjin’s ethnic minority population (65%) (china.com.cn, 2009). Tianjin’s sexual minority community is relatively robust with numerous online and physical venues. From 2008-2009, HIV prevalence estimates for Tianjin MSM have ranged from 5.9% to 8.6% (Ning et al., 2011; Yunhua Zhou et al., 2014).  2.4 Qualitative in-depth interviews Elucidating contextually-rich accounts of embodied experiences was imperative to the development and credibility of propositions advanced within the dissertation. To that end, I conducted exploratory in-depth interviews with 41 MSM of various ethnic and religious affiliations. Interview transcripts were then analysed with an interpretive thematic approach that involved investigator and methodological triangulation (Mathison, 1988). The following sub-sections of 2.4 describe methods of qualitative data collection and analysis in further detail.  2.4.1 Maximum variation sampling In-depth interviewees were recruited using a purposive maximum variation sampling scheme (Clark & Creswell, 2010, p. 174). The rationale for maximum variation was to facilitate a more nuanced understanding of ethnocultural minority MSM experiences, while also producing accounts that could be triangulated with results from the ethnoculturally diverse cross-sectional survey. Specifically, maximum variation was sought for five interviewee characteristics: ethnic affiliation, religious affiliation, age, education, and migrant status.     62    However, population imbalances between ethnic and religious minority groups in North China presented two recruitment challenges. First, the population size of MSM from larger ethnic minority and religious groups required imposing a recruitment cap of five to six individuals for certain ethnic and religious groups. Without such a cap, the interviewee sample would have quickly been dominated by ethnic Han and Buddhists. Second, the sparse population of MSM from smaller ethnic minority and religious groups constrained the ability to ensure maximum variation by age, education, or migrant status. For example, most Christian participants had higher levels of educational attainment because of challenges to locate Christians who had not completed college. Hence, obtaining maximum variation by ethnic and religious affiliation assumed highest priority. Within each ethnic or religious affiliation group, maximum variation by age, education, and migrant status was pursued to the greatest extent possible.   Originally, the aim was to enroll 30 in-depth interview participants; however, based on emergent needs to seek out disconfirming evidence and augment sample variability, the sample size was increased and a total of 41 men were eventually interviewed. One interview transcript was excluded from analysis because project staff and I determined that the interviewee in question likely misrepresented himself as an ethnic Hui minority, when in fact responses during the interview indicated otherwise. Demographic statistics for the 40 eligible interviewees are presented in Table 2.1.      63     Table 2.1: In-depth interview participant characteristics (n=40)    No. (%) Ethnic affiliation     Han 14 (35)    Hui    6 (15)    Manchu 5 (13)    Mongol 5 (13)    Tujia/Maonan/Yi/Li 4 (10)    Uyghur/Salar 3 (8)    Zhuang 3 (8) Religious affiliation     Areligious 16 (40)    Buddhist 10 (25)    Christian 7 (18)    Muslim 7 (18) Age     < 30 years old 20 (50)    > 30 years old 20 (50) Educational attainment     Less than vocational college 13 (33)    Vocational college& above 27 (68) Hukou status     Official Beijing resident 10 (25)    Migrant       30 (75)    2.4.2 Recruitment & enrollment   Interviewees were recruited by four methods. First, staff from the Chaoyang CDC and two tongzhi advocacy organisations introduced the interviewer to ethnic and religious minority MSM in their existing social networks who had expressed interest in participating. The interviewer then contacted these individuals to explain the study in greater detail and arrange interview logistics. Second, those who completed the interview were provided my contact  64   information and encouraged to refer their eligible peers. Third, potentially eligible interviewee participants were contacted via several gay social networking websites and chatrooms and invited to participate in the study. Fourth, during the cross-sectional survey, eligible individuals who had come to receive HIV testing services on their own volition were invited to participate in the in-depth interview. Proportional modes of recruitment methods are presented in Figure 2.4.  Face-to-face interviews were conducted at a safe and quiet location determined by the interviewer and interviewee, and included venues such as restaurants, coffee shops, private homes, and office workplaces. Due to potential privacy concerns, no true identifiers were requested of participants. Men were eligible to participate in the survey if they self-reported ever having sex with another man (oral or anal sex), were at least 18 years old, were willing and able to provide written informed consent, and satisfied the purposive maximum variation sampling requirements at the time of enrollment. All interviewees received an honorarium of ¥50 CNY (~ $8 USD as of July 1, 2013) in order to compensate for their time and transportation costs. As a point of reference, a 10 kilometer taxi trip in 2013 cost approximately 29 CNY in Beijing (~ $4.7 USD as of July 1, 2013) and 22 CNY in Tianjin (~ $3.6 USD as of July 1, 2013). Subway tickets in both cities costed between 2-5 CNY (~ $0.3-0.8 USD as of July 1, 2013).   Introduced by affiliates of health department  35% Introduced by sexual minority advocacy NGO 20% Participant referral 10% MSM social networking website 7% Cross-sectional survey participant 28% Figure 2.4: In-depth interviews by mode of recruitment (n=40)  65   2.4.3 Qualitative data collection  Qualitative data originated from in-depth open-ended interviews conducted in Mandarin Chinese. In-depth open-ended interviewing was favoured because its loosely structured format gave the interviewees freedom to liberally contextualize their experiences, attitudes, and decision-making processes within broader sociocultural influences of interest (i.e., ethnicity, religion, and sexuality). Moreover, the exploratory nature of the qualitative data collection phase necessitated interviewer flexibility to probe for further information and pursue new directions extemporaneously during each interview. Prior to data collection, I received four months of formal coursework training in qualitative research methods. Throughout the qualitative data collection phase, I conscientiously applied well-established qualitative interviewing techniques in each interview (e.g., establishing rapport, strategically ordering conversation topics, avoiding closed-ended questions, providing positive reinforcement, requesting clarification of ambiguous references, etc.).  In practice, these in-depth interviews were essentially “guided conversations,” whereby I steered the general orientation of the interview, but interviewees maintained considerable autonomy to elaborate upon or redirect conversational topics as they saw fit. Memorization of key interview mile markers and conversational topics helped facilitate natural transitions between topics and enhance the interviewee’s narrative “flow,” but a topical guide was occasionally referenced as a means of sparking deeper conversational dialogue with less gregarious participants. The topical guide was designed specifically for the dissertation and touched upon topics such as perceived significance of ethnic and religious affiliations, personal sexual milestones (e.g., sexual debut), and sexual decision-making processes (e.g., considerations when seeking sex partners). However, given the nature of the in-depth open- 66   ended interview, not all questions in the topical guide were addressed in the interview, and vice-versa. In order to avoid potential distractions and to focus attention on the interviewee, no notes were taken during the interview. Rather, all interviews were audio recorded and lasted an average of 62 minutes (range: 21 – 96 minutes).  2.4.4 Transcription All interviews were transcribed verbatim into simplified Chinese by China CDC project staff. In an effort to faithfully represent interviewees’ verbal responses, all verbal utterances (e.g., “um”) were transcribed. To ensure fidelity of the transcriptions, I personally conducted systematic quality control checks on each completed interview transcript and ensured that transcripts which failed to capture the interview verbatim were corrected accordingly. In order to retain the linguistically sensitive nuances of participant expressions, all interview transcripts were analysed in the original Mandarin Chinese.   2.4.5 Data analyses Interpretive thematic analysis  (J. Green & Thorogood, 2009, pp. 13–14; Starks & Trinidad, 2007) was conducted for all qualitative data and occurred iteratively throughout the data collection and write-up phases of the dissertation project. Through recursive de-contextualisation and re-contextualisation of each qualitative datum vis-a-vis broader themes and patterns, interpretive thematic analysis enabled deeper understanding of ethnic and religious minority MSM embodied experiences and decision-making processes relevant to HIV acquisition and transmission. Interview transcripts were read multiple times for the interrelated purposes of coding, interpretive analysis (Starks & Trinidad, 2007), and verification (Morse, Olson, & Spiers, 2002). In the first coding cycle, I sought firmer understanding of “what was going on” by coding  67   interview transcripts with line-by-line descriptive and process coding approaches (Saldaña, 2009, pp. 70, 77). First-cycle coding was conducted with pen and paper-based interview transcripts. Each participant’s first cycle codes were then enumerated into a double-columned, 3-5 page code summary word processing document. Consolidating each individual’s codes into a visually compressed format enhanced my ability to postulate thematic patterns, recognise deviant cases, and perform “constant comparisons” within and between interviewees (Charmaz, 2007, p. 187). Based on the first cycle codes, reflexive analytic memos were drafted with the explicit aim of contextualizing interview findings in relation to evolving theories derived from other data in the study. As a form of investigator triangulation (Mathison, 1988) to guard against any of my undue subjectivities, two bilingual China CDC project staff verified my coding decisions and analyses by reviewing all coded interview transcripts, code summary documents, and analytic memos. Both staff members were graduate students studying HIV vulnerabilities among MSM in China. Disagreements in coding or analyses were resolved by document revisions and face-to-face discussions. Preliminary themes and hypotheses were also subjected to evaluation during data feedback presentations to a sexual minority advocacy NGO and public health working-group on August 5, 2013. Attendants of the presentation anecdotally corroborated certain findings (e.g., interethnic discrimination), but none of the emergent hypotheses were openly contested. After all first cycle codes had been finalized, second cycle pattern coding was conducted for the purposes of revising, distilling, and classifying preliminary codes into higher order thematic categories of interest (Saldaña, 2009, p. 152). First, a qualitative data analysis software package, Atlas.TI (Scientific software, Berlin), was used to digitally code interview transcripts. Then, each second cycle code was printed onto strips of paper which were shuffled and grouped with other thematically similar codes. This process of physically organising and reorganising  68   codes augmented my ability to recognise conceptual themes and infer their relational patterns. After September 2013, development of thematic patterns and coding decisions were also informed by the extent to which triangulated qualitative and quantitative findings were convergent, inconsistent, or contradictory. These hierarchical themes subsequently formed the scaffolding of the qualitative write-up, supported by illustrative quotes and observations. Chinese-to-English translations of quotes were completed by me. Given that English was my most proficient language, all codes and memos were written in English.  2.4.6 The candidate as the qualitative research instrument Since project conception, I remained vigilant about how my own embodied identities and pre-existing beliefs uniquely shaped the qualitative data and its analyses. Frequent memo writing throughout the data collection phase enabled me to reflexively examine the study implications of my own identity and subjectivities. Pre-existing beliefs were bracketed to the extent possible (Starks & Trinidad, 2007), but my social identities inevitably rendered profound effects on participant recruitment and how participants discussed certain interview topics.     As a dual national (United States & Republic of China), ethnically Han, Christian man who had never had sex with another man, I was naturally regarded by some participants as an unequivocal “outsider” in certain respects, but very much an “insider” in others. In this way, the multi-dimensional nature of intersectional identities precluded any grand delineations of a singular “in-group” or “out-group,” but implications of my ethnicity, religion, and sexual experience warrant elaboration. First, as a visible ethnic Han, my social interactions were heavily influenced by individual attitudes towards ethnic Han Chinese in general. In light of the broader Uyghur-Han interethnic tensions in China, my ethnic Han heritage may have helped dissuade some potential Uyghur participants from being interviewed. Anecdotes during the fieldwork  69   indicated that even among MSM, Uyghur and Han men had very limited social interaction. On the other hand, if I were ethnically Uyghur, many negative comments about Uyghurs by participants would likely have been moderated or completely self-censored. Second, my Christian religious affiliation naturally enhanced my ability to build social rapport and relationships with other Christian participants. As a Christian supporter of sexual minority rights, I was invited to participate in numerous sexual minority Christian social events, during which I was able to meet and interact with many Christian MSM. Third, my lack of sexual experience with other men limited my understanding of MSM sex to that of sexually explicit media and personal accounts. I made socially tactful intimations of my lack of sexual experience with other men when appropriate during social introductions, but my sexual orientation appeared to have limited influence on recruitment of participants’ or their degree of openness during the interview. Generally, most male public health researchers of MSM HIV vulnerabilities in China do not identify as MSM.     2.5 Questionnaire design From March to August 2013, China CDC staff (Drs. Yuhua Ruan, Yiming Shao, and Yunan Xu) and I collaboratively designed the structured survey questionnaire based on ongoing findings from the qualitative interviews and literature. The following section describes the process by which the phase three data collection instrument was ultimately finalized.     2.5.1 Designing the survey data collection instrument Design of the structured questionnaire was an iterative process that required balancing comprehensive measurements with minimization of respondent fatigue. Informed by preliminary analyses of the qualitative interviews and extant literature, an initial draft of the questionnaire was developed consisting of 16 distinct sections and maximum of 363 response items drawn  70   from pre-existing questionnaires and scales. However, in order to minimize participant fatigue, the questionnaire was eventually trimmed to 11 sections and a maximum of 168 response items. Additional details about questionnaire measurements are discussed in section 2.6.4.        2.5.2 Focus group pilot testing of the questionnaire  Study staff and I pilot tested the questionnaire to identify conceptual gaps and illogical, redundant, ambiguous, or poorly worded sections before the cross-sectional survey was fully implemented. To that end, two groups of four MSM (n=8) were recruited by convenience sample from a popular MSM venue and invited to provide feedback about how the questionnaire could be improved. Inclusion criteria for the focus group were: (1) self- identification as an MSM, (2) at least 18 years old, and (3) able and willing to provide written informed consent. After the context of the larger study and objectives of the pilot test were explained to the group, each participant then completed the questionnaire by themselves and then provided oral feedback in an open group setting. I facilitated the feedback discussions while a study staff member recorded the discussion with an audio recorder. The first and second focus groups lasted for 57 and 39 minutes, respectively; both sessions were conducted at a private area agreed upon by all focus group participants. Most participant feedback was concerned with issues of ambiguous phrasing, which was later addressed accordingly in questionnaire revisions. For compensation of time, each participant received an honorarium of ¥ 50 CNY (~ $8 USD as of July 1, 2013). Demographic details of focus group participants are presented in Table 2.2.    71    Table 2.2 : Sociodemographics of focus group participants (n=8) Variable Percentage Ethnicity     Han 88%    Hui 12% Age     20-29 25%    30-39 38%    40-49 25%    50-59 13% Migrant     Yes 63%    No 37% Religious affiliation     Areligious 50%    Buddhist 25%    Christian 0%    Communist 13%    Muslim 13% Highest level of educational attainment (n=7)     Elementary/ middle school 29%    High school 29%    Vocational college 43%  2.6 Quantitative survey Quantitatively measuring constructs of interest in a sizable MSM population sample was critical for triangulating qualitative findings and generating novel hypotheses. To that end, a cross-sectional survey involving a blood-based HIV testing and a structured questionnaire was implemented in Beijing and Tianjin from July 2013 until April 2014. The following sub-sections of 2.6 elaborate upon the cross-sectional survey methods.   2.6.1 Participant enrollment The cross-sectional survey of MSM in Beijing and Tianjin was conducted at five HIV voluntary counseling and testing (VCT) sites devoted to serving MSM. Survey participants were  72   recruited passively and actively by four methods. First, eligible individuals who had come to receive VCT services on their own volition were invited on-site to participate in the study. Second, successfully enrolled participants were encouraged to recruit their eligible MSM friends. Third, MSM-oriented HIV prevention CBOs distributed study recruitment information to potential participants via a smart phone social networking application. Fourth, potential study participants were contacted via several tongzhi social networking websites and chatrooms. Due to potential privacy concerns, no true identifiers were requested of participants. Men were eligible to participate in the survey if they self-reported ever having sex with another man, were at least 18 years old, had never tested positive for HIV, and were willing and able to provide written informed consent. Due to resource constraints precluding unconditional recruitment of all VCT clinic patrons, a 1:1 design ratio was maintained throughout the survey enrollment period between the designated comparison group (ethnic or religious minority MSM) and referent group (non-religious ethnic majority Han MSM); in effect, only a proportion of otherwise eligible non-religious ethnic majority Han MSM patrons of the VCT sites were enrolled into the study. Of 407 men who met the eligibility criteria and were invited to participate in the study, seven refused enrollment (1.7%). Once study staff ascertained eligibility, obtained written informed consent, and provided instructions on how to complete the survey, participants proceeded to complete the self-administered survey in private at the VCT site; for participants who were functionally illiterate (<5%), surveys were administered by study staff. Thereafter, participants received confidential HIV pre-test counseling, a blood-based HIV rapid test, and appropriate post-test counseling from trained public health staff. Upon completion of the survey and HIV test, participants received an honorarium of ¥50 CNY.   73   2.6.2 HIV diagnostics          HIV-1 serostatus was initially determined by blood-based rapid testing (Determine HIV-1/2, Abbott Japan Co., Japan) and confirmed by HIV-1/2 Western Blot (HIV Blot 2.2 WBTM, Genelabs Diagnostics, Singapore). 2.6.3 Data management Completed paper-based survey questionnaires were checked for completeness and logical responses by at least two project staff members. Then, surveys were entered into computerized databases by China CDC public health graduate students using EpiData Entry 3.1 (The Epidata Association, Odense). In order to ensure fidelity of data entry, all questionnaire data were double-entered by two different individuals. I then used the PROC COMPARE function in SAS 9.3 (SAS Institute, Cary, NC) to identify discrepant entries, which were ultimately resolved by examination of the original paper-based questionnaire. Ultimately, less than 1% of data fields were missing. 2.6.4 Measures  2.6.4.1 Dependent variables Participants reported information about sexual histories (age at same-sex sexual debut and circumcision), number of male sex partners in the past six months, sexual behaviours in the past three months (engaged in male group sex, any anal sex, UAI with casual and steady male partners, and anal sex position), HIV testing history, and substance use (frequency of alcohol consumption and use of illicit drugs). Continuous variables were dichotomized or categorized into tertiles because of extreme outlier values and to facilitate clearer data interpretation. Sexual partner characteristics were assessed using a modified version of the UNAIDS Sexual Network Questionnaire (UNAIDS, 1998). Using a multiple-choice format, each  74   individual was asked to report the gender, ethnic affiliation, and religious affiliation of their three most recent sex partners in the past six months. Dichotomous dummy variables were then created to indicate whether or not the participant’s last three sex partners in the past six months included a male member of a specific ethnic or religious affiliation.  2.6.4.2 Independent variables 2.6.4.2.1 Ethnic affiliation & ethnic identity affirmation  Ethnic affiliation was reported by multiple-choice selection. Perceived ethnic discrimination was measured as having ever experienced ethnic discrimination. Mandarin language proficiency was determined by asking participants if Mandarin was their most proficient spoken language.  Ethnic identity affirmation was assessed using a modified version of the Multigroup Ethnic Identity Measure’s (MEIM) five-item “Affirmation and Belonging” sub-scale, which was designed to measure “ethnic pride, feeling good about one’s background, and being happy with one’s group membership” (Phinney, 1992). The MEIM uses a five-point Likert scale ranging from “1 = Strongly disagree” to “5 = Strongly agree,” and respondents are asked to report their degree of endorsement for statements such as “I have a strong sense of belonging to my own ethnic group” and “I have a lot of pride in my ethnic group and its accomplishments.”  The raw scores for ethnic identity affirmation were each calculated by averaging the values of all their respective sub-scale items.  Three other bilingual project staff from China CDC and I collectively translated the ethnic identity affirmation sub-scales after thorough discussions about how best to faithfully represent the original English meanings in written Chinese. The Cronbach’s alpha level for the ethnic identity affirmation subscales was 0.89, indicating strong scale reliability.  75   2.6.4.2.2 Religious affiliation & religiosity Religious affiliation was assessed by a multiple choice question with the following options: none, Buddhism, Catholicism, Daoism, Islam, Protestantism, and other. Selections for religious affiliation were chosen according to the State Administration for Religious Affairs’ rubric of officially institutionalized major religions within China (Goossaert & Palmer, 2011, p. 153), but “Catholicism” and “Protestantism” were combined as “Christian” in the present study. Religious discrimination was measured as having ever experienced religious discrimination. Religiosity measures were assessed using a version of the Duke University Religion Index (DUREL) (Koenig & Büssing, 2010) previously translated, adapted, and validated for mainland China (Eric Yang Liu & Koenig, 2013). Essentially, the DUREL is a five question index that captures three dimensions of religiosity: organisational religiosity (assessed by a single-item measuring frequency of participation in public religious activities on a scale of 1-6), private religiosity (assessed by a single-item measuring frequency of participation in solitary religious activities on a scale of 1-6), and intrinsic religiosity (assessed by three items measuring personal commitment to religion on a scale of 3-15) (Koenig & Büssing, 2010). The intrinsic religiosity sub-scale demonstrated excellent reliability in the present study, yielding a Cronbach’s alpha value of 0.91.  The following Spearman correlation coefficients were calculated for each pair of religiosity indicators: organisational & private religiosity, 0.69; organisational and intrinsic religiosity, 0.61; private and Intrinsic religiosity, 0.55.  76   2.6.4.3 Control variables Five measures were assessed as control variables in data analyses: Age, migrant status, city (Beijing or Tianjin), rural vs. urban household registration status, and highest level of educational attainment.  2.6.4.4 Mediator variables Internalized and perceived external homonegativity were identified as putative mediator variables, and were measured using modified versions of homosexual stigma scales previously translated and validated among MSM in mainland China (H Liu, Feng, & Rhodes, 2009). In order to mitigate participant fatigue, the internalized and perceived external homonegativity scales were truncated from ten to four items and eight to four items, respectively; the four items with the highest factor loadings for each sub-scale, as reported by the scales’ creator, were selected to be used in the dissertation project. Studies with MSM in the US have also used shortened versions of the homonegativity scales (Fredriksen-Goldsen et al., 2012; Ha, Ross, Risser, & Nguyen, 2014). Based on a four-point Likert scale, raw scores for internalized and perceived external homonegativity were calculated by averaging the values for all items of the respective sub-scale (range: 4-16). The internalized and perceived external homonegativity scales respectively yielded Cronbach’s alpha values of 0.85 and 0.87, implying strong internal consistency. 2.6.5 Quantitative analyses Chi-square tests were used to compare descriptive distributions of categorical independent variables across HIV status and testing history, sexual behaviours & experiences, substance use, and sociodemographics. Analysis of variance (ANOVA) was used to compare the mean scores of continuous independent variables across the aforementioned dependent and  77   control variables. Pooled and Satterthwaite methods were used to calculate significance tests for two-group samples with equal and unequal variances, respectively. Welch ANOVA tests were used to assess the significance of mean differences between three or more groups with unequal variance.  2.6.5.1.1 Testing for correlations between independent and dependent variables  Each dependent variable was regressed on each independent variable in unadjusted binary or ordinal logistic regression models. Parameter estimates involving ordinal logistic regression models which did not satisfy assumptions of proportionality were not reported, as such estimates did not have sufficient model fit. Ordinal logistic regression was used for models regressed on “more male sex partners” and “more frequent alcohol intake”. Then, in order to obtain adjusted measures of association between independent and dependent variables, control variables were added as covariates to each regression model.   In order to assess for possible effect modification by ethnocultural affiliation, the following interaction terms were created and modeled in unadjusted logistic and ordinal regression: (1) ethnic affiliation X ethnic identity affirmation, (2) religious affiliation X organisational religiosity, (3) religious affiliation X private religiosity, and (4) religious affiliation X intrinsic religiosity. In addition, stratum-specific measures of association were calculated from significant interaction terms in order to compare directionality and magnitude of correlations between ethnic and religious affiliation groups. Statistical significance was defined as p<0.05. Marginal significance was defined as 0.05<p<0.1. 2.6.5.1.2 Power calculation Given that the dissertation examined multiple outcomes of interest and had comparison groups of variable sizes, a generic power calculation was conducted prior to data collection using  78   SAS 9.3 (Cary, NC). Assuming that the true prevalence of a characteristic among the index and comparison group was 5% and 25%, and that the size of the index and comparison group was respectively n=250 and n=25, the study would have had 86% power to detect a statistically significant difference (alpha=.05) between the two groups.   2.6.5.1.3 Testing for mediation Mediation analyses were conducted in order to test the hypothesis that homonegativity significantly accounted for associations between independent and dependent variables. According to Baron & Kenny, satisfying the following conditions would indicate mediation: (1) the explanatory variable must affect the putative mediator, (2) the explanatory variable must affect the response variable, (3) the putative mediator must affect the response variable while simultaneously including  the independent variable as a covariate, and (4) the independent variable’s magnitude of effect on the dependent variable must be weaker when simultaneously including the putative mediator as a covariate (1986) (Figure 2.5). For the purposes of this dissertation, the PROCESS SAS macro (A. F. Hayes, 2013) was used to assess measures of association and 95% CIs pertaining to conditions one and three. In addition, PROCESS was used to calculate the indirect effect of independent variables on dependent variables, as well as the corresponding 95% CIs derived by bootstrapping with 5000 iterations. Indirect effects were calculated as the product of beta coefficient “a” and beta coefficient “b”. Figure 2.5: Conceptual model for mediation     79   In order to satisfy the PROCESS macro parameter requirements, “number of male sex partners in the past 6 months” was dichotomized to less than or more than four because the macro could not properly handle multi-group categorical outcome variables. All statistical tests, measures of association, and 95% confidence intervals were calculated with SAS 9.3. 2.7 Ethical reviews and considerations To help ensure that the rights and welfare of study participants were adequately protected, dissertation study protocols were reviewed and approved by the National Center for AIDS/STD Control and Prevention, Chinese Centers for Disease Control and Prevention (Project number: X120717232, Co-PIs: Stephen W. Pan, Yuhua Ruan, and Yiming Shao), and the Providence Health Care Research Institute, University of British Columbia (Reference number: PHC REB H12-00975, PI: Patricia M. Spittal). All ethical board approvals were obtained prior to the start of any data collection activities. As a former HIV/STI disease intervention specialist, I was able to provide participants clinical knowledge about HIV transmission and how to reduce their risk of HIV acquisition. In addition, I provided participants with information about where they could receive free HIV testing and information about relevant services.  80   Chapter 3: Narratives of intersecting ethnic, religious, and sexual identities: implications for HIV vulnerability and testing  3.1 Introduction In 2011, MSM are estimated to have accounted for 29.4% of all 48,000 new HIV infections in China (Ministry of Health of the People’s Republic of China, 2012), yet only represented 0.3% of the country’s population (UNDP, 2012). The expeditious spread of HIV among MSM in China is complex and can be attributed to multiple factors. Due to the biological fragility of the rectum’s tissue lining, HIV is approximately 18 times likelier to be transmitted by receptive anal sex than vaginal sex (Boily et al., 2009), a risk that extends equally to both males and females alike (Baggaley et al., 2010; Beyrer et al., 2012; Boily et al., 2009). Homonegativity (social stigma against homosexuality) and HIV-associated stigma have also deterred MSM from learning about or accessing HIV testing and prevention services (Xuefeng Li et al., 2012; Santos et al., 2013; Y. Song et al., 2011; Wei et al., 2014; Z. Wu et al., 2006). Early diagnosis of HIV not only leads to better health outcomes for HIV-positive individuals (Kilmarx & Mutasa-Apollo, 2012), but also reduces the likelihood of HIV being unwittingly transmitted by undiagnosed individuals (Metsch et al., 2008). It is estimated that approximately 61 – 87% of HIV-infected MSM in China are undiagnosed (E. P. F. Chow et al., 2014).  In light of these challenges, health departments and MSM-affiliated CBOs in China have aggressively sought to expand HIV testing and safer sex interventions among MSM (E. L. Fan, 2014; Miller, 2013, p. 210; Z. Wu et al., 2010). However, the ability of CBOs to successfully build ties with MSM has been challenged by the diversity of sub-groups under the broad MSM rubric (Chapman et al., 2009; H. H. Li et al., 2010; Tong Ge, 2005, p. 551). Two important social  81   groups which have received little to no consideration in the voluminous MSM HIV prevention literature in China are ethnic and religious minorities. 3.1.1 Ethnicity in the PRC Currently, the PRC recognises exactly 56 official ethnic groups within China that range in population from several thousand to over 1 billion (ethnic Han majority) (National Bureau of Statistics of China, 2013). Collectively, China’s 55 ethnic minority groups total about 111 million persons (~9% of the national population) and are characterized by highly heterogeneous histories, religions, socioeconomic statuses, languages, diet, acculturation, and geographic homelands  (National Bureau of Statistics of China, 2013; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b). The following sections briefly describe six major ethnic groups in the PRC of relevance in the present study. 3.1.1.1 Han In terms of population and political clout, Han are the dominant ethnic group in mainland China and closely associated with notions of “Chinese” culture and identity (Gladney, 1998, p. 11). Confucian ideals of family lineage and gender roles retain strong influence within ethnic Han society, despite efforts by the government to eliminate Confucian influences during the first 30 years of the PRC (Adamczyk & Cheng, 2015; Steward et al., 2013). Buddhism and atheism are arguably the most popular official religious orientations among ethnic Han, but Christianity is also rapidly becoming adopted throughout the cities and countryside (“Cracks in the atheist edifice,” 2014; Pew Research Center, 2012).  3.1.1.2 Manchus Progenitors of present day ethnic Manchus originated from Northeast China  and established China’s Qing dynasty (1644-1911) (Spence, 2013, p. 33). Ironically, the new  82   governing Manchus began to adopt cultural practices of their ethnic Han subjects, and the Manchu language was eventually displaced by Chinese (Crossley, 1997, p. 127; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b). Similarly, Manchu shamanistic religious practices gave way in favour of Tibetan Buddhism during their centuries of rule (Dede, n.d.; Goossaert & Palmer, 2011, p. 32). With generations of heavy intermarriage between Han and Manchus, ethnic Manchus of contemporary China have become culturally indistinguishable from Northern ethnic Han in all but name (Mackerras, 2003, p. 16).  3.1.1.3 Mongols Compared to the Manchus, cultural assimilation to Han culture has been historically more limited for the Mongols. As of 1994, 80% of ethnic Mongols could still speak their heritage language, compared to less than 1% of ethnic Manchus (M. Zhou, 2003). However, heritage language and religion is fading among ethnic Mongols of contemporary China. Practice of Tibetan Buddhism has become relatively weak within contemporary Mongol culture in China (Goossaert & Palmer, 2011, p. 369), and the dominance of Mandarin Chinese in the economy is rapidly marginalizing the use of Mongolian, as well as other minority languages (Bulag, 2003, pp. 234–5). Relations between ethnic Mongols and Han throughout periods of the 20th century were marked by tensions concerning issues of political autonomy and national identity (Bulag, 2003, p. 229), but ethnic Mongol nationalism had largely subsided by the 21st century (Goossaert, 2011, p. 369).  3.1.1.4 Uyghurs The Uyghurs of Xinjiang province are among the most culturally distinct and least assimilated into Han society (Mackerras, 2003, p. 16). In addition to their Central Asian features and Turkic-based language system which continues as the mother tongue of most today (Dwyer,  83   2005), Uyghurs boast a long Islamic tradition tracing back to at least the 16th century (Gladney, 2004, p. 213). In reality, many ethnic Uyghurs may not be practicing Muslims who abide by religious rituals such as calls to prayer, but Muslim traditions are inextricable from Uyghur culture and sense of ethnic identity within China. Unfortunately, recent concerns with violent Islamic extremism have prompted officials in Xinjiang province to issue restrictions on religious cultural practices such as fasting during Ramadan (“Beijing bans Xinjiang officials and schools from participating in Ramadan,” 2014). Controversial social policies such as this have fueled Uyghur discontent and exacerbated growing interethnic tensions over the past decade.    3.1.1.5 Hui The Hui lack their own distinct heritage language and primarily use the local language, most often Chinese (Gladney, 1998, p. 37). Essentially, the Hui ethnic classification was created based on lineage to Persian, Arab, Turkish, and Mongol migrants who settled throughout China prior to the fourteenth century and who retained their Islamic faith, but adopted local cultural practices such as attire and language (Gladney, 1998, p. 49). Due largely to their eclectic histories and geographic dispersal throughout most of the PRC, ethnic Hui of different regions engage in dissimilar cultural practices and subscribe to dissimilar notions of what defines Hui ethnicity (Gladney, 1998, pp. 166–7). Consequently, conventional Islamic practices such as male circumcision vary widely among the Hui, and are generally less prevalent when compared with ethnic Uyghurs (Mackerras, 1995, p. 117). Nonetheless, in spite of their differences, ethnic Hui in China readily endorse the Hui ethnic classification and identify with other Hui as ethnic peers (Gladney, 1998, p. 167). Relations between ethnic Hui and Han in the 21st century have for the most part been amicable and devoid of interethnic violence (Crane, 2014).     84   3.1.1.6 Zhuang Unlike other ethnicities, the Zhuang lack a unified religious system and language dialect. Each Zhuang village paid homage to local animistic deities which may or may not have been recognised by ethnic Zhuang in neighbouring villages (Kaup, 2000, pp. 43–4), but Chinese Buddhism has  also taken root among the Zhuang (Wickeri & Tam, 2011, p. 55). Myriad Zhuang language dialects also contribute to social divisions among the Zhuang, whereby Mandarin will often be the language of choice between Zhuang who speak different dialects (Kaup, 2000, p. 176). Hence, despite being China’s most populous ethnic minority group, the Zhuang possess relatively weak ethnic solidarity.  3.1.2 Religion in the PRC   The Marxist-Leninist principles upon which the Chinese Communist Party was founded are unequivocal in its commitment to atheism. According to the Chinese Communist Party’s Official Policy Statement on Religion, “We Communists are atheists and must unremittingly propagate atheism” (Chinese Communist Party, 1982). Nonetheless, leaders within the communist party recognised that sweeping attempts to abolish religion were impractical and counterproductive to their efforts at consolidating political support during the formative years of the PRC, and in 1954 established the Religious Affairs Bureau to oversee policies and activities of what would become the five officially state-sanctioned religions: Islam, Buddhism, Protestantism, Daoism (also known as Taoism), and Catholicism (Goossaert & Palmer, 2011). Public expressions of religion were later severely repressed during the Cultural Revolution in the 1960s and 1970s (F. Yang, 2012, pp. 72–3), but since the early 1980s, Chinese citizens have essentially been free to affiliate with and openly practice “normal religions” as defined by the state (Laliberté, 2011; F. Yang, 2012). In conjunction with China’s broader socioeconomic  85   reforms since the late 1970s, these religious freedoms have helped spur exponential growth in religious affiliations among Chinese citizens. National survey data indicate that China was home to 68 million Christians (5.1%), 244 million Buddhists (18.2%), and 21 million Muslims (1.6%) as of 2009 (Pew Research Center, 2012). Personal religious affiliation is often a marker for specific religious social organisational structures and beliefs. Generally, religions in China can be crudely classified into two categories: churched religions and unchurched religions (Stark et al., 2005). Churched religions are characterized by social congregations of members who subscribe to explicit institutionally sanctioned religious creeds. In China, Islam, Catholicism, and Protestantism constitute the largest churched religions, and each maintains specific normative ideals about moral behaviour. Given that members are typically expected to believe specific religious creeds and partake in social religious activities, churched religions naturally have considerable capacity to instill a strong sense of religious identity and exert social pressures among their congregants. In effect, the religious community to which an individual claims affiliation with serves as an important social reference group (Bock et al., 1983). In contrast, unchurched religions impose far fewer expectations on its members. Buddhism and Daoism are the main unchurched religions officially recognised by the Chinese state, and neither requires its members to acknowledge specific religious creeds or partake in religious social activities. Buddhist and Daoist venues (e.g., temples and shrines) primarily cater to individualized religious practices and are thus less well suited for supporting religious communities and religious social norms (Eric Y Liu, 2011, p. 145).  In addition to differences in social organisations, religious groups in China fundamentally differ in their views of the supernatural. Christianity and Islam both espouse beliefs in an  86   omnipotent higher being consciously concerned with the moral rectitude of humanity (i.e., the Lord or Allah, respectively), while Buddhism and Daoism are religions primarily founded upon beliefs in a “natural order” to the universe driven by karma and “the way,” respectively. Essentially, the core principle of karma holds that “deeds by all forms of sentient beings create life circles and consistently influence past, present, and future experiences” (Eric Y Liu, 2011, p. 144). Daoism can be understood as “the paramount force behind the natural order that keeps the universe ordered and balanced” (Eric Y Liu, 2011, p. 144). As neither Daoism nor Buddhism demand exclusive religious recognition of its members, the distinction between self-proclaimed Daoists and Buddhists in China are often blurred, with worldviews and practices being routinely and unconsciously blended (L. Fan & Whitehead, 2011, p. 21).   3.1.3 Ethnicity, religion, and HIV among MSM One important observation from the small but growing body of minority ethnic and racial MSM HIV research is that neither racial nor ethnic minority status per se is associated with greater HIV risk. The breadth of diversity between ethnic and racial groups and cultural contexts simply precludes any such universal theories. For example, in the US, African Americans represent 35% of incident HIV infections among MSM (Maulsby et al., 2014), but only ~14% of the total US population (Rastogi et al., 2011). On the other hand, Asian American MSM have consistently had lower HIV prevalence and incidence than MSM of all other US racial groups (Wei et al., 2011). Similarly divergent trends have also been observed between minority “Blacks” and “Asians” in the United Kingdom (Elford et al., 2012). In Malaysia, one cross-sectional study found that HIV infection and UAI were both significantly more prevalent among ethnic majority Malay MSM than ethnic minority Chinese MSM (5.3% vs. 1.8% and 49.4% vs. 39.8%, respectively) (Kanter et al., 2011).   87   Nonetheless, cultural values and norms within ethnic groups have the potential to exert tremendous influence on HIV vulnerabilities and testing patterns. Values such as filial piety, obedience, religious devoutness, and heteronormative masculinity all serve as the moral foundations that delineate which sexual behaviours are to be socially celebrated, tolerated, or stigmatized. Given the range in cultural values by ethnic group, it is unsurprising that homonegativity manifests more acutely in some ethnicities than others. In the US, black respondents were twice as likely as white respondents to report that homosexuality is “always wrong” (S. N. Glick & Golden, 2010), due in part to heteronormative Christian codes of sexual morality and traditionally masculine archetypes of the “Strong Black Man” within African American communities (Dillon & Basu, 2013). In light of the harmful implications of homonegativity on HIV vulnerabilities discussed in section 1.3.2.3, affiliation with ethnic groups that espouse strong homonegative values can be potentially detrimental to the sexual health of MSM. Results from one Canadian study indicated that South Asian MSM in Toronto and Vancouver had greater internalized homonegativity than white MSM, and that endorsement of South Asian heritage cultural values was associated with UAI  (Ratti, Bakeman, & Peterson, 2000).  For many men, religion is a fundamental framework around which their lives and world perspectives are organised, and which is often intertwined with ethnic affiliation. Due to heteronormative and austere religious codes of sexual morality, MSM of certain religious faiths (e.g., Islam or Christianity) may be more likely to encounter and internalize stigmatization of their sexual orientation and sexual behaviours (Keogh, n.d.; Kugle & Chiddy, 2009; Siker, 2007; Smallwood, 2013; Wagner et al., 2012; Wilson et al., 2011). Homonegative religious sexual norms in particular might inhibit same-sex sexual behaviours, but may also limit information and  88   access to health services such as HIV testing and counseling by socially isolating MSM of faith from other MSM (Andrinopoulos et al., 2014; Santos et al., 2013). Currently though, epidemiological data of HIV vulnerabilities among MSM by religious affiliation remains scarce and inconsistent. Results of the few such studies indicate that Muslim MSM in West Africa appear less likely to test for HIV and more likely to have sex with females when compared with their Christian counterparts (Lorente et al., 2012; Sheehy et al., 2013). In Israel, Muslim MSM appear to have lower HIV prevalence than their Jewish counterparts (Mor et al., 2013). However, no significant differences in HIV vulnerabilities were observed between religious groups among Malaysian MSM (Muslim, Hindu, Buddhist, Christian, no religion, and other religion) (Kanter et al., 2011) or between Christian denominations among US MSM (Garofalo et al., 2014). 3.1.4 Ethnic and religious MSM in China Ethnic and religious minority MSM remain a neglected and poorly understood sub-group within Chinese society. Despite burgeoning social science research of sexual and ethnocultural minorities in China, both populations are often conceptualized with little consideration of the other. Studies of male sexual minorities in China have rarely accounted for ethnocultural minority experiences, and vice versa. Nonetheless, limited evidence indicates that Christian- and Muslim-affiliated MSM in China are prone to greater internalized and perceived external homonegativity (Chan & Huang, 2014; Kamaliti, 2011; Tong Ge, 2005, pp. 130–1). In addition to MSM stigmatization in mainstream Chinese society, many Christian and Muslim MSM in China must also contend with the religious-based stigmatization of same-sex sexual behaviours within their religious social reference groups. As elsewhere in the world, mainstream Christian and Muslim institutions in China view same-sex sexual behaviours as inherently immoral acts which violate religious values (Chan & Huang, 2014; Kamaliti, 2011). All major Buddhist sects  89   discourage acts of “sexual misconduct” in a generic sense (Corless, 1998, p. 253; Mitchell, 2008, p. 121; Sweet, 2007, p. 76), but conflicting interpretations of what constitutes “sexual misconduct” has led to differing contemporary Buddhist perspectives on the moral significance of same-sex sexual behaviours in China, whereby Buddhism within China today generally lacks a dominant religious view on the moral implications of same-sex sexual behaviour (Bao, 2012). To date, few if any public health studies have yet specifically addressed HIV vulnerabilities among ethnic or religious minority MSM in mainland China. 3.2 Objectives and rationale In response to the stark lack of information concerning the intersection of ethnicity, religion, sexuality, and HIV in China, this qualitative inquiry sought to generate contextually-rich accounts of “embodied experiences” (Starks & Trinidad, 2007) that could enhance understanding of how ethnicity and religion are shaping HIV vulnerabilities and uptake of HIV testing services among MSM.  3.3 Methods Between February 2013 to December 2013, I conducted and analysed qualitative in-depth interviews with 40 ethnic and religious minority MSM in Beijing, China.  3.3.1 Study site: Beijing Beijing ranks among China’s most prosperous and prestigious cities. In 2013, Beijing’s gross regional product (GRP) per capita was 2.2 times higher than the national average (~$15, 000 USD vs. ~$7,000 USD), and post-secondary educated residents as a proportion of the total regional population was almost quadruple the national rate (32% vs. 9%) (National Bureau of Statistics of China, 2014). Its robust economy and cosmopolitan social scene draws millions of migrants from across the country and around the world, whereby 37% of Beijing’s 20 million  90   inhabitants were migrant residents in 2010 (C. Huang, 2014). Although all 55 ethnic minority groups of China are represented among Beijing’s population, 96% of Beijing residents are ethnically Han (Beijing Statistical Information Net, 2011), and standard Mandarin is the dominant dialect in most personal and professional spheres. Manchus, Hui, Mongols, and Koreans constitute the vast majority of Beijing’s ethnic minority population (87%) (Beijing Statistical Information Net, 2011). Hundreds of Buddhist, Christian, and Muslim religious venues are located throughout the city, though most residents do not have any specific religious affiliation (China.com.cn, 2008). Beijing’s sexual minority community is among the most vibrant and socially active in China, with an estimated MSM population size between 115,731 and 239,258 persons (Guiying et al., 2014). Since 2009, HIV prevalence estimates for Beijing MSM have ranged from 2.5% to 8.0% (Yunhua Zhou et al., 2014).   3.3.2 Recruitment & enrollment   Interviewees were recruited using purposive maximum variation sampling based on ethnic affiliation, religious affiliation, age, education, and migrant status (Clark & Creswell, 2010, p. 174). Most study informants were introduced to me by health departments, sexual minority advocacy organisations, and other study participants. A minority of informants were recruited via gay social networking websites and MSM HIV testing sites. Proportional modes of recruitment and participant sociodemographics are presented in Figure 2.4 and Table 2.1, respectively.  Face-to-face interviews were conducted at a safe and quiet location determined by the interviewee and me, and included venues such as restaurants, coffee shops, private homes, and office workplaces. Due to potential privacy concerns, participants were free to use pseudonyms, and no true identifiers were requested. Men were eligible to participate in the survey if they self- 91   reported ever having sex with another man (oral or anal sex), were at least 18 years old, were willing and able to provide written informed consent, and satisfied the purposive maximum variation sampling requirements at the time of enrollment. Based upon local transportation and meal prices, all interviewees received an honorarium of 50 CNY. Study protocols were reviewed and approved by the National Center for AIDS/STD Control and Prevention, Chinese Centers for Disease Control and Prevention, and the Providence Health Care Research Institute, University of British Columbia. All names in the present study are fictionalized pseudonyms.  3.3.3 Data collection Qualitative data originated from in-depth open-ended interviews conducted in Mandarin. All interviews were audio recorded, and lasted an average of 62 minutes (range: 21 – 96 minutes).  A topical guide covered issues such as perceived significance of ethnic and religious identities, personal sexual milestones (e.g., sexual debut), and sexual decision-making processes (e.g., considerations when seeking sex partners). However, given the nature of the in-depth open-ended interview, not all questions in the topical guide were addressed in the interview, and vice-versa.  3.3.4 Transcription All interviews were transcribed verbatim into simplified Chinese by China CDC project staff. In order to retain the linguistically sensitive nuances of participant expressions, all interview transcripts were analysed in the original Mandarin Chinese.    92   3.3.5 Data analyses Interpretive thematic analysis  (J. Green & Thorogood, 2009, pp. 13–14; Starks & Trinidad, 2007) was conducted for all qualitative data (i.e., interview transcripts and field observations) and occurred iteratively throughout the data collection and write-up phases of the dissertation project. Through recursive decontextualization and recontextualization of each qualitative datum vis-a-vis broader themes and patterns, interpretive thematic analysis enabled deeper understanding of participants’ lived experiences and decision-making processes relevant to HIV vulnerability and testing. Interview transcripts were read multiple times for the interrelated purposes of coding, interpretive analysis (Starks & Trinidad, 2007), and verification (Morse et al., 2002). In the first coding cycle, I sought firmer understanding of “what was going on” by coding interview transcripts with line-by-line descriptive and process coding approaches (Saldaña, 2009, pp. 70, 77). Each participant’s first cycle codes were then enumerated in a multi-columned word processing document. Consolidating each individual’s codes into a visually compressed format enhanced my ability to postulate thematic patterns, recognise deviant cases, and perform “constant comparisons” within and between interviewees (Charmaz, 2007, p. 187). Based on the first cycle codes of each interview, analytic memos were drafted with the explicit aim of contextualizing interview findings in relation to evolving theories derived from other data sources in the study. As a form of investigator triangulation (Mathison, 1988), two bilingual China CDC project staff verified my coding decisions and analyses by reviewing all coded interview transcripts, code summary documents, and analytic memos. Disagreements in coding or analyses were resolved by document revisions and face-to-face discussions. Preliminary  93   themes and hypotheses were also subjected to evaluation during a data feedback presentation to a sexual minority advocacy group in Beijing on August 5, 2013.  After all first cycle codes had been finalized, second cycle pattern coding was conducted for the purposes of revising, distilling, and classifying preliminary codes into higher order thematic categories of interest (Saldaña, 2009, p. 152). First, a qualitative data analysis software package, Atlas.TI (Scientific software, Berlin), was used to digitally code interview transcripts. Then, each second cycle code was printed onto strips of paper which were shuffled and grouped with other thematically similar codes. This process of physically organising and reorganising codes augmented my ability to recognise conceptual themes and infer their relational patterns. These hierarchical themes subsequently formed the scaffolding of the qualitative write-up, supported by illustrative quotes and observations. Chinese-to-English translations of quotes were completed by me.  3.3.6 The researcher as the qualitative research instrument Prior to data collection, I received four months of formal training in qualitative research methods.  Since project conception, I remained vigilant about how my own embodied identities and pre-existing beliefs uniquely shaped the qualitative data and its analyses. Frequent memo writing throughout the data collection phase enabled me to reflexively examine the study implications of my own subjectivities and identity as a dual national (United States & Republic of China), ethnically Han, Christian man who had never had sex with another man. Pre-existing beliefs were bracketed to the greatest extent possible (Starks & Trinidad, 2007).   94   3.4 Results Three main themes relevant to the research question emerged from the qualitative data analyses: (a) Sociocultural worldviews, (b) Negotiating ethnicity, religion, and sexuality, and (c) Ethnocultural bias. 3.4.1 Sociocultural worldview One of the most profound impacts of ethnic and religious affiliation was its potential to influence how interviewees constructed their social identities and assigned meaning to human experiences. Ethnic and religious identities served as powerful interpretive frameworks through which interviewees understood deeply personal philosophical questions concerning personal social identity and ultimate causality of events. 3.4.1.1 Ethnic social identities  When explicitly asked about how their ethnic group differed from the Han in contemporary China, many ethnic minorities characterized their ethnic group as essentially the same as ethnic Han. To Akhun, a young 20+ year old Muslim Uyghur, “The only difference is religion. Other than that, there aren’t any differences [between Han and Uyghurs].” Nonetheless, interviewees’ conceptualizations about personal ethnic identity were revealing about how ethnicity shaped their everyday social life. No cohesive “pan-ethnic minority” group identity emerged, but each ethnic minority interviewee invariably reached some type of understanding about the personal significance and meaning of his personal ethnic minority status. 3.4.1.1.1 Devalued, yet undeniable ethnic identities “Han-icization” (hanhua) was the slightly pejorative term used by participants to describe the process of ethnic minorities losing their cultural distinctiveness as they are assimilated into Han Chinese society. “Han-icization” was believed to be much stronger in urban cities where  95   ethnic Han typically represented a greater proportion of the population. Indeed, ethnic minority interviewees who grew up in major urban cities tended to exhibit a weaker sense of ethnic pride, and described their ethnic status matter-of-factly as a bureaucratic designation which held little personal meaning to them.  One ethnic group commonly cited for their heavy assimilation into Han society were the Manchus. Hu, a stout no-nonsense middle-aged man who grew up speaking Mandarin even disassociated himself from his Manchu ethnicity at one point during the interview. Now, I don’t even think of myself as a Manchu. I feel like we’re about the same as ethnic Han. At most, the only difference is what’s written on our household registration booklet.   [Hu, 40+ years old]   Similar comments diminishing the personal significance of ethnicity were commonly echoed by respondents of other ethnicities to varying degrees, not just ethnic Manchus. According to Zhu, a taciturn ethnic Hui man in his twenties who grew up in Beijing, being Hui represented little more than practical educational benefits provided by the government.  Interviewer: What does your Hui ethnic identity mean to you?  Zhu: There isn’t any meaning.  Interviewer: There’s no meaning at all? Zhu: What meaning is there? I’m not going to school now, so I’m not getting bonus points or special privileges. There’s none of that. I’m working now. It doesn’t make any difference whether I’m Hui or not.   [Zhu, 20+ years old]  Devaluation of one’s own ethnic status typically began at an early age and was at times actively encouraged by the individual’s family, as in the case of Cheng, a sanguine and reflexive young ethnic Hui. My parents would point out to me when I was acting like an ethnic Han, because [they felt] ethnic Hui were not that modern, not that open, not that capitalistic, and not that egalitarian or even-handed … My mom is an ethnic Han, and she doesn’t approve of  96   some ethnic Hui customs and the faith. That filled me with a lot of negative messages … Before I turned 18 and before I went to college, I identified with my parents’ view point, [I would] avoid interacting with ethnic Hui and avoid thinking about religion.  [Cheng, 20+ years old] These remarks from Hu, Zhu, and Cheng are telling, in that they reveal how strongly some ethnic minorities have internalized social marginalization of their ascribed ethnic identities. The ethnocultural lineages from which they descend are perceived to be meaningless or even useless, and yet, their immutable ethnic minority status acts as an impermeable barrier that permanently excludes them from formally becoming a member of the ethnic Han majority. Regardless of their personal opinions or practices, ethnic minorities are legally prohibited from taking on an ethnic status different from that of their parents (Mullaney, 2011, p. 123). Moreover, because of government sponsored social benefits designated for ethnic minorities (United Nations’ Committee on the Elimination of Racial Discrimination, 2009b), parents with discordant Han-ethnic minority statuses may be more likely to assign ethnic minority status to their children, rather than Han status. For those who are unable or unwilling to conform to the social norms and expectations of the ethnic minority group to which they have been categorized, weak ethnic identity affirmation and confusion about competing ethnocultural reference groups can ensue.  3.4.1.1.2 Evolving attitudes towards ethnic identity   While participants’ formal ethnic status may have been immutable, it was quickly apparent that perceptions and attitudes towards personal ethnic identity affirmation were often in flux, sometimes to the point of apparent contradiction. Hu explicitly dismissed any significance of his ethnicity early on during the interview, but towards the end he proceeded to express discontent that “society doesn’t really appreciate Manchus.”   97   For those who grew up without strong affirmation of ethnic identity, actively developing ethnic identity affirmation seemed to provide ethnic minority interviewees with an enhanced sense of purpose and understanding of where they came from. Jiang described the relatively late process by which he cultivated a more personally meaningful Zhuang ethnic identity. I was in my thirties when I finally identified with my ethnic status … I gradually understood Zhuang history and culture, and only then found out what it was all about. Like, why, where did I come from, and what’s our historical origin … now I’ll constantly be introducing others to Zhuang history and culture whenever I get the chance. And so my friends are like, “How do you have such a sense of purpose?”  [Jiang, 30+ years old] 3.4.1.2 Religious worldview Compared to ethnic identities which were largely an inherited and static status, religious affiliations were self-determined and subject to change over time. Overall, religion proved to exert a discernably stronger influence on participants’ interpretations of human experience than ethnicity. Across Buddhists, Christians, and Muslims alike, three transcendent themes emerged: religious social identities, belief in a higher being, and cosmic moral order. 3.4.1.2.1 Sense of religious social identity Whether initially indoctrinated by family or later in life by other means, religious affiliations marked interviewees’ sense of membership within a larger community of others with similar values and beliefs in the sacred, regardless if they participated in religious social activities or not. Compared to ethnic status, religiously-affiliated interviewees generally projected a notably stronger sense of religious pride, as expressed by Li, an ethnic Hui. This [ethnicity] isn’t for me to determine. My father’s ancestors, they themselves were Hui. I can only inherit this from them. I don't feel any special honor or anything. If there is something to be honored about, it’s that I am a Muslim. [Li, 40+ years old]  98   Among religiously affiliated men, religious communities functioned as a key social reference group from which interviewees learned and negotiated religious normative beliefs, membership expectations, and simply what it meant to be a Buddhist, Christian, or Muslim. Attending a traditional Christian church for many years, Sōng (30+ year old) spoke of constantly “experiencing pressure” from other Christian “brothers and sisters” to get married, according to the church’s heteronormative values. However, as Sōng spent more time interacting with a sexual minority Christian organisation, he began to adopt new social expectations of what it meant to be a Christian. In Sōng’s own words, “I stopped placing so much weight on the critical and discriminatory perspectives of the traditional church.” Among Muslim participants, practices of male circumcision and abstention from alcohol varied widely and were not closely linked with sense of religious identity. Multiple Muslim interviewees did acknowledge that Muslim males “must be circumcised” and that Muslims “are not permitted to drink alcohol,” but such proscriptions seemed inconsistent with the self-reported behaviours of many Muslim participants. When asked to explain why they were not circumcised, several Muslim interviewees cited “fear of pain” as the main reason. Moreover, male circumcision was not essential to many participants’ sense of Muslim identity and was often viewed as optional, as in Xie’s experience. Regarding circumcision, your family might require you to, maybe if you have an elder in the house who places a lot of weight on this, that is, if they’re really strict about it, then you get cut [circumcised], as a co-worker told me. But I didn’t get cut. [Xie, 30+ years old]    Similarly, few Muslim participants completely abstained from consuming alcohol. Given the pervasiveness of alcohol in male homosocial culture in China (G. Song & Hird, 2014, pp. 193–5), many Muslims would acquiesce to social pressures and “occasionally drink some alcohol.”  99   One young Hui Muslim man described “drinking everyday” during a time of extreme duress, while another Muslim Uyghur man spoke of “drinking too much” and then having anal sex with another male friend. In both cases, the interviewee did not perceive that becoming inebriated threatened their strong sense of Muslim religious identity. 3.4.1.2.2 Omnipotent higher power  Religious interviewees, especially Christians & Muslims, expressed convictions in an omnipotent and omnipresent higher being who consciously monitored and regulated the affairs of their everyday lives.  For Ma, a devout Muslim man, this supernatural entity was manifested in the form of Allah. If one day you’re on a road and a leaf falls from a tree and brushes upon your hair, or falls beneath your feet and you step on it – these minute events, all of them have already been arranged by Allah. It includes your marriage, your health, your fortune – whatever. Before the world existed, Allah already predetermined it for you. [Ma, 30+ years old] In light of such beliefs, it was understandable that individual sexual orientation was seen as but one of infinite issues governed by karma, God, or Allah. Whether same-sex attraction was perceived to be determined by genes, the social environment, or some combination thereof, most religious participants ultimately attributed their sexual orientation to supernatural circumstances.  So, me being a homosexual for my entire life, liking men, this is also something that Allah has let me become, something that he has arranged and did. So then what am I able to do? What can I do to relieve or change these restraints? It can’t be changed. In China, there’s a saying: Take things as they come. Allah has already determined me to be a homosexual, so I can only accept it.  [Ma, 30+ years old]      100   3.4.1.2.3 Cosmic moral order Belief in some type of cosmic moral order to humanity was ubiquitous among all religiously affiliated interviewees, and critically informed their explanatory understandings of illness. Such moral order was believed to be adjudicated by karmic balance among Buddhists, God among Christians, and Allah among Muslims. Cao, a middle-age Buddhist man, viewed his mother’s death as the direct consequence of her transgressions earlier in life. My mother and father, they were never harmonious. When I was small, they fought very intensely, “You won’t yield to me, then I won’t yield to you.” In the end, my mom died … I would exhort them “don’t always be fighting, don’t fight”, I would say to them, “what good is this fighting and anger and anxiety going to do?” And look, in the end … after ten days, [she’s ]dead. Isn’t this also a karmic cycle? [Cao, 40+ years old]  The suggestion that Cao’s mother died prematurely because of her marital quarreling reflected a more general belief that negative experiences were a logical function of morally reprehensible behaviours. When asked what he thought about people infected with HIV, Cao again viewed the situation in the context of a broader cosmic moral order.  Infected people, well, they admit to being doomed, so what can I do to help? If I had a cure, I’d give it to them right away. So, what can they do? This is called karma. This is the expected retribution – so, why is it that other people can’t get infected but it’s just you that was infected?  [Cao, 40+ years old] Sun, a young and passionate young Christian, applied similar moral interpretive frameworks and perceived HIV infection as “a means by which God speaks to you” to discourage wayward sexual behaviour. Supernatural retribution, however, was not always seen as inevitable. Several individuals believed they could redress their moral debt by “repenting” (hui gai), doing “good deeds” (zuo hao shi), or atoning for their sins (shu zui), irrespective of a natural causative relationship between the moral behaviour and positive outcome.   101   3.4.2 Negotiating ethnicity, religion, and sexuality When I first started having sex, I had three conflicts. The first kind of conflict was the conflict between my ethnic faith and the Christian faith. The second kind of conflict was the conflict between being tongzhi and Christian faith. The third type of conflict was the conflict between tongzhi and my ethnicity. The relationships between these three were like a triangle. They were incompatible.  [Sun, 20+ years old]  The intersectionality of ethnic, religious, and sexual minority identities engendered a diverse array of cognitive and behavioural responses relevant to HIV vulnerability. Overlapping ethnocultural and sexual identities appeared to have nominal implications for HIV vulnerability among some (e.g., areligious Manchu), while were of enormous consequence for others (e.g., Muslim Hui). But despite these variations, the one unifying experience shared by all ethnocultural minority MSM was their common process of negotiating what it meant to be an ethnic and/or religious minority MSM. With respect to HIV vulnerability, two key themes emerged: (a) Reconciling religious identity and sexual behaviours, and (b) Religion and sexual behaviours.    3.4.2.1 Reconciling religious identity and sexual behaviours The ability to harmoniously reconcile religious identity and sexual behaviours depended upon personal views concerning the moral significance of same-sex sexual behaviours. Those who felt that sex between men inherently violated religious values spoke of fear and frustration about the immutability of their erotic attraction to other men and its implications for their spiritual well-being. On the other hand, such negative affect was less evident among participants who contested orthodox religious proscriptions of sex among men and thereby did not perceive dissonance between their religious identity and sexual activities. Of course, this binary moral vs. immoral characterization is an imperfect representation of all interviewees’ experiences. In  102   reality, several individuals were ambivalent if having sex with men contravened their personal religious beliefs, and many described perspectives that evolved non-linearly over time.  3.4.2.1.1 Perceived incompatibilities of religion and sexuality Intense homonegativity was expressed by many who felt that having sex with men violated their codes of religious moral conduct. To Li and several other religious interviewees, being MSM was portrayed as being despised by your own community and suffering eternal spiritual condemnation.    According to principle, we Hui hate these type of people the most, especially homosexuals. Hate them the most. After you die, you can’t go to heaven and you can’t go to hell. You are put in a small black room tortured without anywhere to go. According to the sacred texts, this seems like what happens. Hate these homosexuals the most. But at the same time, I can't help myself. [Li, 40+ years old]  The fear of divine retribution for having sex with other men was also palpable for Ma. Eventually, perceived incompatibility of his religious beliefs and sexual inclinations became a continual source of mental anguish that even led to self-inflicted harm.    Before, I wasn’t able to accept anal sex because it was really painful the first time I tried it. Also, because of our Muslim faith, a part of me felt scared and that it was taboo ... When I first realized that I liked this, when another man gave me oral sex and I sensed that I liked men, I got really scared, psychologically. I was scared of Allah’s punishment. Islam forbids men from having sexual relations with other men. It’s a really serious sin. At the time, I was really scared, but I just wasn't able to suppress the longing desires and needs. For a while, I tried to stop. I told myself that if I kept doing this, Allah would severely punish me and send me to hell after I died. Later on, I just wasn’t able to quit. Another issue is what happens if my family or people back home find out? How could I ever go back and face my family or people back home? I never smoke, but for a couple of days back then, I was really conflicted so asked a buddy for a pack of cigarettes. I smoked the cigarettes, and then used the lit cigarettes to burn my arm, vowing to never do it with men, never think about men, and never have sex with men. I smoked five or six cigarettes in a row, resolutely making burns here [motions toward arm], making burn scars right here.” [Ma, 30+ years old]  103   Against their broader religious beliefs in an omnipotent higher power and cosmic moral order, Christian and Muslim interviewees often grappled with the uneasy question of why God or Allah intended for them to be sexually attracted to other men. Ma went on to speak of his ongoing quest to find greater meaning in his attraction to other men, and the toll it had taken on him.    The only thing I'm afraid of or worried about is my religious faith. I haven't been able to get over the fear. I've continuously been searching for answers about what to do. Does Allah insist on me being a homosexual? Is he testing me? I feel like this is a test that is too painful and too heavy to bear. The temptations are a bit too unbearable. [Ma, 30+ years old] Belief in an omnipotent higher being also became an acute source of painful frustration for some because their immutable erotic attraction to other men was often interpreted as a sign of Allah or God’s refusal to grant their requests for a more heteronormative sexual orientation. It’s been very hard for me to accept my own [sexual] identity, and I’ll want God to change me. I’ve cried and prayed many times over this, prayed incessantly. But there’s never been a real change. It feels really, really difficult. And, it also affects my confidence. I’ll wonder why God is letting me experience such difficult circumstances and won’t change me. I beg, from this kind of situation, I beg that if only my sexual orientation can be changed, many problems would be easily resolved. I’m very willing to become part of mainstream society.  [Sōng, 30+ years old] The preceding vignettes illustrated how some participants suppressed their sexual inclinations as a means of coping with the perceived conflict of being a Christian or Muslim MSM. However, several participants who held orthodox religious moral views about homosexual sex employed alternative strategies that appeared to alleviate the internalized mental guilt and shame of violating their religious moral tenets. For example, in alignment with his beliefs in a cosmic moral order and a fungible nature of moral transgressions and virtues, Xie, a  104   Muslim ethnic Hui, described how performing “good deeds” counterbalanced what he considered to be moral lapses.  I feel that, when it comes to religious rules, maybe it says that you shouldn’t do it [men having sex with men], but it’s already happened, so you just use other ways – for example doing some public service to make-up for it, or going to more religious ceremonies and doing good deeds and the like. [Xie, 30+ years old] In contrast to suppressing sexual inclinations or redressing moral debts, some Christians and Muslims coped with the perceived discord between religious faith and sexual practice by simply accepting their sexual behaviours as moral transgressions. Unwilling to conform to heteronormative expectations, Yuan consciously embraced his sexuality, in spite of beliefs that doing so could lead to punitive repercussions in his afterlife.     I consider myself a Muslim, but so what if I am a Muslim? I was born as this type of person, “gay” that is. But I’m unwilling to force myself to do something I don’t like. What I’m saying is that, how do you say, when asking about sin, some things will happen after death, the religion says so, doesn’t it? But I feel that, I came into the world and that I should think for myself. It’s better to be happy. Happiness is all you need. [Yuan, 20+ years old]  3.4.2.1.2 Identity integration Virtually all Christian and Muslim participants felt that their religion’s mainstream organisations stigmatized same-sex sexual behaviours based on religious codes of sexual morality, but many personally dissented from this orthodox position and were able to successfully cultivate an identity that integrated their religious faith and sexual behaviours. For Cheng, an effervescent Muslim graduate student, Islamic proscriptions against sex between men were simply an anachronism in the 21st century.  The Koran’s original text forbids [incoherent audio]. It indicates that a man is forbidden from copulating with another man, and that a man is forbidden from loving another man. So, from the wording, it seems like this issue is pretty serious. But it’s like I said before, I  105   feel that this has historical reasoning … I believe this was written at that time in order to maintain population levels, whereas today we have already attained good economic and health conditions, and the peace time population has stabilized … When Muhammad was creating or spreading Islam, he was considering society’s conditions of the time, but today’s society has changed. So, I don’t think this is Allah’s intention, because when Allah creates people, he already decides their fate … So if a person’s fate has been decided, how can you say this isn’t God’s will? It is God’s will that I like people of the same sex. [Cheng, 20+ years old]  Reinterpreting religious texts was the means by which several participants were able to reconcile their religious identity and sexual practices without any notions of cognitive identity dissonance. According to such alternative interpretations, men having sex with other men per se never was or was no longer a breach of proper religious moral comportment. Xu believed that rules governing behaviour in the Bible had “been added on by man,” and Ding similarly seemed to question the divine origins of the Bible. This Bible, it was written by many people, different people wrote it. Thus, it will definitely be influenced by the time within which people lived [when it was written]. They will include their social perspectives. They’ll set out a few regulations and write them in. Hence, I am skeptical about the requirements that were written in the end. [Ding, 30+ years old]  Ultimately for Xu, God’s unconditional love for humanity subverted the dominant homonegative religious narrative. Those straight Christians have a respect for, reverence of traditional religion. They always think that there are multiple places in the Bible that say “tongzhi” behaviour is not right, that it’s wrong, that it’s something which displeases God. Yeah. So they use that kind of reasoning to argue against those Christians like us, “tongzhi” Christians. But, we will refute them. Actually, God loves all people the same, he won’t cast unscrupulous blame on you or wish for you to go to hell later just because you are a minority. No, he won’t have that attitude. [Xu, 30+ years old]   106   Several Christians and Muslims who were not raised in religious environments did not face any difficulties adopting a harmonized Christian or Muslim MSM identity. Typically, awareness of their sexuality was relatively well established by the time they adopted a Christian or Muslim affiliation. On the other hand, reconciling religion and sexual practices proved to be a lengthy and painful process for many who were exposed to heteronormative religious social norms at an earlier age. Sun became a Christian in high school and struggled for several years with what it meant to be a Christian MSM.  Back when I thought that being “tongzhi” was a sin, I felt particularly tortured, particularly unworthy to God, and I would go repent. My relationship with God wasn’t very good. I never wanted to read the Bible or feel like praising God because I felt like, if you think that me being “tongzhi” is a sin, then why did you let me become a “tongzhi,” and then say that you don’t like “tongzhi?” ... Now I joined this Beijing fellowship, a fellowship where everyone is a “tongzhi” Christian, because there isn’t a church in Beijing that openly accepts homosexuals … Lots of people might be Christians, but after finding out they are homosexual, will leave the church and leave God. Because they can’t change this part of themselves, they can only leave the church. The church is not accepting and intolerant, so they leave. So we feel that that this is a kind of misinterpretation of God. God loves all people, it doesn’t matter who. As long as you are willing to come to God and repent, willing to accept his values as your own values, I think that is enough … So, I feel like I am doing very well now. I don’t feel tortured from being “gay.” [Sun, 20+ years old]  Like Sun, several other Christian interviewees attended a tongzhi Christian fellowship and described its importance as a source of support during their quest to understand what it meant to be a Christian MSM. For Sōng, the atmosphere of the tongzhi Christian fellowship was inimitable.  I feel like the fellowship has given me a lot of support psychologically. Because after all, we have the same [tongzhi] status and the same faith. We have a lot in common, and there’s a lot of things I can openly talk with them about. It can help get rid of this kind of loneliness, yeah, it’s a really good way. At the same time, you can get a lot of support and encouragement from everyone.  [Sōng, 30+ years old]   107   3.4.2.1.3 Religion in context: mainstream heteronormativity and homonegativity Regardless of whether participants believed that having sex with men contravened their religious moral tenets, all interviewees, religious and non-religious alike, experienced pressures of social conformity and stigmatization from heteronormativity in mainstream Chinese society. For Xu and many other interviewees, the main source of distress came not from an internal conflict between religious faith and sexuality, but from the personal desires and perceived social obligations as a son to marry, procreate, and continue their family lineage.  At home, I always express how I don’t want to get married, this attitude about not wanting to get married, and it really grieves my parents. Oftentimes while we’re eating together, they will shed tears and suddenly start to cry, and I will feel like a very unfilial child letting them become so hurt. Yeah, this is the biggest part … I attach great importance to my parents as the one and only child. I wish for them to be happy. They really just want me to get married and have children, to live a normal person’s life. So, I want to satisfy their needs. Moreover, I’m also envious about those who can have a family and know that feeling … I still want to have my own descendants, a life that I can pour my love into…   [Xu, 30+ years old]  It is worth noting that none of the Buddhist participants described or implied personal conflict between their religion and having sex with men per se. According to Deng, a 20+ year old pious Buddhist, “there is no mention of this [sex between men] within the classic scriptures from Buddha’s teachings during his time on Earth.” Nonetheless, Buddhist MSM were not immune from the negative effects of mainstream homonegativity. Shen, a devout Buddhist man, spoke of his ongoing struggle of coming to terms with his sexuality.  Before, I just didn’t accept myself, it was that kind of a tendency. There were also thoughts about suicide … Why did I have to be like that? Why couldn’t I be like normal people? It was like this not being able to accept myself that led to self-hating …. If there were a pill that could change my sexual orientation, I would buy it, no matter how expensive. Now, if I’m honest, I really would still want to. But, this can’t be changed, I can’t be changed.   108   [Shen, 30+ years old]  3.4.2.2 Religion and sexual behaviours Among Buddhists, Christians, and Muslims alike, sex was closely tied with religious codes of conduct and laden with deep moral significance. Participants described how personal religious beliefs about sexuality strongly influenced their attitudes and decisions about sex in general, with or without condoms. Embedded within their broader worldviews about an omnipotent higher power and cosmic moral order, two major themes concerning religion and sexual behaviour emerged from the interviews: (a) Boundaries of a moral sex life and (b) Sexual inhibitions.  3.4.2.2.1 Defining the boundaries of a moral sex life In terms of sexual behaviour, the prevailing effect of religion, regardless of affiliation, was to inhibit participants from acting upon their sexual desires. However, the form and degree of such religious sexual inhibitions varied considerably given the diversity of theological perspectives about what constituted religiously sanctioned sexual conduct. Moreover, each individual developed their own dynamic personal understanding of what defined a moral sex life, among which three distinct perspectives emerged: (a) total abstention from sex with other men, (b) monogamy with either a man or woman, and (c) avoidance of undefined promiscuity. Christian and Muslim interviewees overwhelmingly perceived that their respective mainstream religious organisations deemed sex among men as inherently immoral. Speaking on MSM and mainstream Christian churches, Luo (20+ years old) felt that “most traditional churches still don’t accept it. They will only accept you, but they won’t accept this behaviour …. They’ll urge you to change.” Similar sentiments were echoed by Akhun: “The religion doesn’t allow it [sex between men]. It’s because we, Muslims all say that the Koran, it says so in the  109   Koran.” In short, according to this orthodox theological perspective that numerous Christians and Muslims subscribed to, it was not possible to have sex with men and live a religiously moral sex life. In contrast to the orthodox definition of a moral sex life, members of a sexual minority Christian organisation believed that the sex of one’s sex partner had no bearing on one’s moral standing. Rather, they contended that a moral Christian sex life for MSM was characterized by remaining committed and faithful to one long-term partner centered on God. Thus, the sexual minority Christian fellowship actively promoted monogamous same-sex sexual relationships, but also discouraged casual sex on moral grounds. Sun’s narrative summarized this emerging theological perspective that in-part repudiated the orthodox Christian tenets of sexual morality. Notably, no analogous sexual minority Muslim organisations were mentioned during interviews.  My enlightened pastor accepts me. He said that being “tongzhi” isn’t a sin. If you want to talk about sin, then everyone has sin, everyone’s a sinner. So, I asked him if “tongzhi” behaviour is a sin, and he said “It isn’t a sin, but if you do 4-1-9 [one night stands] or have multiple partners, then I don’t approve. What I do support is you needing to have a steady partner, putting safety first. [Sun, 20+ years old] General non-promiscuity was the  third religious view of how MSM should live a moral sex life, and  only stipulated that MSM should avoid a “promiscuous” or “licentious” (luan) lifestyle, without consensus on what “promiscuous” or “licentious” behaviour entailed. This differed from the monogamy perspective in that concurrent sexual partnerships were not necessarily a violation of sexual morality. This type of perspective was most common among Buddhists. When asked about how their religion impacted their sex life, respondents of the “non-promiscuity” (bu luan) perspective felt that religion “restrained” their sexual activity and “during sex, [religion] prevented indulgence.”  However, due to its abstractness and lack of  110   explicit thresholds delineating moral and immoral behaviour, the “non-promiscuity” position of sexual morality was almost totally subject to personal interpretation. Unlike Christian and Muslim interviewees, Buddhist participants did not suggest that maintaining a moral sex life required abstention from sex with other men or monogamy. 3.4.2.2.2 Sexual inhibitions Based partly upon these myriad religious perspectives of how MSM should live a moral sex life, numerous interviewees depicted how personal religious beliefs affected their sexual behaviours, especially among those who believed that sex among men or sex with multiple partners was immoral. Unable to reconcile his religious beliefs with his sexual inclinations, Ma decided to abstain from sex with men. So I found out that I liked men and thought ‘How can this be?’ ‘How is this possible?’ It’s not permitted, our religion doesn’t allow this, Islam despises this type of thing. I had a lot of [psychological] pressure, so I burned those scars [on myself] wanting to quit. At the time, I didn’t want to, or didn’t go to these kinds of venues, and didn’t interact with these kinds of people. It [lasted] three months, almost three or four months, just less than half a year. [Ma, 30+ years old] On the other hand, some men limited their number of sex partners because of religious proscriptions against sexual excess in general. Xie had no intentions to stop having sex with other men, yet still felt that religious faith inhibited his sexual activities. I feel that Islam might help keep people from doing promiscuous things. There is a doctrine within which it says not to do promiscuous things. Moreover, it says that people need to control their desires …. [Xie, 30+ years old] In either case, compliance to one’s personal codes of religious sexual morality was typically self-enforced by both fear of supernatural reprisals and desire to satisfy the self-perceived behavioural expectations tied to their religious identity. Believing that “The body is  111   not made for sexual indulgence, but for God and the glory of God,” Luo often coped with internal fears of imminent divine retribution brought on by his casual sexual encounters.    Interviewer:Have you ever had group sex? Luo:No, but I have hoped to try it in the past because I’ve always wanted to be intimate with lots of people …. I still think about doing it, but I suppress such thoughts as soon as I start having fears about it. Interviewer:What are you fearful of? Luo: God punishing me. It’s obvious that this type of thing is wrong. Regardless of what type of excuse or reasoning you use, it’s still wrong …. I feel very guilty every time I meet up with people from the internet …. I feel like I shouldn’t do this sort of thing. Interviewer: Why do you feel that way? Luo: Many reasons. I am fearful of, scared of God’s wrath.   [Luo, 20+ years old] For many Christian & Muslim participants, belief in an omnipresent sentient higher being implied that they were continuously under a form of supernatural surveillance. Ding described his experience as that of an external being literally commanding him to suppress his sexual urges.      Although you sometimes get really horny, you’re still beholden to religious discipline. It’s like there’s really another voice saying “No, you can’t do that today, you mustn’t do that, not this time.” And then, you’ll think, “Ok. If I can’t do it I can’t do it. It’s OK. I’ll find something else to do.” Or, sometimes it [sexual desire] will just naturally dissipate …. Before even meeting up with someone, I’ll have been able to control myself.  [Ding, 30+ years old] However, such forms of perceived supernatural surveillance were by no means always effective at suppressing sexual inclinations. Many religious participants described a type of cycle whereby periods of sexual restraint would inevitably be punctuated by moments of acute sexual urges.    112   I’ll usually try to suppress it [sexual desires] and hold, hold, hold, hold, hold out until all of a sudden I lose all rational thinking…[Then] I’ll just get online and start chatting …. It always ends up as a one-night stand.  [Luo, 20+ years old] For Ma, religious meanings associated with certain body parts directly influenced his decisions about sexual position.  In Islam, we have rules. A person’s mouth is for eating. Besides eating and speaking, the third function is to praise Allah, to praise holy Muhammad …. I feel like the mouth is a sacred organ, a very clean and sacred organ …. To go and give another man oral sex down there with all the urine, to me is dirty and a bit disgusting …. I won’t give other people oral. [Ma, 30+ years old] It is worth emphasizing that religious beliefs were but one of many other considerations that informed religious interviewees’ sexual decision making processes. Complex relations and mainstream heteronormativity contextualized the broader social environment for all participants, and proved to be a powerful sexual inhibitor among religious and non-religious participants alike. To Liang, a non-religious man, family was the primary source of his sexual inhibitions.  I’m already a father, I need to have control. You need to abandon everything else. At minimum, you need to be respectable in the eyes of your children. A father needs to have the dignity of a father. … The first time I went out and did it with a man and went back home and saw my son, I felt like I committed some major offense. I couldn’t shake that feeling for a long time…. I thought “what’s going on?” and didn’t do it [have sex] for a long time.  [Liang, 40+ years old] Religious influences also had limited impact on decisions surrounding condom use. Having protected sex, as opposed to unprotected sex, neither diminished nor enhanced the perceived religious moral significance of same-sex sexual behaviours or having multiple sex partners. Among religious and non-religious participants alike, concern about personal health was the  113   primary rationale for using condoms; reasons to forego condoms included desire for greater partner intimacy and physical pleasure.   3.4.3 Ethnocultural bias Participants’ intersecting minority identities were invariably contextualized by the dominant non-religious ethnic Han cultural environment that characterized Beijing and transcended sexual minority communities. The following sections focus on how mainstream ethnocultural hierarchies structured interpersonal ethnic relations, particularly those involving members of the Muslim community. With respect to HIV vulnerabilities, three main themes emerged: (a) Cultural devaluation in heteronormative society, (b) Ethnocultural status & sex partner selection, and (c) Uyghur men and social segregation.  3.4.3.1 Cultural devaluation in heteronormative society Instances of Han ethnocentrism were apparent from numerous participant narratives in both heteronormative and homonormative settings. Most expressions of ethnocentrism involved cultural devaluation of Muslims, especially those who were ethnically Uyghur.     In recent years, ethnic tensions between Uyghurs and Han have been widely publicized in the mass media and were reflected in interviews. When Dong, an ethnic Han participant from Western China was asked why he never interacted with ethnic minorities, he said,  Because they’re too dirty for my liking …. As natives of Xinjiang [province in Western China], they eat lamb. Their bodies have that fox smell. It’s too strong …. Anyways, I don’t like those, them. I don’t like those Uyghurs, let’s just say that … Uyghurs and Kazaks.  [Dong, 30+ years old]  However, negative Uyghur stereotypes were subscribed by other ethnic minorities as well, not just the Han majority. Cheng, a non-Uyghur Muslim said,  114   I’ll definitely think that – it’s not just Uyghurs, all of the ethnic minorities in Xinjiang will have some – especially in the hinterland – some dishonest behaviour, some unrestrained behaviour …. Regarding Uyghurs, it’s commonly agreed upon by everyone that there are lots of thieves. Stealing cell phones, dealing drugs. [Cheng, 20+ years old]  Such pejorative sentiments were corroborated by the negative experiences of Uyghur participants. Akhun, described his experience of being adversely stereotyped in public. Many people say “purse snatcher”, think that you’re a thief or something …. So, for this religious ceremony, we went to a Mosque, and at the time we all wore our [religious] caps. You’d know what kind of cap it was if you saw it … and when they saw us get on the bus, they hid away their bags … those shopping bags that you carry around. There were a lot of people on the bus, and everyone said “this is my bag,” because [supposedly] we were thieves. [Akhun, 20+ years old] For Arzigul, another young Uyghur migrant from West China, mainstream bias against Uyghurs engendered not only emotional stress, but also myriad practical obstacles to living in Beijing.   It’s hard for me to get a hotel cause the hotel says that I am Uyghur, they won’t let me stay there. And when looking for a job, it’s hard because the employer says he doesn’t want me because I am Uyghur … I really felt bad when he told me this. I didn’t want to stay in Beijing anymore. They think that the young men from Xinjiang are thieves… Sometimes it’s not fair. Emotionally, I sometimes feel like it’s really hard to bear. Thus, I feel like it’s better if I go home. [Arzigul, 20+ years old]  Liu  also encountered difficulties conducting ordinary business in Beijing, which he attributed to not just his ethnicity, but his geographic background as well. From his narrative, it also became clear that negative ethnocultural experiences for Muslims may also be fueled by concerns about terrorism. When I pass through airport security, it takes much much longer than ethnic Han. And so, lots of times the people at the airport will give me disposable slippers, have me wear disposable slippers on the plane, and then check my shoes in as luggage. [Liu, 20+ years old]   115   3.4.3.2 Ethnocultural status & sex partner selection During anonymous sexual encounters, exchange of personal information was limited, and ethnicity of a potential sex partner was largely perceived of as irrelevant. Moreover, as many ethnic minorities in China are superficially indistinguishable from the ethnic Han majority, determining ethnicity of a stranger was often difficult. This was articulated by Lin, a gregarious ethnic Han migrant. Interviewer: So you never had an ethnic minority [sex partner]? Lin: I’m not sure. I haven’t asked anyone. You can’t go and ask people if they’re an ethnic minority. They’ll think it’s weird to ask such a question …. it’s sex, you’re having sex, why would you ask if someone is an ethnic minority or not? [Lin, 20+ years old] However, for paoyou 1  and “B.F.” (i.e., boyfriend) relationships that entailed more social interaction, ethnic markers of a partner became more pertinent. For Jiang, speaking Mandarin with an ethnic Zhuang accent was a clear disadvantage to attracting potential partners.   Society has different levels of status. They might think that someone who speaks with a strong Zhuang accent is from the countryside, so it will therefore have an effect. Of course, the most important factor is still how handsome you are…. But if you’re not handsome and you speak with a Zhuang accent, then you’ll be of the lower grades, or you’ll be one of those who aren’t chosen, one of those people who aren’t selected. [Jiang, 30+ years old]  Given the presence of anti-Muslim sentiments in mainstream society, it was unsurprising to note that several participants actively avoided Muslims when seeking new sex partners. This was the case for Xiao, an ethnic Zhuang man from Southern China.  Interviewer: When seeking a [sex] partner, will you consider the person’s ethnic status? Xiao: Sure I will… I won’t look for a Muslim.                                                  1 Paoyou is a term used to describe a person with whom one has sex on multiple occassions, but in the absence of any long-term commitments to the relationship. Can be literally translated as “ejaculation buddy”.   116   Interviewer: Won’t look for a Muslim. Why? Xiao: Because you can’t eat together, that’s the main reason. There aren’t other [reasons]. Interviewer: So if they’re able to eat pork then there wouldn’t be an issue? Xiao: It’s not if they can eat pork, it’s just if they’re not too fussy – for example, if they can go into any restaurant, it’s fine if they don’t eat pork but I [still] can, in that case it wouldn’t really matter. But I’ll still want to avoid [them], it’s a real hassle. Moreover, they’re so adamantly opposed to it [sex between men]. Like, you’ll think about how they’re family is opposed to it, and you won’t even want to socialize with them….  [Xiao, 20+ years old] In addition to practical inconveniences involving dietary restrictions and family conflicts, avoidance of Muslim ethnic minorities was also motivated by strong negative stereotypes. Despite, or perhaps because of his limited contact with ethnic Hui, Wu, a Mongol man, found ethnic Hui to be both inscrutable and intolerable, Interviewer: Will you consider ethnicity [when looking for a sex partner]? Would you prefer to find another Mongol? Wu: I’ve never thought about looking for another Mongol…but I won’t look for an ethnic Hui…. I just feel like they’re really – their lifestyle is really inconvenient. Second, I don’t really like to interact with them. And then, in my heart, I’ll have these kinds of thoughts: if I need to have a person to date, and if I find out that they’re an ethnic Hui, I won’t continue … Lots of us just don’t like Hui, Hui people. So, the reason I don’t like to interact with them is, they’re just one of these very fussy types of people. You can’t do this, and you can’t do that. I just don’t like interacting with them…. You’ll feel that their way of thinking is different. The differences between Hui and Mongol might be even bigger [than Han]…. I think it has something to do with Mongols’ personality. Most Mongol personalities are more casual and outgoing…. But when it comes to ethnic Hui, we [Mongols] get scared. With those kind, communicating with them is more of a hassle. [Wu, 20+ years old]  117   3.4.3.3 Uyghur men and social segregation The challenges of recruiting Uyghur MSM throughout the data collection process underscored the unique degree of social distance between Uyghur and non-Uyghur MSM, which was also confirmed by prominent MSM opinion leaders and multiple interviewees. Indeed, potential Uyghur interviewees may have been more reluctant to discuss intimate personal experiences with an ethnic Han interviewer. As a result of such de facto ethnic segregation, access to HIV services may be more challenging for Uyghur MSM. Born and raised in far Western China among Uyghur, Hui, and Kazak minority groups, Liu found status quo HIV prevention programs for MSM to be ineffective for many ethnic minorities.    I feel like if the health departments are going to keep working as if they were [just] servicing ethnic Han, then it’s going to be difficult to reach ethnic minorities. Really difficult … If you want to do an ethnic minority intervention, you have to do it using an ethnic minority framework.  [Liu, 20+ years old] 3.4.4 Discussion  Intersecting ethnic, religious, and sexual minority statuses have profound implications for individual- and population-level HIV vulnerabilities. Yet, notions of ethnocultural and sexual intersectionality remain largely absent from China’s HIV prevention literature, despite the country’s pluralistic society. This study is arguably the first to explicitly explore, ethnicity religion, and HIV vulnerabilities and testing among MSM in mainland China. Study findings illuminated myriad mechanisms by which ethnicity and religion may be shaping HIV-related decision-making processes and social contexts. Specific expressions of ethnic and religious affiliation varied between social groups and individuals, but three broad overarching messages emerged: (a) Ethnic and religious affiliations anchored individuals’ personal identity and sense of social norms within larger sociocultural reference groups and networks, (b) Conscious  118   decisions about sexual behaviours were guided by larger beliefs about cosmic moral order and religious codes of sexual morality, and (c) ethnocultural biases shaped decisions relevant to sexual partnerships and uptake of HIV health care services.  3.4.4.1 Ethnocultural affiliations, social norms, and sense of social identity Among North American MSM, poor attachment to heritage ethnic identity has been associated with riskier behaviours such as UAI and less HIV testing (Donnell et al., 2002; Vu et al., 2011). The qualitative nature of the present study precluded statistical testing, but accounts of indifferent and negative attitudes toward one’s own ethnic identity suggest that similar trends may be occurring among ethnic minority MSM in Beijing. In a sobering sense, devaluation of personal ethnic minority identity reflects larger trends in the marginalization of ethnic minority languages and histories in the PRC’s educational system (M. Zhou, 2003, pp. 27–28). On the other hand, ethnic minorities who culturally assimilate to Han society remain subject to pejorative labels such as “ethnic minorities in name only” (shen fen shao shu min zu) if they fail to maintain some modicum of their ethnic cultural heritage. Future studies should assess how ethnic identity affirmation may be impacting HIV vulnerabilities among ethnic minority MSM in China. Developing a healthy sense of ethnic identity may help MSM and non-MSM alike to augment individual self-worth and in turn improve resolve and capacity to adopt behaviours to reduce HIV vulnerability. Ethnic and religious minority groups served as an important social reference group by which its members could develop, compare, and evaluate their own behaviours and attitudes, even for those who consciously dissented from mainstream ethnic and religious social norms. Perceived social expectations of ethnocultural groups created potential pressures for participants to avoid behaviours viewed as socially deviant. However, the relatively low population of ethnic  119   minorities in Beijing and the consequent limited interaction with ethnic peers suggests that ethnic minority social norms are likely weaker in Beijing than other regions where ethnic minorities represent a higher proportion of the population. For example, in Xinjiang province, ethnic Han only represent 40% of the population (Statistic Bureau of Xinjiang Uygur Autonomous Region, 2014). In contrast, the congregational religious social structure of Christianity and Islam afforded an institutionalized mechanism for enforcement of religious-based social norms among its members. Participation in religious social activities helped establish and reinforce normative boundaries of what constituted moral and immoral behaviours, though to be sure, level of engagement in religious social activities varied considerably between and within religious affiliations. Few Buddhists routinely participated in religious social activities, and numerous Christians and Muslims acknowledged only nominal interactions with religious peers. Given their reduced exposure to religious social norms, such individuals may be less inhibited from behaviours deemed as immoral by the respective religious community (e.g., men having sex with men). Unfortunately, the lack of behavioural survey data disaggregated by ethnic or religious affiliation in China has made it difficult to estimate how prevalence of such behaviours varies by ethnicity, religious group, and level of engagement in religious activities.   3.4.4.2 Cosmic moral order, supernatural punishment, & sexual behaviour Among other potential mediating factors, findings from this study suggest that beliefs in supernatural retribution for perceived sexual transgressions may be an important consideration in sexual decision making among Christian, Buddhist, and Muslim men in China and elsewhere. Fear of karmic retribution, God/Allah’s wrath, and eternal damnation served as powerful incentives for individuals to maintain what they understood as moral sexual conduct. For those who felt that having sex with men or having concurrent sex partners violated their religious  120   values, beliefs in supernatural surveillance and punishment persuaded many to limit their sexual activities with other men. These results accord well with more general empirical and experimental studies which demonstrated that belief in moralizing supernatural forces deters behaviours perceived as being immoral (Schloss & Murray, 2011). That said, expectations of divine retribution for putative immoral sexual behaviour varied widely between religious affiliations. Lacking well-established religious traditions that explicitly stigmatized same-sex sexual behaviour or sexual concurrency (Sweet, 2007), Buddhist participants exhibited relatively limited concern about incurring supernatural retribution for same-sex sexual behaviours or sexual concurrency. Without credible threats of supernatural punishment for sexual transgressions, frequency of same-sex sexual activities among Buddhist MSM may be comparable with that of non-religious MSM. In contrast, most Muslim participants felt that having sex with men was in clear violation of their religious moral tenets, even among those who continued to have sex with men. Moreover, Muslim informants were in general relatively more conscientious of divine retribution for same-sex sexual behaviours than those of other religious persuasions. In light of experimental evidence which has shown that perceived belief in a vengeful higher being (vs compassionate higher being) discourages behaviours seen as immoral (Shariff & Norenzayan, 2011), it is possible that the inhibiting effects of perceived supernatural punishment on sexual behaviours are more pronounced among Muslims.  Dominant religious norms of sexual morality, however, are subject to changes over time. Sexual minority Christian social organisations in China signify the growing resonance of alternative religious perspectives which do not stigmatize same-sex sexual behaviours per se, but which nonetheless continue to promote specific sexual practices (e.g., monogamy) based on religious values. These sexual minority religious social organisations also offer religious MSM  121   an important source of unique social support that would be more difficult to attain in mainstream religious organisations or non-religious MSM social groups. Public health departments and HIV prevention advocacy groups should consider partnering with popular opinion leaders of sexual minority religious social organisations, particularly in regions with large populations of Christians and Muslims. Popular opinion leaders of Buddhist sexual minority social groups may also be a promising channel for promoting HIV testing and prevention activities among MSM, though Buddhism’s decentralized social structure and nominal expectations of lay-members may complicate peer-driven behavioural interventions.   Study findings also caution against overgeneralizing beliefs surrounding supernatural punishment. Diverse understandings of sexual morality and supernatural punishment among members of a common religious affiliation underscored the salience of personal agency. Compartmentalization of religious and sexual identities has been identified as a strategy for reconciling supposedly conflicting identities among Christian MSM in North America (Rodriguez & Ouellette, 2000), and appears to be prevalent among religious MSM of North China as well. By cognitively segregating religious identity from same-sex attraction, religious MSM who have difficulty reconciling religious identity and sexual behaviours are able to avoid perceiving potential threats of supernatural punishment associated with their sexual practices. But even in the absence of identity compartmentalization strategies, religious participants may not necessarily feel that perceived acts of sexual immorality inevitably trigger supernaturally-mediated negative outcomes. Several informants believed in fungible personal moral accounts whereby moral debts created by sexual transgression could be counterbalanced with moral credits earned through virtuous deeds. Consequently, beliefs in supernatural punishment may be  122   less effective at deterring putatively immoral behaviours if the individual believes that they are able to elude supernatural retribution through compensatory behaviours. Beliefs in supernatural surveillance and punishment may also have limited effects on sexual behaviours among religious MSM who successfully achieved identity integration, whereby religious and sexual identities are viewed positively and free of mutual conflict (Rodriguez & Ouellette, 2000). Similar to Christian MSM in North America (Rodriguez & Ouellette, 2000), the most common means of achieving identity integration among Christian and Muslim MSM was to critically re-interpret religious texts in a way that no longer proscribed same-sex sexual behaviours. Achieving religious and sexual identity integration, however, was often a protracted process which entailed considerable psychological distress for many. Psychological distress caused by discord between sexual behaviour and religious beliefs may be leading to increased UAI or alcohol consumption. Future quantitative studies in China should examine if and how religion impacts MSM HIV vulnerability via pathways of negative coping.  3.4.4.3 Acculturation and ethnocultural bias  Ethnocultural biases in sex partner selection at the population level can constrain sex partner choices for stigmatized groups, and lead to distinct sexual networks divided along ethnic and racial boundaries (Maulsby et al., 2014; H Fisher Raymond & McFarland, 2009). Candid participant accounts indicated that reluctance to developing steady sexual partnerships with Muslim MSM may be increasing isolation of Muslim MSM sexual networks in North China. Surveys assessing religious sexual mixing patterns of MSM are needed to gauge the degree of interconnectivity between Muslim and non-Muslim MSM sexual networks in China. Study findings indicated that lack of fluency in Chinese literacy and spoken Chinese potentially hinders access to HIV testing and prevention literature in Beijing, where such  123   information is communicated almost exclusively in Mandarin Chinese. In addition, language discrimination also emerged as a stigmatizing status symbol that complicated sexual power dynamics by disadvantaging the speaker’s social standing. Conceivably, MSM who speak Mandarin with heavy stigmatized accents may be less confident and persuasive when trying to negotiate safer sex with potential partners. In light of the fact that ethnic minorities from all over China come to Beijing to study Chinese as a second language each year (Hasmath, 2011), HIV interventions should be sensitive to the linguistic diversity of their target populations. Unfortunately, few if any studies in North China have been conducted in languages other than Chinese. Multi-lingual survey assessments are needed to identify how well current Chinese language HIV prevention messaging is reaching and resonating with linguistic minority MSM. 3.4.4.4 Limitations   Study findings should be prudently interpreted with several limitations in mind. First, certain sub-populations of ethnic and religious MSM in North China may not have been represented in the study sample. By recruiting participants exclusively via Chinese language mediums, the sample may have been biased towards those with stronger Chinese language skills. Nevertheless, by seeking maximum variation in multiple attributes, project partners and I were able to successfully recruit achieve strong sample diversity in ethnicity, language, religion, and sociodemographics.  Second, the individualized personal accounts of the present study cannot be used to make generalizations about the broader ethnocultural minority MSM population in North China, let alone all of China. Nonetheless, qualitative study findings remain instructive for theory-generation and the future development of quantitative surveys better suited to assessing issues of generalizability.    124   3.4.4.5 Conclusion With few exceptions, HIV-related public health and social science literature in China has tended to frame MSM and ethnocultural minorities as mutually exclusive social groups, whereby ethnic and religious minorities are largely absent from MSM-oriented discourse, and vice versa. To the contrary, findings from this study demonstrated that MSM in North China are of diverse ethnic and religious persuasions with particular histories, cultural norms, and social structures. Such cultural influences powerfully shape decision making processes related to HIV, though they are by no means deterministic. The diversity of accounts in this study highlights the importance of appreciating individuals’ unique agency in constructing meaning in their world, even as quantitative research methods inevitably reduce human experiences to manipulable variables and parameters for generalizability.    125   Chapter 4: Ethnicity, acculturation, and ethnic discrimination among men who have sex with men in North China: Implications for HIV vulnerability and testing 4.1 Introduction MSM in China are disproportionately affected by HIV. In 2011, MSM are estimated to have accounted for 29.4% of all 48,000 new HIV infections nationwide (Ministry of Health of the People’s Republic of China, 2012) but only 0.3% of the overall population (UNDP, 2012). In response, public health organisations have sought to augment HIV testing and safer sex practices among MSM, and HIV prevention research with MSM in China has expanded dramatically (J. T. Lau et al., 2014). However, few studies have explicitly acknowledged the ethnic diversity of MSM in China, or examined how the intersectionality of sexual and ethnic minority status may be shaping HIV vulnerability and testing. 4.1.1 Ethnic affiliation & HIV vulnerabilities  Ethnic affiliations define fundamental sociocultural boundaries (Eriksen, 2010, p. 45) and are often a powerful predictor of individual vulnerability to HIV. Invariably, each potential incident of HIV transmission is situated within particular sociocultural, historical, and geographical contexts which indirectly affect the probabilities of transmission (Aral et al., 2008; Friedman et al., 2009). An important observation from the small but growing body of minority ethnoracial MSM HIV research is that ethnic minority status per se is not necessarily indicative of greater HIV risk. The breadth of diversity between ethnoracial groups and cultural contexts simply precludes any such generalizations. For example, in the US, ethnoracial minority African Americans represent 35% of incident HIV infections among MSM (Maulsby et al., 2014), but  126   only ~14% of the total US population (Rastogi et al., 2011). On the other hand, minority Asian American MSM have consistently had lower HIV prevalence and incidence than MSM of all other US ethnoracial groups (Wei et al., 2011).  Currently, the PRC recognises exactly 56 official ethnic groups within China that range in population from several thousand  to over 1 billion (ethnic Han majority) (National Bureau of Statistics of China, 2013). Collectively, China’s 55 ethnic minority groups total about 111 million persons (~9% of the national population) and are characterized by highly heterogeneous histories, religions, socioeconomic statuses, languages, diet, geographic homelands, and acculturation to ethnic majority Han culture  (National Bureau of Statistics of China, 2013; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b). During the earlier phases of the HIV epidemic in China, certain ethnic minority groups were highly overrepresented in HIV incident cases. By the year 2000, ethnic minorities only comprised 8% of the total population, but accounted for 36% of all reported HIV cases (Jing & Huan, 2010). Since the turn of the century, HIV disparities between Han and ethnic minorities have become less pronounced as the epidemic has percolated beyond ethnic minority social circles and public health surveillance has improved, but the toll of HIV has remained strikingly uneven between different ethnicities. Certain highly assimilated ethnic minorities have socioeconomic status (SES) and cultural practices and values comparable with that of the majority Han (Mackerras, 2003) and there exists no evidence to suggest that they significantly differ from ethnic Han in terms of HIV burden. However, many less assimilated ethnic minority groups in Southern and Western China have been grossly overrepresented in HIV burdens. Within some ethnic minority communities in Southwest China, HIV prevalence in the general  127   population has risen as high as 11% (Dong et al., 2014), compared to 0.058% nationwide (Ministry of Health of the People’s Republic of China, 2012).  The HIV vulnerabilities of ethnic minority MSM in China are poorly understood and understudied. One exceptional study indicated that many ethnic Uyghur MSM in Xinjiang province continue to engage in unprotected anal and vaginal sex (70% and 47%, respectively), despite being well informed about HIV/AIDS (J. Zhang et al., 2010). One reason ethnic minorities’ HIV vulnerabilities are poorly understood in China is because conventions of crudely dichotomizing ethnic status into “Han majority” and “other” have obfuscated any meaningful interpretations of ethnicity as a sociocultural construct. Epidemiological HIV surveys in China typically aggregated all 55 ethnic minority groups into a single category. As a result, significant associations for specific ethnic groups become masked and comparisons between studies left untenable because the ethnicities of participants classified under the pan-ethnic minority rubric can drastically change from study to study. For example, one Beijing study in 2008 indicated that HIV was spreading 4.7 times significantly faster among ethnic minority MSM (S. Li et al., 2011), while another Beijing study in 2009 did not detect any significant disparity between majority Han and ethnic minority MSM (D. Li et al., 2012). In Yunnan province, HIV incidence rates have been 5.7 times higher among ethnic minority MSM (vs majority Han), but again, without disaggregated data by ethnicity, the interpretive and applicative utility of such findings becomes disputable (Xu et al., 2013).  4.1.2 Acculturation: implications for HIV vulnerability and testing  Acculturation has been defined as “the changes that take place as a result of contact with culturally dissimilar people, groups, and social influences”, and encompasses multiple unique domains (Schwartz et al., 2010). Specifically, HIV-related acculturation research has identified  128   cultural practice (e.g., language usage) and cultural identity (e.g., sense of attachment to a particular ethnic group) as two distinct acculturation domains with distinct effects on HIV-sexual vulnerability (Schwartz et al., 2011, 2014). For example, among Hispanic adolescent males in the US, ethnic identity affirmation was associated with more oral sex partners, but Hispanic cultural practices was actually associated with fewer partners (Schwartz et al., 2014). Moreover, cultural practices and identity do not necessarily correlate. It is entirely plausible that an individual or ethnic group can exhibit weak cultural practices of their ethnic group, but simultaneously have strong ethnic identity affirmation. Language skills have strong potential to impact ethnic minorities’ access and uptake of HIV health services such as HIV testing. For example, poor proficiency in the dominant culture’s language can constrain “the extent to which people can access health information, communicate their health problems and questions to healthcare providers, understand health instructions, and participate in interventions” (Unger & Schwartz, 2012, p. 353). One study with Asian MSM in the US found that odds of HIV testing was almost three times lower for those who did not speak English at home (Toleran et al., 2013), and a study with Latino MSM in the US linked greater use of Spanish with UAI among non-main sex partners (Donnell et al., 2002). Ethnic identity affirmation can be defined as “the extent to which one is attached to one’s racial/ethnic group and views that group positively” (Schwartz et al., 2011). Along with any sense of identity derived from one’s sexual behaviours, ethnic identity affirmation among ethnocultural minority MSM has important implications for individual self-worth and mental health, which in turn directly impact one’s resolve and capacity to adopt HIV preventative behaviours (Chae & Yoshikawa, 2008; Nemoto et al., 2003; Vu et al., 2011). Despite stronger heteronormative and homonegative values among certain ethnocultural groups, evidence  129   suggests that ethnic identity affirmation protects MSM against various HIV sexual vulnerabilities. Among MSM in the US, greater Latino ethnic affirmation was associated with less UAI (Donnell et al., 2002), and pride in Asian/Pacific Islander ethnic identity was correlated with more frequent HIV testing (Vu et al., 2011).  Notably, frequent HIV testing was even more likely when Asian/Pacific Islander MSM in the US reported strong ethnocultural identity in combination with strong sexual identity (Vu et al., 2011). Within most urban settings in mainland China, Han represent the dominant ethnic group with whom ethnic minority groups interact. Chinese language proficiency is arguably one of the most conspicuous examples of ethnic Han cultural practice, and in fact, the term “Han language” (han yu) is commonly used interchangeably with “Chinese” (zhong wen). Naturally, Chinese serves as the native language for ethnic minorities who lack their own distinct heritage language or whose heritage language has become socially obsolete. Ethnic minorities raised in majority Han social environments also typically acquire strong Chinese literacy. By contrast, mastery of Chinese can be considerably more challenging for ethnic minorities who have limited Chinese language exposure in daily life. Nonetheless, evidence suggests that access and uptake of HIV-related health services are being inhibited by language barriers among some ethnic minorities. Up until the late 2000s, official dissemination of HIV prevention information in far West China was only conducted in Chinese, despite the fact that local ethnic minorities represented the majority of the population and often had limited proficiency in Chinese. Naturally, such health promotion interventions would have been ineffective at augmenting HIV prevention knowledge, attitudes, and practices for ethnic minorities who lacked sufficient Chinese comprehension (A. Hayes, 2012).   130   Over the past twenty years, many of China’s ethnic minority groups have been experiencing a growing sense of ethnic consciousness and pride (Demick, 2013; Gladney, 1995). In 2010, a multi-site survey of ethnic minority adolescents in far West China showed that all surveyed ethnic minority groups reported stronger sense of closeness to their ethnicity than ethnic Han (Tang & He, 2010). Similarly, another survey among adults in West China showed that 91% of Uyghurs reported being proud of their ethnicity, compared to just 67% of ethnic Han respondents (Yee, 2003). However, despite China’s diverse and sizable ethnic minority population, few if any studies in the country have examined how language skills or sense of ethnic identity affirmation may be impacting HIV vulnerabilities.  4.1.3 Discrimination and HIV Discrimination has been defined as a “behavioural manifestation of a negative attitude, judgement, or unfair treatment toward members of a group” (Pascoe & Richman, 2009), and can occur at the interpersonal and/or structural levels (Link & Phelan, 2001). Interpersonal discrimination includes personally-mediated acts of devaluation and exclusion, while structural discrimination involves social norms and institutionalized practices which undermine the welfare of stigmatized groups, with or without interpersonal discrimination (Angermeyer et al., 2014). Acts of interpersonal and structural discrimination can be committed intentionally or unintentionally, and may or may not be perceived by the putatively discriminated individual (Jones, 2001).  Current research indicates that discrimination based on race/ethnicity, religion, and sexual orientation have negative impacts on a broad spectrum of health indicators including substance use, psychological distress, and HIV sexual vulnerability (Krieger, 2014; Rippy & Newman, 2006). Compelling empirical evidence suggests four possible mechanisms linking  131   ethnocultural or sexual orientation discrimination and HIV vulnerability. First, instances of perceived discrimination can trigger psychological distress which prompts negative coping strategies such as unprotected sex with casual partners (Díaz et al., 2004; Earnshaw et al., 2013). Second, discrimination can constrain access to important HIV prevention information and services. For example, discriminatory language policies can impede access to HIV prevention literature among linguistic minority groups (Wohl et al., 2009), while perceptions of discrimination can dissuade marginalized ethnocultural minorities from establishing contact with health service institutions associated with the dominant social group (Earnshaw et al., 2013). Third, ethnocultural discrimination can create interpersonal power imbalances that disadvantage low-social status individuals leading up to and during sexual situations, whereby safer sex such as condom use and insertive versus receptive sexual positioning becomes more difficult to negotiate and practice (A. I. Green, 2008; C. S. Han et al., 2014; C. Han, 2008). Fourth, ethnic/racial discrimination when selecting sex partners can create ethnoculturally segregated social and sexual networks, and constrain one’s ability to partner with members outside their ethnocultural group. If HIV prevalence varies between such ethnoculturally segregated sexual networks, then members of disparate ethnocultural groups may have different risks of exposure to HIV, independent of individual sexual behaviours (Maulsby et al., 2014; B. Mustanski et al., 2014).  Among ethnoracial minority MSM in the US, perceived racism alone and in combination with perceived external homonegativity has consistently been linked to UAI (Ayala et al., 2012; C. S. Han et al., 2014; Huebner et al., 2013; Mizuno et al., 2012). But even without cognizant acknowledgement, ethnoracialized social structures and notions of sexual desirability among MSM can lead to interracial sexual power imbalances that disproportionately place certain  132   marginalized ethnic and racial minorities at elevated vulnerability to HIV infection (Arnold et al., 2014; C. Han, 2008; McAdams-Mahmoud et al., 2014; Ro et al., 2013). Driven in-part by de facto racial residential segregation (H. F. Raymond et al., 2014) and racial bias against black MSM by non-black MSM (H Fisher Raymond & McFarland, 2009), black MSM sexual networks have become more racially homogenous, interconnected, and segregated (Maulsby et al., 2014; B. Mustanski et al., 2014). As a consequence of racialized sexual network patterns and racially distinct background prevalence, African-American MSM may face greater risk for HIV infection, independent of any individual behaviours. Discrimination of ethnic minority groups in China has been reported by numerous sources and is of concern to both the United Nations and Chinese government (United Nations’ Committee on the Elimination of Racial Discrimination, 2009a, 2009b). At the interpersonal level, experiences of ethnocultural discrimination in China vary significantly between social groups and has ranged from cultural adolescent bullying (S. W. Pan & Spittal, 2013) to large-scale interethnic violence (“Innocent civilians make up 156 in Urumqi riot death toll,” 2009).  In recent years, several ethnographies have examined the sociopolitical determinants of health among deprived ethnic minority groups in Southwest China (Hyde, 2007; S. Liu, 2011; Lyttleton et al., 2011), but few studies have specifically examined interpersonal ethnocultural discrimination and its health implications. One such study which showed that adolescents in urban Chinese cities were more likely to experience depression, suicidal ideation, and interpersonal violence if they perceived being bullied because of their ethnicity or religion (S. W. Pan & Spittal, 2013). Though quantitative research concerning ethnocultural discrimination and HIV vulnerability in China remains underdeveloped, the aforementioned literature strongly suggests a decidedly positive relationship.     133   4.2 Objectives and rationale  In response to this stark knowledge gap concerning the intersectionality of ethnicity, sexuality, and HIV vulnerability in China, the present study sought to assess how ethnic affiliation, acculturation, and discrimination may be impacting HIV vulnerabilities, sexual networks, and HIV testing among MSM in North China. Five hypotheses were tested:  (1)  Ethnic affiliation for some but not all ethnic minority groups is associated with increased HIV vulnerabilities and poorer HIV testing patterns, compared to ethnic majority Han  (2) Ethnic affiliations of sex partners is associated with ethnic affiliation of the participant  (3)  Stronger ethnic identity affirmation is associated with reduced HIV vulnerabilities and more frequent HIV testing (4)   Some, but not all, ethnic minority groups have a stronger magnitude of association between ethnic identity affirmation and HIV vulnerabilities/testing than ethnic majority Han  (5) Poorer Mandarin language proficiency is associated with greater HIV vulnerabilities and reduced HIV testing (6)  Perceived ethnic discrimination is associated with greater HIV vulnerabilities and reduced HIV testing 4.3 Methods 4.3.1 Study setting: Beijing & Tianjin Beijing ranks among China’s most prosperous and prestigious cities. Its robust economy and cosmopolitan social scene draws millions of migrants from across the country and around  134   the world, whereby 37% of Beijing’s 20 million inhabitants were migrant residents in 2010 (C. Huang, 2014). Although all 55 ethnic minority groups of China are represented among Beijing’s population, 96% of Beijing residents are ethnically Han (Beijing Statistical Information Net, 2011), and standard Mandarin is the dominant dialect in most personal and professional spheres. Manchus, Hui, Mongols, and Koreans constitute the vast majority of Beijing’s ethnic minority population (87%) (Beijing Statistical Information Net, 2011). Beijing’s sexual minority community is among the most vibrant and socially active in China, with an estimated MSM population size between 115,731 and 239,258 persons (Guiying et al., 2014). Since 2009, HIV prevalence estimates for Beijing MSM have ranged from 2.5% to 8.0% (Yunhua Zhou et al., 2014).   Tianjin is a relatively wealthy Chinese municipality that attracts millions of domestic migrants and plays a vital role in the North China economic zone; of its 14 million inhabitants, 25% are migrants (C. Huang, 2014). The vast majority of residents in Tianjin are ethnically Han (97%) and ethnic Muslim Hui represent the bulk of Tianjin’s ethnic minority population (65%) (china.com.cn, 2009). Tianjin’s sexual minority community is relatively robust with numerous online and physical gay venues. From 2008-2009, HIV prevalence estimates for Tianjin MSM have ranged from 5.9% to 8.6% (Ning et al., 2011; Yunhua Zhou et al., 2014).   4.3.2 Four ethnic groups of North China The following sections provide a succinct and selective overview of four ethnic groups who hold strong presence within the national consciousness and who represent the bulk of ethnic minorities in North China.  135   4.3.2.1 Han In terms of population and political clout, Han are the dominant ethnic group in mainland China and closely associated with notions of “Chinese” culture and identity (Gladney, 1998, p. 11). Confucian ideals of family lineage and gender roles retain strong influence within ethnic Han society, despite efforts by the government to eliminate Confucian influences during the first 30 years of the PRC (Adamczyk & Cheng, 2015; Steward et al., 2013). Buddhism and atheism are arguably the most popular official religious orientations among ethnic Han, but Christianity is also rapidly becoming adopted throughout the cities and countryside (“Cracks in the atheist edifice,” 2014; Pew Research Center, 2012).  4.3.2.2 Manchus Progenitors of present day ethnic Manchus originated from Northeast China and established China’s Qing dynasty (1644-1911) (Spence, 2013, p. 33). Ironically, the new governing Manchus began to adopt cultural practices of their ethnic Han subjects, and the Manchu language was eventually displaced by Chinese (Crossley, 1997, p. 127; United Nations’ Committee on the Elimination of Racial Discrimination, 2009b). Similarly, Manchu shamanistic religious practices gave way in favour of Tibetan Buddhism during their centuries of rule (Dede, n.d.; Goossaert & Palmer, 2011, p. 32). With generations of heavy intermarriage between Han and Manchus, ethnic Manchus of contemporary China have become culturally indistinguishable from Northern ethnic Han in all but name (Mackerras, 2003, p. 16).  4.3.2.3 Mongols Compared to the Manchus, cultural assimilation to Han culture has been historically more limited for the Mongols. As of 1994, 80% of ethnic Mongols could still speak their heritage language, compared to less than 1% of ethnic Manchus (M. Zhou, 2003). However, heritage  136   language and religion is fading among ethnic Mongols of contemporary China. Practice of Tibetan Buddhism has become relatively weak within contemporary Mongol culture in China (Goossaert & Palmer, 2011, p. 369), and the dominance of Mandarin Chinese in the economy is rapidly marginalizing the use of Mongolian, as well as other minority languages (Bulag, 2003, pp. 234–5). Relations between ethnic Mongols and Han throughout periods of the 20th century were marked by tensions concerning issues of political autonomy and national identity (Bulag, 2003, p. 229), but ethnic Mongol Nationalism had largely subsided by the 21st century (Goossaert, 2011, p. 369).  4.3.2.4 Hui The Hui lack their own distinct heritage language and primarily use the local language, most often Chinese (Gladney, 1998, p. 37). Essentially, the Hui ethnic classification was created based on lineage to Persian, Arab, Turkish, and Mongol migrants who settled throughout China prior to the 14th century and who retained their Islamic faith, but adopted local cultural practices such as attire and language (Gladney, 1998, p. 49). Due largely to their eclectic histories and geographic dispersal throughout most of the PRC, ethnic Hui of different regions engage in dissimilar cultural practices and subscribe to dissimilar notions of what defines Hui ethnicity (Gladney, 1998, pp. 166–7). Consequently, conventional Islamic practices such as male circumcision vary widely among the Hui and are generally less prevalent when compared with ethnic Uyghurs (Mackerras, 1995, p. 117). Nonetheless, in spite of their differences, ethnic Hui in China readily endorse the Hui ethnic classification and identify with other Hui as ethnic peers (Gladney, 1998, p. 167). Relations between ethnic Hui and Han in the 21st century have for the most part been amicable, and devoid of interethnic violence (Crane, 2014).  137   4.3.3 Ethical reviews To help ensure that the rights and welfare of study participants were adequately protected, dissertation study protocols were reviewed and approved by the National Center for AIDS/STD Control and Prevention, Chinese Centers for Disease Control and Prevention (CDC) (Project number: X120717232, Co-PIs: Stephen W. Pan, Yuhua Ruan, and Yiming Shao), and the Providence Health Care Research Institute, University of British Columbia (Reference number: PHC REB H12-00975, PI: Patricia M. Spittal) prior to any data collection.  4.3.4 Participant enrollment and survey data collection Study participants were recruited for a cross-sectional survey of ethnically and religiously diverse MSM conducted from July 2013 to April 2014 at five HIV VCT clinics devoted to serving MSM throughout Beijing and Tianjin. Participants were recruitment in four ways. First, men who had come to receive VCT services on their own volition were invited on-site to participate in the study. Second, successfully enrolled participants were encouraged to recruit their eligible MSM friends. Third, MSM-oriented HIV prevention CBOs distributed study recruitment information to potential participants via a smart phone social networking application. Fourth, potential study participants were contacted via several gay social networking websites and chatrooms. Participants were free to use pseudonyms, and no true identifiers were requested. Men were eligible to participate in the survey if they self-reported ever having sex with another man, were at least 18 years old, had never tested positive for HIV, and were willing and able to provide written informed consent. Due to resource constraints precluding unconditional recruitment of all VCT clinic patrons, a 1:1 design ratio was maintained throughout the survey enrollment period between the designated comparison group (ethnic or religious minority MSM) and referent group (non-religious ethnic majority Han MSM); in effect, only a proportion of  138   otherwise eligible non-religious ethnic majority Han MSM patrons of the VCT sites were enrolled into the study. Of 407 men who met the eligibility criteria and were invited to participate in the study, seven refused enrollment (1.7%). Once study staff ascertained eligibility, obtained written-informed consent, and provided instructions on how to complete the survey, participants proceeded to complete the self-administered, pen-and-paper survey in private at the VCT site. For participants who were functionally illiterate (<5%), surveys were administered by study staff. Upon completing the survey, participants received confidential HIV pre-test counseling, a blood-based HIV rapid test, and appropriate post-test counseling from trained public health staff. Participants received an honorarium of ¥50 CNY (~ $8 USD as of July 1, 2013).  4.3.5 HIV diagnostics          HIV-1 serostatus was initially determined by blood-based rapid testing (Determine HIV-1/2, Abbott Japan Co., Japan) and confirmed by HIV-1/2 Western Blot (HIV Blot 2.2 WBTM, Genelabs Diagnostics, Singapore).  4.3.6 Data management Completed paper-based survey questionnaires were checked for completeness and logical responses by at least two project staff members. Then, surveys were entered into computerized databases by China CDC public health graduate students using EpiData Entry 3.1 (The Epidata Association, Odense). In order to ensure fidelity of data entry, all questionnaire data were double-entered by two different individuals. The PROC COMPARE function in SAS 9.3 (SAS Institute, Cary, NC) was then used to identify discrepant entries, which were ultimately resolved by examination of the original paper-based questionnaire. At the completion of data collection, less than 1% of data fields were missing.  139   4.3.7 Measures 4.3.7.1 Dependent variables Participants reported information about sexual histories (age at same-sex sexual debut and circumcision), number of male sex partners in the past six months, sexual behaviours in the past three months (engaged in male group sex, any anal sex, UAI with male partners, and anal sex position), HIV testing history, and substance use (frequency of alcohol consumption and use of illicit drugs). Continuous variables were dichotomized or categorized into tertiles because of extreme outlier values and to facilitate clearer data interpretation.  Ethnic steady sex partner preferences was assessed by asking respondents to report if it was important for a steady sex partner to be of their same ethnicity. Sexual partner characteristics were assessed using a modified version of the UNAIDS Sexual Network Questionnaire (UNAIDS, 1998). Using a multiple-choice format, each individual was asked to report the gender and ethnic affiliation of their three most recent sex partners in the past six months. Dichotomous dummy variables were then created to indicate whether or not the participant’s last three sex partners in the past six months included a male member of a specific ethnic affiliation. Due to small cell sizes, sex partners reported as “Manchu” or “Mongol” were combined with the “other” category of ethnic minorities from North China, and sex partners belonging to an ethnic minority group from South China were excluded from the analysis. 4.3.7.2 Independent variables Ethnic affiliation was reported by multiple-choice selection. Perceived ethnic discrimination was measured as having ever experienced ethnic discrimination. Mandarin language proficiency was determined by asking participants if Mandarin was their most proficient spoken language.   140   Ethnic identity affirmation was assessed using a modified version of the Multigroup Ethnic Identity Measure’s (MEIM) five-item “Affirmation and Belonging” sub-scale, which was designed to measure “ethnic pride, feeling good about one’s background, and being happy with one’s group membership” (Phinney, 1992). The MEIM uses a five-point Likert scale ranging from “1 = Strongly disagree” to “5 = Strongly agree,” and respondents are asked to report their degree of endorsement for statements such as “I have a strong sense of belonging to my own ethnic group” and “I have a lot of pride in my ethnic group and its accomplishments.”  The raw scores for ethnic identity affirmation were each calculated by averaging the values of all their respective sub-scale items.  Three other bilingual project staff from China CDC and I collectively translated the ethnic identity affirmation sub-scales after thorough discussions about how best to faithfully represent the original English meanings in written Chinese. The Cronbach’s alpha level for the sense of ethnic identity affirmation subscale was 0.89, indicating strong scale reliability. 4.3.7.3 Control variables Five measures were assessed as control variables in data analyses: Age, migrant status, city (Beijing or Tianjin), rural vs. urban household registration status, and highest level of educational attainment.  4.3.8 Data analyses Chi-square and Fisher’s exact tests were used to compare distributions of ethnic affiliation, Mandarin language proficiency, and perceived discrimination across HIV status and testing history, sexual behaviours & experiences, substance use, and sociodemographics. Analysis of variance was used to compare the mean scores of ethnic identity affirmation across the aforementioned dependent and control variables. Pooled and Satterthwaite methods were  141   used to calculate significance tests for 2-group sample comparisons with equal and unequal variances, respectively. Then, binary or ordinal logistic regression was employed to evaluate measures of associations between independent and dependent variables. Each dependent variable was regressed on each independent variable in unadjusted binary or ordinal logistic regression models. Assumptions of proportionality were satisfied for all ordinal logistic regression models. Then, in order to obtain adjusted measures of association between independent and dependent variables, control variables were added as covariates to each regression model. To assess if ethnic affiliation moderated the strength of associations between ethnic identity affirmation and HIV vulnerabilities/testing, two-way interaction terms were created and modeled in logistic regression by crossing ethnic affiliation with ethnic identity affirmation.  To assess how the ethnicity of sex partners varied between ethnic minority and ethnic majority Han men, four separate binary logistic regression models were used to calculate if ethnic minority affiliation was significantly associated with odds of having a sex partner from one of four ethnic minority group categories: (a) ethnic Han, (b) ethnic Hui, (c) pan-Northern ethnic minority, and (d) unknown ethnicity. Statistical significance was defined as p<0.05. Marginal significance was defined as 0.05<p<0.1. 4.4 Results  4.4.1 Descriptive findings  Descriptive findings stratified by ethnic affiliation are presented in Table 4.1. Ethnic Han respondents comprised the preponderance of the sample (69%), followed by Hui (11%), Manchu (7%), Mongol (5%), other ethnic minorities of Southern China (5%), and other ethnic minorities of Northern China (3%). Most participants were younger than 30 years of age (63%), single  142   never-married (84%), and not a permanent resident of the interview city (63%). Educational attainment was relatively high: 54% completed four-year college, 41% completed high school or vocational college, and only 6% had less than high school education.  Overall HIV prevalence was 9.8%, and varied significantly by ethnic affiliation. Other-Southern ethnic minorities had the highest prevalence of HIV (37%), followed by Manchus (18%), Mongols (10%), Hui (9%), Other-Northerners (9%), and the Han majority (7%). Most participants reported previously receiving an HIV test (78%). Approximately one-third of participants experienced their same-sex sexual debut before their 20th birthday, 66% reported fewer than four male sex partners in the past six months, and only 16% had been circumcised. Prevalence of circumcision among ethnic Muslim Hui (36%) was significantly higher than all other ethnic groups. Within the past three months, 85% of participants had engaged in anal sex with another man, and over half reported recent UAI (57%). Among only those who had anal sex in the past three months, “versatile” (i.e., insertive and receptive) was the most commonly reported sexual position (45%), followed by “insertive” (31%), and “receptive” (24%).  The mean score of ethnic identity affirmation was 3.92 and significantly varied by ethnic affiliation. Ethnic majority Han reported the lowest scores for ethnic identity affirmation, followed by Manchus, Other-Southerners, Mongols, Hui, and Other-Northerners. Most participants reported Mandarin as their most fluent spoken language (92%), but prevalence was significantly lower among Southern ethnic minorities (79%) and Other-Northern ethnic minorities (73%). Compared to ethnic Han, prevalence of ethnic discrimination was significantly higher among ethnic Hui (15% vs 3%) and Northern ethnic minorities (36% vs 3%). Preference for steady sex partners of the same ethnicity significantly differed by ethnic affiliation. Over one fourth of ethnic Han reported preference for ethnic Han steady sex partners,  143   but Manchus and ethnic minorities of Southern China were largely indifferent to the ethnicity of their steady sex partner. Conversely, 41% of Hui and 45% of Other Northern ethnic minorities felt that it was important that steady male sex partners were of the same ethnicity. Interethnic sexual partnership patterns also differed according to participants’ ethnic affiliation. Ethnic Manchu participants were less likely to report sex with ethnic Han men, and ethnic Hui participants were far more likely than other ethnic groups to report sex with ethnic Hui men.     144   Table 4.1: Survey participant characteristics, by ethnic affiliation  Han (n=275)    Manchu (n=28) Mongol (n=20) Hui (n=45) Other –Southern (n=19) Other – Northern (n=11) p-value  n (%) n (%) n (%) n (%) n (%) n (%)   HIV STATUS & TESTING        Ever tested for HIV 222 (81) 20 (71) 15 (75) 34 (76) 11 (58) 8 (72) 0.24 HIV positive  20 (7) 5 (18) 2 (10) 4 (9) 7 (37) 1 (9) <0.01  ACCULTURATION        Ethnic identity affirmation Mean score (SD)  2..67 (0.77) 3.33 (0.85) 3.71 (0.72) 3.77 (0.83) 3.54 (0.88) 4.27 (0.74) 0.03 Ever felt ethnically discriminated 8 (3) 0  (0) 0  (0) 7  (15) 2  (11) 4  (36) <0.0001 Most fluent spoken language is Mandarin 257 (93) 28 (100) 20 (100) 40 (87) 15 (79) 8 (73) <0.01  SOCIODEMOGRAPHICS        City           Beijing 228 (83) 25 (89) 14 (70) 24 (52) 16 (84) 11 (100) <0.0001    Tianjin 48 (17) 3 (11) 6 (6) 22 (48) 3 (16) - Age           <25 years 78 (28) 6 (21) 7 (35) 15 (33) 11 (58) 3 (27) 0.24    25-29 years 86 (31) 9 (32) 8 (40) 17 (38) 5 (26) 4 (37)    >30 years 111 (40) 13 (46) 5 (25) 13 (29) 3 (16) 4 (37) Educational attainment        Less than secondary school 15 (5) 5 (18) - 2 (4) 2 (11) - 0.06 Secondary school or vocational college 115 (42) 13 (46) 4 (20) 19 (42) 8 (42) 3 (27) Four-year college 145 (53) 10 (36) 16 (80) 24 (53) 9 (47) 8 (73) Migrant status (vs. official permanent resident)  175 (64) 19 (68) 16 (80) 18 (40) 13 (68) 6 (55) 0.02 Urban (vs. Rural) household registration status 178 (65) 17 (61) 4 (20) 40 (89) 13 (68) 7 (64) 0.03 Marital status        Single 232 (84) 18 (64) 19 (95) 39 (87) 18 (95) 9 (82) 0.01 Married, living with wife 16 (6) 5 (18) - 3 (7) 1 (5) - Married, not living with wife 22 (8) 2 (7) - 2 (4) - - Divorced, in a “sham” marriage, or other 5 (2) 3 (11) 1 (1) 1 (2) - 2 (18)  SEXUAL BEHAVIOURS & EXPERIENCES Age at same-sex sexual debut            <20 years  92 (34) 9 (32) 5 (25) 22 (48) 8 (42) 3 (30) 0.55    20-22 years 90 (33) 10 (36) 6 (30) 16 (35) 7 (37) 3 (30)    >22 years old 92 (34) 9 (32) 9 (45) 8 (17) 4 (21) 4 (40) Circumcised 42 (15) 1 (4) 1 (5) 16 (36) 1 (5) 2 (18) <0.01 Bought or sold sex a 11 (4) 1 (4) 1 (5) 1 (2) - 2 (18) 0.22 Engaged in male group sex a 31 (11) 3 (11) 2 (10) 7 (16) - 2 (18) 0.58 Anal sex with another male a 234 (85) 23 (82) 19 (95) 34 (76) 3 (16) 10 (91) 0.42 Unprotected anal intercourse a 161 (58) 12 (43) 14 (70) 19 (41) 13 (68) 7 (64) 0.09  145    Table 4.1 (continued): Survey participant characteristics, by ethnic affiliation  Han (n=275)    Manchu (n=28) Mongol (n=20) Hui (n=45) Other –Southern (n=19) Other – Northern (n=11) p-value  n (%) n (%) n (%) n (%) n (%) n (%)   SEXUAL BEHAVIOURS & EXPERIENCES Anal sex positioning a, c              Insertive only 74 (32) 9 (39) 3 (16) 12 (35) 1 (6) 5 (50) 0.19       Versatile 103 (44) 9 (39) 12 (63) 15 (44) 7 (44) 4 (40)       Receptive only 56 (24) 5 (22) 4 (21) 7 (21) 8 (50) 1 (10) Number of male sex partners b        Fewer than two  83 (30) 11 (39) 4 (20) 19 (41) 6 (32) 3 (27) 0.55 Two or three 100 (36) 7 (25) 10 (50) 15 (33) 4 (21) 5 (45) Four or more 93 (34) 10 (36) 6 (30) 12 (26) 9 (47) 3 (27) Ever experienced forced sex 35 (13) 1 (4) 4 (20) 8 (18) 6 (32) 2 (18) 0.11  SUBSTANCE USE        Ever used illicit drugs 13 (5) 1 (4) 1 (5) 1 (2) 1 (5) 1 (9) 0.94 Frequency of alcohol intake b        Never 74 (27) 6 (21) 7 (35) 17 (37) 4 (21) 2 (18) 0.74 Less than once per week 173 (63) 21 (75) 11 (55) 25 (54) 12 (63) 7 (64) More than once per week 28 (10) 1 (4) 2 (10) 4 (9) 3 (16) 2 (18) ETHNIC AFFILIATION & SEX PARTNERS Important that steady sex partner is of the same ethnicity 72 (26) 1 (4) 4 (20) 19 (41) 1 (5) 5 (45) <0.01 Had sex with an ethnic Han man b 229 (83) 17 (61) 14 (70) 38 (83) 17 (89) 9 (82) 0.06 Had sex with an ethnic Hui man b 4 (1) 1 (4) 0 (0) 7 (15) 1 (5) 1 (9) <0.0001 Had sex with an ethnic minority Northern man b 4 (1) 1 (4) 0 (0) 1 (2) 1 (5) 1 (9) 0.41 Had sex with a man of unknown ethnic affiliation b 65 (24) 11 (39) 9 (45) 10 (22) 5 (26) 2 (18) 0.16 a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months;   Cross-tabulations of ethnic identity affirmation are presented in Table 4.2. Ethnic affirmation scores were significantly higher among participants who reported anal sex in the previous three months, had lower Mandarin language proficiency, or were permanent residents of the interview city. However, mean scores of ethnic identity affirmation were comparable across all other sociodemographic, acculturation, and HIV-related variables.  146   Table 4.2: Survey participant characteristics, by ethnic identity affirmation   Ethnic identity affirmation  p-value  Mean score (SD)   HIV STATUS & TESTING   Ever tested for HIV   No 3.76 (0.82) 0.21 Yes 3.64 (0.79) HIV positive    No 3.68 (0.79) 0.29 Yes 3.54 (0.83)  ACCULTURATION   Ever felt ethnically discriminated   No 3.66 (0.79) 0.25 Yes 3.87 (0.92) Most fluent spoken language is Mandarin   No 3.98 (0.81) 0.02 Yes 3.64 (0.79)  SOCIODEMOGRAPHICS   City   Beijing 3.67 (0.78) 0.95 Tianjin 3.67 (0.86) Age   <25 years 3.70 (0.69) 0.20 25-29 years 3.57 (0.86) >30 years 3.73 (0.81) Educational attainment   Less than secondary school 3.60 (0.68) 0.58 Secondary school or vocational college 3.72 (0.76) Four-year college 3.64 (0.83) Residency status    Official permanent resident 3.77 (0.75) 0.04 Migrant 3.61 (0.81) Household registration status   Urban 3.65 (0.80)  0.52 Rural 3.70 (0.77) Marital status   Single 3.65 (0.80) 0.24 Married, living with wife 3.72 (0.94) Married, not living with wife 3.75 (0.64) Divorced, in a “sham” marriage, or other 4.10 (0.63)        147    Table 4.2 (continued): Survey participant characteristics, by ethnic identity affirmation  Ethnic identity affirmation  p-value  Mean score (SD)   SEXUAL BEHAVIOURS & EXPERIENCES   Age at same-sex sexual debut    <20 years  3.67 (0.77) 0.76 20-22 years 3.63 (0.80) >22 years old 3.70 (0.82) Circumcised   No 3.70 (0.85) 0.71 Yes 3.66 (0.79) Bought or sold sex a   No 3.67 (0.79) 0.68 Yes 3.75 (0.95) Engaged in male group sex a   No 3.69 (0.76) 0.27 Yes 3.52 (1.02) Anal sex with another male a   No 3.94 (0.66) <0.01 Yes 3.62 (0.81) Unprotected anal intercourse a   No 3.70 (0.78) 0.50 Yes 3.65 (0.81) Anal sex positioning a, c   Insertive only 3.71 (0.78) 0.31 Versatile 3.55 (0.82) Receptive only 3.63 (0.83) Number of male sex partners b    Fewer than two  3.77 (0.77) 0.10 Two or three 3.69 (0.82) Four or more 3.56 (0.78) Ever experienced forced sex   No 3.69 (0.79) 0.31 Yes 3.57 (0.80)  SUBSTANCE USE   Ever used illicit drugs   No 3.68 (0.79) 0.35 Yes 3.50 (0.98) Frequency of alcohol intake b   Never 3.67 (0.76) 0.99 Less than once per week 3.67 (0.80) More than once per week 3.68 (0.84) a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months   148   Mandarin language proficiency and ethnic discrimination cross-tabulations are presented in Table 4.3. Strong Mandarin language proficiency was evenly distributed across most measures, with the exceptions of forced sex and experience of ethnic discrimination. Participants with weaker Mandarin fluency were twice as likely to report history of forced sex (28% vs. 13%), and four times more likely to report history of ethnic discrimination (16% vs. 4%). Aside from Mandarin language proficiency, history of perceived ethnic discrimination was not significantly associated with any assessed measures in the study.     149   Table 4.3: Survey participant characteristics, by language use & ethnic discrimination  Primary spoken language Ever perceived ethnic discrimination  Mandarin n (%) Other n (%) p-value Yes n (%) No n (%) p-value  HIV STATUS & TESTING       Ever tested for HIV 287 (78) 25 (78) 0.99 17 (81) 295 (78) 0.74 HIV positive  34 (9) 5 (16) 0.25 2 (10) 37 (10) 0.97  SOCIODEMOGRAPHICS       City          Beijing 295 (80) 23 (72) 0.27 18 (86) 300 (79) 0.47    Tianjin 73 (20) 9 (28) 3 (14) 79 (21) Age          <25 years 114 (31) 7 (22)  6 (29) 115 (30)     25-29 years 118 (32) 12 (38) 0.55 7 (33) 123 (32) 0.99    >30 years 136 (37) 13 (41)  8 (38) 141 (37)  Educational attainment       Less than secondary school 22 (6) 2 (6)  0 (0) 24 (6)  Secondary school or vocational college 145 (39) 17 (53) 0.29 8 (38) 154 (41) 0.44 Four-year college 201 (55) 13 (41)  13 (62) 201 (53)  Migrant status (vs. official permanent resident)  226 (61) 23 (72) 0.24 15 (71) 234 (62) 0.37 Urban (vs. Rural) household registration status 142 (39) 9 (28) 0.24 18 (86) 254 (67) 0.07 Marital status       Single 310 (84) 27 (84) 0.61 18 (86) 319 (85) 0.94 Married, living with wife 23 (6) 2 (6) 1 (5) 24 (6) Married, not living with wife 25 (7) 1 (3) 1 (5) 25 (7) Divorced, in a “sham” marriage, or other 10 (3) 2 (6) 1 (5) 11 (3)  SEXUAL BEHAVIOURS & EXPERIENCES Age at same-sex sexual debut           <20 years  131 (36) 8 (25)  7 (35) 132 (35)     20-22 years 118 (32) 14 (44) 0.34 7 (35) 125 (33) 0.98    >22 years old 116 (32) 10 (31)  6 (30) 120 (32)  Circumcised 55 (15) 8 (25) 0.13 6 (29) 57 (15) 0.10 Bought or sold sex a 16 (4) 0 (0) 0.23 1 (5) 15 (4) 0.85 Engaged in male group sex a 39 (11) 6 (19) 0.16 4 (19) 41 (11) 0.25 Anal sex with another male a 309 (84) 29 (91) 0.32 17 (81) 321 (85) 0.64 Unprotected anal intercourse a 205 (56) 21 (66) 0.28 14 (67) 212 (56) 0.33 Anal sex positioning a, c             Insertive only 93 (30) 12 (41)  5 (29) 100 (31)        Versatile 140 (45) 11 (38) 0.46 9 (53) 142 (44) 0.75       Receptive only 75 (24) 6 (21)  3 (18) 78 (24)  Number of male sex partners b       Fewer than two  119 (32) 7 (22)  6 (29) 120 (32)  Two or three 129 (35) 12 (38) 0.44 7 (33) 134 (36) 0.89 Four or more 120 (33) 13 (41)  8 (38) 125 (33)  Ever experienced forced sex 48 (13) 9 (28) 0.02 6 (29) 51 (13) 0.05  150    Table 4.3 (continued): Survey participant characteristics, by language use & ethnic discrimination  Primary spoken language Ever perceived ethnic discrimination  Mandarin n (%) Other n (%) p-value Yes n (%) No n (%) p-value  SUBSTANCE USE       Ever used illicit drugs 18 (5) 0 (0) 0.20 1 (5) 17 (4) 0.95 Frequency of alcohol intake b       Never 103 (28) 8 (25)  5 (24) 106 (28)  Less than once per week 228 (62) 21 (66) 0.92 14 (67) 235 (62) 0.91 More than once per week 37 (10) 3 (9)  2 (10) 38 (10)  a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months;  4.4.2 Hypothesis 1: Association between ethnic affiliation and HIV vulnerabilities & testing  The first study hypothesis stated that some but not all ethnic minority groups have greater HIV vulnerabilities and poorer testing practices compared to the Han ethnic majority. Results presented in Table 4.4 supported this hypothesis, indicating that while Manchu and Mongol HIV vulnerabilities were comparable to the Han ethnic majority, the same was not true of ethnic Hui, ethnic minorities of Southern China, or other ethnic minorities of Northern China. Compared to ethnic Han, ethnic minorities of Southern China had significantly higher odds of HIV infection (adjusted odds ratio (AOR): 7.40, 95% confidence interval (CI): 2.33–23.47) and forced sex (AOR: 3.27, 95% CI: 1.12-9.52), while other ethnic minorities of Northern China had almost six-fold greater odds of buying or selling sex (AOR: 5.97, 95% CI: 1.04–34.22).  However, in contradiction to the hypothesis, ethnic affiliation among ethnic minority Hui was inversely associated with HIV vulnerabilities. Compared to the Han majority, ethnic Hui had significantly lower odds of same-sex anal intercourse (AOR: 0.43, 95% CI: 0.19-0.98) and UAI  151   (AOR: 0.41, 95% CI: 0.21-0.82). Odds of circumcision were also significantly higher for ethnic Hui than the Han majority (AOR: 2.62, 95% CI: 1.24-5.51). 4.4.3 Hypothesis 2: Association between participant ethnic affiliation and Ethnicity of sex partner Table 4.4 also presents evidence in support of the second study hypothesis, which postulated that participant ethnic affiliation is associated with the ethnic affiliations of their sex partners. Compared to Han participants, Manchu participants had significantly lower odds of reporting sex with ethnic Han men (AOR: 0.37, 95% CI: 0.16 – 0.87), but higher odds of reporting sex with men of an unknown ethnicity (AOR: 2.63, 95% CI: 1.01 – 6.84). Ethnic Hui participants had 16-fold higher odds of reporting sex with an ethnic Hui man (AOR: 16.23, 95% CI: 3.83 – 68.80), but odds of partnering with men of Han, Northern, or unknown ethnicity were otherwise comparable with ethnic Han participants. Ethnic Mongol, Southern-, and Northern ethnic minority participants did not significantly differ from Han participants’ ethnic partnership patterns.            152   Table 4.4: Ethnic affiliation and dependent variable correlations  Dependent  variable Ethnic affiliation Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI)  HIV STATUS & TESTING    Ever tested for HIV Han 1.00 1.00  Manchu 0.59 (0.25 – 1.42) 0.59 (0.23 – 1.54)  Mongol 0.71 (0.25 – 2.05) 0.64 (0.20 – 2.05)  Hui 0.76 (0.36 – 1.59) 1.18 (0.52 – 2.71)  Other – South China 0.33 (0.13 – 0.85)* 0.43 (0.15 – 1.22)  Other – North China 0.63 (0.16 – 2.47) 0.46 (0.10 – 2.08) HIV infection Han 1.00 1.00  Manchu 2.77 (0.95 – 8.07)† 2.28 (0.72 – 7.29)  Mongol 1.42 (0.31 – 6.54) 1.37 (0.26 – 7.16)  Hui 1.24 (0.41 – 3.82) 1.49 (0.43 – 5.23)  Other – South China 7.44 (2.64 – 20.99)* 7.40 (2.33 – 23.47)*  Other – North China 1.28 (0.16 – 10.47) 1.98 (0.23 – 17.33)  SEXUAL BEHAVIOURS & EXPERIENCES Same-sex anal intercourse a Han 1.0 1.00  Manchu 0.80 (0.29 – 2.23) 0.81 (0.28 – 2.31)  Mongol 3.32 (0.43 – 25.4) 3.24 (0.41 – 25.45)    Hui 0.56 (0.26 – 1.18) 0.43 (0.19 – 0.98)*  Other – South China 0.93 (0.26 – 3.34) 0.97 (0.26 – 3.64)  Other – North China 1.75 (0.22 – 14.00) 2.08 (0.26 – 16.79) Unprotected anal intercourse a Han 1.00 1.00  Manchu 0.54 (0.24 – 1.18) 0.50 (0.22 – 1.12)  Mongol 1.67 (0.62 – 4.47) 1.82 (0.66 – 5.00)  Hui 0.50 (0.27 – 0.95)* 0.41 (0.21 – 0.82)*  Other – South China 1.55 (0.57 – 4.19) 1.64 (0.59 – 4.53)  Other – North China 1.25 (0.36 – 4.37) 1.49 (0.42 – 5.31) More male sex partners b Han 1.00 1.00  Manchu 0.84 (0.41 – 1.72) 0.74 (0.35 – 1.53)  Mongol 1.13 (0.49 – 2.61) 1.15 (0.49 – 2.70)  Hui 0.65 (0.36 – 1.15) 0.54 (0.29 – 1.01)†  Other – South China 1.38 (0.58 – 3.26) 1.50 (0.62 – 3.62)  Other – North China 0.92 (0.30 – 2.79) 1.12 (0.37 – 3.41) Ever experienced forced sex Han 1.00 1.00  Manchu 0.25 (0.03 – 1.87) 0.23 (0.03 – 1.77)  Mongol 1.67 (0.53 – 5.27) 1.89 (0.57 – 6.24)  Hui 1.40 (0.61 – 3.25) 1.85 (0.75 – 4.57)  Other – South China 3.08 (1.10 – 8.61)* 3.27 (1.12 – 9.52)*  Other – North China 1.48 (0.31 – 7.13) 1.27 (0.26 – 6.25) MSM sexual debut before 23 years of age Han 1.00 1.00 Manchu 1.06 (0.46 – 2.43) 1.36 (0.53 – 3.47)  Mongol 0.61 (0.25 – 1.53)  0.35 (0.12 – 1.00)†  Hui 2.38 (1.07 – 5.30)* 1.59 (0.65 – 3.91)  Other – South China 1.88 (0.61 – 5.81)  0.92 (0.26 – 3.28)  Other – North China 0.88 (0.25 – 3.07)  0.91 (0.22 – 3.68)        153   Table 4.4 (continued):  Ethnic affiliation and dependent variable correlations Dependent  variable Ethnic affiliation Crude odds ratio  (95% CI) Adjusted odds ratio d (95% CI)  SEXUAL BEHAVIOURS & EXPERIENCES Receptive anal sex only a ,c Han 1.00 1.00  Manchu 0.88 (0.31 – 2.49) 0.97 (0.33 – 2.82)  Mongol 0.85 (0.27 – 2.66) 0.86 (0.26 – 2.83)  Hui 0.80 (0.33 – 1.92) 0.46 (0.18 – 1.18)  Other – South China 3.18 (1.14 – 8.86)* 2.87 (0.97 – 8.51)†  Other – North China 0.35 (0.04 – 2.85) 0.38 (0.05 – 3.09) Circumcised Han 1.00 1.00  Manchu 0.21 (0.03 – 1.56) 0.23 (0.03 – 1.76)  Mongol 0.29 (0.04 – 2.25) 0.26 (0.03 – 2.04)  Hui 2.97 (1.49 – 5.92)* 2.62 (1.24 – 5.51)*  Other – South China 0.31 (0.04 – 2.38) 0.32 (0.04 – 2.48)  Other – North China 1.24 (0.26 – 5.93) 1.17 (0.24  -5.66) Bought or sold sex a Han 1.00 1.00  Manchu 0.89 (0.11 – 7.18) 1.00 (0.12 – 8.31)  Mongol 1.27 (0.16 – 10.35) 1.50 (0.17 – 13.21)  Hui 0.54 (0.07 – 4.25) 0.70 (0.08 – 6.08)  Other – South China not calculable    not calculable  Other – North China 5.35 (1.03 – 27.78)* 5.97 (1.04 – 34.22)* Engaged in male group sex a Han 1.00 1.00  Manchu 0.95 (0.27 – 3.33) 0.79 (0.21 – 2.98)  Mongol 0.88 (0.19 – 3.97) 0.92 (0.19 – 4.47)  Hui 1.42 (0.58 – 3.44) 1.51 (0.57 – 3.99)  Other – South China not calculable    not calculable  Other – North China 1.76 (0.36 – 8.50) 2.32 (0.46 – 11.72)  SUBSTANCE USE    Ever used illicit drugs Han 1.00 1.00  Manchu 0.75 (0.09 – 5.93) 0.92 (0.11 – 7.81)  Mongol 1.06 (0.13 – 8.55) 0.99 (0.11 – 9.08)  Hui 0.46 (0.06 – 3.59) 0.29 (0.03 – 2.43)  Other – South China 1.12 (0.14 – 9.05) 1.32 (0.15 – 11.32)  Other – North China 2.02 (0.24 – 16.96) 3.25 (0.15 – 11.32) More frequent alcohol intake Han 1.00 1.00  Manchu 1.67 (0.70 – 3.99) 1.86 (0.77 – 4.52)  Mongol 0.71 (0.29 – 1.71) 0.60 (0.24 – 1.49)  Hui 0.65 (0.35 – 1.19) 0.76 (0.40 – 1.45)  Other – South China 1.10 (0.42 – 2.86) 1.10 (0.41 – 2.97)  Other – North China 1.16 (0.33 – 4.05) 1.03 (0.29 – 3.70)  ETHNIC AFFILIATION & SEX PARTNERS  Had an ethnic Han male partner Han 1.00 1.00  Manchu 0.32 (0.14 – 0.72)* 0.37 (0.16 – 0.87)*  Mongol 0.48 (0.18 – 1.31) 0.39 (0.14 – 1.13)†  Hui 0.98 (0.43 – 2.22) 0.94 (0.39 – 2.25)  Other – South China 1.75 (0.39 – 7.81) 1.99 (0.43 – 9.31)  Other – North China 0.92 (0.19 – 4.41) 0.82 (0.17 – 4.01)   154   Table 4.4 (continued): Ethnic affiliation and dependent variable correlations Dependent  variable Ethnic affiliation Crude odds ratio  (95% CI) Adjusted odds ratio d (95% CI)  ETHNIC AFFILIATION & SEX PARTNERS Had an ethnic Hui male partner Han 1.00 1.00  Manchu 2.52 (0.27 – 23.35) 2.17 (0.22 – 21.03)  Mongol Not calculable    Not calculable  Hui 12.21 (3.42 – 43.61)* 16.23 (3.83 – 68.80)*  Other – South China 3.78 (0.40 – 35.58) 3.27 (0.31 – 34.08)  Other – North China 6.80 (0.70 – 66.51) 5.41 (0.50 – 58.10) Had an ethnic minority Northern male partner Han 1.00 1.00  Manchu 2.52 (0.27 – 23.35) 3.36 (0.32 – 35.04)  Mongol Not calculable Not calculable  Hui 1.51 (0.17 – 13.83) 2.45 (0.23 – 26.08)  Other – South China 3.78 (0.40 – 35.58) 2.28 (0.21 – 24.17)  Other – North China 6.80 (0.70 – 66.51) 6.40 (0.59 – 69.96) Had a partner of unknown ethnic affiliation Han 1.00 1.00 Manchu 2.10 (0.94 – 4.71)† 2.63 (1.01 – 6.84)*  Mongol 2.66 (1.05 – 6.69)* 2.81 (0.98 – 8.00)†  Hui 0.90 (0.42 – 1.92) 1.03 (0.43 – 2.47)  Other – South China 1.16 (0.40 – 3.34) 1.27 (0.39 – 4.08)  Other – North China 0.72 (0.15 – 3.42) 0.75 (0.14 – 3.86) a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months;  d adjusted models control for age, migrant status, city, rural/urban status, and education;  * p<0.05; † p<0.1   4.4.4 Hypotheses 3-4: Association between ethnic identity affirmation and HIV vulnerabilities & testing; Association between ethnic identity affirmation × ethnic affiliation and HIV vulnerabilities & testing  Results presented in Table 4.5 partially support the third study hypothesis that stronger ethnic identity affirmation correlates with reduced HIV vulnerabilities and more testing. Each one-point increase in the ethnic identity affirmation score was significantly associated with 48% lower odds of same-sex anal intercourse (AOR: 0.52, 95% CI: 0.35-0.78), and 24% lower odds of having more male sex partners (AOR: 0.76, 95% CI: 0.60-0.96). However, contrary to the fourth hypothesis, ethnic affiliation did not moderate the magnitude of association between ethnic identity affirmation and any measure of HIV vulnerability or testing (results not shown).   155   Table 4.5: Ethnic identity affirmation and dependent variable correlations Dependent  variable Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI)  HIV STATUS & TESTING   Ever tested for HIV 0.82 (0.61 – 1.12) 0.76 (0.54 – 1.08) HIV infection 0.81 (0.54 – 1.21) 0.91 (0.57 – 1.43)  SEXUAL BEHAVIOURS & EXPERIENCES  Same-sex anal intercourse a 0.57 (0.39 – 0.83)* 0.52 (0.35 – 0.78)* Unprotected anal intercourse a 0.92 (0.71 – 1.18) 0.87 (0.67 – 1.14) More male sex partners b 0.78 (0.62 – 0.98)* 0.76 (0.60 – 0.96)* Ever experienced forced sex 0.84 (0.59 – 1.18) 0.79 (0.55 – 1.15) MSM sexual debut before age 23 0.92 (0.71 – 1.21) 0.95 (0.70 – 1.28) Receptive anal sex only a ,c 1.03 (0.76 – 1.41) 1.06 (0.75 – 1.49) Circumcised 1.07 (0.76 – 1.50) 1.01 (0.71 – 1.44) Bought or sold sex a 1.14 (0.60 – 2.18) 1.03 (0.53 – 1.98) Engaged in male group sex a 0.77 (0.52 – 1.12) 0.72 (0.48 – 1.06)  SUBSTANCE USE   Ever used illicit drugs 0.76 (0.43 – 1.35) 0.70 (0.39 – 1.28) More frequent alcohol intake 0.99 (0.77 – 1.27) 1.06 (0.81 – 1.37) a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months; d adjusted models control for age, migrant status, city, rural/urban status, and education;  * p<0.05    4.4.5  Hypothesis 5: Association between mandarin proficiency and HIV vulnerabilities & testing Results presented in Table 4.6 offer limited support for the hypothesis that poorer Mandarin language proficiency is associated with greater HIV vulnerabilities and less HIV testing. Aside from a marginally significant association with experience of forced sex (AOR: 2.43, 95% CI: 0.99-6.00), speaking a primary language other than Mandarin was not correlated with any other measure of HIV vulnerability or testing.   156   4.4.6 Hypothesis 6: Association between ethnic discrimination and HIV vulnerabilities & testing The sixth study hypothesis postulated that ethnic discrimination is significantly associated with greater HIV vulnerabilities and less HIV testing, but no supporting evidence was detected (Table 4.6). Ethnic discrimination was not associated with HIV testing or any measure of HIV vulnerability.  157   Table 4.6: Language proficiency, ethnic discrimination, and dependent variable correlations   Most proficient spoken language is not Mandarin Ever experienced ethnic discrimination Dependent  variable Crude odds ratio  (95% CI) Adjusted odds ratio d  (95% CI) Crude odds ratio (95% CI) Adjusted odds ratio d  (95% CI)  HIV STATUS & TESTING     Ever tested for HIV 1.01 (0.42 – 2.42) 0.88 (0.33 – 2.34) 1.21 (0.40 – 3.69) 1.19 (0.31 – 4.54) HIV infection 1.80 (0.65 – 5.00) 1.27 (0.38 – 4.26) 1.02 (0.23 – 4.59) 1.21 (0.22 – 6.55)  SEXUAL BEHAVIOURS & EXPERIENCES   Same-sex anal intercourse a 1.85 (0.54 – 6.26) 1.96 (0.54 – 7.06) 0.77 (0.25 – 2.36) 0.96 (0.28 – 3.26) Unprotected anal intercourse a 1.52 (0.71 – 3.24) 1.42 (0.64 – 3.18) 1.58 (0.62 – 3.99) 2.15 (0.77 – 6.04) More male sex partners b 1.52 (0.76 – 2.96) 1.42 (0.71 – 2.86) 1.21 (0.54 – 2.72) 1.41 (0.59 – 3.37) Ever experienced forced sex 2.61 (1.14 – 5.97)* 2.43 (0.99 – 6.00)† 2.57 (0.95 – 6.94) 2.27 (0.75 – 6.82) MSM sexual debut before age 23 1.03 (0.47 – 2.23) 1.11 (0.45 – 2.75) 1.09 (0.41 – 2.91) 1.00 (0.33 – 3.11) Receptive anal sex only a ,c 0.81 (0.32 – 2.07) 0.79 (0.28 – 2.20) 0.67 (0.19 – 2.38) 0.82 (0.20 – 3.33) Circumcised 1.90 (0.81 – 4.44) 1.92 (0.77 – 4.75) 2.26 (0.84 – 6.07) 1.73 (0.57 – 5.22) Bought or sold sex a Not calculable Not calculable 1.21 (0.15 – 9.65) 0.80 (0.08 – 8.49) Engaged in male group sex a 1.95 (0.76 – 5.02) 1.76 (0.64 – 4.84) 1.94 (0.62 – 6.04) 2.15 (0.61 – 7.61)  SUBSTANCE USE     Ever used illicit drugs Not calculable Not calculable 1.07 (0.14 – 8.41) 1.11 (0.11 – 10.74) More frequent alcohol intake 1.17 (0.56 – 2.47) 1.30 (0.60 – 2.82) 1.23 (0.49 – 3.07) 1.33 (0.50 – 3.55) a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months; d adjusted models control for age, migrant status, city, rural/urban status, and education;  * p<0.05; † p<0.1  158   4.5 Discussion This study is arguably the first quantitative study in China to: (a) specifically examine HIV vulnerabilities and testing patterns among ethnic minority MSM, and (b) assess the HIV-related effects of ethnic acculturation and ethnic discrimination. In accordance with the study hypotheses, several key findings emerged.  4.5.1 HIV vulnerabilities vary substantially by ethnic affiliation  Results partly supported hypothesis 1, which postulated that HIV vulnerabilities and HIV testing are associated with some but not all ethnic minority groups. HIV testing practices did not significantly vary by ethnicity, but numerous measures of HIV vulnerability were strongly correlated with specific ethnic minority groups. Compared to ethnic Han, ethnic minorities of South China were not only five times more likely to be HIV-positive, but were also over twice as likely to report experiences of forced sex. The association between HIV serostatus and affiliation with a Southern ethnic minority group may be partly attributed to community background prevalence. Ethnic minority groups in Southwest China have long been recognised as having higher HIV prevalence than their ethnic Han counterparts (S. Qin et al., 2014; Ruan et al., 2007; Xiao, Kristensen, Sun, Lu, & Vermund, 2007). Thus, the elevated HIV prevalence among Southern ethnic minorities could be a reflection of regional disparities in HIV background prevalence among MSM in China. Given that MSM HIV prevalence in certain major cities of Southwest China are twice that of the national average (13.28% vs 5.98%) (L. Zhang, Chow, Jing, et al., 2013), it may be that MSM migrating from Southern China to Beijing/Tianjin in general have greater overall HIV prevalence compared with migrants from the North. Unfortunately, current research practices in China typically dichotomize individuals as having  159   official local household registration status or not. Future studies should also consider assessing regional origins of domestic migrants.   However, hometown background prevalence alone unlikely fully accounts for the disproportionate burden of HIV among Southern ethnic minority MSM living in Beijing and Tianjin. Given that sexual violence has been associated with HIV infection (Buller, Devries, Howard, & Bacchus, 2014), greater HIV vulnerability among Southern ethnic minority MSM in Beijing/Tianjin may also be a function of sexual power imbalances. Qualitative studies should explore how geographic background and regionalism may be impacting negotiation of safer sex among MSM in China. Lower prevalence of riskier sexual behaviours among ethnic minority Hui is in-part likely attributable to ethnoreligious homonegativity and religious codes of sexual morality. Qualitative studies suggest that ethnic groups with Islamic traditions in China stigmatize same-sex sexual behaviours more severely than Buddhist, Christian, or non-religious communities, whereby such acts are viewed as morally reprehensible by divine intervention (see Chapter four). Religious-based stigma enforced by beliefs in supernatural punishment may be deterring ethnic Hui MSM from engaging in riskier sexual practices with men.  To be sure, ethnic and religious affiliations are distinct and it should not be assumed that all ethnic Hui are Muslim or that all ethnic Hui have similar ethnocultural practices. Reports show that ethnic Hui communities in Southeast China do not identify as Muslim or even follow basic Muslim practices such as abstaining from consumption of pork (Gladney, 1998, pp. 166–7). Religious customs of male circumcision among ethnic Hui also seem to be weak in urban North China, as only 36% of ethnic Hui reported being circumcised. Future studies in China  160   should explicitly assess the influence of religious affiliation on MSM HIV vulnerabilities and testing.   4.5.2 Ethnic sexual mixing patterns do not occur at random Analysis of sex partner ethnicities provided strong support for hypothesis 2, which postulated that ethnic affiliations of sex partners is associated with ethnic affiliation of the participant. Ethnic Hui men were disproportionately more inclined to report sex with other ethnic Hui, indicating that sexual networks of ethnic Hui MSM may be more ethnically homogenous when compared with Han counterparts. This finding appears to be explained at least partly by Hui participants’ preference for other Hui men, as many Hui and Other-Northern ethnic minorities expressed the importance of a steady partner being of concordant ethnicity. That said, exclusion by other ethnicities cannot be ruled out as another reason why Hui are more likely to partner with members of their own ethnic group.  It was unclear why Manchus were less likely to have ethnic Han partners and more likely to have unknown partners, but Manchus may be less likely to assume that a partner of undetermined ethnicity is Han. In any circumstance, ethnically segregated sexual networks have profound implications for HIV transmission networks and HIV intervention effectiveness. Additional studies are needed to elucidate how ethnicity may be influencing MSM sexual networks in China.  4.5.3 Ethnic identity affirmation associated with less sex with men  Study findings lent partial support for hypothesis 3, which stated that ethnic identity affirmation is associated with reduced HIV vulnerabilities and more frequent HIV testing. Stronger ethnic identity affirmation was correlated with less anal sex with men and fewer male sex partners, though no significant association was detected with UAI. It was not apparent what accounted for the association between ethnic identity affirmation and less sex with men, but  161   homonegativity may be one possible explanation. Conceivably, if homonegativity is stronger among traditional ethnic cultures, then men with stronger ethnic identity affirmation may be less inclined to have sex with other men. With respect to hypothesis 4, it was unclear why ethnic affiliation did not significantly moderate the relationship between ethnic identity affirmation and same-sex sexual behaviours, given supposedly greater homonegativity within Hui communities. Additional research on the behavioural health implications of ethnic identity affirmation in China is needed. The proportion of participants who had never tested for HIV was relatively high across all ethnic affiliations (22%), but in contrast to findings among minority Asian MSM in the US (Vu et al., 2011), no association was detected between ethnic identity affirmation and HIV testing. This discrepant finding may be explained by different sampling methods. Unlike the US study which sampled men at sexual minority venues and did not require participants to undergo an HIV test, all participants of the current study completed the questionnaire and were tested at HIV voluntary testing and counseling clinics designated for MSM.  4.5.4 Language proficiency and ethnic discrimination each/respectively have no significant associations with HIV vulnerability or testing  Analyses yielded little evidence to support hypotheses 4 and 5, which postulated that poorer Mandarin language proficiency and ethnic discrimination would each be correlated with greater HIV vulnerabilities and less HIV testing. History of ethnic discrimination was not significantly correlated with HIV testing history or any measurement of HIV vulnerabilities. One possible explanation may be that the definition of ethnic discrimination was too vague and simplistic. Measuring ethnic discrimination dichotomously across the lifespan may have precluded detection of any association between ethnic discrimination and outcome measures.  162   Isolated incidents of ethnic discrimination from years past may have little consequence for current health behaviours. Futures studies in China should consider measuring the frequency and/or severity of ethnic discrimination across a more restricted time frame, as well as provide participants specific examples of what would constitute ethnic discrimination. The marginally significant association between poorer Mandarin language proficiency and forced sex could not be explained in the current study, but weaker oral communication may have disadvantaged participants’ ability to negotiate safer sex. Outreach and content of safer sex skills training in North China should take into account variation in Mandarin language proficiencies among their clientele.  4.5.5 Limitations Several study limitations merit noting. First, the non-random sample may not have been representative of MSM in North China. Effects of ethnic affiliation, acculturation, and ethnic discrimination on HIV vulnerabilities may manifest in qualitatively different ways within rural settings or communities with different concentrations of ethnic minorities or migrants. However, by actively recruiting minority MSM via multiple channels and measuring indicators of acculturation, I am confident that this study sample constitutes the most robust survey of acculturation and ethnic minority MSM in China to date. Second, by recruiting participants exclusively via Chinese language mediums, the sample may have been biased towards those with stronger Chinese language skills. It is entirely plausible that ethnic minorities with weaker Chinese language skills had lower probabilities of being recruited into the study. Future research should consider multiple languages and modes of enrollment (e.g., internet-based questionnaires) to reduce sampling bias. Third, the study may have been underpowered to detect significant correlations between dependent variables and (a) ethnic discrimination and (b) poorer Mandarin  163   language proficiency. Despite the small number of individuals who reported experiencing ethnic discrimination or primary languages other than Mandarin, descriptive results can serve as a baseline of comparison for future studies.  4.5.6 Re-conceptualizing indicators of ethnocultural difference With few exceptions, HIV-related public health and social science literature in China has tended to frame MSM and ethnic minorities as mutually exclusive social groups, whereby ethnic minorities are largely absent from MSM-oriented discourse, and vice versa. Striking interethnic differences in heritage cultural practices, values, sense of identity, and experiences of discrimination, call into question the utility of current epidemiological practices which aggregate all 55 ethnic minority groups together as a single non-Han category. Arguably more informative for the purposes of HIV prevention would be disaggregated reporting of ethnic affiliations combined with measures that indicate degree of acculturation to Han society (e.g., proficiency in Mandarin Chinese) and preservation of ethnic cultural heritage (e.g., ethnic identity affirmation). Even in quantitative studies and surveillance data with relatively few ethnic minorities, crudely dichotomizing ethnic minority groups along theory-driven cultural boundaries (e.g., Northern or Southern geographic origins) may reveal significant measures of association that would not have been detectable otherwise. By critically examining current practices of collecting and reporting ethnic affiliation, and developing innovative models of delineating ethnocultural differences pertinent to HIV prevention, researchers and health policy makers in China will be better informed about the specific needs of ethnocultural sub-groups among MSM.   164   Chapter 5: Religious affiliation, religiosity, and religious discrimination among men who have sex with men in North China  5.1 Introduction  During the Maoist era (1949-76) of China, incidence of sexually transmitted diseases (STD) was virtually reduced to zero, due primarily to aggressive STD elimination campaigns, limited geographic mobility, and a socialist planned economy (M. S. Cohen et al., 1996; Dikötter, 1997; S. Pan, 2006). However, China’s pivotal introduction of “reform and opening up” socioeconomic liberalization policies in the late 1970s enabled profound social changes such as massive internal migration, commercial sex work, and shifting sexual norms that became conducive to the rapid proliferation of blood-borne and sexually transmitted infections (Sutherland & Hsu, 2012a; L. Zhang, Chow, Jahn, et al., 2013). In 1985, HIV/AIDS was confirmed for the first time in mainland China (World Health Organization, 1986), and has since become endemic to all 31 provinces, infecting millions and claiming no fewer than 136,000 lives (National Health and Family Planning Commission of The People’s Republic of China, 2014). The burden of HIV in China, however, has not been equally shared throughout the population. 5.1.1 HIV & men who have sex with men in mainland China  MSM represent 29% of all new HIV infections in China, and are becoming infected faster than any other behavioural sub-group (Ministry of Health of the People’s Republic of China, 2012). Between 2001 and 2010, the rate of MSM transmission in China increased from 0.39 to 0.98 per 100 person-years, while national HIV prevalence among MSM has risen from 1.77% to 5.98% (L. Zhang, Chow, Jing, et al., 2013). Moreover, due in large part to homonegativity and HIV-associated stigma that deter MSM from learning about or accessing HIV testing and prevention services (Xuefeng Li et al., 2012; Santos et al., 2013; Y. Song et al.,  165   2011; Wei et al., 2014; Z. Wu et al., 2006), 61-87% of HIV-infected MSM in China are estimated to be undiagnosed (E. P. F. Chow et al., 2014).  The rapid spread of HIV among MSM in China has been closely linked to China’s “reform and opening up” policies and the unprecedented social environments within which new sexually-oriented identities and communities could be formed. As geographic travel restrictions were rolled back, men and women with erotic same-sex attraction gained opportunities to relocate and join larger sexual minority communities. By the 2000s, gay bars could be found in most major cities (W.-S. Chou, 2000; Rofel, 2007), homosexual behaviour was decriminalized, and homosexuality was no longer classified as a mental illness (Y. Li, 2006). The internet and new modes of telecommunications also played a key role in the MSM population’s exponential growth by offering new opportunities for men with same-sex attraction to discreetly and efficiently meet MSM throughout the country (W.-S. Chou, 2000; Ho, 2010; Miège, 2009). In turn, the scope and connectivity of MSM sexual networks increased dramatically, thereby accelerating HIV transmission within the MSM population (Helleringer & Kohler, 2007).  5.1.2 Religious affiliation and religiosity in the People’s Republic of China The growth of China’s sexual minority communities have risen in parallel with the increasing popularity of religion. Historically, most denizens of dynastic China (220 BC - 1912 AD) engaged in fluid Daoist, Confucian, and Buddhist religious practices with little if any notions about mutually exclusive religions and discrete religious identities. Such practices are sometimes referred to as “popular Chinese religion” (minjian xinyang) (Goossaert, 2011). However, beginning with its introduction from the West in the early 20th century and formally institutionalized by the Chinese government’s Religious Affairs Bureau in the 1950s (Goossaert, 2011), concepts of religion (zongjiao) and religious faith (zongjiao xinyang) have gained  166   increasing traction within the consciousness of Chinese society over the past century (Goossaert, 2011). Religious practices were roundly suppressed during the cultural revolution (F. Yang, 2012), but since the early 1980s, Chinese citizens have essentially been free to affiliate with and openly practice “normal religions” as defined by the state, so long as activities are conducted within officially state-sanctioned religious apparatuses and are not seen as provoking social unrest (Laliberté, 2011; F. Yang, 2012). These liberties, in conjunction with broader social and economic reforms, helped facilitate new opportunities for public religious engagement among the Chinese citizenry (F. Yang, 2011). By 2010, China had approximately 244 million Buddhists, 68 million Christians, and 24 million Muslims (Pew Research Center, 2012).  In addition to serving as a form of social identity, personal religious affiliation is often a marker for specific religious social organisational structures and beliefs. Generally, religions in China can be crudely classified into two categories: churched religions and non-churched religions (Stark et al., 2005). Churched religions are characterized by social congregations of members who subscribe to explicit institutionally sanctioned religious creeds. In China, Islam, Catholicism, and Protestantism constitute the largest officially recognised churched religions, and each maintains its own religious codes of moral comportment. Given that members are typically expected to believe specific religious creeds and partake in social religious activities, churched religions naturally have considerable capacity to instill a strong sense of religious identity and exert social pressures among their congregants. In effect, the religious community to which an individual claims affiliation with serves as an important social reference group (Bock et al., 1983). In contrast, non-churched religions impose far fewer expectations on its members. Buddhism and Daoism are the main unchurched religions officially recognised by the Chinese state, and neither requires its members to acknowledge specific religious creeds or partake in  167   religious social activities. Buddhist and Daoist venues (e.g., temples and shrines) primarily cater to individualized religious practices and are thus less well suited for supporting religious communities and religious social norms (Eric Y Liu, 2011, p. 145).  In addition to differences in social organisations, religious affiliations in China fundamentally differ in their views of the supernatural. Christianity and Islam both espouse beliefs in an omnipotent higher being consciously concerned with the moral rectitude of humanity (i.e., Allah or the Lord), while Buddhism and Daoism are religions primarily founded upon beliefs in a “natural order” to the universe driven by karma and “the way,” respectively. Essentially, the core principle of karma holds that “Deeds by all forms of sentient beings create life circles and consistently influence past, present, and future experiences” (Eric Y Liu, 2011, p. 144). Daoism can be understood as “The paramount force behind the natural order that keeps the universe ordered and balanced” (Eric Y Liu, 2011, p. 144). As neither Daoism nor Buddhism demand exclusive religious recognition of its members, the distinction between self-proclaimed Daoists and Buddhists in China are often blurred, with worldviews and practices being routinely and unconsciously blended (L. Fan & Whitehead, 2011, p. 21). While individual religious affiliation is defined by membership in a religious social group, religiosity is an inherently multi-dimensional construct that can be conceptualized as “A level of adherence to, participation in, influence of, or identification with a set of [religious] beliefs” (Shaw & El-Bassel, 2014). Within the sexual HIV prevention literature, religiosity has been commonly operationalized as religious faithfulness (i.e., importance of religion in one’s life), frequency of participation in religious activities, influence of religion during childhood/adolescence, influence of religion on behaviour, and self-appraised degree of being religious (Shaw & El-Bassel, 2014). As with any religiously pluralistic society, manifestations of  168   religiosity in China vary by religious affiliation. For example, frequency of participation in religious activities might include sutra chanting for Buddhists, Bible studies for Christians, or Friday prayers for Muslims.  In recent decades, religiosity in China has increased markedly. As a proxy national metric of participation in religious activities, the number of Buddhist/Daoist temples increased from ~14,557 in 1995 to ~23,000 in 2009; within that same time frame, the number of Christian churches and meeting points rose from ~41,377 to ~64,000 (F. Yang, 2012). Due to both shifting concepts of religion and greater religious faithfulness, Chinese citizens increasingly view religion as an important element of their lives (Figure 1.4).  5.1.3 Religious affiliation, religiosity, and religious discrimination: implications for HIV vulnerability Religion wields strong potential to influence individual and group-level sexual HIV vulnerabilities via myriad pathways (Shaw & El-Bassel, 2014). Religious affiliations delineate boundaries of distinct organisational social structures and cultural norms between given religious groups (Stark et al., 2005), and have been shown to correlate with population HIV prevalence, individual HIV infection, and riskier sexual behaviours (Shaw & El-Bassel, 2014). Lower HIV prevalence and HIV risk among Muslim communities and Muslim individuals have respectively emerged as the most consistent finding within international research concerning HIV and religious affiliation (P. B. Gray, 2004; Nattrass, 2009; Obermeyer, 2006; Shaw & El-Bassel, 2014), while effects associated with Christian, Hindu, and other sub-denominational affiliations have been more varied by setting and study (Shaw & El-Bassel, 2014).  Religiosity has generally been protective against HIV infection and HIV risk factors such as unprotected sex (Shaw & El-Bassel, 2014), but associations can vary by how religiosity is  169   operationalized. Among other methods, religiosity has been measured as personal importance of religion, participation in religious activities, or frankly as how religious a respondent perceives themselves (Shaw & El-Bassel, 2014). The implications of such religious growth on the HIV epidemic in China are considerable, particularly in regions with large religious populations. However, research concerning religion and HIV vulnerability in China is essentially non-existent. Discrimination has been defined as a “behavioural manifestation of a negative attitude, judgement, or unfair treatment toward members of a group” (Pascoe & Richman, 2009), and can occur at the interpersonal and/or structural levels (Link & Phelan, 2001). Interpersonal discrimination includes personally-mediated acts of devaluation and exclusion, while structural discrimination involves social norms and institutionalized practices which undermine the welfare of stigmatized groups, with or without interpersonal discrimination (Angermeyer et al., 2014). Acts of interpersonal and structural discrimination can be committed intentionally or unintentionally, and may or may not be perceived by the putatively discriminated individual (Jones, 2001).  Current research indicates that discrimination based on race/ethnicity, religion, and sexual orientation have negative impacts on a broad spectrum of health indicators including substance use, psychological distress, and HIV sexual vulnerability (Krieger, 2014; Rippy & Newman, 2006). Compelling empirical evidence suggests four possible mechanisms linking ethnocultural or sexual orientation discrimination and HIV vulnerability. First, instances of perceived discrimination can trigger psychological distress which prompts negative coping strategies such as unprotected sex with casual partners (Díaz et al., 2004; Earnshaw et al., 2013). Second, discrimination can constrain access to important HIV prevention information and  170   services. For example, perceptions of religious discrimination can dissuade marginalized religious minorities from establishing contact with health service institutions associated with the dominant social group (Earnshaw et al., 2013). Third, ethnocultural discrimination can create interpersonal power imbalances that disadvantage discriminated individuals leading up to and during sexual situations, whereby safer sex becomes harder to practice (C. S. Han et al., 2014; C. Han, 2008). Fourth, discrimination against certain ethnocultural groups when selecting sex partners can create socially segregated sexual networks that place ethnocultural minorities at greater risk of exposure to HIV (Maulsby et al., 2014; B. Mustanski et al., 2014).  Research regarding the health implications of religious discrimination in China has been scanty. One exception is a study which showed that adolescents in urban Chinese cities were more likely to experience depression, suicidal ideation, and interpersonal violence if they perceived being bullied because of their religion (S. W. Pan & Spittal, 2013). Currently, there exist few if any studies concerning religious discrimination and HIV vulnerability in China.    5.1.4 HIV, religion, and men who have sex with men For many men, religion is a fundamental framework around which their lives and world perspectives are organised, and which often intertwines with ethnic affiliation. Due to heteronormative and austere religious codes of sexual morality, MSM of certain religious faiths (e.g., Islam or Christianity) may be more likely to encounter and internalize stigmatization of their sexual orientation and sexual behaviours (Keogh, n.d.; Kugle & Chiddy, 2009; Siker, 2007; Smallwood, 2013; Wagner et al., 2012; Wilson et al., 2011). Homonegative religious sexual norms in particular might inhibit same-sex sexual behaviours, but may also limit information and access to health services such as HIV testing and counseling by socially isolating MSM of faith from other MSM (Andrinopoulos et al., 2014; Santos et al., 2013). Currently though,  171   epidemiological data of HIV vulnerabilities among MSM by religious affiliation remains scarce and inconsistent. Results of the few such studies indicate that Muslim MSM in West Africa appear less likely to test for HIV and more likely to have sex with females when compared with their Christian counterparts (Lorente et al., 2012; Sheehy et al., 2013). In Israel, Muslim MSM appear to have lower HIV prevalence than Jewish MSM (Mor et al., 2013). However, no significant differences in HIV vulnerabilities were observed between religious groups among Malaysian MSM (Muslim, Hindu, Buddhist, Christian, no religion, and other religion) (Kanter et al., 2011) or between Christian denominations among US MSM (Garofalo et al., 2014).   Compared to religious affiliation, effects of religiosity on HIV vulnerabilities have been less studied, but results have been more consistent. Profound differences in doctrine, practices, and norms present challenges for generalizing about religiosity across religious affiliations, but religious faithfulness and participation in religious activities have been linked to less risky sexual behaviours among MSM and transgender women in Chicago (Dowshen et al., 2011; Garofalo et al., 2014). Unfortunately, mechanisms of association between religiosity and reduced HIV-related sexual vulnerabilities remain poorly understood. In theory, greater religious faithfulness and participation in religious activities may reduce HIV-related sexual vulnerabilities by promoting social support and personal coping skills (Garofalo et al., 2014), but religious social stigmatization of homosexual and extramarital sex may also be compelling religiously faithful and religiously active MSM to suppress same-sex behaviours in accordance with their beliefs about sexual morality (Chapter three).  Research of religion and MSM remains extremely limited in mainland China. Cultural studies indicate that Christian MSM in China may face heightened homonegativity and psychological distress trying to reconcile ostensibly incompatible sexual and religious social  172   identities (Chan & Huang, 2014), but there is currently no quantitative data about the HIV implications of religious affiliation or religiosity among MSM in China. Attempts to estimate the population size of religiously affiliated MSM in China have yet to be published, but extrapolating MSM (F. Y. Wong et al., 2009) and religious affiliation (Pew Research Center, 2012) population estimates translates to approximately of 0.9-1.8 million Buddhist MSM, 300-600 thousand Christian MSM, and 32-160 thousand Muslim MSM within China today.  5.2 Objectives and rationale In response to the stark lack of information concerning the intersectionality of religion, sexuality, and HIV in China, the present study sought to assess how religious affiliation, religiosity, and religious discrimination may be impacting HIV vulnerabilities, sexual networks, and uptake of HIV testing services among MSM in North China. There were five study hypotheses: (1) HIV vulnerabilities and testing patterns among Buddhist, Christian, and Muslim MSM are significantly different from areligious counterparts (2) Religious affiliation of sex partners is associated with religious affiliation of the participant (3)  Stronger religiosity is inversely correlated with HIV vulnerabilities and testing  (4) Associations between religiosity and HIV vulnerabilities/testing are moderated by religious affiliation (5) Perceived religious discrimination is associated with greater HIV vulnerabilities and reduced HIV testing (6) The magnitude of association between perceived religious discrimination and HIV vulnerabilities/testing is stronger among Buddhists, Christians, and Muslims than areligious men  173   5.3 Methods 5.3.1 Study design and recruitment Data for the present study originate from a larger cross-sectional survey of religious and ethnic minority MSM which was conducted from July 2013 to April 2014 in Beijing and Tianjin, China at five HIV VCT sites devoted to serving MSM. Survey participants were recruited by four methods. First, men who had come to receive VCT services on their own volition were invited on-site to participate in the study. Second, successfully enrolled participants were encouraged to recruit their eligible MSM friends. Third, a non-governmental organisation dedicated to MSM HIV prevention distributed study recruitment information to potential participants via a widely used smart phone social networking application. Fourth, potential study participants were contacted via several MSM social networking websites and chatrooms. Due to potential privacy concerns, participants were free to use pseudonyms, and no true identifiers were requested. Men were eligible to participate in the survey if they self-reported ever having sex with another man, were at least 18 years old, had never tested positive for HIV, and were willing and able to provide written informed consent. Due to resource constraints precluding unconditional recruitment of all VCT clinic patrons, a 1:1 design ratio was maintained throughout the survey enrollment period between the designated index group (ethnic or religious minority MSM) and referent group (areligious ethnic majority Han MSM); in effect, only a proportion of otherwise eligible non-religious ethnic majority Han MSM patrons of the VCT sites were enrolled into the study. Once study staff ascertained eligibility, obtained written-informed consent, and provided instructions on how to complete the survey, participants proceeded to complete the self-administered survey in private at the VCT site; for participants who were functionally illiterate (<5%), surveys were administered by study staff. Thereafter,  174   participants received confidential HIV pre-test counseling, a blood-based HIV rapid test, and appropriate post-test counseling from trained staff. Upon completion of the survey and HIV testing, participants received an honorarium of ¥50 CNY. Study protocols were reviewed and approved by the National Center for AIDS/STD Control and Prevention, Chinese Centers for Disease Control and Prevention (CDC) (Project number: X120717232, Co-PIs: Stephen W. Pan, Yuhua Ruan, and Yiming Shao), and the Providence Health Care Research Institute, University  of British Columbia (Reference number: PHC REB H12-00975, PI: Patricia M. Spittal).  5.3.2 HIV diagnostics  HIV-1 serostatus was initially determined by blood-based rapid testing (Determine HIV-1/2, Abbott Japan Co., Japan) and confirmed by HIV-1/2 Western Blot (HIV Blot 2.2 WBTM, Genelabs Diagnostics, Singapore).  5.3.3 Measures 5.3.3.1 Dependent variables Participants reported information about sexual histories (age at same-sex sexual debut and circumcision), number of male sex partners in the past six months, sexual behaviours in the past three months (engaged in male group sex, any anal sex, UAI with male partners, and anal sex position), HIV testing history, and substance use (frequency of alcohol consumption and use of illicit drugs). Continuous variables were dichotomized or categorized into tertiles because of extreme outlier values and to facilitate clearer data interpretation. Sexual partner characteristics were assessed using a modified version of the UNAIDS Sexual Network Questionnaire (UNAIDS, 1998). Using a multiple-choice format, each individual was asked to report the gender and religious affiliation of their three most recent sex partners in the past six months. Dichotomous dummy variables were then created to indicate  175   whether or not the participant’s last three sex partners in the past six months included a male member of a specific religious affiliation.  5.3.3.2 Independent variables Religious affiliation was assessed by a multiple choice question with the following options: none, Buddhism, Catholicism, Daoism, Islam, Protestantism, and other. Selections for religious affiliation were chosen according to the State Administration for Religious Affairs’ rubric of officially institutionalized major religions within China (Goossaert & Palmer, 2011), but “Catholicism” and “Protestantism” were combined as “Christian” in the present study. Religious discrimination was measured as having ever experienced religious discrimination. Religiosity measures were assessed using a version of the Duke University Religion Index (DUREL) (Koenig & Büssing, 2010) previously translated, adapted, and validated for mainland China (Eric Yang Liu & Koenig, 2013). Essentially, the DUREL is a five question index that captures three dimensions of religiosity: organisational religiosity (assessed by a single-item measuring frequency of participation in public religious activities on a scale of 1-6), private religiosity (assessed by a single-item measuring frequency of participation in solitary religious activities on a scale of 1-6), and intrinsic religiosity (assessed by three items measuring personal commitment to religion on a scale of 3-15) (Koenig & Büssing, 2010). The intrinsic religiosity sub-scale demonstrated excellent reliability in the present study, yielding a Cronbach’s alpha value of 0.91.  The following Spearman correlation coefficients were calculated for each pair of religiosity indicators: Organisational & Private religiosity, 0.68; Organisational and intrinsic religiosity, 0.61; Private and Intrinsic religiosity, 0.55.     176   5.3.3.3 Control variables Five measures were assessed as control variables in data analyses: Age, migrant status, city (Beijing or Tianjin), rural vs. urban household registration status, and highest level of educational attainment.  5.3.4 Data analyses  Completed questionnaires were double-entered in Epidata 3.1 (The Epidata Association, Odense) and cleaned using SAS 9.3 (SAS Institute, Cary, NC). Chi-square tests were used to compare distributions of religious affiliation, religiosity, and religious discrimination across HIV status and testing history, sexual behaviours & experiences, substance use, and sociodemographics. Analysis of variance (ANOVA) was used to compare the mean scores of ethnic identity affirmation across the aforementioned dependent and control variables. Pooled and Satterthwaite methods were used to calculate significance tests for two-group samples with equal and unequal variances, respectively. Welch ANOVA tests were used to assess the significance of mean differences between three or more groups with unequal variance. Then, binary or ordinal logistic regression was employed to evaluate measures of associations between independent and dependent variables. Each dependent variable was regressed on each independent variable in unadjusted binary or ordinal logistic regression models. Parameter estimates involving ordinal logistic regression models which did not satisfy assumptions of proportionality were not reported, as such estimates did not have sufficient model fit. Ordinal logistic regression was used for models regressed on “more male sex partners” and “more frequent alcohol intake”. Then, in order to obtain adjusted measures of association between independent and dependent variables, control variables were added as covariates to each regression model. To assess if religious affiliation  177   statistically moderated the strength of associations between religiosity measures and HIV vulnerabilities/testing, two-way interaction terms were created and modeled in logistic regression by crossing each religiosity measure with religious affiliation. In addition, religious affiliation-specific measures of association were calculated from interaction terms in order to compare magnitude of correlations between religious affiliation groups.   To assess how the religious affiliation of sex partners varied between Buddhist, Christian, Muslim, and areligious participants, five separate binary logistic regression models were used to calculate if religious affiliation was significantly associated with odds of having a sex partner from one of five religious affiliation categories: (a) areligious, (b) Buddhist, (c) Christian, (d) Muslim, (e) unknown religious affiliation. Statistical significance was defined as p<0.05.  5.4 Results 5.4.1 Descriptive findings Descriptive results stratified by religious affiliation are presented in Table 5.1. Areligious respondents comprised the preponderance of the sample (63%), followed by Buddhists (16%), Muslims (12%), and Christians (9%). Two respondents who selected “Daoism” and “other” as religious affiliations were excluded from the analyses, yielding a final sample size of n=398. The majority of participants were younger than 30 years old (63%), had completed four-year college (53%), lacked official residency status (62%), originated from urban districts (68%), and had never been married (84%).  HIV prevalence was just below 10%, and about one in five participants had never been tested for HIV (22%). Neither HIV infection nor testing history significantly varied by religious affiliation. Same-sex sexual debut before age 20 was relatively common among Buddhists, compared to areligious participants (48% vs. 29%). Only one in three Muslims were circumcised  178   (34%), but prevalence of circumcision overall was far lower (16%). Approximately 70% of respondents reported fewer than four male sex partners in the past six months, but proportions varied by religious affiliation. Almost half of Buddhists reported four or more sex partners in the past six months (48%), compared to 26% among Christians, 33% among Muslims, and 30% among areligious respondents. Within the previous three months, the vast majority of participants had engaged in anal sex with men (84%), but Muslims and Christians were twice as likely as areligious participants to have refrained from any anal sex with men (28% and 27% vs. 14%, respectively). Recent UAI was reported by over half of the participants (56%), but was least commonly reported by Muslims (41%). Among those who had anal sex with men, “versatile” was the most commonly reported position (45%), followed by “insertive” (31%), and “receptive” (24%). History of any forced sex was 14%, and 4% of participants reported buying or selling sex in the previous three months.  The mean score of organisational, private, and intrinsic religiosity was 2.01, 1.51, and 2.79, respectively. Christians scored highest in all three religiosity measures, followed by Buddhists, Muslims, and areligious participants. History of perceived religious discrimination also varied significantly by religious affiliation; Christians and Muslims were over three times more likely to experience religious discrimination than Buddhists (19% vs. 5%).   Concordant religious sexual mixing patterns were observed for all religious affiliation groups. Areligious participants were most likely to report sex with an areligious man (70%), Buddhist participants were most likely to report sex with a Buddhist man (26%), Christian participants were most likely to report sex with a Christian man (22%), and Muslim men were most likely to report sex with a Muslim man (17%). However, the proportion of men reporting sex with a man of unknown ethnicity was comparable across all religious affiliations (38%).   179   Table 5.1: Survey participant characteristics, by religious affiliation  Overall (n=398)    Areligious (n=249) Buddhist (n=65) Christian (n=37) Muslim (n=47) p-value  n (%) n (%) n (%) n (%) n (%)   HIV STATUS & TESTING       Ever tested for HIV 308 (78) 190 (76) 53 (82) 31 (84) 36 (77) 0.64 HIV positive  39 (10) 25 (10) 8 (12) 3 (8) 3 (7) 0.77  RELIGIOSITY & DISCRIMINATION      Organisational religiosity    Mean (SD) 2.01 (1.30) 1.34 (0.71)  2.86 (0.93) 3.81 (1.50) 3.02 (1.34) <0.0001 Private religiosity    Mean (SD) 1.51 (1.09) 1.09 (0.53) 2.05 (1.26) 2.76 (1.53) 2.02 (1.39) <0.0001 Intrinsic religiosity    Mean (SD) 2.79 (1.22) 2.22 (0.98) 3.65 (0.87) 4.13 (0.77) 3.55 (1.19) <0.0001 Ever felt religiously discriminated 22(6) 3 (1) 3 (5) 7 (19) 9 (19) <0.0001  SOCIODEMOGRAPHICS       City          Beijing 314 (79) 202 (81) 55 (85) 33 (92) 24 (52) <0.0001    Tianjin 72 (21) 47 (19) 10 (15) 3 (8) 22 (47) Age          <25 years 119 (30) 76 (31) 19 (29) 10 (27) 15 (32) 0.91    25-29 years 129 (33) 78 (31) 20 (31) 14 (38) 18 (38)    >30 years 148 (37) 95 (38) 26 (40) 13 (35) 14 (30) Educational attainment       Less than secondary school 24 (6) 16 (6) 5 (8) 2 (5) 1 (2) 0.46 Secondary school or vocational college 161 (41) 93 (37) 27 (42) 20 (54) 21 (45) Four-year college 211 (53) 140 (56) 33 (51) 15 (41) 25 (53) Migrant status (vs. official permanent resident)  246 (62) 156 (63) 42 (65) 28 (76) 22 (47) 0.05 Urban (vs. Rural) household registration status 269 (68) 163 (65) 48 (74) 20 (54) 39 (83) 0.02 Marital status       Single 334 (84) 210 (84) 55 (85) 31 (84) 40 (85) 0.95 Married, living with wife 25 (6) 16 (6) 5 (8) 1 (3) 3 (6) Married, not living with wife 26 (7) 17 (7) 3 (5) 4 (11) 2 (4) Divorced, in a “sham” marriage, or other 11 (3) 6 (2) 2 (3) 1 (3) 2 (4)  SEXUAL BEHAVIOURS & EXPERIENCES          Age at same-sex sexual debut           <20 years  136 (35) 72 (29) 31 (48) 15 (42) 18 (40) 0.03    20-22 years 131 (33) 90 (36) 15 (23) 8 (22) 18 (40)    >22 years old 126 (32) 85 (34) 19 (29) 13 (36) 9 (20) Circumcised 63 (16) 35 (14) 8 (12) 4 (11) 16 (34) <0.01 Bought or sold sex a 16 (4) 8 (3) 2 (3) 4 (11) 2 (4) 0.17 Engaged in male group sex a 45 (11) 26 (10) 10 (15) 8 (3) 6 (13) 0.63 Anal sex with another male a 334 (84) 213 (86) 62 (95) 27 (73) 34 (72) <0.01 Unprotected anal intercourse a 223 (56) 137 (55) 46 (71) 21 (58) 19 (41) 0.02  180    Table 5.1 (continued): Survey participant characteristics, by religious affiliation  Overall (n=398)    Areligious (n=249) Buddhist (n=65) Christian (n=37) Muslim (n=47) p-value  n (%) n (%) n (%) n (%) n (%)   SEXUAL BEHAVIOURS & EXPERIENCES          Anal sex positioning a, c             Insertive only 104 (31) 63 (30) 19 (31) 9 (35) 12 (36) 0.06       Versatile 151 (45) 98 (46) 21 (34) 16 (61) 15 (45)       Receptive only 80 (24) 51 (24) 22 (35) 1 (4) 6 (18) Number of male sex partners b        Fewer than two  126 (32) 79 (32) 14 (22) 14 (39) 19 (41) 0.08 Two or three 140 (36) 95 (38) 20 (31) 10 (28) 15 (33) Four or more 130 (33) 75 (30) 31 (48) 12 (33) 12 (26) Ever experienced forced sex 56 (14) 29 (12) 11 (17) 7 (19) 9 (19) 0.34  SUBSTANCE USE       Ever used illicit drugs 18 (5) 11 (4) 0 (0) 4 (11) 3 (6) 0.08 Frequency of alcohol intake b       Never 110 (28) 70 (28) 14 (22) 10 (27) 16 (34) 0.37 Less than once per week 248 (62) 159 (64) 40 (62) 22 (59) 27 (57) More than once per week 40 (10) 20 (8) 11 (17) 5 (14) 4 (9) RELIGIOUS AFFILIATION & SEX PARTNERS     Had sex with an areligious man b 249 (63) 174 (70) 33 (51) 18 (49) 24 (51) <0.01 Had sex with a Buddhist man b 33 (8) 6 (2) 17 (26) 5 (14) 5 (11) <0.0001 Had sex with a Christian man b 14 (4) 0 (0) 4 (6) 8 (22) 2 (4) <0.0001 Had sex with a Muslim man b 13 (3) 4 (2) 0 (0) 1 (3) 8 (17) <0.0001 Had sex with a man of unknown religious affiliation b 152 (38) 90 (36) 29 (45) 20 (43) 13 (35) 0.55 a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months  Cross-tabulations of religiosity are presented in Table 5.2. Organisational religiosity was significantly higher among participants who experienced forced sex, while intrinsic religiosity was significantly higher among those who had bought or sold sex and had fewer than two male sex partners. However, private religiosity did not significantly vary by any measure of HIV vulnerability or testing.     181   Table 5.2: Survey participant characteristics, by religiosity   Organisational religiosity Private religiosity Intrinsic religiosity  Mean score (SD) p-value Mean score (SD) p-value Mean score (SD) p-value  HIV STATUS & TESTING       Ever tested for HIV       No 2.00 (1.24) 0.92 1.49 (1.14) 0.84 2.79 (1.24) 0.99 Yes 2.02 (1.32) 1.52 (1.08) 2.79 (1.22) HIV positive        No 2.03 (1.31) 0.56 1.53 (1.11) 0.36 2.80 (1.24) 0.66 Yes 1.90 (1.21) 1.36 (0.93) 2.71 (1.16)  DISCRIMINATION       Ever felt religiously discriminated       No 1.93 <0.0001 1.46 (1.01) 0.03 3.62 (1.01) <0.01 Yes 3.36 2.41 (1.82) 2.74 (1.22)  SOCIODEMOGRAPHICS       City       Beijing 2.03 (1.33) 0.50 1.53 (1.14) 0.44 2.85 (1.20) 0.08 Tianjin 1.93 (1.16) 1.44 (0.88) 2.58 (1.30) Age       <25 years 2.05 (1.38)  1.45 (1.09)  2.86 (1.16) 0.59 25-29 years 2.00 (1.25) 0.93 1.54 (1.00) 0.75 2.82 (1.21) >30 years 1.99 (1.29)  1.54 (1.18)  2.71 (1.29) Educational attainment       Less than secondary school 1.83 (1.17)  1.55 (1.18)  2.24 (1.40) <0.05 Secondary school or vocational college 1.94 (1.26) 0.46 1.54 (1.28) 0.66 2.76 (1.27) Four-year college 2.08 (1.34)  1.45 (0.93)  2.88 (1.15) Residency status        Official permanent resident 2.02 (1.20)  0.93 1.57 (1.12) 0.40 2.79 (1.25) 0.96 Migrant 2.01 (1.36) 1.47 (1.07) 2.79 (1.21) Household registration status       Urban 2.08 (1.29) 0.13 1.56 (1.17) 0.13 2.88 (1.19) 0.03 Rural 1.87 (1.32) 1.40 (0.89) 2.60 (1.27) Marital status       Single 2.03 (1.32) 0.41 1.50 (1.07) 0.90 2.82 (1.19) 0.36 Married, living with wife 2.04 (1.40) 1.52 (0.96) 2.68 (1.35) Married, not living with wife 1.62 (1.06) 1.50 (1.27) 2.41 (1.43) Divorced, in a “sham” marriage, or other 2.25 (1.06) 1.75 (1.48) 2.97 (1.40)  SEXUAL BEHAVIOURS AND EXPERIENCES     Age at same-sex sexual debut        <20 years  2.09 (1.30)  1.56 (1.10)  2.92 (1.19) 0.24 20-22 years 1.89 (1.22) 0.43 1.39 (1.02) 0.34 2.70 (1.12) >22 years old 2.03 (1.35)  1.57 (1.15)  2.73 (1.35)          182   Table 5.2 (continued): Survey participant characteristics, by religiosity   Organisational religiosity Private religiosity Intrinsic religiosity  Mean score (SD) p-value Mean score (SD) p-value Mean score (SD) p-value  SEXUAL BEHAVIOURS & EXPERIENCES     Circumcised       No 1.99 (1.29) 0.45 1.49 (1.07) 0.13 2.75 (1.25) 0.16 Yes 2.13 (1.37) 1.60 (1.23) 2.99 (1.10) Bought or sold sex a       No 2.00 (1.30) 0.46 1.51 (1.11) 0.84 2.76 (1.23) 0.03 Yes 2.25 (1.39) 1.56 (0.73) 3.46 (1.02) Engaged in male group sex a       No 2.01 (1.30) 0.86 1.53 (1.12) 0.31 2.82 (1.22) 0.25 Yes 2.04 (1.31) 1.36 (0.88) 2.59 (1.23) Anal sex with another male a       No 1.96 (1.22) 0.14 1.66 (1.34) 0.32 2.89 (1.32) 0.50 Yes 2.29 (1.64) 1.48 (1.04) 2.77 (1.21) Unprotected anal intercourse a       No 2.06 (1.35)  0.54 1.51 (1.14) 0.95 2.81 (1.24) 0.84 Yes 1.98 (1.27) 1.51 (1.06) 2.78 (1.22) Anal sex positioning a, c       Insertive only 2.07 (1.28)  1.54 (1.04)  2.86 (1.22) 0.71 Versatile 1.96 (1.27) 0.46 1.41 (0.96) 0.58 2.74 (1.22) Receptive only 1.84 (1.05)  1.49 (1.12)  2.76 (1.17) Number of male sex partners b        Fewer than two  2.22 (1.48)  1.65 (1.32)  3.04 (1.23) 0.02 Two or three 1.91 (1.25) 0.09 1.46 (1.03) 0.21 2.65 (1.23) Four or more 1.92 (1.14)  1.43 (0.89)  2.70 (1.19) Ever experienced forced sex       No 1.95 (1.27) 0.03 1.45 (1.02) 0.05 2.75 (1.23) 0.08 Yes 2.37 (1.42) 1.84 (1.44) 3.05 (1.15)  SUBSTANCE USE       Ever used illicit drugs       No 2.01 (1.29) 0.89 1.51 (1.11) 0.73 2.80 (1.22) 0.57 Yes 2.06 (1.59) 1.44 (0.78) 2.63 (1.34) Frequency of alcohol intake b       Never 2.04 (1.29)  1.64 (1.31)  2.72 (1.34) 0.41 Less than once per week 2.18 (1.22) 0.39 1.44 (1.01) 0.26 2.78 (1.19) More than once per week 1.88 (1.36)  1.58 (0.93)  3.03 (1.11)   a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months              Prevalence of perceived religious discrimination did not significantly vary by any assessed measures in the study (Table 5.3), though forced sex was marginally significantly higher among those who reported history of religious discrimination (27% vs. 13%). 183 Table 5.3: Survey participant characteristics, by religious discrimination Ever perceived religious discrimination Yes (n=22) n (%) No (n=376) n (%) p-valueHIV STATUS & TESTING Ever tested for HIV 20 (91) 290 (77) 0.13 HIV positive 0 (0) 39 (10) 0.13 SOCIODEMOGRAPHICS City    Beijing 20 (91) 296 (79) 0.78    Tianjin 2 (9) 80 (31) Age    <25 years 8 (36) 112 (30) 25-29 years 6 (27) 124 (33) 0.78 >30 years 8 (36) 140 (37) Educational attainment Less than secondary school 0 (0) 24 (6) Secondary school or vocational college 10 (45) 151 (40) 0.46 Four-year college 12 (54) 201 (53) Migrant status (vs. official permanent resident) 16 (73) 232 (62) 0.30 Urban (vs. Rural) household registration status 19 (86) 251 (67) 0.06 Marital status Single 19 (86) 317 (84) 0.58 Married, living with wife 0 (0) 25 (7) Married, not living with wife 2 (9) 24 (6) Divorced, in a “sham” marriage, or other 1 (5) 10 (3) SEXUAL BEHAVIOURS & EXPERIENCES Age at same-sex sexual debut    <20 years 8 (37) 130 (34) 20-22 years 6 (27) 125 (34) 0.82 >22 years old 8 (36) 118 (32) Circumcised 6 (27) 57 (15) 0.13 Bought or sold sex a 0 (0) 16 (4) 0.32 Engaged in male group sex a 18 (82) 335 (89) 0.29 Anal sex with another male a 17 (77) 319 (85) 0.34 Unprotected anal intercourse a 12 (55) 212 (56) 0.87 Anal sex positioning a, c       Insertive only 6 (35) 98 (31)       Versatile 10 (59) 141 (44) 0.19       Receptive only 1 (6) 79 (25) Number of male sex partners b Fewer than two 7 (32) 119 (32) Two or three 9 (41) 132 (35) 0.81 Four or more 6 (27) 125 (33) Ever experienced forced sex 6 (27) 50 (13) 0.07  184   Table 5.3 (continued): Survey participant characteristics, by religious discrimination  Ever perceived religious discrimination   Yes (n=22) n (%) No (n=376) n (%) p-value  SUBSTANCE USE    Ever used illicit drugs 1 (5) 17 (5) 1.00 Frequency of alcohol intake b    Never 6 (27) 104 (28)  Less than once per week 12 (55) 236 (63) 0.42 More than once per week 4 (18) 36 (10)   185   5.4.2 Hypothesis 1: Association between religious affiliation and HIV vulnerabilities & testing  The first study hypothesis stated that Buddhist, Christian, and Muslim MSM have significantly different HIV vulnerabilities and testing patterns compared to areligious counterparts. Results presented in Table 5.4 generally support hypothesis 1. No significant associations were detected between religious affiliation and HIV infection or testing history, but specific religious affiliations were predictive of several measures of HIV vulnerability. Compared to respondents with no religious affiliation, odds of both any anal sex and UAI were significantly higher among Buddhists and lower for Muslims. Additionally, Buddhists had significantly higher odds of more sex partners, and Muslims had significantly higher odds of circumcision.  5.4.3 Hypothesis 2: Association between participant religious affiliation and sex partner religious affiliation Table 5.4 presents evidence in support of the second study hypothesis, which postulated that participant religious affiliation is associated with religious affiliation of their sex partners. Compared to areligious participants, Buddhists, Christians, and Muslims were all significantly less likely to report sex with an areligious man, but more likely to report sex with a Buddhist man. Muslim participants had 20-fold higher odds of reporting sex with a Muslim man, but odds of reporting sex with a man of unknown religious affiliation did not significantly vary by religious affiliation.    186   Table 5.4: Religious affiliation and dependent variable correlations Dependent  variable Ethnic affiliation Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI)  HIV STATUS & TESTING    Ever tested for HIV Areligious 1.00 1.00  Buddhist 1.37 (0.69 – 2.74) 1.40 (0.67 – 2.91)  Christian 1.60 (0.64 – 4.03) 1.62 (0.62 – 4.28)  Muslim 1.02 (0.49 – 2.12) 1.23 (0.55 – 2.74) HIV infection Areligious 1.00 1.00  Buddhist 1.23 (0.54 – 2.94) 1.33 (0.54 – 3.27)  Christian 0.82 (0.23 – 2.85) 0.73 (0.20 – 2.68)  Muslim 0.63 (0.18 – 2.16) 0.68 (0.18 – 2.56)  SEXUAL BEHAVIOURS & EXPERIENCES Same-sex anal intercourse a Areligious 1.0 1.00  Buddhist 3.49 (1.04 – 11.73) * 3.69 (1.09 – 12.50) *  Christian 0.44 (0.20 – 0.99) * 0.48 (0.21 – 1.11) †  Muslim 0.43 (0.21 – 0.89) * 0.33 (0.15 – 0.73) * Unprotected anal intercourse a Areligious 1.00 1.00  Buddhist 1.98 (1.10 – 3.57) * 2.06 (1.13 – 3.75) *  Christian 1.15 (0.56 – 2.32) 1.19 (0.58 – 2.45)  Muslim 0.58 (0.30 – 1.09) † 0.47 (0.24 – 0.93)* More male sex partners b Areligious 1.00 1.00  Buddhist 1.93 (1.16 – 3.23)* 1.95 (1.16 – 3.27)*  Christian 0.91 (0.48 – 1.72) 0.94 (0.49 – 1.82)  Muslim 0.73 (0.41 – 1.30) 0.62 (0.34  - 1.14) Ever experienced forced sex Areligious 1.00 1.00  Buddhist 1.54 (0.73 – 3.29) 1.54 (0.72 – 3.31)  Christian 1.77 (0.71 – 4.39) 1.61 (0.64 – 4.08)  Muslim 1.80 (0.79 – 4.09) 2.25 (0.95 – 5.38) † MSM sexual debut before 23 years of age Areligious 1.00 1.00 Buddhist 1.27 (0.70 – 2.30) 1.33 (0.69 – 2.56)  Christian 0.93 (0.45 – 1.92)  0.99 (0.43 – 2.21)  Muslim 2.10 (0.97 – 4.56) †  1.66 (0.70 – 3.89) Receptive anal sex only a ,c Areligious 1.00 1.00  Buddhist 1.74 (0.95 – 3.19) † 1.75 (0.92 – 3.32) †  Christian 0.13 (0.02 – 0.96)* 0.14 (0.02 – 1.09) †  Muslim 0.70 (0.27 – 1.79) 0.47 (0.17 – 1.27) Circumcised Areligious 1.00 1.00  Buddhist 0.86 (0.38 – 1.95) 0.88 (0.39 – 2.03)  Christian 0.76 (0.26 – 2.29) 0.89 (0.29 – 2.71)  Muslim 3.26 (1.61 – 6.59) * 3.04 (1.45 – 6.40) * Bought or sold sex a Areligious 1.00 1.00  Buddhist 0.96 (0.20 – 4.62) 1.00 (0.20 – 4.91)  Christian 3.65 (1.04 – 12.80) * 3.10 (0.85 – 11.39) †  Muslim 1.34 (0.28 – 6.51) 1.48 (0.29 – 7.65) Engaged in male group sex a Areligious 1.00 1.00  Buddhist 1.56  (0.71 – 3.43) 1.68 (0.75 – 3.77)  Christian 0.78 (0.22 – 2.72) 0.80 (0.22 – 2.89)  Muslim 1.29 (0.50 – 3.32) 1.32 (0.48 – 3.61)       187   Table 5.4 (continued): Religious affiliation and dependent variable correlations Dependent  variable Ethnic affiliation Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI)      SUBSTANCE USE    Ever used illicit drugs Areligious 1.00 1.00  Buddhist Not calculable    Not calculable  Christian 2.71 (0.81 – 9.00) 2.85 (0.79 – 10.27)  Muslim 1.51 (0.40 – 5.64) 0.97 (0.24 – 3.92) More frequent alcohol intake Areligious 1.00 1.00  Buddhist 1.03 (0.59 – 1.79) 1.07 (0.61 – 1.89)  Christian 0.97 (0.48 – 1.96) 0.95 (0.46 – 1.95)  Muslim 0.73 (0.39 – 1.35) 0.86 (0.45 – 1.65)  RELIGIOUS AFFILIATION & SEX PARTNERS  Had an areligious male partner Areligious 1.00 1.00  Buddhist 0.45 (0.26 – 0.78) * 0.48 (0.27 – 0.86) *  Christian 0.41 (0.20 – 0.82) * 0.42 (0.21 – 0.88) *  Muslim 0.45 (0.24 – 0.85) * 0.40 (0.20 – 0.79) * Had an Buddhist male partner Areligious 1.00 1.00  Buddhist 14.34 (5.38 – 38.25) * 15.98 (5.56 – 45.88) *  Christian 6.33 (1.83 – 21.93) * 7.48 (2.01 – 27.77) *  Muslim 4.82 (1.41 – 16.51) * 6.10 (1.63 – 22.90)* Had a Christian male partner Areligious 1.00 1.00  Buddhist Not calculable Not calculable  Christian Not calculable Not calculable  Muslim Not calculable Not calculable Had a Muslim male partner Areligious 1.00 1.00  Buddhist Not calculable  Not calculable  Christian 1.70 (0.19 – 15.65) 1.46 (0.15 – 14.03)  Muslim 12.56 (3.60 – 43.72)* 20.26 (5.04 – 81.50)* Had a male partner of unknown religious affiliation Areligious 1.00 1.00 Buddhist 1.42 (0.82 – 2.48) 1.14 (0.62 – 2.09)  Christian 0.96 (0.47 – 1.97) 0.97 (0.43 – 2.17)  Muslim 1.31 (0.70 – 2.47) 1.52 (0.73 – 3.16) a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months;  d adjusted models control for age, migrant status, city, rural/urban status, and education;  * p<0.05; † p<0.1     188   5.4.4 Hypothesis 3: Association between religiosity and HIV vulnerabilities & testing  Results presented in Table 5.5 both contradict and weakly support the third study hypothesis that stronger religiosity is inversely correlated with HIV vulnerabilities and testing. Contrary to expectations, higher scores of organisational and private religiosity were associated with greater odds of forced sex, and higher scores of intrinsic religiosity was correlated with greater odds of commercial sex. However, higher organisational religiosity score was  marginally associated with less anal intercourse. Higher scores in organisational, private, or intrinsic religiosity were also marginally correlated with fewer male sex partners.        189   Table 5.5: Religiosity and dependent variable correlations  ORGANISATIONAL RELIGIOSITY PRIVATE RELIGIOSITY INTRINSIC RELIGIOSITY Dependent  variable Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI) Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI) Crude odds ratio (95% CI) Adjusted odds ratio d (95% CI)  HIV STATUS & TESTING      Ever tested for HIV 1.01 (0.84 – 1.21) 1.01 (0.83 – 1.23) 1.02 (0.82 – 1.28) 0.99 (0.78 – 1.25) 1.00 (0.82 – 1.21) 0.99 (0.81 – 1.23) HIV infection 0.92 (0.71 – 1.21) 0.96 (0.74 – 1.26) 0.84 (0.58 – 1.22) 0.89 (0.60 – 1.31) 0.94 (0.72 – 1.23) 1.03 (0.78 – 1.36)  SEXUAL BEHAVIOURS & EXPERIENCES     Same-sex anal intercourse a 0.84 (0.69 – 1.01) † 0.84 (0.69 – 1.02) † 0.88 (0.70 – 1.09) 0.88 (0.70 – 1.10) 0.93 (0.74 – 1.16) 0.97 (0.77 – 1.22) Unprotected anal intercourse a 0.95 (0.82 – 1.11) 0.97 (0.83 – 1.13) 1.01 (0.84 – 1.21) 1.02 (0.85 – 1.22) 0.98 (0.84 – 1.16) 1.02 (0.86 – 1.20) More male sex partners b 0.88 (0.76 – 1.01) † 0.89 (0.77 – 1.02) † 0.87 (0.74 – 1.03) 0.87 (0.73 – 1.03) † 0.85 (0.73 – 0.98)* 0.87 (0.75 – 1.01) † Ever experienced forced sex 1.25 (1.03 – 1.52)* 1.25 (1.02 – 1.52)* 1.30 (1.05 – 1.60)* 1.30 (1.04 – 1.61)* 1.23 (0.97 – 1.55) † 1.22 (0.96 – 1.55) MSM sexual debut before age 23 0.98 (0.84 – 1.16) 0.96 (0.80 – 1.15) 0.93 (0.77 – 1.12) 0.94 (0.76 – 1.16) 1.06 (0.89 – 1.26) 1.03 (0.85 – 1.24) Receptive anal sex only a ,c 0.89 (0.72 – 1.10) 0.86 (0.68 – 1.08) 1.03 (0.81 – 1.31) 1.00 (0.78 – 1.28) 0.98 (0.80 – 1.21) 0.97 (0.78 – 1.21) Circumcised 1.08 (0.89 – 1.32) 1.07 (0.87 – 1.32) 1.09 (0.87 – 1.37) 1.08 (0.85 – 1.36) 1.17 (0.94 – 1.46) 1.17 (0.93 – 1.47) Bought or sold sex a 1.14 (0.80 – 1.63) 1.17 (0.82 – 1.66) 1.04 (0.68 – 1.61) 1.07 (0.68 – 1.68) 1.63 (1.05 – 2.53)* 1.64 (1.05 – 2.56)* Engaged in male group sex a 1.02 (0.81 – 1.29) 1.06 (0.83 – 1.35) 0.84 (0.59 – 1.19) 0.84 (0.59 – 1.21) 0.86 (0.67 – 1.11) 0.92 (0.71 – 1.19)  SUBSTANCE USE      Ever used illicit drugs 1.03 (0.72 – 1.47) 1.06 (0.73 – 1.53) 0.94 (0.59 – 1.50) 0.97 (0.58 – 1.61) 0.89 (0.61 – 1.32) 0.92 (0.62 – 1.36) More frequent alcohol intake 1.07 (0.92 – 1.25) Not calculable 0.86 (0.72 – 1.03) 0.86 (0.72 – 1.03) 1.01 (0.86 – 1.19) Not calculable a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months;  d adjusted models control for age, migrant status, city, rural/urban status, and education;  * p<0.05; † p<0.1  190   5.4.5 Hypothesis 4: Association between religiosity × Religious affiliation and HIV vulnerabilities & testing  Results from interaction analyses strongly supported hypothesis 4, which postulated that the directionality and magnitude of associations between religiosity and HIV vulnerabilities/testing will vary by religious affiliation. Three measures of association were significantly moderated by religious affiliation: (1) UAI and organisational religiosity, (2) anal sex and private religiosity, and (3) circumcision and intrinsic religiosity.   Correlations between organisational religiosity and UAI were significantly different between (a) Buddhists and Christians, and (b) Muslims. Among Buddhists and Christians, increasing organisational religiosity was predictive of lower probabilities of UAI; however, among Muslims, increasing organisational religiosity actually predicted higher probability of UAI (Figure 5.1). Figure 5.1: Predicted probabilities of unprotected anal intercourse by organizational religiosity and religious affiliation  191    Correlations between private religiosity and anal sex were significantly different between Muslims and Christians. Among Muslims, increasing private religiosity was predictive of lower probabilities of anal sex with men. However, among Christians, probability of anal sex with men dropped rapidly with each increasing score of private religiosity (Figure 5.2).   Figure 5.2: Predicted probabilities of anal sex by private religiosity and religious affiliation               Correlations between intrinsic religiosity and circumcision were significantly different between Buddhists and Areligious participants. Among areligious participants, probability of circumcision rose as intrinsic religiosity scores increased, while the inverse trend was observed among Buddhists (Figure 5.3).   192   Figure 5.3: Predicted probabilities of circumcision by intrinsic religiosity and religious affiliation              Table 5.6 presents religious affiliation-specific measures of association between religiosity and dependent variables, derived from interaction models. Among areligious participants, each one-point increase in intrinsic religiosity was associated with 60% higher odds of circumcision. Among Buddhists, each one-point increase in organisational and intrinsic religiosity was correlated with a 48% and 54% decline in odds of UAI and circumcision, respectively. Among Christians, greater private religiosity was significantly associated with an approximately 50% decline in odds of anal sex. Parameter estimates were not reported if the interaction term was not significant.     193   Table 5.6: Religiosity and dependent variable correlations, by religious affiliation  ORGANISATIONAL RELIGIOSITY PRIVATE RELIGIOSITY INTRINSIC RELIGIOSITY Dependent  variable Religious    affiliation Adjusted odds ratio b (95% CI) Adjusted odds ratio b (95% CI) Adjusted odds ratio b (95% CI) Same-sex anal intercourse a    Areligious   0.98 (0.49 – 1.57)  Buddhist  3.18 (0.36 – 28.31)  Christian  0.49 (0.27 – 0.88)*  Muslim  1.55 (0.82 – 2.92)  Unprotected anal intercourse a    Areligious  0.99 (0.69 – 1.41)   Buddhist 0.52 (0.27 – 1.02) †   Christian 0.67 (0.41 – 1.10)   Muslim 1.42 (0.89 – 2.25)   Circumcised    Areligious    1.62 (1.11 – 2.36)* Buddhist   0.46 (0.20 – 1.06) † Christian   0.90 (0.23 – 3.50) Muslim   0.84 (0.49 – 1.44) a past three months; b adjusted models control for age, migrant status, city, rural/urban status, and education;   * p<0.05; † p<0.1  194   5.4.6 Hypothesis 5: Association between religious discrimination and HIV vulnerabilities & testing The fifth hypothesis postulated that religious discrimination is associated with greater HIV vulnerabilities and less testing, but little supporting evidence was detected (Table 5.7). Experience of religious discrimination was marginally associated with forced sex, but not significantly associated with any measured outcomes of HIV vulnerability or testing.   Table 5.7: Religious discrimination and dependent variable correlations   Ever experienced religious discrimination Dependent  variable Crude odds ratio (95% CI) Adjusted odds ratio d  (95% CI)  HIV STATUS & TESTING   Ever tested for HIV 2.94 (0.68 – 12.84) 3.04 (0.67 – 13.82) HIV infection Not calculable Not calculable  SEXUAL BEHAVIOURS & EXPERIENCES   Same-sex anal intercourse a 0.60 (0.21 – 1.70) 0.69 (0.24 – 1.97) Unprotected anal intercourse a 0.92 (0.39 – 2.18) 1.01 (0.42 – 2.44) More male sex partners b 0.86 (0.39 – 1.90) 0.89 (0.40 – 1.99) Ever experienced forced sex 2.40 (0.90 – 6.43) † 2.47 (0.90 – 6.77) † MSM sexual debut before age 23 0.79 (0.32 – 1.95) 0.60 (0.22 – 1.66) Receptive anal sex only a ,c 0.19 (0.02 – 1.44) 0.19 (0.03 – 1.52) Circumcised 2.11 (0.79 – 5.63) 2.30 (0.84 – 6.32) Bought or sold sex a Not calculable Not calculable Engaged in male group sex a 1.83 (0.59 – 5.66) 2.42 (0.74 – 7.88)  SUBSTANCE USE   Ever used illicit drugs 1.01 (0.13 – 7.97) 1.19 (0.14 – 9.84) More frequent alcohol intake 1.24 (0.53 – 2.88) 1.26 (0.53 – 3.02) a past three months; b past six months; c only among men who engaged in anal intercourse in the past three months; d adjusted models control for age, migrant status, city, rural/urban status, and education;  * p<0.05; † p<0.1  5.5 Discussion Religion can profoundly impact individual sexual behaviours and the complex sociocultural contexts within which they do or do not occur. Results from this study suggest that  195   religious affiliation and religiosity are shaping the HIV vulnerabilities of MSM in North China in several ways.  5.5.1 HIV vulnerabilities vary significantly by religious affiliation Study results partly supported hypothesis 1, which postulated that HIV vulnerabilities and HIV testing among Buddhist, Christian, and Muslim MSM are significantly different from areligious counterparts. First, Christian and Muslim MSM appeared less likely to engage in certain riskier behaviours such as anal sex, possibly due to religious moral stigmatization of homosexual behaviour and fear of provoking divine retribution, as has been observed among Christian MSM (Balaji et al., 2012; Foster, Arnold, Rebchook, & Kegeles, 2011; Jeffries et al., 2014; Kubicek et al., 2009; Severson, Muñoz-Laboy, & Kaufman, 2014) and Muslim MSM (Eidhamar, 2014; Yip, 2004) . Compared to areligious participants, Muslims and Christians were twice as likely not to have had any anal sex within the past three months. Second, although odds of circumcision were significantly higher for Muslims, results indicate that the majority of urban Muslim men in Beijing and Tianjin remain uncircumcised, an observation which sharply contrasts the high prevalence of male circumcision in Muslim communities outside of China (Drain, Halperin, Hughes, Klausner, & Bailey, 2006; Tram & Bertrand, 2014). Hence, any possible protective benefits conferred by male circumcision appear to have limited applicability to the Muslim MSM population in North China. Male circumcision rates may be higher in communities with a higher proportion and longer history of Muslims and Muslim traditions.  Third, Buddhist religious affiliation was linked with riskier sexual behaviours, even after controlling for sociodemographic factors. All major Buddhist sects discourage acts of “sexual misconduct” in a generic sense (Corless, 1998, p. 253; Mitchell, 2008, p. 121; Sweet, 2007, p.  196   76), but conflicting interpretations of what constitutes “sexual misconduct” has led to differing contemporary Buddhist perspectives on the morality of same-sex sexual behaviours in China. One prominent Chinese Buddhist master has averred that homosexuality is neither inherently right or wrong (Hsing Yun, 2001), while another felt that homosexuality “planted the seeds which lead to rebirth in the lower realms of existence” (Corless, 1998, p. 255). Thus, Buddhism within China today currently lacks a dominant religious view on the moral implications of same-sex sexual behaviour. Historically though, sex between Buddhist men in practice was highly prevalent and well-tolerated throughout much of Dynastic China (Hinsch, 1990, p. 97). As a continuation of this legacy, MSM in China today appear to encounter relatively limited religious-based sexual stigmatization from Buddhist populations when compared with Christian or Muslim communities. In light of ethnographic evidence that Buddhism has become an increasingly popular religion among MSM in China (Miller, 2013, pp. 343–5; Tong Ge, 2005, pp. 131–2), additional research is needed concerning the mechanisms of association between Buddhist religious affiliation and HIV vulnerabilities among MSM in China. 5.5.2 Religious sexual mixing patterns do not occur at random Analysis of sex partner religious affiliations provided strong support for hypothesis 2, which postulated that religious affiliations of sex partners is associated with religious affiliation of the participant. More specifically, results suggest that MSM in North China are more likely to report sexual partnerships with men who share their religious affiliation. Areligious participants were most likely to report sex with areligious men, Buddhists were most likely to report sex with Buddhist men, and Muslims were most likely to report sex with Muslim men. Buddhist and Muslim participants were more likely than areligious men to respectively report Buddhist and Muslim sex partners, indicating that sexual networks of Buddhist and Muslim MSM may be  197   more religiously homogenous when compared with areligious counterparts. On the other hand, the fact that Christian and Muslim participants were more likely to partner with Buddhist men than areligious participants, underscores the complexity of inter-religious sexual mixing patterns among MSM. As religious affiliations continue to rise in China, additional studies are needed to elucidate how religion may be impacting MSM sexual networks. 5.5.3 Religiosity may have both protective and deleterious impacts on HIV vulnerability Study findings both partly supported and contradicted the hypothesis that stronger religiosity is inversely correlated with HIV vulnerabilities and testing. Measures of religiosity were marginally associated with less anal sex and fewer male sex partners, but also significantly associated with forced and commercial sex. Precise mechanisms could not be determined from this study, but social control, gender norms, and homonegativity within religious social groups are possible factors. Correlation between MSM stigmatization and religious affiliation among the Hong Kong general population intimates the existence of similar homonegative attitudes by non-MSM religious peers in mainland China (J. T. F. Lau, Choi, Tsui, & Su, 2007). Conceivably, religiosity may increase levels of perceived external and internalized homonegativity, which could in-turn discourage men from seeking out new male partners, forming social bonds with other MSM, and developing stable sexual relationships. As a consequence of reduced social connectedness with other sexual minorities, MSM with high religiosity scores may be more likely to partner with commercial sex workers and unfamiliar men inclined to perpetrate forced sex. Moreover, less social connectedness with other MSM might also inhibit exposure to important HIV prevention messaging such as skills about avoiding forced sex, which has previously been shown to correlate with UAI among MSM in North China (S. W. Pan et al., 2014).   198   5.5.4 Directionality and magnitude of associations between religiosity and HIV vulnerabilities vary by religious affiliation Results partly supported the hypothesis that associations between religiosity and HIV vulnerabilities/testing are moderated by religious affiliation. Organisational and private religiosity was linked with reduced HIV vulnerabilities among Buddhists and Christians, but not Muslims or Areligious participants. Additionally, greater intrinsic religiosity was correlated with more circumcision among Areligious participants, but was actually marginally associated with less circumcision among Buddhists.   It is worth stressing that while augmenting religiosity for specific religious groups may appear to be a promising means of reducing HIV vulnerabilities among religious MSM in future interventions, it should most certainly not be endorsed if lower HIV vulnerability is achieved through increased exposure to homonegativity and stigmatizing social environments. As opposed to mainstream religious organisations that may be less tolerant of homosexual activities, nascent sexual minority Buddhist and Christian organisations in Taiwan and mainland China may be a promising platform to enhance religiosity without potential subjection to sexual stigmatization (Chan & Huang, 2014; C. J. Huang, Valussi, & Palmer, 2011). Publicized sexual minority Muslim organisations, however, are virtually non-existent in mainland China. Additional empirical research about the health benefits of religiosity among MSM in China is warranted. 5.5.5 No evidence that religious discrimination increases HIV vulnerability or reduces testing practices Analyses yielded little evidence to support hypotheses 5, which postulated that religious discrimination would be significantly correlated with greater HIV vulnerabilities and less HIV testing. History of religious discrimination was marginally associated with forced sex, but not  199   significantly correlated with any measurement of HIV vulnerabilities. With only 22 individuals (6%) reporting history of religious discrimination, the study may have lacked power to detect significant associations. Another possible explanation may be that the definition of religious discrimination was too vague and simplistic. Measuring religious discrimination dichotomously across the lifespan may also have precluded detection of any association between religious discrimination and outcome measures. Isolated incidents of religious discrimination from years past may have little consequence for current health behaviours. Futures studies in China should consider measuring the frequency and/or severity of religious discrimination across a more restricted time frame, as well as provide participants specific examples of what would constitute religious discrimination. 5.5.6 Limitations  Several study limitations merit noting. First, the non-random sample may not have been representative of MSM in North China. In particular, effects of religious affiliation and religiosity on HIV vulnerabilities may manifest in qualitatively different ways within rural settings or communities with higher concentrations of religiously affiliated MSM. However, by actively recruiting religious minority MSM via multiple channels and documenting religious affiliations and religiosity, I am confident that this study sample constitutes the most robust survey of Buddhist, Christian, and Muslim MSM in China to date. Second, modestly sized sub-samples and small cell counts may have precluded detection of otherwise significant differences by religious affiliation or religiosity (e.g., HIV infection). Nonetheless, the sample was still sufficiently powered to detect significant differences for an array of important HIV vulnerabilities (e.g., UAI and greater numbers of sex partners).   200   Despite the rapid growth of religion in mainland China, little is currently known about how religious affiliation and religiosity may be influencing HIV vulnerabilities among the Chinese general population, let alone MSM. This cross-sectional study has demonstrated possible mechanisms by which religion may be impacting HIV vulnerability among Buddhist, Christian, and Muslim MSM in China, and is arguably the first quantitative study to examine religion and religiosity with respect to primary HIV infection in China. However, the study perhaps raises more questions than it is capable of answering: Why was Buddhism associated with anal sex? How does religion influence the lives of HIV-positive MSM and their potential for subsequent HIV transmission? These issues represent but a small sliver of the larger lacuna of religion and HIV in China. Future studies are urgently needed.     201   Chapter 6: Religion, ethnicity, homonegativity, and HIV vulnerabilities: a mediation analysis  6.1 Introduction  Ethnic affiliations are socially constructed identities delineated by a political consensus that particular groups of people share some distinct degree of common history, practices, and/or values (Eriksen, 2010, pp. 15–7). Ethnic affiliations are often also a powerful predictor of individual vulnerability to HIV. Invariably, each potential incident of human-to-human HIV transmission is situated within particular sociocultural, historical, and geographical contexts which indirectly affect the probabilities of transmission (Aral et al., 2008; Friedman et al., 2009). Since the beginnings of the global epidemic, such structural elements often engender skewed distributions of HIV disease and vulnerabilities at the societal level, whereby marginalized ethnic minorities endure disproportionate suffering due to HIV (Piot et al., 2001). One of the important observations from the small but growing body of minority ethnoracial MSM HIV research is that neither religious nor ethnic minority status per se is associated with greater HIV risk. The breadth of diversity between ethnoracial groups and cultural contexts simply precludes any such universal theories. For example, in the US, African Americans represent 35% of incident HIV infections among MSM (Maulsby et al., 2014), but only ~14% of the total US population (Rastogi et al., 2011). On the other hand, minority Asian American MSM have consistently had lower HIV prevalence and incidence than MSM of all other US racial groups (Wei et al., 2011). Similarly divergent trends have also been observed between minority “Blacks” and “Asians” in the United Kingdom (Elford et al., 2012). In Malaysia, one cross-sectional study found that HIV infection and UAI were both significantly more prevalent among ethnic majority Malay MSM than ethnic minority Chinese MSM (5.3%  202   vs. 1.8% and 49.4% vs. 39.8%, respectively) (Kanter et al., 2011). In North China, Southern ethnic minorities appear to have significantly greater HIV vulnerabilities compared to the Han majority, while the obverse is true among ethnic minority Hui (Chapter five). Ethnic identity affirmation can be defined as “the extent to which one is attached to one’s racial/ethnic group and views that group positively” (Schwartz et al., 2011). Along with any sense of identity derived from one’s sexual behaviours, ethnic identity affirmation among ethnic minority MSM has important implications for individual self-worth and mental health, which in turn directly impact one’s resolve and capacity to adopt HIV preventative behaviours (Chae & Yoshikawa, 2008; Nemoto et al., 2003; Vu et al., 2011). Despite stronger heteronormative and homonegative values among certain ethnocultural groups, evidence suggests that ethnic identity affirmation among MSM is inversely associated with various HIV sexual vulnerabilities. Among MSM in the US, greater affirmation of Latino ethnic identity was associated with less UAI (Donnell et al., 2002), and pride in Asian/Pacific Islander ethnic identity was correlated with more frequent HIV testing (Vu et al., 2011).  Notably, frequent HIV testing was even more likely when Asian/Pacific Islander MSM in the US reported strong ethnocultural identity in combination with strong sexual identity (Vu et al., 2011). In North China, ethnic identity affirmation appears to be associated with less anal sex and fewer sex partners (Chapter five). In addition to ethnoracial affiliation and ethnic identity affirmation, religious affiliation and religiosity have gained increasing recognition as critical influences on HIV vulnerabilities and testing among MSM.     Religious affiliations delineate boundaries of distinct organisational social structures and cultural norms between given religious groups, and have been shown to correlate with population HIV prevalence, individual HIV infection, and risker sexual behaviours (Shaw & El-Bassel,  203   2014). Empirical studies also indicate that religious affiliations have profound implications for the HIV vulnerabilities of men who have sex with men (MSM) (Lorente et al., 2012; Sheehy et al., 2013) (see Chapter six). In West Africa, Muslim MSM appear less likely to test for HIV and more likely to have sex with females when compared with their Christian counterparts (Lorente et al., 2012; Sheehy et al., 2013). In North China, Muslim MSM appear to engage in fewer riskier behaviours, while Buddhist MSM are more likely to report riskier sexual practices (see Chapter six).  In contrast to religious affiliation which is defined by group membership, religiosity is an inherently multi-dimensional construct that can be conceptualized as “a level of adherence to, participation in, influence of, or identification with a set of [religious] beliefs” (Shaw & El-Bassel, 2014). Differences in doctrine, practices, and norms present challenges for generalizing about religiosity across religious affiliations, but religious faithfulness and participation in religious social activities have been correlated with less risky sexual behaviours among MSM and transgender women in the US (Dowshen et al., 2011; Garofalo et al., 2014). In one study with MSM in North China, private religious activities among Christians appear to inhibit riskier sexual behaviours, while various forms of religiosity were associated with forced sex and commercial sex, respectively (Chapter six).   Mechanisms of association between religious affiliation, religiosity, ethnic affiliation, and ethnic identity affirmation with MSM HIV vulnerabilities remain poorly understood, but homonegativity (social stigma against homosexuality) may be an important contributing factor.  Homonegativity can be understood as being internalized or perceived externally. Internalized homonegativity refers to the internalization of heteronormative prejudices against sexual minorities by sexual minorities (Williamson, 2000), while perceived external  204   homonegativity alludes to the expectation of stigmatization and discrimination based on sexual minority status (Meyer, 1995). Internationally, internalized homonegativity has been correlated with UAI (Andrinopoulos et al., 2014; Ross, Kajubi, et al., 2013; Ross, Berg, et al., 2013), more sexual partners (Neilands et al., 2008), and less frequent HIV testing (Pyun et al., 2014; Santos et al., 2013). Similarly, perceived external homonegativity has also been associated with UAI among MSM in China and the US (Ayala et al., 2012; K. Choi et al., 2008). In theory, associations between homonegativity and HIV-related behaviours may be a function of negative stress coping (Williamson, 2000) and reduced socialization with other MSM , whereby MSM avoid interacting with other MSM who may otherwise be important sources of social support and information about HIV knowledge, prevention skills, and testing services (Ayala et al., 2012; Huebner et al., 2002).  Studies from the US indicate that MSM of certain religious faiths may be more likely to experience internalized and perceived external homonegativity (Keogh, n.d.; Kugle & Chiddy, 2009; Siker, 2007; Smallwood, 2013; Wagner et al., 2012; Wilson et al., 2011), which in turn engenders greater HIV vulnerability such as UAI (Ross et al., 2008; Smallwood, 2013). However, little is known about the relationship between religion, homonegativity, and HIV vulnerabilities among MSM in the East Asian context.   This study sought to test the corresponding mediational effects of homonegativity among a sample of MSM in North China.  Based upon results from earlier analyses (Chapters four and five), I hypothesized that internalized and perceived external homonegativity would significantly mediate the following statistically significant measures of association identified in chapters four and five:  1. Association between Muslim religious affiliation and anal sex in past three months  205   2. Association between Buddhist religious affiliation and anal sex in past three months 3. Association between Muslim religious affiliation and unprotected anal sex in past three months 4. Association between Buddhist religious affiliation and unprotected anal sex in past three months 5. Association between Buddhist religious affiliation and more than three sex partners in past six months 6. Association between organisational religiosity and ever experienced forced sex 7. Association between private religiosity and ever experienced forced sex 8.  Association between intrinsic religiosity and bought or sold sex in past three months 9.  Association between private religiosity among Christians and anal sex in past three months 10. Association between intrinsic religiosity among areligious participants and circumcision 11. Association between Southern ethnicity minority affiliation and HIV infection 12. Association between Hui ethnic affiliation and anal sex in past three months 13. Association between Hui ethnic affiliation and unprotected sex in past three months 14. Association between Southern ethnic minority affiliation and ever experienced forced sex 15. Association between Hui ethnic affiliation and circumcision 16. Association between Northern ethnic minority affiliation and Bought or sold sex 17. Association between ethnic identity affirmation and anal sex in past three months 18. Association between ethnic identity affirmation and more than three sex partners in past six months    206   6.2 Methods  6.2.1 Study design and recruitment Data from the present study originate from a larger cross-sectional survey of religious and ethnic minority MSM which was conducted from July 2013 to April 2014 in Beijing and Tianjin, China at five HIV voluntary counseling and testing (VCT) sites devoted to serving MSM. Survey participants were recruited passively and actively by four methods. First, men who had come to receive VCT services on their own volition were invited on-site to participate in the study. Second, successfully enrolled participants were encouraged to recruit their eligible MSM friends. Third, a non-governmental organisation dedicated to MSM HIV prevention distributed study recruitment information to potential participants via a widely used smart phone social networking application. Fourth, potential study participants were contacted via several gay social networking websites and chatrooms. Due to potential privacy concerns, participants were free to use pseudonyms, and no true identifiers were requested. Men were eligible to participate in the survey if they self-reported ever having sex with another man, were at least 18 years old, had never tested positive for HIV, and were willing and able to provide written informed consent. Due t