University of Southern California Dissertations and Theses

Marital matching in West Africa: an examination of interethnic and interreligious first marriages in Benin

(USC Thesis Other)

Marital matching in West Africa: an examination of interethnic and interreligious first marriages in Benin

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Marital matching in West Africa: An examination of interethnic and interreligious first marriages in Benin by Alli Coritz A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (SOCIOLOGY) May 2020 ii ACKNOWLEDGEMENTS I would like to sincerely thank all the people who made my time at USC a memorable one and helped me make it through the dissertation process. While I am not able to name everyone, please know that so many of you have helped shape me as a person and as a scholar by offering me your time. For that, I thank you all. Thank you to my dissertation chair, Dr. Emily Smith-Greenaway. It has been an honor and privilege to work with you and learn from you. Your mentorship has been invaluable and I appreciate it more than words can express. Thank you for your time, effort, advice, and support. I cannot imagine making it through a Ph.D. program without you as my advisor. I cannot wait to see what the future holds in store for your career and research. Thank you. Thank you to Dr. Lynne M Casper for offering me support, guidance, and life advice throughout my time in the department. Your support and advice, particularly for my career aspirations, were invaluable. I appreciate the hours you spent talking with me. Thank you to Dr. Dowell Myers for offering me your time and support during my stay at USC. Your feedback and questions were a great asset in my development and always framed in such a way that I felt bolstered by them. Thank you to all of the friends I have made during my time in this program. Mary Ippolito: thank you for all of your support, love, and friendship – I truly would not have survived without you. Thank you Birgan Gokmenoglu for being my sanity and social life my first year and continuing to be my friend from London. Thank you to my cohort for going through the years and coursework with me: Michelle Demus, Briana Jex, Minwoo Jung, Jihye Lee, and especially LaToya Council, my good friend, confidant, and cat-sitter. Thank you to Yael Findler, Ben Weiss, Meas Meng for being amazing friends (and dissertation editors). Thank you to Eunjeong Paek, Oded Maron, Matthew Ripley, and Rachelle Wang for your friendship and support. Thank you to my wise elders, Kushan Dasgupta and Jennifer Candipan, for getting me through the dissertation phase. Thank you to my friends who helped see me through life before the Ph.D. and offered me support during this process. Thank you to my Peace Corps friends: Jocelyn Brousseau and Cara McGraw, I am so grateful we all ended up in Los Angeles together, Jess Fell for your continual support of both my Ph.D. and my move to D.C., and Raney Lee, thank you so much for being a consistent and amazing friend – your friendship is invaluable to me. Thank you to my amazing Arizona State friends: Heidi TenPas, Maria Polletta, and Scott Ross. And thank you to Robyn Blust for being an amazing friend for nearly two decades: I am honored to call you my friend. Thank you to the sociology office staff for keeping the whole place running, as well as for offering me lessons in friendship and sanity. Stachelle Overland: you are a true gem and I could not have survived without you. There is no way to thank you enough. Thank you Amber Thomas, Melissa Hernandez, Lisa Losorelli, and Angie Mendoza for all of your support, friendship, and the time you spent talking to me about random topics. Thank you for all that you do for the department. Thank you to all of my friends, students, and loved ones in Benin. I am truly grateful and honored for the hospitality, warmness, and openness with which you welcomed me in your homes. The interethnic and interreligious relationships I witnessed in Benin helped inspire this research and I hope I did them justice. I would like to offer a special thank you to Madame Pema, Benin’s most inspiring and hardworking woman who I was fortunate to have as my iii maman. Thank you for feeding, clothing, and naming me, while also ensuring I was safe and integrated into our local community. May you rest in peace. Thank you to my family for offering consistent support and love during this process. To my mom, Bobbie, and sister, Sari: thank you for your support and unwavering optimism (and listening to my random stories). Thank you to my Dad, Al, for your encouragement both of the Ph.D. process and my desire to get a job quickly. Thank you to my partner, Fabian, for believing in me, feeding me, listening to me, and ignoring my rambles as needed. And lastly, thank you to my cat, Chester, for being my sanity for seven years: it might be ridiculous that I brought you home from Benin but I’ve never regretted it. iv TABLE OF CONTENTS ACKNOWLEDGEMENTS ii LIST OF TABLES vi LIST OF FIGURES ix ABSTRACT x CHAPTER 1 INTRODUCTION 1 Overview of dissertation 4 CHAPTER 2 STUDY CONTEXT AND DATA 10 Study Context 10 Ethnicity in Benin 11 Religion in Benin 12 Geographic clustering in Benin 13 Gender, marriage, and family life in Benin 16 Macro-context of Benin 20 Data 22 CHAPTER 3 INTERETHNIC AND INTERRELIGIOUS MARRIAGE IN BENIN: INDIVIDUAL- AND COUPLE-LEVEL TRENDS AND PATTERNS 27 Background 29 Measures 32 Dependent variables 32 Key individual-level independent variables 33 Key couple-level independent variables 35 Analytic Approach 37 Results 38 Changes in interethnic and interreligious marriage over time 38 Spatial distribution of interethnic and interreligious marriage 40 Characteristics of interethnic and interreligious couples and marriages 47 Discussion 61 CHAPTER 4 FERTILITY PREFERENCE CONCORDANCE AND FERTILITY OUTCOMES IN INTERETHNIC AND INTERRELIGIOUS MARRIAGES 74 Social Groups and Fertility Preferences and Outcomes 75 Ethnicity and fertility 76 Religion and fertility 77 Couples’ Fertility Preference Concordance 78 Heterogamy, Couples, and Fertility Preferences and Outcomes 80 Fertility preference hypotheses 80 Fertility outcomes hypotheses 82 Study Context 84 Ethnicity and fertility preferences and outcomes in Benin 84 Religion and fertility preferences and outcomes in Benin 85 Sample 86 v Measures 86 Dependent variables 86 Key independent variables 88 Analytic Approach 91 Results 92 Individual fertility preferences 94 Couple concordance on fertility preferences 104 Completed fertility 118 Discussion 122 CHAPTER 5 MARITAL INVESTMENT AND INTERETHNIC AND INTERRELIGIOUS MARRIAGES 130 Measuring Marital Investment 131 Study Context 135 Extramarital sex and polygyny 135 Ethnicity, religion, extramarital sex, and polygyny in Benin 137 Samples 138 Measures 139 Dependent variables 139 Control variables 140 Analytic Approach 144 Results 144 Extramarital sex 144 Polygyny 153 Discussion 162 CHAPTER 6 CONCLUSION 172 Summary 172 Contributions to current research 175 Limitations 178 Discussion and implications 180 REFERENCES 183 APPENDIX CHAPTER 2 203 APPENDIX CHAPTER 3 212 APPENDIX CHAPTER 4 220 APPENDIX CHAPTER 5 222 vi LIST OF TABLES Table 2.1: A comparison of married women’s characteristics in the full DHS sample, first marriages in the full DHS, and first marriages in the couples’ data (Survey years: 1996, 2001, 2006, 2011, and 2018) ................................................................................................. 25 Table 3.1: Most common interethnic and interreligious pairings .................................................. 44 Table 3.2: Individual characteristics and difference testing of married men and women in Benin by ethnic composition of marriage ........................................................................................ 50 Table 3.3: Individual characteristics and difference testing of married men and women in Benin by religious composition of marriage .................................................................................... 51 Table 3.4: Odds ratios of individual-level characteristics associated with entry into interethnic marriages ............................................................................................................................... 52 Table 3.5: Odds ratios of individual-level characteristics associated with entry into interreligious marriages ............................................................................................................................... 53 Table 3.6: Characteristics and difference testing of couples in Benin by ethnic composition of marriage ................................................................................................................................. 55 Table 3.7: Logistic regressions of couple-level characteristics associated with entry into interethnic marriages (N=9828) ............................................................................................ 56 Table 3.8: Characteristics and difference testing of couples in Benin by religious composition of marriage ................................................................................................................................. 58 Table 3.9: Logistic regressions of couple-level characteristics associated with entry into interreligious marriages (N=9828) ........................................................................................ 59 Table 4.1: Means and percentages of fertility variables of interest by ethnic composition of marriage (Standard deviations in parentheses) ...................................................................... 93 Table 4.2: Means and percentages of fertility variables of interest by religious composition of marriage (Standard deviations in parentheses) ...................................................................... 94 Table 4.3: Men’s ideal family size — Interethnic ......................................................................... 96 Table 4.4: Men’s ideal family size — Interreligious ..................................................................... 98 Table 4.5: Women’s ideal family size — Interethnic .................................................................. 101 Table 4.6: Women’s ideal family size — Interreligious ............................................................. 103 vii Table 4.7: Gender gap in ideal family size .................................................................................. 105 Table 4.8: Absolute ideal family size gap ................................................................................... 107 Table 4.9: Likelihood of reporting the same ideal family size .................................................... 110 Table 4.10: Likelihood of reporting an ideal family size within two children of each other ...... 113 Table 4.11: Likelihood of reporting an ideal family size within four children of each other ..... 115 Table 4.12: Completed fertility for women aged 40 to 49 .......................................................... 120 Table 5.1: Odds ratio of men reporting engaging in extramarital sex within last 12 months – Couples’ characteristics ....................................................................................................... 148 Table 5.2: Odds ratio of men reporting engaging in extramarital sex within last 12 months – Men’s characteristics interethnic marriages ........................................................................ 150 Table 5.3: Odds ratio of men reporting engaging in extramarital sex within last 12 months – Men’s characteristics interreligious marriages .................................................................... 152 Table 5.4: Means and percentages of characteristics of women in polygynous marriages by ethnic composition of marriage (Standard deviations in parentheses) ................................ 155 Table 5.5: Odds ratio of marriage transitioning to polygyny – Couples’ characteristics ............ 157 Table 5.6: Odds of transition to polygyny — Men’s characteristics interethnic ......................... 159 Table 5.7: Odds of transition to polygyny — Men’s characteristics interreligious .................... 161 Appendix 3 Table 1: Pairings by ethnicity and gender of spouse by absolute number of pairings and percentage of all interethnic marriages ......................................................................... 212 Appendix 3 Table 2: Pairings by religion and gender of spouse by absolute number of pairings and percentage of all interreligious marriages ..................................................................... 214 Appendix 3 Table 3: Individual characteristics and difference testing of recently married men and women in Benin by ethnic composition of marriage .................................................... 216 Appendix 3 Table 4: Individual characteristics and difference testing of recently married men and women in Benin by religious composition of marriage ................................................ 217 Appendix 3 Table 5: Couple characteristics and difference testing of recent marriages in Benin by ethnic composition of marriage ...................................................................................... 218 viii Appendix 3 Table 6: Couple characteristics and difference testing of recent marriages in Benin by religious composition of marriage .................................................................................. 219 Appendix 4 Table 1: Fertility preferences and outcomes by ethnicity of wife in Benin ............. 220 Appendix 4 Table 2: Fertility preferences and outcomes by religion of wife in Benin .............. 221 Appendix 5 Table 1: Extramarital sex and polygyny by ethnicity of husband ........................... 222 Appendix 5 Table 2: Extramarital sex and polygyny by religion of husband ............................. 223 Appendix 5 Table 3: Means and percentages of characteristics of women in general sample, monogamous marriages, and polygynous marriages (Standard deviation in parentheses) . 224 Appendix 5 Table 4: Likelihood of being in a polygynous marriage .......................................... 225 ix LIST OF FIGURES Figure 2.1: A map of Benin, highlighting its location in sub-Saharan Africa. .............................. 11 Figure 2.2: Percent of ethnic fractionalization in each department of Benin, measured using a Herfindahl-Hirschman index. ................................................................................................ 14 Figure 2.3: Percent of religious fractionalization in each department of Benin, measured using a Herfindahl-Hirschman index. ................................................................................................ 15 Figure 3.1: The probability of urban and rural couples being interethnic by decade of marriage 39 Figure 3.2: The probability of urban and rural couples being interreligious by decade of marriage ................................................................................................................. 40 Figure 3.3: Percent of marriages that are interethnic by department ............................................. 41 Figure 3.4: Percent of marriages that are interreligious by department ........................................ 42 Appendix 2 Figure 1: Beninese languages in the Niger-Congo language family ....................... 208 Appendix 2 Figure 2: Beninese language in the Nilo-Saharan language family ......................... 209 Appendix 2 Figure 3: A map of Benin’s departments ................................................................. 210 Appendix 2 Figure 4: Beninese population pyramids ................................................................. 211 x ABSTRACT Family formation patterns are in flux globally including assortative partner selection. Prior research notes the historical importance of shared social identities — specifically ethnicity and religion — in partner selection, with most couples sharing these social identities. Yet, in many contexts globally, large structural changes, such as increasing education, urbanization, and migration have created diverse sites for people of varying social groups to interact extensively, increasing the frequency of intermarriages along ethnic and religious lines. However, due to the importance of these social identities in value formation, marriages in which spouses’ social identities do not match often experience higher levels of marital strain. Despite sub-Saharan Africa’s changing socioeconomic structures, there is minimal research on the frequency of interethnic and interreligious marriages and the marital strain experienced by these couples in the region. In this dissertation, I analyze the trends in interethnic and interreligious first marriages in Benin, West Africa. I use couples’ data from five Demographic and Health Surveys (DHS) collected between 1996 and 2018 to analyze the temporal and spatial trends in interethnic and interreligious marriages and the characteristics that are associated with entry into these marriages. I then examine the correlation between these marital types and indicators of marital strain, including fertility preference discordance and men’s engagement in extramarital sex. Three empirical chapters provide an examination of interethnic and interreligious marriages in Benin. The first empirical chapter, Chapter 3, explores the changes in interethnic and interreligious marriage over time by comparing marital cohorts (pre-1980s to the 2010s), followed by a spatial exploration of the geographic spread of interethnic and interreligious marriages. I use binary and multivariate logistic models to explore the relationship between xi intermarriage and men’s and women’s individual characteristics, as well as their characteristics relative to each other. These analyses indicate that interethnic first marriages have increased rapidly and are tied to increasing education, urbanization, and wealth. The increase in interethnic marriage seems to correlate with structural changes affecting the marriage market. Specifically, interethnic marriages appear to correlate with higher levels of potential interethnic interactions and the prioritization of matching human capital over shared group identity. I found no such patterns for interreligious marriages: the socioeconomic and demographic characteristics of interreligious first marriages mirror those of the average intrareligious first marriage and these unions do not display consistent signs of increasing in frequency. My findings suggest that these two forms of intermarriage are not uniformly tied to larger socioeconomic changes and that they are distinct marital formation patterns. Chapter 4, the second empirical chapter, compares the fertility preferences and outcomes of interethnic and interreligious couples to their intraethnic and intrareligious counterparts. Ethnicity and religion both play important roles in shaping individuals’ fertility preferences and couples’ fertility outcomes. However, much of the literature on fertility is centered on women, and we know little about how different social identities within couples are linked to fertility preferences and completed fertility in sub-Saharan contexts. I find that men and women in interethnic marriages have significantly lower ideal family sizes than their intraethnic counterparts and that they are more likely to be concordant in their fertility preferences than intraethnically married couples. Contrarily, interreligious couples do not have distinct fertility preferences and they show only minimally increased concordance compared to intrareligious couples. The results for fertility outcomes among women ending their reproductive years indicate that interethnic couples do not have fewer children despite their lower preferences, xii however, interreligious couples have more children despite no difference in their preferences. These findings suggest that couples of different religious backgrounds experience fertility differently than intrareligious couples. Further, the differing outcomes between interethnic and interreligious couples indicate that interethnic and interreligious marriages are not uniformly tied to couples’ fertility preferences and outcomes although both involve partners of different social identities. The third empirical chapter, Chapter 5, examines the presence of marital destabilizing behaviors that indicate low marital investment in intact marriages. Specifically, I examine two locally salient marital destabilizing behaviors — extramarital sex and polygyny. I ask whether men in interethnic or interreligious marriages are less invested in their unions by estimating if they are more likely to engage in behaviors that are disruptive to the couple-unit and that cause negative health outcomes for their wives. Using logistic regression, I find that interethnically married men have higher odds of engaging in extramarital sex, however, when controlling for educational attainment or wealth, this difference is non-significant. This indicates that it is not being interethnically married, per se, that results in this higher likelihood of engaging in extramarital sex, rather interethnically married men are disproportionately in higher socioeconomic brackets where extramarital sex is more common. Importantly, interethnic marriages are no more likely to transition to polygyny. Interreligiously married men, on the other hand, are significantly more likely to engage in extramarital sex, although not polygyny, than intrareligiously married men even when controlling for educational attainment and wealth. This suggests that men in interreligious marriages are less invested in their unions as they are more likely to engage in marital destabilizing behaviors that are linked to negative outcomes for their wives and children. That these findings differ across the two types of intermarriage suggests that xiii interethnic couples do not experience similar levels of marital strain as interreligious couples, despite both being marriages across social groups. This further supports interethnic and interreligious marriages as being distinct marital patterns with different social meanings. In sum, the results of this dissertation indicate that marriage markets are in flux in Benin, West Africa in ways that are shifting the salience of shared social identities in partner selection. One of the main findings of this dissertation is that selection into and the outcomes of being in an intermarriage are not the same for the average interethnic and interreligious couple, indicating that intermarriage is not a uniform experience across social groups. Interethnic marriage in Benin is correlated with recent structural changes, such as increased access to school and urbanization. This supports research in other regions of the world finding that social identity homogamy often decreases with structural changes, especially those that diversify networks. Although interethnic marriages are uncommon, their ties to higher levels of education and wealth appear to be protective against higher marital strain, both in terms of their relative fertility preferences and the husband’s engagement in marital destabilizing behaviors. This may be because interethnic couples, on average, have more resources to weather familial discontent and are more likely to exhibit the characteristics that correlate with more power for wives in the relationship and higher levels of communication at the couple-level. Contrarily, in Benin, the average interreligious marriage mirrors the average intrareligious marriage outside of their choice to marry interreligiously. These couples are neither richer, nor more educated, nor more urban — and there have been relatively stable high rates of interreligious couplings across survey waves. Yet, interreligious marriages appear to experience more marital strain regarding fertility and extramarital sex than their intrareligious counterparts. This suggests that interreligious marriages experience higher levels of strain in the Beninese xiv context and, unlike the interethnic couples, they are less likely, on average, to have the resources to help buffer this increased marital strain. These findings tie into rich literatures showing the benefits of increased resources in staving off marital strain and indicate that although both types of intermarriage involve partners marrying across social groups, interethnic and interreligious marriages are distinct in their correlation with structural changes and their marital outcomes. 1 CHAPTER 1 INTRODUCTION Assortative marriage, or the patterned selection of spouses with specific characteristics, is common globally and is particularly pronounced along the lines of social identities (Qian and Lichter 2018). Historically, marriages have been homogamous along ethnic/racial and religious lines, indicating that shared social backgrounds are prized on the marriage market (Kalmijn 1998; Qian 2005; Qian and Lichter 2018). There are several reasons for the persistence of shared social backgrounds in marriages. Social groups play a key role in organizing networks (Leszczensky and Pink 2019; McPherson, Smith-Lovin, and Cook 2001) and social groups are often geographically clustered (Maïga and Baya 2011; Martin, Hacker, and Francesco 2018; Smith 2011; Torabi and Baschieri 2010). Further, extended kin’s expectations of appropriate marital partners often include social homogamy (Hosseini and Erfani 2014; Maïga and Baya 2011; Torabi and Baschieri 2010). Additionally, there are often historic and structural divisions between social groups that may make certain types of pairings particularly unlikely (Kalmijn and Van Tubergen 2010; Qian 2005; Safi and Rogers 2008). In the United States, for example, there are relatively few marriages across racial lines and marriages among whites and Blacks are especially uncommon due to racism, including the enduring social, physical, and spatial exclusion of Black Americans by whites (Bialik 2017; Qian 2005). Globally, however, assortative marriage is in flux. Whereas historically social group homogamy has been prized, there is an increasing shift towards homogamy along achieved characteristics or markers of human capital (Esteve et al. 2016; Gonalons-Pons and Schwartz 2017; Hu and Qian 2016; Kashyap, Esteve, and García-Román 2015; Schwartz and Mare 2005). Large-scale structural and economic shifts have made it increasingly possible and often favorable 2 for people to select partners whose achieved characteristics match their own, weakening historically favored patterns of heterogamy (Esteve et al. 2016; Hu and Qian 2016; Kashyap et al. 2015; Schwartz and Gonalons-Pons 2017; Schwartz and Mare 2005). One important example is that of educational assortment in marriage: increased education for women has decreased traditional heterogamy that favored husbands being more educated than their wives (Esteve et al. 2016; Frye and Urbina 2019; Hu and Qian 2016; Schwartz and Mare 2005). Couples who have higher levels of human capital are often increasingly better positioned to accumulate economic resources, marry and remain married, and have overall better family outcomes (Amato 2010; Bumpass, Sweet, and Cherlin 1991; Esteve et al. 2016; Killewald 2016; McLanahan and Schwartz 2002; Oppenheimer 1997; Schwartz and Gonalons-Pons 2017). One of the ways in which these changes correspond to shifting social group homogamy is that these changes have opened up new social locations, such as schools and the workforce, for people to meet partners of different social backgrounds (Calvès and Gnoumou Thiombiano 2014; Kruse and Kroneberg 2019; Leszczensky and Pink 2019; Maas and Zijdeman 2010; Rosenfeld and Thomas 2012). Social groups, both ethnic and religious, are frequently geographically clustered (Behrman, Kohler, and Watkins 2002; Maïga and Baya 2011; Smith 2011; Torabi and Baschieri 2010). As a result, there were often limited chances for people of different backgrounds to meet and form substantive relations, increasing marital homogamy along social group lines. These new social sites diversify networks, allowing for more persistent relationships — both platonic and romantic — across social groups, particularly among individuals who may not identify strongly with their primary social groups (Behrman et al. 2002; Kruse and Kroneberg 2019; Leszczensky and Pink 2019; Maïga and Baya 2011). 3 Additionally, these broader shifts frequently correlate with an increasing level of individualization in which marriage and partner selection are imbued with more personal meaning (Calvès and Marcoux 2007; Lesthaeghe 2014). As a result, people are increasingly selecting partners in social contexts that emphasize individual preferences in marital partners, rather than partner selection remaining a family affair (Andro, Hertrich, and Robertson 2002; Calvès 2016a; Calvès and Marcoux 2007; Lesthaeghe 2014; Rosenfeld and Kim 2005). The decreasing importance of familial involvement and familial approval in partner selection opens up possibilities for marriages outside of the extended family’s shared social group. Combined, these changes create new possibilities for people to select partners of different social backgrounds while also altering the importance of shared human capital in several global contexts. As matching levels of human capital become a more salient feature on the marriage market, people are often increasingly willing and able to marry across social groups in order to achieve this newly prized form of homogamy (Calvès and Gnoumou Thiombiano 2014; Çelikaksoy 2016; Furtado 2012; Qian and Lichter 2011; Schwartz, Zeng, and Xie 2016). The outcome of this in many countries is increasing rates of intermarriage across social identities, changing traditional forms and patterns of assortative marriage. However, because these social identities often correlate with distinct values, preferences, and social networks, interethnic and interreligious marriages are not always neutral pairings. These heterogamous couples illustrate the continued importance of social boundaries — both how they can shift over time and how crossing them leads to distinct outcomes — making them important marriages to study (Kalmijn 1998). At the couple-level, interethnic and interreligious marriages are frequently less stable and often experience a higher risk of divorce, indicating that marrying across social boundaries may serve as a destabilizing force in marriages (Locoh and Thiriat 4 1995; Maïga and Baya 2011; Smith, Maas, and Tubergen 2012). Further, there are often implications for the children of these unions, as their social identity is often harder to form and less stable (Awusabo-Asare 1990; Gnoumou Thiombiano 2017; Gorenburg 2006; Maïga and Baya 2011). Additionally, the children of these unions may have worse health outcomes than children of intramarriages (Choi and Reichman 2019). At a more macro-level, interethnic and interreligious unions between minority and majority groups also frequently result in a greater loss of the minority culture (i.e. language, customs) than the majority culture (Botha 2008; Langer and Ukiwo 2009; Nuworsu 2015). As a result, interethnic and interreligious marriage is often linked to cultural homogenization that favors majority groups under the guise of assimilation (Gorenburg 2006; Martin et al. 2018; Qian, Glick, and Baston 2015; Safi and Rogers 2008). Additionally, because social identity homogamy is presumed in many places, some countries’ laws have had to adjust to accommodate the existence of children from intermarriages (Awusabo-Asare 1990). Thus, the couples that marry across social groups often do so in the face of cultural and structural realities that can make these unions more challenging than intramarriages. Overview of dissertation The main goal of this dissertation is to examine interethnic and interreligious marriage in a contemporary West African context, combining literatures on demography, marriage, family, fertility, ethnicity, and religion in the region. Interethnic and interreligious marriage is an important topic to study as these social identities are strongly linked to norms, preferences, and beliefs and these couples often receive less familial support than couples whose social identities match (Kalmijn 1998). West Africa is experiencing the shifts that corresponded to increasing intermarriage in other global contexts, yet there is minimal research on intermarriage in the 5 region (Calvès and Gnoumou Thiombiano 2014; Çelikaksoy 2016; Crespin-Boucaud 2018; Furtado 2012; Qian and Lichter 2011; Schwartz et al. 2016). This is particularly surprising given the region’s high levels of ethnic and religious diversity and the historic importance of social identity in family life (Caldwell 1996; Caldwell and Caldwell 1987). The little data we have on this topic offer mixed conclusions for whether interethnic and interreligious marriages are becoming more common (Crespin-Boucaud 2018). Studies focusing on West African capital cities indicate that this may be a burgeoning trend correlating with the prioritization of human capital on the marriage market (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014). Further, we know little about the individuals who marry interethnically and interreligiously in West Africa, let alone what the implications of these union types are for couples. I use Benin as my case study as ethnicity and religion have remained especially salient in marital formation there compared to other West African countries (Crespin-Boucaud 2018). Benin is also experiencing many of the social and economic changes — increasing women’s education and urbanization, in particular — that correlate with broader changes in marital patterns in the region (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Clark and Brauner-Otto 2015; Clark, Koski, and Smith-Greenaway 2017; Koski, Clark, and Nandi 2017; Maïga and Baya 2014). Additionally, compared to other countries in the region, Benin experiences low levels of ethnic and religious conflict, which lowers some of the barriers to interethnic and interreligious pairings (Adotevi, Law, and Ronen 2017; Dickovick 2008; Houngnikpo and Decalo 2013). Despite the lack of outright conflict, tensions exist between 6 social groups 1 which may be why intermarriages are less common, on average, in Benin than several other sub-Saharan African countries (Crespin-Boucaud 2018). In this dissertation, I analyze: (1) the spatial and temporal trends of interethnic and interreligious marriage, (2) the individual- and couple-level characteristics associated with entry into interethnic and interreligious marriages, (3) fertility preference concordance in interethnic and interreligious marriages, (4) the fertility outcomes of interethnic and interreligious couples, (5) and men’s investment in their marriages using locally salient indicators: extramarital sex and polygyny. To accomplish this, I use the Beninese Demographic and Health Surveys (BDHS), specifically the couples’ data which pair husbands’ and wives’ data. Collected every five years since 1996, I have data on couples married over the course of five decades. Chapter 2 offers more information on the Beninese context and datasets used in these studies. In Chapter 3, I provide an exploration of the social, economic, and demographic correlates of interethnic and interreligious marriage in Benin. Specifically, I explore trends in their geographic spread across the country, how and if their frequency has changed across marital cohorts, and which characteristics are associated with entry into these unions. Spatially, intermarriage correlates with diversity – interethnic marriage is more common in ethnically diverse areas and the same is true for interreligious marriage and religious diversity. Additionally, ethnic and religious groups that reside in proximity frequently intermarry. However, the results of this chapter suggest average differences between interethnic and interreligious couples. Interethnic marriage correlates strongly with urbanicity, whereas interreligious couples reside in urban and rural locations at the same frequency as intrareligious couples. Temporally, interethnic marriage 1 For example, the Bariba historically raided the Betamaribé and sold them as slaves, resulting in present-day tensions (Houngnikpo and Decalo 2013). 7 is increasingly common throughout marital cohorts and survey waves, suggesting that this is a burgeoning marital formation pattern. Contrarily, interreligious marriages are relatively stable across marital cohorts and survey waves, only minimally increasing, suggesting that these unions are not increasingly or decreasingly common in Benin. I also examine the characteristics that correlate with entry into interethnic and interreligious marriages. I find that interethnic marriage correlates strongly with education for both men and women, as well as wealth and urbanicity, suggesting that these are not marriages of people unable to find partners on the marriage market – rather, these attributes suggest that these are marital formations correlated with structural changes. Contrarily, interreligious couples are not distinct socioeconomically and demographically, on average, from their intrareligious counterparts minus their choice to marry interreligiously. My findings do not suggest ties between interreligious unions and recent structural changes in the Beninese context. These results indicate that intermarriage across social groups is not uniformly tied to broader social changes and that the boundaries between certain social groups may be more sensitive to changes in the marriage market than others depending on the social context. Thus, the correlates of either interethnic or interreligious marriage cannot be broadly applied to different types of intermarriage as these are distinct forms of partner selection. I examine the fertility preferences, concordance, and outcomes of couples in intermarriages in Chapter 4, as ethnicity and religion are strongly tied to fertility in West Africa. Additionally, as couples in interethnic and interreligious unions have already transgressed social norms by selecting a partner from a different social group, they may also be at the vanguard of fertility changes in West Africa. I explore these questions asking whether these couples exhibit different fertility preferences than intramarried couples, whether they have different likelihoods of fertility 8 preference concordance, and whether these couples have distinct fertility outcomes. The results of this study again vary by the type of intermarriage. I find that interethnically married men and women have lower ideal family sizes and higher levels of couple-level fertility preference concordance, especially at higher gaps, suggesting increased communication in these couples. They do not, however, have distinct fertility outcomes. In all, these results indicate that the choice to marry interethnically correlates with other distinct preferences. Contrarily, interreligious couples do not have distinct fertility preferences and they are only minimally more likely to be concordant in their preferences. They do, however, have higher completed parities than intrareligious couples. This suggests that interreligious couples face more marital strain, less successful compromises, and perhaps a desire to validate their relationship with more children, resulting in higher fertility levels than for couples who share a religious background. Thus, interethnic and interreligious unions are not homogenous in their salience in couples’ fertility preferences and outcomes. Rather, interreligious marriages appear to be more strained, and this is likely due to the lower selectivity into interreligious unions than interethnic unions. Although neither type of intermarriage exhibits lower completed fertility, interethnic couples have distinctly lower fertility preferences, indicating that these are couples who may have lower fertility than average in future cohorts and be at the vanguard of fertility declines. In Chapter 5, I use two locally salient marital destabilizing behaviors to see if intermarried men are more likely to act in ways that harm the couple-unit, indicating they are less invested in their primary couple unit than men who are intramarried. Specifically, I examine extramarital sex and taking on an additional wife, transitioning a monogamous union into a polygynous union, both of which are used as alternatives to divorce in Benin. Extramarital sex and polygyny are both indicators and causes of low marital investment in the Beninese context, providing a metric 9 to compare the relative investment of men in interethnic and interreligious marriages versus their intramarried counterparts. As with the earlier chapters, the results of this study indicate that interethnic and interreligious unions are not equally at risk of husbands engaging in marital destabilizing behaviors. Among interethnic unions, the results of this study offer strong evidence of selection-effects — I find that interethnically married men are more likely to engage in extramarital sex at baseline but once controls for education or wealth are included, this effect disappears. This suggests that the capital levels of interethnically married men put them in the demographic of men who are most likely to engage in extramarital sex – young, educated and wealthy. Contrarily, interreligiously married men are significantly more likely to engage in extramarital sex regardless of controls. This finding strongly suggests that men in interreligious marriages are less invested in their marriages than their intrareligious counterparts, with husbands being more likely to engage in marital destabilizing behaviors, increasing the likelihood of negative well-being for the household. There is no evidence that either type of intermarriage is at increased risk of becoming polygynous. The results of this study suggest that interreligious marriages, in particular, experience higher levels of marital strain as the husbands in these unions are likely less invested in their marriages than men who are intrareligiously married. This increases their likelihood of exhibiting behaviors that may harm — mentally and physically — the couple-unit. Once again, these results suggest that the high levels of positive selection into interethnic marriages may serve to stave off marital strain, whereas couples in interreligious marriages are less likely to have the resources to do so. 10 CHAPTER 2 STUDY CONTEXT AND DATA Study Context This study focuses on the Republic of Benin, shown in Figure 2.1. Benin is a diverse West African country that is undergoing the social and economic shifts that often correlate with changes in social identity homogamy in family formation. Specifically, I study two important social categories that shape daily life and family formation in Benin: ethnicity and religion. In Benin, marriages between partners of the same ethnic background (86%) or same religious affiliation (72%) comprise the majority of marriages and 62% of marriages are homogamous in both ethnicity and religion, with less than 3% of couples reporting different ethnic and religious affiliations 2 . Given the role ethnicity and religion appear to play in partner selection this study examines the characteristics of interethnic and interreligious marriages in Benin. 2 The results shown in this study are robust to the exclusion of couples who are both interethnic and interreligious. These couples’ characteristics are more similar to the average interethnic couple rather than the average interreligious couple. 11 Figure 2.1: A map of Benin, highlighting its location in sub-Saharan Africa. Map source: Wikimedia Ethnicity in Benin Benin’s high levels of ethnic diversity make it an ideal case study for interethnic marriage. Benin is among one-third of countries globally that do not have a majority ethnic group (Fearon 2003). Benin’s L’Institut National de la Statistique et de L’Analyse Economique (INSAE) uses shared linguistic and geo-cultural histories to group Benin’s 60 ethnolinguistic groups into the nine overarching ethnic groups used in the Demographic and Health Surveys (Amadou Sanni and Atodjinou 2012). These nine ethnic groups are: Adja (16%), Bariba (9%), Betamaribé (6%), Dendi (4%), Fon (43%), Peulh (5%), Yoa/Lokpa (3%), Yoruba (12%), and other (1%). A more 12 in-depth description of these ethnic groups is located in Appendix 2.A and their linguistic family trees are located in Appendix 2 Figures 1 and 2. Using these nine categories, Benin has one of the highest levels of linguistic fractionalization in sub-Saharan Africa (Buzasi 2015, 2016). Unlike many countries in the region, Benin does not have a national language; regional languages, such as Fon and Dendi, and the official language, French, are used for communication. As a result, many of the Beninese do not share a language with their co-nationals from other regions, despite speaking multiple languages on average (Buzasi 2016). Compared to many countries in West Africa, Benin’s ethnic diversity is not a source of major conflict (Adotevi et al. 2017; Dickovick 2008). However, there are boundaries and tensions between various ethnic groups due to historic conflicts and violence along ethnic lines (Hahonou 2009, 2015; Houngnikpo and Decalo 2013; Law and Lovejoy 1999; Ojo 2008). Religion in Benin Benin is a religiously diverse context, with the INSAE grouping religions in Benin into seven large categories: Traditional religions (15%), including Vodun 3 , Islam (24%), Catholicism (30%), Protestantism (6%), other Christian denominations (18%), other religions (1%), and no religion (6%). A more in-depth description of these religious categories is located in Appendix 2.B. Unlike in neighboring countries, such as Nigeria, religion is not a source of conflict in Benin (Adotevi et al. 2017; Houngnikpo and Decalo 2013). Beninese religious groups are known for their syncretic nature, specifically as it relates to incorporating Traditional beliefs and practices into monotheistic religions (Falen 2011, 2016; Houngnikpo and Decalo 2013). A good example 3 Referred to as “Voodoo” in Anglophone countries. 13 of this syncretism is the Celestial Church of Christ (categorized as “other Christian”) which is a church founded in Benin that incorporates Christian beliefs with Traditional practices and recognizes polygynous marriage (Falen 2008a). However, unlike ethnicity, there is a gendered component to religious identification in Benin, with women being more likely to identify as adherents of Christian religions than men who often favor identifying with Traditional beliefs (Falen 2008a). Geographic clustering in Benin Geographically, Benin is comprised of 12 departments of unequal size (Appendix 2 Figure 3). The four northern departments 4 are significantly larger than the eight southern departments. The northern departments comprise 29% of the Beninese population and the smaller southern departments are more densely populated, making up 71% of the Beninese population. Ethnic groups in Benin are largely geographically clustered within specific departments, with the north being more ethnically diverse than the south. Similarly, religious affiliation is also highly clustered due to its correlation with ethnicity: the south has strong ties to Traditional beliefs and Catholicism while the north is predominantly Muslim, rendering the south more religiously diverse than the north (Adotevi et al. 2017). Figure 2.2 illustrates the ethnic fractionalization of each department in Benin using a Herfindahl-Hirschman index 5 . This index measures the likelihood that two randomly selected people would come from different ethnic backgrounds. In Benin, there is a 75% likelihood that any two randomly selected Beninese individuals would come from different ethnic groups. The most diverse department is the Borgou, with a 74% likelihood of selecting two individuals from 4 The four northern departments are the Alibori, Atacora, Borgou, and Donga. 5 𝐻= 𝑠 ! ! ! !!! 14 different ethnic backgrounds and the Zou is the least diverse, with only a 9% likelihood of selecting two individuals from different backgrounds. Littoral, home to Cotonou, the economic capital and largest city, has a 54% likelihood of selecting individuals of different ethnic backgrounds. Figure 2.2: Percent of ethnic fractionalization in each department of Benin, measured using a Herfindahl-Hirschman index. Note: Uses data from BDHS 2006, 2011-2012, and 2017-2018 due to changes in Benin’s departments. Map source: The Demographic and Health Surveys Program’s Spatial Data Repository. Similarly, Figure 2.3 illustrates the religious fractionalization of Benin. In Benin, there is a 79% chance that any two randomly drawn Beninese individuals would identify themselves as adherents of different religions. The most religiously diverse departments are Plateau and the Atacora Alibori Alibori Donga Borgou Collines Zou Mono Couffo Atlantique Ouémé Plateau Littoral <30% 30-59% 60%+ 15 Atacora, where there is an 80% likelihood of selecting two people of differing religions. The least religiously diverse department is the Alibori, where there is a 28% chance of selecting two people of different religions. Figure 2.3: Percent of religious fractionalization in each department of Benin, measured using a Herfindahl-Hirschman index. Note: Uses data from BDHS 2006, 2011-2012, and 2017-2018 due to changes in Benin’s departments. Map source: The Demographic and Health Surveys Program’s Spatial Data Repository. Power and resources in the country are not evenly distributed amongst ethnic and religious groups. The Fon, Yoruba, and Bariba ethnic groups are the most politically dominant (Adida 2015; Adotevi et al. 2017). Additionally, many institutional resources, such as education, favor the southern region of the country, due to the legacy of Catholic and, to a lesser extent Protestant, 16 missions that prioritized educating and converting people of Traditional religions rather than expending resources on the Muslim north (Adotevi et al. 2017; Nunn 2010). This colonial legacy rendered northern, predominantly Muslim, ethnic groups, on average, poorer and less formally educated than southern ethnic groups that adopted Christianity, such the Fon and Yoruba (Adotevi et al. 2017; Nunn 2010). Due to the lack of formal education in the north, fewer northerners than southerners are literate in French (Kouta 2016). Gender, marriage, and family life in Benin Life in Benin, as in all other countries in the world, is shaped by gender. According to the United Nations Development Programme’s Gender Inequality Index, Benin is one of the most gender unequal countries in the world and it is more unequal than the average sub-Saharan country (United Nations Development Programme 2018). This inequality begins in childhood and is particularly marked in education, although educational inequality has declined over time. Women born in the 1940s attended an average of 0.6 years of school as compared to an average of 4.8 for women born in the 2000s. Much of this low educational attainment is because 88% of women born in the 1940s never attended school as compared to only 28% of women born in the 2000s (author’s calculations). Girls’ education is markedly lower than boys’ in Benin despite overall increases in access to education for both (Tabutin and Schoumaker 2004). For example, of the population 25 years and older in 2010- 2017, only 18% of women had attended secondary school versus 33% of men (United Nations Development Programme 2018). This is due to several factors: girls are significantly less likely to be enrolled in primary school (88% versus 100% of boys) and the girls who do attend primary school are slightly less likely to make it through than boys (97% versus 99%). Further, girls are less likely to be subsequently enrolled in lower secondary school (41% versus 49% for boys). 17 After lower secondary school, there is a precipitous drop in enrollment for girls because schooling is no longer free for them (Adida et al. 2018). Since parents often cannot afford to pay for all of their children’s fees, they prioritize sending their sons (Adida et al. 2018). Contrary to girls who attend school for free through 9 th grade (lower secondary) and only pay upon entering 10 th grade (higher secondary), there is a significant but lesser drop in school enrollment for boys at the entry of lower secondary when they start paying fees after completing primary school (Adida et al. 2018). Relatedly, marriage 6 in Benin occurs at a young age for women and is universal. Between 2010 and 2016, approximately 7% of Beninese girls married before the age of 15 and 26% married by the age of 18 (United Nations Children’s Fund 2017). However, the average age at first marriage is slowly rising as education for women increases: for marriages occurring in the early 1990s, women were aged 18.9 years on average, as compared to marriages occurring in the 2010s, where women were approximately 20.5 years old (author’s calculation). This contrasts with men whose first marriage occurred, on average, at approximately 25 years of age regardless of the year married between 1991 and 2018 (author’s calculations). Marriage and partner selection have undergone large shifts in Benin. Marriages are largely considered a merging of families rather than the union of solely two people although this view is 6 As in much of sub-Saharan Africa, marriages in Benin do not always have clear start dates and marriage is often more of a process than a singular event (Meekers 1992). Many couples consider themselves married if they cohabit, regardless of whether there was a wedding ceremony, in part because cohabiting prior to a wedding ceremony allows them to avoid or delay the ceremonial expenses required, such as bride prices (Calvès 2016a; Gnoumou Thiombiasno 2017; Otite 1991). In much of West Africa, there are three types of wedding ceremonies: religious ceremonies, civil ceremonies involving the government, and “traditional” ceremonies which are ethnically specific (Gnoumou Thiombiano 2017; Otite 1991). Some couples may never have a ceremony and some may have all three. These may occur before and/or after cohabitation. 18 decreasingly common in urban areas (Falen 2011). Parents used to select their children’s marital partner, including betrothing girls as babies, but this custom is becoming unfavorable to both men and women (Falen 2011). Instead, parents now frequently go through the motions of selecting their children’s spouse with the man’s family contacting the family of their son’s partner of choice to arrange the couple formally (Falen 2011). Polygyny is common in Benin, although the frequency has decreased over time. Among all marriages in 1996, 33% of men and 50% of women were married polygynously in contrast to 2018 when 22% of men and 38% of women were polygynously married (INSAE and ICF International 2013; Ministère de la Santé du Bénin/INSAE 1996). Polygyny used to be considered favorably in Benin by both husbands and wives, but this view has largely fallen out of favor (Falen 2008a, 2011). Generally, men view polygyny more favorably than women and, as a result, men are less likely to identify as Catholic or Christian in Benin due to these religions’ opposition to polygyny (Falen 2008a, 2011). As in much of sub-Saharan Africa, parenthood is central to adulthood and the legitimization of marriages in Benin, with infertility being grounds to end a union (Falen 2011). Although Benin began to have noticeable fertility declines starting in the 1980s for urban areas and the 1990s for rural areas (Garenne and Joseph 2002), fertility declines have largely stalled. The total fertility rate (TFR) for women dropped to its lowest in 2011-2012 at 4.9 children per woman before rebounding to 2006 levels in 2018 for a TFR of 5.9 (INSAE and ICF International 2013, 2019). Women tend to give birth for the first time at young ages in Benin, with 20% of teens 15- 19 being mothers in 2018 (The World Bank 2019). The median age at first birth for women aged 25-29 has increased by nearly a year between 1996 and 2018 from 19.6 years of age to 20.5, indicating that marriage and first birth are closely aligned (INSAE and ICF International 2019). 19 Women in Benin also experience high levels of unintended pregnancies at 26% of births, although this is below the 29% average for sub-Saharan Africa (Kwabena et al. 2019). This is in part due to low levels of contraceptive usage among married women in Benin, who cite side effects, health concerns, and their husbands’ opposition as the key reasons for not using contraception (Guttmacher Institute 2015). Compared to other regions of Africa, West African countries, including Benin, are unique in that married women historically earn their own money and maintain their own budgets separate from their husband’s — including being responsible for various household costs independently from their husband, particularly related to their children (Caldwell 1996; Duflo and Udry 2004). These distinct household budgets may be attributable to historic and current high levels of polygyny that centered the family unit around a mother and her children (Caldwell 1996). While the gender work gap is small in Benin (Nordman, Robilliard, and Roubaud 2011; United Nations Development Programme 2018), there is a large gender income gap because of the coding of certain lower-paid occupations and crops as suitable for women rather than men (Caldwell 1996; Duflo and Udry 2004; Nordman et al. 2011). Approximately 13% of households in Benin are headed by women 7 and these female headed- households are less likely to live in poverty, on average, than male-headed households — 28% versus 38%, respectively (Attanasso 2005; The World Bank 2019). However, poor male-headed households have more resources, on average, than poor female-headed households due to men’s ability to access credit and other assets more easily than women (Attanasso 2005). The variation in the poverty female-headed households experience is linked to ethnicity — women-headed 7 These are households where women identified themselves as the head of the household rather than naming a man. This category likely includes many non-remarried widows or households where the husband moved out, akin to desertion. 20 households from ethnic groups that allow women to engage more heavily in agriculture are less likely to experience poverty than women-headed households from groups that curtail women’s agricultural activity (Attanasso 2005). Historically, marriages in West Africa have been considered less stable than marriages in other regions of sub-Saharan Africa due to women retaining strong connections to their family of origin and relative financial independence from their husbands (Caldwell 1996; Falen 2011). However, a recent study shows that compared to many sub-Saharan African countries, Benin has a low divorce rate and is one of six countries with a divorce rate that appears to be declining (Clark and Brauner-Otto 2015). Eighty-six percent of women are in their first unions and only an estimated 14% of unions end in divorce after 15-19 years of marriage as compared to 25% for the region (Clark and Brauner-Otto 2015). Early widowhood is also uncommon in Benin, with less than 4% of first unions ending due to the death of a spouse after 15-19 years of marriage (Clark and Brauner-Otto 2015). In the event of union dissolution, Beninese women usually remarry. In the BDHS women’s surveys, 70% of women who experienced a union dissolution are currently remarried (author’s calculations). Of the women who have not remarried, 45% report that their husband moved out and 38% are widowed (author’s calculations). Macro-context of Benin Benin is home to approximately 11 million people and is undergoing the demographic changes that often correlate with shifts in partner selection. Benin’s population has nearly quintupled from 1960 to 2017 due to a high TFR that has largely stalled — peaking at seven in the 1980s and lowering to nearly six in 2018 (INSAE and ICF International 2019; The World Bank 2019). This population growth is reflected in its triangularly shaped population pyramid (Appendix 2 21 Figure 4). Benin has also urbanized rapidly: in 1960, 9% of the population was urban as compared to 47% in 2017 (The World Bank 2019). Benin is a low-income country and its economy relies heavily on agriculture with its main export being cotton (The World Bank 2019). More than 40% of the population worked in agriculture in 2018 — including nearly half of men (49%) and over one-third (34%) of women (The World Bank 2019). Benin had a gross domestic income per capita 8 of $870 in 2018 (The World Bank 2019). In 2016, 43% of the population lived on less than $1.90 a day and a full 89% lived on less than $5.50 a day (The World Bank 2019). Benin became independent from France in 1960 and retains important legacies of colonization. French is the official language of Benin and its education system is based on the French school system (Adotevi et al. 2017; Houngnikpo and Decalo 2013). Benin struggles to balance the use of French as their official language and the local languages learned at home. Benin has tested and promoted the use of local languages in primary schools at several points in its history, although they have never successfully implemented it nationally (Baba-Moussa 2013). After independence, Benin’s government was unstable and the country experienced a military regime under the guise of Marxism until 1990 (Adotevi et al. 2017; Houngnikpo and Decalo 2013). Benin’s current government legal code is modeled on the French governmental system and includes democratic elections (Adotevi et al. 2017; Houngnikpo and Decalo 2013). 8 Atlas method using 2018 United States Dollars. 22 Data The Demographic and Health Surveys (DHS) started in the early 1980s as a follow-up survey to the World Fertility Surveys and Contraceptive Prevalence Survey. Spearheaded by the United States Agency for International Development (USAID), these surveys were conducted in the late 1970s and the early 1980s (Croft, Marshall, and Allen 2018). The goal of these early surveys was to make population data — particularly related to fertility — readily available and comparable between countries (Sprehe 1974). They covered topics such as contraceptive usage, birth histories, and other topics focused on reproductive health (Sprehe 1974). The DHS was launched in 1984 as an expansion and replacement to these surveys by including data not only on fertility but health and nutrition as well (Boerma and Sommerfelt 1993). Demographic and Health Surveys have been conducted in over 90 countries across Asia, Africa, Europe, Latin America, and Oceania (Croft et al. 2018). The DHS uses standard questionnaires across countries to allow for comparability across countries. These surveys are focused predominately on reproductive- aged women (15-49 years old), asking about their household and socioeconomic characteristics, as well as their reproductive careers and contraceptive usage. The information gathered from these surveys is used for monitoring and evaluation purposes in the fields of population, health, and nutrition. The surveys are designed to be nationally representative, using a stratified cluster design. The DHS uses existing sampling frames, usually a country’s census, to cover the target population of women in their reproductive years, children under five, and, in most surveys, men aged 15-59 (ICF International 2012). The DHS uses a two-stage clustering sample using (1) enumeration areas that usually match clustering areas of the census, which are selected for survey based on a probability proportional to population size, followed by (2) a complete household listing of the 23 selected clusters, after which households are selected for survey based on equal probability. This ensures that the surveyed houses are randomly selected (ICF International 2012). They also use a multilevel stratification design in which the population is stratified at the regional level, as well as by urbanicity within regions (ICF International 2012). This study uses five DHS datasets collected in Benin in five-year intervals between 1996 and 2018. The BDHS relies on Benin’s most recent census data (1992 for 1996 and 2013 for 2018) to determine enumeration areas and they use Benin’s departments as their regional stratification before determining urban and rural stratifications. The BDHS has high response rates ranging from 97% to 99% of households in 2001 and 2006 respectively. Similarly, response rates are high for women: ranging from 94% in 2006 to 98% in 2018. Men’s responses are generally lower than for women and households, with 91% of men answering surveys in 2006 and a high of 98% responding in 2018. In this study, I use the BDHS’s couples’ surveys, which include information on both partners’ self-reported ethnicities and religious affiliations, allowing me to identify interethnic and interreligious marriages. The DHS interviews available men and women in each household and matches the independent surveys of men and women partnered with each other, creating the couples’ data. This results in a dataset of self-reported responses for couples; we are not relying on one person’s recall of their partner’s background characteristics. The surveys contain a variety of socio-demographic indicators such as educational attainment and urbanicity, as well as self- reported marital information such as the year of marriage and the number of spouses. While the data the DHS collect are nationally representative at the household level, the couples’ data are more select as male partners tend to be more difficult to find and interview (Chae 2016; Dodoo 1993). For example, only male partners who were available and present in 24 the sampled households were interviewed meaning marriages with non-residential husbands — who are more likely to be labor migrants or polygynous — may be underrepresented in the sample. As a result, I restrict the couples’ sample to women who report living with their husbands (99%). Further, due to data limitations, I restrict the couples’ sample to women in first unions (87%). Questions about marital timing (i.e. age, year, duration, etc.) are only asked of first marriages and recall about event timing is more consistent for first marriages than later unions (Chae 2016). Restricting to continuously married women means that the marital lives of remarried widowed and divorced women are not represented in these findings 9 . Table 2.1 shows how the sample used in these studies compares to all married women sampled in the individual DHS, as well as those in their first marriages. 9 This restriction means that cohesive couples are overrepresented in these data and intraethnic and intrareligious couples are likely overrepresented, as well. 25 Table 2.1: A comparison of married women’s characteristics in the full DHS sample, first marriages in the full DHS, and first marriages in the couples’ data (Survey years: 1996, 2001, 2006, 2011, and 2018) Full DHS: All married women 1 Full DHS: All first marriages 1 Couples’ data: All first marriages Interethnic (%) n/a n/a 14.12 Interreligious (%) n/a n/a 27.34 Average years educated 1.57 1.67 1.50 (3.14) (3.27) (3.14) Household Wealth Quintile (%) Poorest 20.83 20.88 24.17 Poor 20.56 20.44 21.50 Middle 20.45 20.38 19.36 Rich 20.15 19.79 17.96 Richest 18.01 18.50 17.01 Region (%) North 33.54 37.41 37.13 South 66.46 62.59 62.87 Current Age 31.13 30.76 30.44 (8.19) (8.27) (7.93) Polygynous (%) 42.12 39.25 37.01 Urban (%) 36.47 37.41 35.56 Resides with Husband (%) 84.95 86.17 99.00 N 55561 48728 10935 Source: The Demographic and Health Surveys Program *Standard deviations in parentheses 1 Author’s calculations from women’s individual surveys Additionally, I do not include couples with non-Beninese spouses (0.9% of men; and 1.2% of women), as well as men and women who report their ethnicity as “other” (2% each) in regression models because they cannot be coded as intraethnically married. Contrarily, I do not do the same for the “other Christian” or “other religion” categories of religious identification because doing so would predominantly remove members of African Initiated Churches and thus an important and identifiable slice of religious life in Benin. As a result of the DHS coding ethnic and religious groups as “other,” interethnic and interreligious marriages may be undercounted. I also include only men and women with complete, numeric, responses on all control variables, such as 26 educational attainment (99.9% men; 100% women) and the number of other wives (98.9% women). In total 3,059 couples are not included in these analyses and my final sample consists of over 9,000 couples. 27 CHAPTER 3 INTERETHNIC AND INTERRELIGIOUS MARRIAGE IN BENIN: INDIVIDUAL- AND COUPLE-LEVEL TRENDS AND PATTERNS Marriage is a central element of adulthood in West Africa, like in other world regions (Hollos and Larsen 2008; Smith 2001). Marriage occurs early in the life course and is near universal in much of the region. However, social and economic changes are increasingly fostering more diverse family formation patterns (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Clark and Brauner-Otto 2015; Clark et al. 2017; Koski et al. 2017; Maïga and Baya 2011). Increased education, urbanization, and internal migration, among other major changes, are altering the patterns by which young people are coupling, including raising the age at first marriage and decreasing the rates of formal unions (Bove and Valeggia 2009; Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Clark et al. 2017; Gnoumou Thiombiano and Legrand 2014; Koski et al. 2017). One distinct way in which broader social changes may be altering the patterning of unions in West Africa is by increasing contact among social groups, diversifying social networks that have traditionally been ethnically and religiously homogenous (Caldwell and Caldwell 1987; Crespin- Boucaud 2018; Falen 2008a; Gage 1995; Locoh and Thiriat 1995; Maïga and Baya 2011). This increased network diversity is potentially altering the frequency of interethnic and interreligious marriages (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). Specifically, with high levels of internal migration away from one’s natal community, increasing urbanicity, and high levels of education, the role of group affiliation is changing in partner selection in certain contexts (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). 28 Despite West Africa experiencing the rapid social changes linked with changes in interethnic and interreligious marriages in other contexts (Birkelund and Heldal 2003; Furtado 2012; Rosenfeld and Thomas 2012; Smits, Ultee, and Lammers 1998), we know relatively little about interethnic and interreligious marriage in the region. The few studies focusing on interethnic marriage examine West African capital cities in Burkina Faso (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Gnoumou Thiombiano 2017) and Togo (Gnoumou Thiombiano 2017). These studies indicate that broader social changes are likely linked to increases in the frequency of interethnic coupling in the region. Similarly, there are few studies measuring changes in the rates of interreligious marriage in West Africa. However, the few that do study interreligious marriage at the country-level find that, compared to other regions of sub-Saharan Africa, West Africa has low levels of interreligious marriage (Crespin-Boucaud 2018; Ntoimo and Mutanda 2017). In contrast to the literature on interethnic marriage in sub-Saharan Africa, the scarce literature on interreligious marriage in sub-Saharan Africa shows a decline in the rates of interreligious marriage, particularly in West Africa (Crespin-Boucaud 2018). Part of this decline may be due to a decrease in people identifying as adherents of Traditional religions and a corresponding increase in identifying with monotheistic religions (Crespin-Boucaud 2018) but part of the decline may also be due to educated people increasingly selecting their partners along religious lines (Maïga and Baya 2011). However, many of the studies of interethnic and interreligious couples in West Africa have intensive designs focused on urban localities or are focused on limited timeframes, raising questions of broader spatial and temporal trends (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Gnoumou Thiombiano 2017). 29 In this study, I examine the temporal and spatial location of interethnic and interreligious couples, as well as how their characteristics differ from intraethnic and intrareligious couples, focusing on Benin, West Africa as my case study. Using five Demographic and Health Surveys, I utilize data on couples in first marriages spanning over five decades. This study shows that interethnic couples are increasingly common, are largely urban but increasingly rural, and exhibit higher levels of human and economic capital than people who are intraethnically married. In contrast, the findings for interreligious marriages indicate that although they are increasing, it is a very modest increase and that these marriages are not tied to urbanicity or educational attainment. Rather, many of the socioeconomic and demographic characteristics of interreligious couples mirror those of their intrareligiously married counterparts. These findings support previous research demonstrating that assortative marriage patterns correlate with broader social and economic changes, although these changes are not equal across social categories, thus indicating that interethnic and interreligious marriages are distinct phenomena. Background The process of selecting marital partners is both varied and changing in sub-Saharan Africa. In some settings, including Benin, families traditionally arranged marriages for their children (Caldwell 1996; Caldwell, Orubuloye, and Caldwell 1991; Calvès 2016a; Falen 2011; Gage 1995; Meekers and Calvès 1997; Mensch, Grant, and Blanc 2006; Smith 2001) and the pace of decline of arranged marriages has varied by ethnicity and religion (Calvès 2016b; Gage 1995; Meekers 1995). Parents’ role in partner selection helped ensure a “suitable” match for their children, the criteria of which usually included a partner of the same ethnic and religious background (Meekers, 1995). However, changing economic, social, and institutional landscapes altered traditional partner selection. Choosing one’s own partner while receiving familial consent 30 became the most common method, although it continues to be less common among Muslims than among other religious groups (Calvès and Gnoumou Thiombiano 2014; Falen 2011; Gage 1995; Meekers 1995). Similar to Western countries, receiving familial consent in the West African context protects familial goodwill and ensures the extended family’s support in the case of marital problems and dissolution (Falen 2011; Meekers 1995). In some places, this change in partner selection from parental choice to parental involvement opened up possibilities for diverse couplings, including altering the prioritization of ethnic and religious homogamy in partner selection. Relative to other contexts, West Africa is distinct for its historical acceptance of intermarriages (Goody 1971). Nevertheless, the majority of marriages in the region are intraethnic and/or intrareligious (Calvès, 2016; Gnoumou Thiombiano, 2017; Locoh & Thiriat, 1995; Ntoimo & Mutanda, 2017). Ethnicity and religion remain salient in partner selection for several reasons, including their important role in shaping norms and networks (Calvès and Gnoumou Thiombiano 2014; Falen 2008a; Locoh and Thiriat 1995; Ntoimo and Mutanda 2017; Smith 2011; Takyi 2001). Throughout West Africa, including Benin, ethnic groups often have distinct languages (Buzasi 2015, 2016; Fearon 2003; Posner 2004) and both ethnic groups and religious groups tend to be geographically clustered, minimizing contact with people of other backgrounds (Crespin-Boucaud 2018; Smith 2011). Further, despite their reduced influence, extended families continue to inform individuals’ choice of marital partners (Bertrand-Dansereau and Clark 2016; Calvès 2016a; Calvès and Meekers 1999; Smith 2011). Thus, people frequently pick spouses who come from areas near their extended family and within the same ethnic and religious communities (Calvès 2016b; Maïga and Baya 2011; Smith 2011). 31 However, broader social changes are altering partnering patterns in sub-Saharan Africa, potentially changing the likelihood that people will select partners from a different ethnic or religious background (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). Sub-Saharan Africa is rapidly urbanizing due to rural-to-urban migration, creating ethnically and religiously diverse cities. This migration away from natal communities also decreases the traditional influence of (rural) extended kin on partner selection (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011, 2014). Further, increased schooling, particularly in urban areas, is altering partner choice through changing networks and diversifying ways of meeting potential partners (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). This confluence of new social networks and the decreasing influence of extended family in partner selection makes it more likely for people to couple with partners of different social backgrounds (Calvès 2016a, 2016b; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). On the other hand, there is evidence that these same social changes may be differentially tied to the likelihood of marrying interethnically versus interreligiously. One study found that interethnic couples are increasing in frequency in at least half of sub-Saharan countries whereas interreligious marriages are decreasing in frequency in many (Crespin-Boucaud 2018). One potential explanation for this is the relative ease of forming social networks along religious lines versus ethnic lines in cities as people can easily locate and identify houses of worship (Crespin- Boucaud 2018). Similarly, religiously-based networks may be larger in cities, offering more resources than smaller ethnically-based networks (Crespin-Boucaud 2018). Another potential explanation is that people of high educational attainment may prioritize shared religious affiliation over ethnic homogamy (Maïga and Baya 2011). If there is increasing network 32 homogamy along religious lines with diversification along ethnic lines, interethnic and interreligious marriages may move in opposite directions (Crespin-Boucaud 2018). Given the evidence that shifting institutional and social landscapes stand to alter the frequency of both interethnic and interreligious marriages, this paper examines the individual- and couple-level characteristics of the men and women who marry interethnically and interreligiously in Benin. Benin is a context where ethnicity and religion shape daily life (Adotevi et al. 2017; Houngnikpo and Decalo 2013). Benin is one of the sub-Saharan African nations that did not experience increasing levels of interethnic marriage between 1960 and 1980, suggesting that ethnicity remains particularly salient in the Beninese context (Crespin-Boucaud 2018). Contrarily, interreligious marriages decreased during that same time point suggesting that religion’s salience in partner selection may be increasing in Benin (Crespin-Boucaud 2018), despite prior research finding that different religious backgrounds are not considered a hindrance to marriages in Benin (Falen 2011). I use Benin as a case study to investigate the temporal and spatial distribution of interethnic and interreligious marriages, as well as the social and economic characteristics that are associated with marrying across social groups. Measures Dependent variables To measure intermarriage, I use two salient group identifications: ethnicity and religion. By using men’s and women’s stated identifications, the DHS captures people’s predominant cultural identifiers at the time of survey. There are no formal data on whether women become more likely to adopt their husband’s ethnic and/or religious identification over time and the DHS does not ask about ethnic and religious histories. As such, these measures may underestimate intermarriages, particularly for couples who have been married for longer durations (Crespin- 33 Boucaud 2018). However, the couples who were intermarried at the time of marriage but whose ethnic and/or religious identifications have merged are likely distinct from couples who continue to identify with different social groups. Interethnic marriage. Interviewers asked all individuals “What is your ethnicity?”. BDHS interviewers required responses in nine groups (Adja, Bariba, Betamaribé, Dendi, Fon, other, Peuhl, Yoa/Lokpa, or Yoruba). Interethnic marriages (=1) are coded as those whose ethnicity does not match their spouse’s. I remove people identified as “other” from statistics and models because they cannot be identified as intraethnically married. This exclusion leads to fewer Betamaribé interethnic couples in the full models due to people from the “other” category being their main choice of interethnic marital partners, as well as fewer Dendis, as people from the “other” category are most likely to intermarry with someone who is Dendi. Interreligious Marriage. The DHS requires a single response to religious affiliation. Therefore, it captures interviewees’ primary religious identification, rather than discrete religious categories. Interviewers asked all individuals “What religion do you practice?”. To capture religiously discordant marriages, I code religious affiliation as one of seven categories: Catholic, Protestant, Muslim, Traditional beliefs, other, no religion, and other Christian denominations, which are the religious categories used in the 1996 and 2001 surveys. I then code couples as interreligiously married (=1) if the spouses’ religious affiliations do not match. Key individual-level independent variables Ethnicity of wife and husband. The likelihood of entering an interethnic marriage is often gendered and variable by the population of ethnic groups. To examine gendered marriage markets, I examine the ethnicity of men and women separately. 34 Religion of wife and husband. Similarly, the likelihood of entering an interreligious marriage is often gendered and variable by the population of religious groups. To examine gendered marriage markets, I examine the religion of men and women separately. Age at marriage. A woman’s age at first marriage shapes various outcomes including her total parity, the age of her partner, and her educational attainment. Age at first marriage is continuously measured. I also measure men’s age at first marriage continuously. Educational attainment. I measure the absolute number of years of schooling attended individually for both husbands and wives. I also measure the proportion having ever attended school, as well as literacy rates due to overall low school attendance in Benin, particularly among girls. Ethnic composition of interethnic marriages. To understand the frequency of inter- and intraethnic marriage among Benin’s different ethnic groups, I sort all matches by the husbands’ and wives’ ethnicities to identify patterns for which ethnic groups marry out and, among those who are interethnically married, which ethnic groups marry each other. Religious composition of interreligious marriages. To understand the frequency of inter- and intrareligious marriage among Benin’s different religious groups, I sort all matches by the husbands’ and wives’ religions to identify patterns for which religious groups marry out and, among those who are interreligiously married, which religious groups marry each other. Premarital pregnancy. A key variable of interest is whether a premarital pregnancy precipitated the intermarriage. In sub-Saharan Africa, premarital pregnancy may lead couples to marry to have a “legitimate” birth. While premarital births are declining in Benin, over one-fifth of women gave birth less than eight months after transitioning into their first union (author’s calculations), indicating that premarital pregnancy remains common. Further, unwed women 35 who are already pregnant or mothers often have different marriage market experiences than women without children or visible pregnancies (Smith-Greenaway and Clark 2018). As a result, women with premarital pregnancies may intermarry at different frequencies than women without premarital pregnancies, as they may be considered less attractive on the marriage market. Therefore, I ask if women are more likely to intermarry if they experience a premarital pregnancy compared to women who do not experience a premarital pregnancy. The survey measures the interval from marriage to first birth by comparing the date of marriage to the date of first birth. Births that occurred before or within seven months of marriage are coded as premarital pregnancies (=1) while babies born anytime starting at eight months after marriage are coded as marital pregnancies (=0). The DHS does not inquire about the male partner involved in the pregnancy meaning the father of the child may or may not be the woman’s current spouse. Key couple-level independent variables Age heterogamy. The average age gap between husbands and wives is largest in sub-Saharan Africa. Age heterogamy is correlated with earlier marital age for women, lower contraceptive usage, and higher parity (Barbieri and Hertrich 2005; Takyi 2001; Togunde 1998). African men overwhelmingly marry at later ages than their female counterparts (88% in the couples’ data) and only 5% of women are older than their husbands. I code age homogamy (=0) as partners who are within five years of each other. When the husband is older, I code age heterogamy as couples where there is a five- to nine-year age gap between spouses (=1) and an age gap of ten or more years (=2). Due to the limited numbers of women five or more years older than their husbands (N=86 or 0.9%), I create only one grouping for women who are older than their husbands (=3). Due to the small size of this last category, results should be interpreted with caution. 36 Educational discordance. Due to the large discrepancy in access to schooling for boys and girls, many more women report zero years of educational attainment (76%) than men in the sample (54%). To help account for the large number of Beninese people who have never attended school, I measure the educational discordance of couples with a categorical variable coded as neither wife nor husband are educated (=0), wife and husband are educated (=1), husband is educated but wife is not (=2), and wife is educated but husband is not (=3). Due to households prioritizing sending their sons to school over their daughters, few women (N=461 or 4.6%) have been to school when their husbands have not, meaning the results of this group should be interpreted with caution. However, this is an interesting category to study as education shapes women’s experience on the marriage market and women with higher levels of education may be more likely to have to seek a partner of a different social background (Hannemann et al. 2018; Hannemann and Kulu 2015). Household wealth. I am interested in current household wealth as intermarried households may have different wealth levels than their intramarried counterparts. To measure household wealth, I use the wealth index score, which is a composite measure of a household’s living standard based on weighting assets and services available in the household, such as water sources and materials used to construct the home (Rutstein and Johnson 2004). It is not a measure of income but resources available to all household members (Rutstein and Johnson 2004). As wealth is measured at the time of the survey, it is not reflective of the wealth of couples at the time of their marriage. This means their wealth may have changed between marriage and the time of survey. It, therefore, has limitations in its use as a predictor of intermarriage, but it remains useful in understanding the characteristics associated with intermarriages. 37 Location of residence. To better understand where interethnic and interreligious couples reside, I measure location of residence in three ways: rural (=0) or urban (=1), northern (=1) or southern (1), and by administrative region to account for differences in the distribution of ethnicity, religious affiliation, and resources in Benin (Adotevi et al. 2017; Houngnikpo and Decalo 2013). Similar to wealth, because location is measured at the time of survey rather than the time of marriage, it is limited as a predictor of intermarriage but contributes to our understanding of the broader characteristics of these couples. Decade married. I am interested in how the correlates of intermarriage have changed over time. To track this, I make use of data on women’s date of marriage to create a categorical indicator of decade married ranging from pre-1980 to 2010 or later based on their response to the following question: “In which month and year did you start to live with your husband/partner?” Analytic Approach To examine the differences between interethnic and interreligious couples and their intramarried counterparts, I compare the temporal, spatial and social characteristics that distinguish these couples using descriptive statistics, as well as binary and robustly controlled multivariate logistic regression. The individual-level models include binary results of the likelihood of being intermarried, as well as the likelihood of being intermarried once all individual-level characteristics are included in a multivariate model. For the couple-level models, the first multivariate model I offer includes only characteristics that are likely unchanged since the time of marriage to try and capture the characteristics that are associated with selection into interethnic or interreligious marriages. I subsequently include wealth and urbanicity to better understand these marriages at the time of survey in a second multivariate model. 38 Results Changes in interethnic and interreligious marriage over time Nine percent of couples in the 1996 survey wave are interethnically married as compared to 20% in 2018 indicating a large growth in the frequency of interethnic marriage. In all models, compared to marriages occurring before the 1980s, all decades are associated with increased odds of interethnic marriage, but significance is only consistently reached for marriages occurring in the 2000s. In contrast to the large increase in interethnic marriage, in 1996, 29% of couples surveyed are interreligiously married and this increases to 31% of the sample in 2018, suggesting only a modest increase in interreligious marriage. Being married in the 2010s is related to a significant increase in the likelihood of being interreligiously married in all logistic models, with no other decade reaching significance. There are also differences in intermarriage between urban and rural locations over time. Overall, interethnic marriages comprise 19% of urban couples sampled and 7% of all rural couples. Restricting to recent marriages — those occurring within five years of the survey (1991- 2018) — 17% of urban couples married in the 1990s are interethnic and this rises to 24% among those married in the 2010s. In rural areas, only 6% of couples married in the 1990s are interethnic however, this increases to 10% for couples married in the 2010s. Even when expanding to couples whose marriages are not recent, interethnic marriages make up at least 15% of urban couples regardless of which decade they married — 18% among those married before 1980 and 22% among couples married in the 2010s. Similarly, the percentage of rural interethnic couples peaks at 10% for those married in the 2000s and is as low as 4% for couples married before 1980. 39 However, this association between urbanicity and interethnic marriage has changed over time: interaction effects in the full multivariate model between urbanicity and decade married show that urbanicity plateaus as a predictor for interethnic marriages occurring since the 1980s (Figure 3.1). Contrarily, there is an increasing likelihood of rural couples being interethnic. These results indicate that the rise in interethnic marriage is occurring more steeply in rural areas than urban areas but that interethnic couples comprise a larger portion of the urban social milieu than the rural one. Figure 3.1: The probability of urban and rural couples being interethnic by decade of marriage Source: The Demographic and Health Surveys Program N=9828 In contrast, there are few changes in the urbanicity of interreligious couples over time. In the total sample, interreligious couples represent 28% of the urban population and 27% of the rural population and these results are similar for recent marriages: interreligious marriages comprise 30% of urban recent marriages and 28% in the rural milieu. Nearly 31% of interreligious couples married before the 1980s reside in an urban location compared to 44% of couples married in the 0 .05 .1 .15 .2 .25 Probability of interethnic marriage Pre-1980 1980s 1990s 2000s 2010s Decade of marriage Urban Rural 40 2010s. Among recent marriages (1991-2018), 34% of interreligious couples reside in urban areas and this increases to 45% of marriages occurring in the 2010s, although these trends map on well to increases in urbanicity for the general population in the same period (33% to 43%). Figure 3.2 illustrates that urban and rural marriages were equally likely to be interreligious for several marital cohorts but their likelihood of being rural has largely plateaued, while ties to urbanicity are increasing, although this does not reach significance. These results indicate that, unlike interethnic marriages, interreligious marriage is not strongly tied to urbanicity, but is equally prevalent in rural settings. Figure 3.2: The probability of urban and rural couples being interreligious by decade of marriage Source: The Demographic and Health Surveys Program N=9828 Spatial distribution of interethnic and interreligious marriage Figure 3.3 presents the percentage of interethnic marriage in each department, which maps well onto Figure 2.2 in that less ethnically diverse departments tend to have lower levels of interethnic marriage. Similarly, Figure 3.4 presents the level of interreligious marriage in each department, .2 .25 .3 .35 .4 Probability of interreligious marriage Pre-1980 1980s 1990s 2000s 2010s Decade of Marriage Urban Rural 41 which also maps on fairly well to Figure 2.3, in that less religiously diverse departments of Benin tend to have lower levels of interreligious marriage. Figure 3.3: Percent of marriages that are interethnic by department Note: Uses data from BDHS 2006, 2011-2012, and 2017-2018 due to changes in Benin’s departments. Map source: The Demographic and Health Surveys Program’s Spatial Data Repository. 42 Figure 3.4: Percent of marriages that are interreligious by department Note: Uses data from BDHS 2006, 2011-2012, and 2017-2018 due to changes in Benin’s departments. Map source: The Demographic and Health Surveys Program’s Spatial Data Repository. Importantly, many of the departments with high levels of interethnic marriage experience low levels of interreligious marriage, and the reverse is also true, suggesting that intermarriage is tied to the type of diversity common in the region. For example, the Couffo and Zou in the south are the two most ethnically homogenous departments and they have the lowest shares of interethnic marriage. In contrast, the Donga and Littoral have the highest percentages of interethnic marriage at 37% each. The Donga is an ethnically diverse department and is home to one of the north’s largest cities (Djougou). Littoral is comprised exclusively of Cotonou, the country’s largest city, and is likely reflective of interethnic marriage’s correlation with 43 urbanicity. In the binary model, urban couples are more than three times more likely to report being interethnically married, however, this is attenuated in the full multivariate model, as urban couples are only 78% more likely to report being interethnic than rural couples controlling for other characteristics. Similarly, three out of four northern departments of Benin — the Alibori, Borgou, and Donga — are the departments in which interreligious marriage is least common (14%, 19%, and 10%, respectively) and they are also the only majority Muslim departments of Benin (72%, 53%, and 81%, respectively). In contrast to the other northern departments, the Atacora is highly diverse religiously and experiences high levels of interreligious marriage (44%). Intermarriage overwhelming occurs between ethnic and religious groups whose populations border each other and, for northern interethnic marriages, between ethnic groups who share religious affiliations. Table 3.1 shows the most common ethnic and religious pairings among intermarried couples and Appendix 3 Tables 1 and 2 show the complete breakdown of interethnic and interreligious marriage by ethnic and religious composition and their respective frequencies. At the country level, 23% of all interethnic marriages are between the Fon and the Adja and another 20% occur between the Fon and Yoruba, all of which are predominantly southern ethnic groups. At the country-level, 42% of interreligious marriages involve a partner who identifies as a Traditional adherent and these are equally common in the north and the south. In the south, interreligious marriages involving a Catholic spouse are common (42%) whereas, in the north, interreligious marriages involving someone who identifies as Muslim are common (41%). Importantly, there are no major differences by gender regarding which ethnic groups pair with each other, whereas there are immense gender differences in interreligious pairings, which are detailed below. 44 Table 3.1: Most common interethnic and interreligious pairings in Benin Ethnicity Ethnicity of Spouse Religion Religion of Spouse Adja Fon Traditional None Bariba Dendi Muslim Catholic Betamaribé Other Catholic Traditional Dendi Yoa/Lokpa Protestant Catholic Fon Adja Other Christian Protestant Peuhl Bariba Other Other Christian Other Dendi None Traditional Yoa/Lokpa Dendi Yoruba Fon Source: The Demographic and Health Surveys Program 1 All couples’ survey waves (1996-2018) At the department level, intermarriages frequently occur between the largest groups. For example, in the Alibori — a predominantly Muslim region of Benin — over two-thirds (67%) of all interreligious marriages involve a partner who is Muslim. Further, people whose social groups are strongly divided between regions in Benin rarely marry each other — for example, only 7% of all of interreligious marriages involve a Catholic and Muslim partner. In the northern departments, which comprise 44% of all interethnic marriages, there are several interethnic marriage combinations. Marriages with “other ethnicities” form a large share of interethnic marriages in three out of four northern departments — including over 50% in the Alibori and Atacora — where they intermarry with the largest ethnic groups. Excluding partnerships with “others”, over three-fourths of interethnic marriages are a combination of the Bariba, Dendi, and the Peuhl in the Alibori. The Bariba-Dendi combinations are most common (31%) followed by Bariba-Peuhl (27%). In the Atacora, the most frequent pairing is the Betamaribé-Yoa/Lokpa (24%), as well as the Bariba-Dendi (11%). In the Borgou, the three most common interethnic pairings include one Bariba partner paired with someone Dendi (22%), Peuhl (16%), or Yoruba (14%). In the Donga, the Yoa/Lokpa intermarry with the Dendi (55%) and the Peuhl (17%). 45 Northern interreligious marriages comprise 31% of interreligious marriages in Benin. Although the frequency of common pairings varies by region, the three predominantly Muslim departments of the north (the Alibori, Borgou, and Donga) have similar pairings. Although Catholic-Muslim pairings are uncommon at the country level (7%), they comprise a large portion of interreligious marriages in the Alibori (28%), the Borgou (16%) and the Donga (28%). However, while this pairing is gender-neutral in the Alibori, Muslim women marrying Catholic men is more common in the Borgou, whereas Muslim men marrying Catholic women is more common in the Donga. In the Alibori and Borgou, marriages between men who identify as Muslim and unaffiliated women are also common, although the reverse is not. In the Atacora, the only northern department which is not predominantly Muslim, the single largest grouping of interreligious marriages is between unaffiliated women and men who adhere to Traditional practices (26%). The reverse, Traditional women with religiously unaffiliated men, is uncommon (1%). In the south, which comprises 56% of interethnic marriages, most interethnic marriages occur among the Adja and Fon (39%) or the Yoruba and Fon (35%). Adja-Yoruba pairings are rare (4%). Adja-Fon pairings occur frequently in the Atlantique (65%), the Couffo (88%), Littoral (46%), the Mono (78%), Ouémé (27%), and Zou (44%). Similarly, Yoruba-Fon pairings are also common in several departments: they comprise a significant proportion of interethnic marriages in the Atlantique (22%), Collines (50%), Littoral (33%), Ouémé (46%), Plateau (61%), and Zou (30%). Among southern interreligious marriages (69% of all interreligious marriages), the most frequent pairing in every department involves a partner who identifies with an “other Christian” religion. Pairings among other Christian women and men who identify as Traditional adherents 46 are the most common in half of the southern departments: Atlantique (11%), Couffo (34%), Mono (22%), and the Zou (11%). The reverse is significantly less common. In the Collines, other Christian women marry Protestant men (21%), and they marry Catholic men in Littoral (16%) and Ouémé (22%). In Plateau, 17% of interreligious marriages are between other Christians and Protestants, but this is gender-neutral. Overall, ethnic and religious groups that comprise large shares of the population within departments tend to be a common partner in intermarriages, again highlighting the role geographic proximity plays in partner selection. Interethnic marriages typically form between ethnic groups that share similar religious compositions, especially in the north where predominantly Muslim ethnic groups intermarry frequently (70% of interethnic marriages in the north consist of two Muslims). The relative lack of Adja-Yoruba pairings further highlights the importance of spatial proximity in partner selection, as these two groups’ predominant regions do not overlap despite both residing in the south of the country. Also highlighted is the disproportionate role Traditional adherents play in interreligious marriage despite their population size. This is likely due to Traditional beliefs being widespread throughout Benin despite it not being most people’s main religious identification. Further, unlike for interethnic marriages, gender plays a large role in determining who marries whom in interreligious marriages. For example, when an interreligious marriage includes a partner who identifies with Traditional beliefs, the partner with Traditional beliefs is more frequently male. This is likely related to there being more men who identify as Traditional adherents in the general population, suggesting that the gender imbalance of certain religious identifications plays a role in determining who interreligiously marries. 47 Characteristics of interethnic and interreligious couples and marriages In this section, I present results from descriptive statistics for interethnic and interreligious marriages, respectively (Tables 3.2 and 3.3) as well as from binary and multivariate logistic regressions (Tables 3.4 and 3.5). In interethnic marriages, men and women from the largest ethnic group — the Fon — are significantly underrepresented. In contrast, Adja, Dendi, Yoa/Lokpa, and Yoruba men and women are all significantly overrepresented in interethnic marriages in every model. Among women, though not men, the Peuhl are also significantly more likely to intermarry and Betamaribé men are significantly less likely to intermarry in the binary model. These findings suggest that there are population dynamics at play: people from smaller ethnic groups are more likely to marry interethnically and, in two ethnic groups, there is evidence of gendered processes. Interethnically married men and women are less frequently adherents of Traditional religions. The adherents of universal religions — Islam, Catholicism, and Protestantism — as well as other Christian denominations, are significantly more likely to be interethnically married than Traditional adherents, as are people who identify as non-religious. These findings offer evidence that interethnically married couples may sort themselves along religious lines as members of universal religions are significantly more likely to intermarry, suggesting some people who identify with these religious backgrounds may see religion as a more salient characteristic to match on than ethnicity. Among interreligious marriages, male Traditional adherents are overrepresented, whereas female Traditional adherents are underrepresented, likely due to the gender imbalance in people identifying as Traditional adherents. This pattern is mirrored among other Christian denominations. Among Protestants, other religions, and no religion both men and women are 48 significantly more likely to marry interreligiously, likely due to comprising a small portion of the overall Beninese population. Contrarily, Muslim men and women are significantly underrepresented in interreligious marriages and are significantly less likely to intermarry. Similarly, Catholics of both genders are underrepresented in interreligious marriages, although compared with women from Traditional backgrounds, Catholic women are as likely to marry interreligiously. These findings suggest that Muslims and Catholics, two large religious categories in Benin that are relatively gender-neutral in their identification, are less likely to interreligiously marry, perhaps because they can find partners from their own religious group. On the other hand, smaller religious groups that frequently have gender imbalances in identification — Protestants, for example — are more likely to interreligiously marry, suggesting that members of these groups have more difficulty finding partners who share their religious identification. Different ethnic groups have different likelihoods of being interreligiously married, particularly among northern ethnic groups. Compared to the Fon, the two other predominantly southern ethnic groups, the Adja and the Yoruba, are as likely to be interreligiously married. Among the northern ethnic groups, the Bariba are significantly less likely to be interreligiously married in the binary model, but this reverses in the multivariate model meaning they are more likely to be interreligiously married than the Fon. The same occurs for Peuhl and Yoa/Lokpa women, although the odds are non-significant for men. In both models, the Betamaribé are significantly more likely to be interreligiously married, whereas the Dendi are consistently less likely to be interreligiously married. This may be because northern ethnic groups — baring the Betamaribé — are less religiously diverse, rendering them overall less likely to interreligiously marry than southern groups. 49 Interethnically married men and women are more highly educated according to multiple metrics: average years of schooling attended, educated at all, as well as measures of literacy. All of these measures of educational attainment significantly increase the likelihood of being interethnically married in models with full controls. In contrast, education is not significantly correlated with odds of interreligious marriage for either men or women. Further, men and women who marry at later ages are significantly more likely to be interethnically married, however, age at marriage is not significantly associated with entry into interreligious marriages. Overall, a higher proportion of interethnically married women report a premarital pregnancy than intraethnically married women, however, it is not significant in the multivariate model. Women who enter into interreligious marriages are no more likely to have experienced a premarital pregnancy. These findings indicate that entry into interethnic marriages may be more heavily tied to increasing education (both in terms of years of education and delayed marriage) than entry into interreligious marriages, whose trends reflect the general population. 50 Table 3.2: Individual characteristics and difference testing of married men and women in Benin by ethnic composition of marriage Married Women Married Men All Intra Inter All Intra Inter Ethnicity (%) Fon 39.57 41.23 26.68 *** 40.26 41.23 32.77 *** Adja 14.97 14.81 16.20 14.81 14.81 14.86 Bariba 11.32 11.67 8.59 ** 11.31 11.67 8.50 ** Betamaribé 6.86 7.26 3.76 *** 6.81 7.26 3.31 *** Dendi 4.34 3.57 10.38 *** 3.95 3.57 6.89 *** Peuhl 8.01 8.10 7.25 7.92 8.10 6.45 Yoa/Lokpa 4.05 3.88 5.37 * 4.62 3.88 10.38 *** Yoruba 10.88 9.48 21.75 *** 10.32 9.48 16.83 *** Education Average years all 1.43 1.24 2.89 *** 3.27 3.00 5.42 *** (3.06) (2.83) (4.23) (4.68) (4.44) (5.79) Average years educated 6.00 5.78 6.89 *** 7.05 6.78 8.54 *** (3.46) (3.30) (3.88) (4.53) (4.37) (5.11) Educated (%) 23.84 21.51 41.99 *** 46.42 44.24 63.38 *** Literate (%) 18.41 16.39 33.49 *** 38.93 37.05 53.44 *** Religion (%) Traditional 16.86 18.05 7.61 *** 21.29 22.67 10.47 *** Muslim 26.59 25.23 37.15 *** 27.77 26.61 36.79 *** Catholic 23.28 22.39 30.26 *** 24.02 23.40 28.92 *** Protestant 6.20 6.20 6.18 5.80 5.71 6.54 Other Christian 18.97 19.54 14.50 *** 15.20 15.60 12.09 ** Other 1.74 1.87 0.72 ** 1.10 1.15 0.72 No religion 6.37 6.73 3.58 *** 4.82 4.87 4.48 Age at Marriage 18.31 18.17 19.43 *** 23.91 23.80 24.77 *** (4.06) (3.98) (4.49) (5.43) (5.40) (5.59) Premarital Pregnancy 34.34 33.75 38.94 ** --- --- --- N 9828 8711 1117 9828 8711 1117 Source: The Demographic and Health Surveys Program Standard deviations are in parentheses Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 51 Table 3.3: Individual characteristics and difference testing of married men and women in Benin by religious composition of marriage Married Women Married Men All Intra Inter All Intra Inter Religion (%) Traditional 16.86 18.36 12.88 *** 21.29 18.36 29.02 *** Muslim 26.59 33.84 7.38 *** 27.77 33.84 11.69 *** Catholic 23.28 24.87 19.07 *** 24.02 24.87 21.78 ** Protestant 6.20 4.28 11.28 *** 5.80 4.28 9.83 *** Other Christian 18.97 16.23 26.20 *** 15.20 16.23 12.47 *** Other 1.74 0.39 5.31 *** 1.10 0.39 2.97 *** No religion 6.37 2.02 17.88 *** 4.82 2.02 12.24 *** Education Average years all 1.43 1.43 1.44 3.27 3.26 3.31 (3.06) (3.06) (3.07) (4.68) (4.68) (4.68) Average years educated 6.00 5.99 6.02 7.05 7.06 7.03 (3.46) (3.46) (3.45) (4.53) (4.54) (4.52) Educated (%) 23.84 23.82 23.90 46.42 46.16 47.09 Literate (%) 18.41 18.50 18.20 38.93 38.96 38.79 Ethnicity (%) Fon 39.57 37.77 44.42 *** 40.26 38.85 44.01 *** Adja 14.97 14.58 15.99 14.81 14.09 16.73 ** Bariba 11.32 11.58 10.64 11.31 11.52 10.76 Betamaribé 6.86 4.57 12.81 *** 6.81 4.54 12.80 *** Dendi 4.34 5.79 0.05 *** 3.95 5.27 0.45 *** Peuhl 8.01 9.62 3.75 *** 7.92 9.38 4.04 *** Yoa/Lokpa 4.05 4.79 2.08 *** 4.62 5.47 2.37 *** Yoruba 10.88 11.33 9.68 * 10.32 10.88 8.83 ** Age at Marriage 18.31 18.33 18.28 23.91 23.94 23.83 (4.06) (3.99) (4.23) (5.43) (5.45) (5.39) Premarital Pregnancy 34.34 34.75 33.25 --- --- --- N 9828 7133 2695 9828 7133 2695 Source: The Demographic and Health Surveys Program Standard deviations are in parentheses Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 52 Table 3.4: Odds ratios of individual-level characteristics associated with entry into interethnic marriages Model 1: Binary Model 2: Multivariate Women Men Women Men OR OR OR OR Ethnicity Ethnicity Fon (ref) --- --- Fon (ref) --- --- Adja 1.69*** 1.26* Adja 2.43*** 1.65*** (0.17) (0.13) (0.26) (0.18) Bariba 1.14 0.92 Bariba 1.14 0.83 (0.14) (0.11) (0.17) (0.12) Betamaribé 0.80 0.57** Betamaribé 1.15 0.76 (0.14) (0.10) (0.20) (0.14) Dendi 4.50*** 2.43*** Dendi 3.96*** 2.12*** (0.56) (0.34) (0.65) (0.37) Peuhl 1.38** 1.00 Peuhl 1.41* 0.99 (0.18) (0.14) (0.24) (0.17) Yoa/Lokpa 2.14*** 3.37*** Yoa/Lokpa 1.95*** 3.03*** (0.33) (0.41) (0.35) (0.46) Yoruba 3.55*** 2.23*** Yoruba 3.13*** 1.86*** (0.34) (0.22) (0.33) (0.20) Education Education Years 1.14*** 1.09*** Years 1.11*** 1.09*** (0.01) (0.01) (0.01) (0.01) Educated 2.64*** 2.18*** (0.17) (0.14) Literate 2.57*** 1.95*** (0.19) (0.00) Religion Religion Traditional (ref) --- --- Traditional (ref) --- --- Muslim 3.50*** 2.99*** Muslim 2.95*** 2.59*** (0.43) (0.33) (0.47) (0.39) Catholic 3.21*** 2.68*** Catholic 2.29*** 2.02*** (0.40) (0.30) (0.37) (0.25) Protestant 2.36*** 2.48*** Protestant 2.11*** 1.95*** (0.40) (0.39) (0.38) (0.33) Other Christian 1.76*** 1.68*** Other Christian 1.78*** 1.47** (0.24) (0.22) (0.26) (0.20) Other 0.91 1.35 Other 0.93 1.09 (0.34) (0.51) (0.36) (0.42) No religion 1.26 1.99*** No religion 1.63* 1.95*** (0.25) (0.35) (0.33) (0.36) Age at Marriage 1.07*** 1.03*** Age at Marriage 1.04*** 1.02** (0.01) (0.01) (0.01) (0.01) Premarital Pregnancy 1.25** --- Premarital Pregnancy 1.09 --- (0.08) (0.08) Constant 0.13*** 0.03*** (0.00) (0.00) N 9828 9289 N 9828 9282 Pseudo R 2 0.09 0.07 Source: The Demographic and Health Surveys Program *p < .05; **p < .01; ***p < .001 Standard errors are in parentheses 53 Table 3.5: Odds ratios of individual-level characteristics associated with entry into interreligious marriages Model 1: Binary Model 2: Multivariate Women Men Women Men OR OR OR OR Religion Religion Traditional (ref) --- --- Traditional (ref) --- --- Muslim 0.31*** 0.22*** Muslim 0.26*** 0.20*** (0.03) (0.02) (0.03) (0.02) Catholic 1.09 0.55*** Catholic 1.07 0.52*** (0.09) (0.04) (0.09) (0.04) Protestant 3.76*** 1.46*** Protestant 3.79*** 1.48*** (0.38) (0.14) (0.40) (0.15) Other Christian 2.30*** 0.49*** Other Christian 2.56*** 0.52*** (0.18) (0.04) (0.21) (0.04) Other 19.28*** 4.79*** Other 17.95*** 5.10*** (4.15) (1.07) (3.91) (1.15) No religion 12.64*** 3.84*** No religion 10.84*** 3.78*** (1.42) (0.42) (1.25) (0.43) Education Education Years 1.00 1.00 Years 1.02 1.01 (0.01) (0.00) (0.01) (0.01) Educated 1.00 1.04 (0.05) (0.05) Literate 0.98 0.99 (0.06) (0.05) Ethnicity Ethnicity Fon (ref) --- --- Fon (ref) --- --- Adja 0.93 1.05 Adja 1.08 0.89 (0.06) (0.07) (0.08) (0.07) Bariba 0.78** 0.82* Bariba 2.20*** 2.00*** (0.06) (0.06) (0.23) (0.20) Betamaribé 2.40*** 2.49*** Betamaribé 2.21*** 2.46*** (0.20) (0.21) (0.22) (0.22) Dendi 0.08*** 0.07*** Dendi 0.52* 0.27*** (0.02) (0.02) (0.15) (0.08) Peuhl 0.33*** 0.38*** Peuhl 1.39* 1.13 (0.04) (0.04) (0.21) (0.16) Yoa/ Lokpa 0.37*** 0.38*** Yoa/Lokpa 1.47* 0.95 (0.05) (0.05) (0.25) (0.15) Yoruba 0.73*** 0.72*** Yoruba 1.07 1.02 (0.06) (0.06) (0.09) (0.09) Age at Marriage 1.00 1.00 Age at Marriage 1.00 1.00 (0.01) (0.00) (0.01) (0.01) Premarital Pregnancy 0.94 --- Premarital Pregnancy 0.96 --- (0.04) (0.05) Constant 0.22*** 0.57*** (0.03) (0.07) N 9828 9289 N 9828 9828 Pseudo R 2 0.17 0.11 Source: The Demographic and Health Surveys Program *p < .05; **p < .01; ***p < .001 Standard errors are in parentheses 54 The interethnic couple-level descriptive statistics and logistic regression (Tables 3.6 and 3.7) results offer similar findings as the individual-level models. Interethnic couples do not appear to be comprised of people disadvantaged on the marriage market. Interethnic marriages are not disproportionately remarriages and they are significantly less likely to be polygynous than intraethnic marriages. Interethnic couples also reside more frequently in urban areas and are wealthier. The couples’ relative educational attainment significantly increases the likelihood of being in an interethnic marriage. The effect is strongest when both partners are educated, although the likelihood increases with only one educated partner. When men’s and women’s individual years of schooling are used instead, each year of schooling increases the likelihood of interethnic marriage by 4% for women and by 2% for men in the full controls model. The husband being older than his wife is not correlated with a change in the likelihood of being interethnically married, however, the wife being five or more years older than her husband is significantly correlated with increased odds. However, due to the small sample of women older than her husband, these results should be interpreted with caution. 55 Table 3.6: Characteristics and difference testing of couples in Benin by ethnic composition of marriage All Intra Inter Decade Married (%) Pre-1980s 6.29 6.55 4.21 ** 1980s 16.51 16.97 12.98 ** 1990s 34.50 35.15 29.45 *** 2000s 31.02 30.09 38.32 *** 2010s 11.67 11.24 15.04 *** Marital Type (%) Polygynous (first marriage) 37.49 38.78 27.48 *** Polygynous (all marriages) 39.00 40.31 29.29 *** Interreligious marriage 27.42 27.18 29.27 Remarried 13.50 13.35 14.87 Education (%) Neither educated 49.05 51.43 30.53 *** Both educated 19.31 17.19 35.90 *** Husband educated/Wife not 27.11 27.06 27.48 Wife educated/Husband not 4.53 4.33 6.09 ** Age Gap (%) <5 years 31.39 31.12 32.95 Husband older 5-9 years 36.76 36.67 37.51 Husband older >10 years 30.96 31.32 28.20 * Wife older >5 years 0.89 0.83 1.34 Household Wealth Quintile (%) Poorest 24.15 25.21 15.85 *** Poor 21.71 22.73 13.79 *** Middle 19.21 19.87 14.06 *** Rich 18.10 17.94 19.43 Richest 16.83 14.26 36.88 *** Urban (%) 35.15 32.29 57.48 *** N 9828 8711 1117 Source: The Demographic and Health Surveys Program Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 56 Table 3.7: Logistic regressions of couple-level characteristics associated with entry into interethnic marriages (Page 1 of 2) Model 1: Binary Model 2: Marital Characteristics Model 3: Full Controls OR OR OR Decade Married Pre-1980s (ref) --- --- --- 1980s 1.19 1.08 1.13 (0.21) (0.19) (0.20) 1990s 1.31 1.11 1.19 (0.21) (0.18) (0.19) 2000s 1.98*** 1.52* 1.53* (0.32) (0.25) (0.25) 2010s 2.08*** 1.32 1.34 (0.36) (0.23) (0.24) Marital Type Polygynous 0.60*** 0.75*** 0.79** (0.04) (0.06) (0.06) Interreligious 1.11 1.10 1.15 (0.08) (0.08) (0.08) Education Neither educated (ref) --- --- --- Both educated 3.52*** 3.16*** 1.88*** (0.28) (0.26) (0.18) Husband educated/Wife not 1.71*** 1.66*** 1.38*** (0.14) (0.14) (0.12) Wife educated/ Husband not 2.37*** 2.19*** 1.93*** (0.34) (0.32) (0.29) Age Gap <5 years (ref) --- --- --- Husband older 5-9 years 0.97 1.07 1.05 (0.07) (0.08) (0.08) Husband older >10 years 0.85 1.10 1.07 (0.07) (0.09) (0.09) Wife older >5 years 1.54 1.89* 2.05* (0.44) (0.56) (0.61) 57 Table 3.7 Continued (Page 2 of 2) Model 1: Binary Model 2: Marital Characteristics Model 3: Full Controls OR OR OR Household Wealth Quintile Poorest (ref) --- --- Poor 0.96 1.00 (0.11) (0.12) Middle 1.13 1.10 (0.13) (0.13) Rich 1.72*** 1.43** (0.19) (0.16) Richest 4.12*** 2.62*** (0.40) (0.30) Urban 2.83*** 1.72*** (0.18) (0.13) Region South (ref) --- --- North 0.91 1.39*** (0.06) (0.11) Constant --- 0.07*** 0.04*** (0.01) (0.01) Pseudo R 2 --- 0.05 0.08 Source: The Demographic and Health Surveys Program *p < .05; **p < .01; ***p < .001 Standard errors are in parentheses N=9828 In the interreligious couple-level descriptive statistics and logistic models, Tables 3.8 and 3.9, interreligious couples largely reflect the trends of intrareligious couples. Descriptively, interreligious marriages are disproportionately remarriages, although they are not more likely to be polygynous. The couple’s relative education is not associated with the likelihood of being interreligiously married regardless of how education is measured, nor is the age of spouses relative to each other. In the full controls model (Model 3), compared to couples in the poorest wealth quintile, people in all other quintiles are significantly less likely to be interreligiously married. This marital type is not significantly associated with urbanicity, and couples residing in the north are significantly less likely to be interreligiously married than southern couples. 58 Table 3.8: Characteristics and difference testing of couples in Benin by religious composition of marriage All Intra Inter Decade Married (%) Pre-1980s 6.29 6.45 5.86 1980s 16.51 16.78 15.81 1990s 34.50 34.81 33.69 2000s 31.02 30.93 31.28 2010s 11.67 11.03 13.36 ** Marital Type (%) Polygynous (first marriage) 37.49 37.49 37.51 Polygynous (all marriages) 39.00 39.15 38.60 Interethnic marriage 11.37 11.08 12.13 Remarried 13.50 12.37 16.31 *** Education (%) Neither educated 49.05 49.38 48.20 Both educated 19.31 19.35 19.22 Husband educated/Wife not 27.11 26.80 27.90 Wife educated/Husband not 4.53 4.47 4.68 Age Gap (%) <5 years 31.39 31.29 31.65 Husband older 5-9 years 36.76 36.69 36.96 Husband older >10 years 30.96 31.19 30.35 Wife older >5 years 0.89 0.83 1.04 Household Wealth Quintile (%) Poorest 24.15 23.19 26.68 *** Poor 21.71 21.86 21.34 Middle 19.21 19.66 18.03 Rich 18.10 18.30 17.59 Richest 16.83 17.01 16.36 Urban (%) 35.15 34.91 35.81 N 9828 7133 2695 Source: The Demographic and Health Surveys Program Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 59 Table 3.9: Logistic regressions of couple-level characteristics associated with entry into interreligious marriages (Page 1 of 2) Model 1: Binary Model 2: Marital Characteristics Model 3: Full Controls OR OR OR Decade Married Pre-1980s (ref) --- --- --- 1980s 1.04 1.04 1.05 (0.11) (0.11) (0.11) 1990s 1.06 1.07 1.08 (0.11) (0.11) (0.11) 2000s 1.11 1.12 1.16 (0.11) (0.11) (0.12) 2010s 1.33* 1.37** 1.47** (0.15) (0.16) (0.17) Marital Type Polygynous 1.00 1.03 1.05 (0.05) (0.05) (0.05) Interethnic 1.11 1.10 1.13 (0.08) (0.08) (0.08) Education Neither educated (ref) --- --- --- Both educated 1.02 0.96 0.91 (0.06) (0.06) (0.07) Husband educated/Wife not 1.07 1.06 1.01 (0.06) (0.06) (0.06) Wife educated/ Husband not 1.07 1.02 0.98 (0.12) (0.11) (0.11) Age Gap <5 years (ref) --- --- --- Husband older 5-9 years 1.00 1.01 1.03 (0.05) (0.06) (0.06) Husband older >10 years 0.96 0.98 1.00 (0.06) (0.06) (0.06) Wife older >5 years 1.24 1.25 1.21 (0.29) (0.29) (0.28) 60 Table 3.9 Continued (Page 2 of 2) Model 1: Binary Model 2: Marital Characteristics Model 3: Full Controls OR OR OR Household Wealth Quintile Poorest (ref) --- --- Poor 0.85* 0.79*** (0.06) (0.05) Middle 0.80** 0.71*** (0.06) (0.05) Rich 0.84* 0.72*** (0.06) (0.05) Richest 0.84* 0.67*** (0.06) (0.06) Urban 1.04 1.10 (0.05) (0.06) Region South (ref) --- --- North 0.74*** 0.65*** (0.04) (0.03) Constant 0.33*** 0.46*** (0.04) (0.05) Pseudo R 2 0.00 0.01 Source: The Demographic and Health Surveys Program *p < .05; **p < .01; ***p < .001 Standard errors are in parentheses N=9828 Using descriptive statistics (see Appendices 3.D-3.F), I also compare only recent marriages and the results are largely consistent for interethnic and interreligious marriages with only a few differences of note. In this population, interethnic marriages nearly doubled from the 1990s (10%) to the 2010s (16%), suggesting that this is an increasingly popular marital pattern. While education remains significantly tied to recent interethnic marriage, there is no longer a significant difference in the marital age for women, although couples with an age gap of fewer than five years occur with a higher frequency among recent interethnic couples. Among recent interreligious marriages, the frequency of this pairing overtime has stayed consistent (around 61 31% in the 1990s and 2010s) and these couples are no longer overrepresented among the poorest wealth quintile. These results indicate that the current average interethnic couple varies significantly from the average intraethnic couple and that interethnic marriage is tied to high levels of human and economic capital. Further, these results hold for recently married couples showing that interethnic marriage has been tied to high levels of capital for several decades. Contrarily, the results for interreligious couples do not suggest that the average currently married interreligious couple varies greatly from the average intrareligiously married couple, although they may be becoming slightly more frequent in recent marital cohorts. There is also no indication that interreligious marriage is associated with higher or lower educational attainment and it is inconsistently associated with lower levels of wealth. The characteristics associated with these two types of intermarriage appear to be distinct and are not uniformly tied to broader structural changes. Discussion As a continent rich in social diversity, work from several disciplines inquires how sub-Saharan Africa’s rapid social, demographic, and economic changes will affect existing social boundaries surrounding family life (Addai and Pokimica 2010; Shapiro and Tambashe 2017; Udogu 1999; Wallerstein 1960). Two major social identities, ethnicity and religion, continue to shape large swaths of daily life, particularly in West Africa. Ethnic and religious affiliation are correlated with access to resources (Weinreb 2001), residential location in both urban and rural settings (Smith 2011), and, importantly, marital formation (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Falen 2011; Gnoumou Thiombiano 2017; Locoh and Thiriat 1995; Maïga and Baya 2011). 62 However, emerging evidence in West Africa indicates that in the face of large-scale social and economic changes, the overall role of these social identities in partner selection is in flux, and potentially in opposite ways. Ethnic homogamy, for example, may be less of a priority for recent couples (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). These findings about the role of ethnic identity are striking in light of past research that found ethnicity to be particularly entrenched in shaping family life in West Africa (Caldwell 1996; Caldwell and Caldwell 1987; Caldwell, Orubuloye, and Caldwell 1992). Contrary to the potential decrease in ethnic homogamy, prior research finds that interreligious marriage may be decreasing in West Africa, in part because educated people may increasingly select partners along religious lines and, in urban areas, religion may provide an easy way to develop a social network (Crespin-Boucaud 2018; Maïga and Baya 2011). These recent studies find that achieved characteristics, such as educational attainment, are being increasingly prioritized on the marriage market, altering the perceived value of shared social identities in partner selection (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). I use Benin as a case study to conduct an exploration of social identity’s role in spousal choice and expand our understanding of interethnic and interreligious marriage in West Africa, a region of sub-Saharan Africa where social identity plays a particularly prominent role in family formation (Caldwell 1996; Caldwell and Caldwell 1987; Crespin-Boucaud 2018). Benin is an ideal case study for examining changes in marital formation along social identity lines. Compared to other countries in West Africa, its economic and social developments have been steady but not rapid (Tabutin and Schoumaker 2004; United Nations Children’s Fund 2017). Further, ethnicity and religion are not a point of conflict in Benin compared to neighboring 63 countries such as Nigeria (Adotevi et al. 2017; Dickovick 2008). Despite this, ethnicity remains more salient in Benin in determining partner selection than in other countries in the region and religion’s role in partner selection may be increasing in salience Benin (Crespin-Boucaud 2018). Importantly, I find that intermarried couples, especially interethnic, are becoming more common in Benin since the 1980s, corresponding to major shifts in Beninese society 10 , illustrating how larger societal trends are linked to family life. This study examines an understudied dimension of women’s lives — partner selection — and helps untangle the question of who selects into these diverse and increasingly common marriages. Akin to prior research (Maïga and Baya 2011; Safi and Rogers 2008), I find that interethnic marriage is strongly associated with higher levels of social and economic capital including more education, higher household wealth, and later marital age. In further support of these findings, interethnic marriages are not more likely to be polygynous, nor are they more likely to be remarriages, indicating that these are unlikely to be unions formed by women who are at a disadvantage on the marriage market. While interethnic marriages do have higher odds of being precipitated by a premarital pregnancy, this effect disappears in multivariate models. In all, these findings indicate that interethnic marriage is associated with shifting social structures that prioritize human capital over ethnic homogamy on the marriage market. Finding an increase in interreligious marriage in Benin stands in contrast to prior research that found interreligious marriages were declining in Benin. However, my religious coding is less reductive and leaves room for a greater number of religious identities (Crespin-Boucaud 10 These changes include a new governmental structure (from Marxism to a republic) in 1989 (Adotevi et al. 2017), the emergence of a countrywide fertility transition starting in the mid to late 1990s (Garenne and Joseph 2002), and shifts in education, including decreased enrollment in the 1980s and 1990s and subsequent increased enrollment since 2000 (Lloyd, Kaufman, and Hewett 2000; Tabutin and Schoumaker 2004; The World Bank 2019). 64 2018). Importantly, my study reconfirms that, in contrast to the results for interethnic marriage, interreligious marriage is not associated with education and may be negatively related to wealth (Crespin-Boucaud 2018). Further, I do not find that interreligious marriages are linked with an urban or rural residential location (Maïga and Baya 2011), which is in contrast to findings in the capital of Burkina Faso that urban, educated people prefer to select partners from a shared religious background (Calvès and Gnoumou Thiombiano 2014). Additionally, interreligious marriages are disproportionately remarriages. This finding, along with the increased likelihood of being in the poorest wealth quintile, suggests that interreligious marriages may be more common among people who are likely to be disadvantaged on the marriage market. In all, interreligious marriage is not clearly linked to changing social structures that prioritize human capital on the marriage market. Spatially, intermarriages are most common in regions where there is greater ethnic and religious diversity and, in the case of interethnic marriage, where there are major urban centers. That interethnic couples are more likely to be found in urban areas may be because people in urban areas are more likely to interact with people of other ethnic groups or because rural areas are less accepting of these couples, causing them to migrate (Rosenfeld and Kim 2005). Regardless, interethnic couples are becoming more common in rural areas over time, suggesting that either new pathways to meet people of other ethnicities have opened up through schools, for example, or that interethnic couples are more willing or able to live in rural areas. Contrarily, interreligious marriages have been common in both urban and rural settings throughout the decades, indicating that this form of intermarriage is either more acceptable in rural areas than interethnic marriage, or that rural areas are more religiously diverse than they are ethnically diverse. Importantly, most interethnic and interreligious marriages occur between social groups 65 who reside in geographic proximity. This is most evident in that two of three southern ethnic groups, the Adja and Yoruba, rarely intermarry: instead, both groups usually intermarry with the Fon, whose main geographic range separates the Adja and Yoruba from each other. The spatial element of interethnic and interreligious marriages in Benin is also visible in the country’s north-south divide. Intermarriage between ethnic groups who are predominantly northern and southern are uncommon. The only one of note is between the Bariba and the Yoruba, who live in adjacent regions and often share a religious affiliation. Similarly, along religious lines, Catholics — largely southern — and Muslims — largely northern — rarely intermarry at the country-level. That geographic proximity is linked to intermarriage is not surprising. Studies from other contexts find similar outcomes as proximity increases the likelihood of both shared religion and interaction between people. This increases the opportunity people have to form social networks and therefore enter into relationships with each other (Hannemann et al. 2018; Leszczensky and Pink 2019; Maas and Zijdeman 2010; McPherson et al. 2001; Puur et al. 2018; Safi and Rogers 2008; Voas 2009). Importantly, there are ethnic, religious, and gendered differences in the likelihood to intermarry. As in other countries (Çelikaksoy 2016; Martin et al. 2018; Puur et al. 2018), members from smaller ethnic and religious groups in Benin are significantly more likely to intermarry than people from larger groups. Among most ethnic groups, there are few gendered differences in the likelihood of interethnic marriage. One notable exception is among the Peuhl. Peuhl women are highly represented in interethnic marriages whereas Peuhl men are not. This may be tied to negative stereotypes and discrimination against the Peuhl in Benin and much of West Africa, which may impact Peuhl men more due to their role in cattle grazing, a source of 66 conflict in many countries (Bukari and Schareika 2015; Majekodunmi et al. 2014; Okello et al. 2014). In contrast to interethnic marriage, there are strong gender differences among interreligious couples. Men of the three predominant religious groups (Traditional, Islam, and Catholicism) are overrepresented in interreligious marriages whereas women of these religions are underrepresented. Although they are all overrepresented, Traditional men are significantly more likely to intermarry than their Catholic and Muslim peers and this may be tied to the gender imbalance in Traditional identities. Contrarily, women are overrepresented and men underrepresented among Protestants, other Christians, other, and no religion. As a result, women who are overrepresented in minority religions are significantly more likely to out-marry than women from major religious groups. Women and small ethnic and religious groups may have higher rates of out-marriage because preferred systems of heterogamy — such as wives having less education than their husbands — may become more difficult to maintain in times of social change, straining marriage markets for smaller groups and women. Relatedly, marriage market constraints may be gendered with women being more likely to have to marry out to meet partners of similar achieved characteristics (Voas 2009). Thus, men and women from smaller groups who prioritize their spouses’ having specific socioeconomic characteristics may do so at the expense of marrying someone from a similar ethnic or religious background (Hannemann et al. 2018). Additionally, religious identification is gendered in Benin (Falen 2008a, 2016), meaning there may be too few potential partners in specific religious traditions. This gender imbalance forces Traditional men, for example, to marry women from other religious backgrounds, whereas Traditional women have more partners with a shared religious identity to choose from. 67 Importantly, these results also illustrate that although both interethnic and interreligious marriages involve marrying someone of another social group, in the BDHS couples’ data, these two types of marriages do not occur among similar sub-populations. In the Beninese context, interethnic marriage appears to be highly correlated with structural changes, with these couples being selected for high levels of education, urbanicity, and wealth, reaffirming findings from other contexts that interethnic partnerships seem to be highly selective (Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). However, my findings offer evidence that interethnic couples are increasingly residing in rural areas, suggesting that the importance of shared ethnicity in rural settings may be decreasing in Benin (Maïga and Baya 2011). Additionally, unlike prior research, my results show that when grouping these couples by marital decade, rather than by the woman’s decade of birth, interethnic marriages are increasing in Benin and have been for many decades (Crespin-Boucaud 2018). These findings suggest that in both urban and rural Benin, the prioritization of ethnic homogamy in partner selection is declining over time, in ways that correlate with structural changes that increase network diversity, decrease the role of families in marital selection, and prioritize achieved characteristics over social identities. Notably, this study helps us better understand how interethnic marriage is not independent of the intersection of ethnic and religious identifications. I find that religious identification is important for interethnic marriages, as interethnic marriages are most likely to occur among people from universal faiths (i.e. Islam or Catholicism). This suggests that shared religion may be of increased importance for couples who marry across ethnic identities, especially as universal faiths continue to be tied to education and wealth as compared to Traditional beliefs (Calvès and Gnoumou Thiombiano 2014; Falen 2016). 68 Contrarily, interreligious marriage does not appear to be correlated with structural changes in the Beninese context, nor does there appear to be high levels of selection into these marriages. The role of religious identity in partner selection appears to be relatively constant over time in Benin. Among intact first marriages, the frequency of interreligious couples has remained relatively stable across marital cohorts, in contrast to findings that it may be declining (Crespin- Boucaud 2018). Additionally, these results do not indicate that religious identity is especially salient in rural (Maïga and Baya 2014) or urban areas (Calvès and Gnoumou Thiombiano 2014). Rather, my findings expand our understanding of the role gender imbalances play in interreligious marriages. Religious identities that are skewed in their gender balance experience higher rates of interreligious marriages among the gender overrepresented. Overall, these results indicate that interethnic and interreligious marriages are distinct partner selection patterns in the Beninese context and that marrying across different ethnic identities is less likely than marrying across religions. One potential explanation for higher levels of interreligious marriage is the gender imbalance in religious identification as there are simply not enough men and women of all religious groups for everyone to marry intrareligiously. This may help explain why marrying across religious identities is of little note in Beninese partnerships historically (Falen 2011) and why interreligious marriages experience less selection than interethnic marriages. Because ethnic groups are balanced, people are not forced to out-marry to find a partner, rather their high levels of human and economic capital indicate that these are couples who prioritize shared achieved characteristics over ethnic homogamy. Another explanation is the religious diversity within several ethnic groups. In the Beninese case, although religions are geographically clustered, their clustering happens to overlap with ethnicity in ways that make it easier for interreligious marriages to occur than for interethnic marriages. 69 Additionally, religion is syncretic in Benin, with people often practicing the traditions of multiple faith backgrounds at once, particularly Traditional practices with monotheistic faiths (Falen 2008a, 2011, 2016). As a result of this mixed bag of religious practices, marrying across religious boundaries may seem less like marrying across an identity category than marrying across ethnic boundaries, which often correspond with languages and other social traditions in ways that are not as fluid as religious practices. These differences, therefore, highlight the role of clustering and broader social context in making some boundaries more difficult to create ties across. These results finding that one type of social identity is more of a barrier to marriage are not unique to Benin. In the United States, for example, interreligious marriage is much more common than interracial marriage, although both are increasing in frequency, suggesting that race plays a larger role in partner selection than religion does in the United States (Bialik 2017; Murphy 2015). Contrarily, in a country such as Israel, where religion is the key social boundary, religion plays a larger role in determining partner selection than ethnicity (Murphy 2016; Okun 2004; Stier and Shavit 1994). Thus, these results highlight our need to further study interethnic and interreligious marriage separately in sub-Saharan Africa as we do in other contexts because although all social identities play a role in partner selection to some extent, not all are equally difficult to marry across in different contexts. This study is not without important limitations, in large part due to the sub-sample of marriages used to measure interethnic and interreligious marriage. The DHS couples’ survey is likely not reflective of interethnic and interreligious marriages at large but the overall magnitude of these differences is unknown due to a lack of data on marital histories in the region. 70 The most important limitation of this sub-sample is that interethnic and interreligious marriages may be more likely to end in divorce in the early years of marriage (Locoh and Thiriat 1995). Due to the data available, the results of this study are based exclusively on intact first marriages. This means I am studying a select subset of marriages, especially as I have no information on how common divorce is among interethnic and interreligious couples. The results of this study would likely change if I had marital history data. For example, interethnic unions with low levels of education and wealth may be more prone to dissolving due to increased reliance on their natal families, which would render these types of interethnic couples less common in my sample. To help mitigate the effect of union dissolution on my results, I compared only recent unions — those occurring within five years of the survey — and found that the larger patterns hold for both interethnic and interreligious unions. Similarly, when I restrict the definition of recent marriages to marriages that occurred within two years of the survey, the overall patterns remain robust. While these restrictions certainly do not eliminate the challenges posed by this limitation, they do at least indicate that the select sample of intact first marriages is unlikely to be completely driving these results. Relatedly, if women do not consider a short-lived failed marriage as a true marriage, they may report their current relationship as their first marriage in the survey. However, because Beninese marriages experience low levels of divorce (Clark and Brauner-Otto 2015), and women’s reports of first marriages are largely accurate (Chae 2016), it is unlikely that these are causing intermarriages to be significantly underreported or undercounted in studying intact first marriages. The potential instability of identities is another limitation of this study. Among interethnic marriages, women who have been married for a long time may become more likely with time to 71 take on their husband’s ethnic identification, meaning they would not be identified as interethnically married (Crespin-Boucaud 2018; Puur et al. 2018). However, I am unaware of any studies or data on this phenomenon in sub-Saharan Africa, but given that women in Benin remain strongly tied to their natal families (Falen 2008a, 2008b, 2011), it is unlikely that many women forgo their own ethnic identity in favor of their husbands’. Along similar lines, marriages that were interreligious at the time of marriage may no longer be interreligious due to men and women converting to their partner’s religion for marriage or later in the marriage — a phenomenon much more likely than a change in ethnic identification (Agadjanian 2017; Crespin-Boucaud 2018). Women converting to match their husbands’ religious affiliation may be more common among educated women, resulting in fewer highly educated couples being identified as interreligious (Agadjanian 2017). An additional challenge of measuring interreligious marriages is that the DHS specifically asks people what religion they practice, rather than what religion they identify with 11 . While people’s responses most likely reflect their religious identification, the question reflects a larger truth of religion in West Africa: people often maintain syncretic beliefs that do not always neatly overlap with their expressed religious identification. In West Africa, converting to Islam or Christianity but maintaining the beliefs of Traditional religions is common (Acquah 2011; Okeke, Ibenwa, and Okeke 2017), indicating that expressed religious identities are not always discrete categories. As a result, the insistence of a single response on surveys renders religious life less complicated on paper than in real life. 11 This is not a function of the French language. The BDHS asks “Quelle religion pratiquez- vous?” whereas, in France, the question would be asked “De quelle confession êtes-vous?” or “Quelle est votre religion?” This phrasing choice may be due to the need to translate this question into various local Beninese languages. 72 Another limitation is the lack of relationship information collected by the Demographic and Health Surveys. Unfortunately, women are not asked extensive questions about their relationship histories, meaning I cannot study the transition to marriage from dating among intermarried couples, for example. Additionally, the DHS generally treats marriage and cohabitation as akin to one another as African marriages often do not have a clear cut marital start date (Meekers 1992). However, couples are increasingly turning to cohabitation, particularly in urban areas, to avoid the issue of familial consent required for formal unions (Calvès 2016a; Falen 2011) and the challenges posed by having a religious ceremony when of different religions (Maïga and Baya 2011). Along these same lines, research shows that interethnic cohabitations may be less likely to transition into marriage and that, when they do transition to marriage, it takes longer than for intraethnic cohabiting unions (Calvès 2016a). This is likely similar for couples of different religious backgrounds. Future research would benefit from exploring relationship histories and marital timing to see if interethnic and interreligious relationships are more likely to end before marriage and/or if they are more likely to never transition to marriage. Additionally, the DHS does not ask interviewees about their family of origin and childhood meaning we do not know the ethnic and religious background of their parents or their own religious histories. While all major ethnic groups in Benin are patrilineal, meaning children adopt the ethnic identity of their father, the children of interethnic couples are more likely to marry interethnically, especially to someone from their mother’s ethnic group — and presumably someone of their father’s background in matrilineal societies (Maïga and Baya 2011). This may also apply to interreligious couples where the children of parents of different religions may be more willing to intermarry. These data limitations mean that using the DHS, a study on 73 intermarriage cannot account for whether intermarried individuals are also the children of intermarriages. Despite these limitations, this study offers important insights into partner selection in West Africa, specifically the likelihood of men and women selecting a spouse who does not share their own key social identities. Benin as a case study illustrates how interethnic first marriages appear to be tied to broader social and structural changes that are linked to prioritizing human capital over shared ethnic identities on the marriage market. It also illustrates how the likelihood of interreligiously marrying is deeply tied to gender dynamics in religious identification. Importantly, religious identity may be especially salient among people who do not select partners of the same ethnic background. These results indicate that the overall role of social identities in partner selection is in flux in Benin. These findings contribute to our understanding of the role of social identities in partner selection in West Africa, a region of sub-Saharan Africa where ethnicity and religion have been more deeply entrenched than other regions (Bongaarts and Casterline 2012; Caldwell 1996; Caldwell and Caldwell 1987; Crespin-Boucaud 2018). It also expands our understandings of the ways marriage formation may be changing in developing countries that are in the midst of structural and demographic changes. Interethnic and interreligious marriages are, therefore, important trends to study in greater depth, to better understand family life and marriage markets in the sub-Saharan African region. 74 CHAPTER 4 FERTILITY PREFERENCE CONCORDANCE AND FERTILITY OUTCOMES IN INTERETHNIC AND INTERRELIGIOUS MARRIAGES The fertility transition in sub-Saharan Africa is the subject of many research studies focusing on the unique features contributing to a relatively slow pace of decline, particularly in West Africa (Bongaarts 2006, 2017; Bongaarts and Casterline 2012; Caldwell 2001; Caldwell et al. 1992; Eloundou-Enyegue and Giroux 2012; Ezeh, Mberu, and Emina 2009; Hayford and Agadjanian 2012; Hertrich 2017). Demographers work to identify why fertility declines have stalled in the region (Bongaarts 2006; Moultrie et al. 2008), and one potential factor is individuals’ maintenance of high fertility preferences. In sub-Saharan Africa, men and women continue to desire, and therefore have larger families than in much of the world (Bongaarts 2006, 2017; Bongaarts and Casterline 2012; Caldwell et al. 1992). Importantly, fertility preferences and outcomes are not homogenous across the region. Fertility preferences and outcomes vary among different social groups, including by ethnicity and religion (Agadjanian and Yabiku 2014; Castle 2001; Gyimah, Takyi, and Tenkorang 2008; Johnson-Hanks 2003, 2006; Shapiro and Tambashe 2017; Takyi and Addai 2002; Takyi and Dodoo 2005). At the same time, there is emerging evidence that the salience of social identities in shaping fertility preferences may be changing among women, particularly in urban areas (Shapiro and Tambashe 2017). However, because most studies focus exclusively on men’s or women’s individual preferences, there are few studies in sub-Saharan Africa focused on how different social identities at the couple-level help shape preferences and outcomes. To examine the role of different social identities on fertility preferences and outcomes, I study interethnically and interreligiously married couples in Benin, West Africa and compare them to their intramarried counterparts. Interethnic and interreligious couples may have distinct 75 fertility preferences and outcomes compared to couples who share ethnic and religious backgrounds. Because each partner in the marriage may bring in ethnically and religiously situated perceptions of the ideal number of children to have, these couples may need to negotiate their fertility expectations more than couples from the same social background, which may lead to distinctive fertility outcomes (Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Isiugo- Abanihe 2014; Lehrer 1996). Additionally, as interethnic and interreligious couples have already gone outside of convention in their marital partner choice, these couples may be more likely to be at the vanguard of fertility declines. Therefore, this chapter examines the fertility preferences and outcomes of interethnically and interreligiously married couples. Social Groups and Fertility Preferences and Outcomes Globally and historically, social groups function as a key organizer of social networks and information and resource flows, resulting in distinctive fertility preferences and behaviors along social group lines, such as ethnicity and religion. Social interactions, which form and reinforce norms and values, happen within ethnic groups due to the linguistic and geographic clustering of ethnicities (Bongaarts and Watkins 1996; Watkins 1987). Similarly, social interaction within religious groups promotes the assumption and maintenance of shared norms and values, especially when geographically clustered (Heaton 1986; McQuilian 2004). Thus, a social group’s ability to have and maintain distinct values and behaviors is attributable to norms and preferences shared through linguistic connection, social networks, and a general sense of group cohesion (Addai 1996; Caldwell 2001; Goldscheider and Uhlenberg 1969; Heaton 1986; Kohler, Behrman, and Watkins 2001; McQuilian 2004; Smith 2011). Additionally, different social groups often have unequal access to resources, which further distinguishes their demographic behaviors. Across the globe, ethnic and religious groups have 76 differential access to resources and information due to linguistic, educational, and other structural barriers. In much of West Africa, resources and information are unevenly distributed across social groups, creating notable inequalities across ethnic and religious lines, including in educational attainment and wealth. (Bukari and Schareika 2015; Houngnikpo and Decalo 2013; Mberu and Reed 2014; Odimegwu and Adedini 2017; Ononokpono et al. 2016). This inequality, in turn, shapes social groups’ distinct fertility preferences and behaviors (Booth 2010; Grace and Sweeney 2016; Gubhaju, Jongstra, and Raikoti 2014; Heaton 1986; Weinreb 2001). Ethnicity and fertility Different ethnic groups in West Africa often have distinct childbearing norms, patterning individuals’ fertility preferences (Castle 2001; DeRose and Ezeh 2005; Odimegwu and Zerai 1996; Takyi and Dodoo 2005) and outcomes (Garenne and Zwang 2006; Mberu and Reed 2014; Oni 2006). These distinct trends are due to both structural inequalities that fall along ethnic lines, as well as due to cultural preferences and practices that are distinct from neighboring ethnic groups (Booth 2010; Goldscheider and Uhlenberg 1969; Sly 1970; Watkins 1991; Weinreb 2001). For example, the Yoruba of Benin and Nigeria historically observed particularly long postpartum abstinence periods, leading to longer birth spacing than for other ethnic groups (Caldwell et al. 1991; Feyisetan 1990). Importantly, in West Africa, ethnicity remains heavily tied to fertility preferences and practices for both men and women even after accounting for structural factors (Addai and Trovato 1999; Caldwell et al. 1992; DeRose and Ezeh 2005; Isiugo- Abanihe 2014; Johnson-Hanks 2003; Mberu and Reed 2014; Nahmias 2007; Smith 2011; Speizer 1999). However, the role ethnicity plays in organizing social life and information flows may be shifting with broader social changes (Kohler et al. 2001; Ononokpono et al. 2016; Shapiro and 77 Tambashe 2017). As ethnic groups become less isolated, whether due to the presence of large markets (Kohler et al. 2001) or increasing urbanization (Ononokpono et al. 2016; Shapiro and Tambashe 2017), fertility preferences and behaviors are homogenizing across ethnic lines. These diverse networks offer new information and references with which to form fertility preferences and enact fertility behaviors (Behrman et al. 2002; Ononokpono et al. 2016) and living in an urban center often increases access to resources. As a result, people living in urban centers often have similar preferences and behaviors regardless of ethnic background, whereas people in rural areas — where ethnic groups remain particularly clustered — maintain ethnically distinct fertility preferences and behaviors (Ononokpono et al. 2016; Shapiro and Tambashe 2017). Thus, ethnic differentials in fertility preferences and outcomes may be in flux. Religion and fertility Fertility preferences and childbearing outcomes frequently differ by religious affiliation in West Africa (Adongo, Phillips, and Binka 1998; Caldwell and Caldwell 1987; Gyimah et al. 2008; Hallgren 1993; Heaton 2011; Johnson-Hanks 2006). These differences partially stem from structural inequalities tied to majority or minority status within a population (Johnson-Hanks 2006). Different preferences are also related to religious notions of the ideal family (Adongo et al. 1998; Caldwell and Caldwell 1987; Gyimah et al. 2008; Hallgren 1993; Heaton 2011), as well as the interaction of religion and various country contexts (Heaton 2011; Johnson-Hanks 2006). However, even after accounting for socioeconomic factors, variation in fertility preferences often remain, indicating religion’s important and salient role in shaping fertility (Gyimah et al. 2008; Heaton 2011). Fertility preferences and outcomes are often distinct within countries between people of different religious affiliations: for example, one study finds that adherents of 78 Traditional religions have significantly higher fertility outcomes than Christians and Muslims in Ghana, even after accounting for socioeconomic differences (Gyimah et al. 2008). Similar to the shifting salience of ethnicity in some contexts, religion’s role in organizing social life and information flows may be in flux with broader social changes (Crespin-Boucaud 2018; Heaton 2011). As people urbanize, religion may become a more important social identity for networking than ethnicity, which in turn might reinforce religiously grounded fertility preferences (Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). However, even in more rural areas where religious clustering is common, Traditional religion’s emphasis on large family sizes and disapproval of contraceptive usage is changing in response to outside information and socioeconomic conditions (Adongo et al. 1998), suggesting that increased network homogamy may not reinforce religious differences. Further, increased education may diminish religious differentials in fertility (Johnson-Hanks 2006), although there is evidence that increased education may not completely eliminate these differences (Heaton 2011). Thus, although religion retains an important role in organizing social lives, especially in urban areas, religious differences in fertility preferences and outcomes may be shifting with broader social changes. Couples’ Fertility Preference Concordance Fertility is a shared project that involves couples negotiating and compromising with each other (Voas 2003). While diverse networks may change individual preferences, we know little about how ethnic and religious diversity within couples shapes fertility. Fertility preferences are most frequently asked of women exclusively. Yet as men and women are both involved in the enactment of these preferences — such as the total number of children a woman has — measuring preferences and whether they diverge at the couple-level is important in 79 understanding how fertility outcomes become a reality (Andro et al. 2002; Gyimah et al. 2008; McNamee 2009; Speizer 1999; Voas 2003). As a result of the coupled-nature of fertility, fertility preferences are sometimes used as an indirect proxy to measure the level and quality of communication within couples (Andro et al. 2002; Feyisetan 2000; Ibisomi and Odimegwu 2011; McNamee 2009; Voas 2003). Fertility preferences are a good proxy for couple-level communication because fertility preferences and desires are fluid. People change their preferences in response to new information, including changing economic circumstances, as well as their partners’ preferences — both assumed and confirmed (Debpuur and Ayaga 2002; Feyisetan 2000; Kodzi, Johnson, and Casterline 2010; Lasee and Becker 1997; Sennott and Yeatman 2018; Yeatman and Sennott 2014; Yeatman, Sennott, and Culpepper 2013). While people do select marital partners based on their fertility preferences, a larger force on fertility preferences is the change that occurs within relationships, with preferences most frequently converging towards their partners’, as opposed to diverging further (Yeatman and Sennott 2014). Despite evidence that partners’ fertility preferences often converge towards each other, in West Africa, husbands and wives frequently report significantly divergent fertility preferences (DeRose, Dodoo, and Patil 2002; Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Isiugo- Abanihe 2014). In situations of discordance, men’s preferences may win out over women’s due to gendered power imbalances in support from family, community, and existing social norms (Andro et al. 2002; Behrman et al. 2002; Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Isiugo-Abanihe 2014; Kritz and Makinwa-Adebusoye 1999; Speizer 1999). 80 Heterogamy, Couples, and Fertility Preferences and Outcomes As most marriages in sub-Saharan Africa are ethnically and religiously homogamous, studies on couples’ fertility compromises are focused on couples who share the same cultural backgrounds. However, the negotiation of fertility decisions may be different for interethnic and interreligious couples (Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Isiugo-Abanihe 2014; Lehrer 1996). Both ethnicity and religion play a role in determining a woman’s authority in the household and what an acceptable compromise between partners might look like (Ibisomi and Odimegwu 2011; Isiugo-Abanihe 2014; Kritz and Makinwa-Adebusoye 1999). Therefore, ethnicity’s and religion’s salient roles in forming ideas around fertility preferences and compromises between spouses may cause interethnic and interreligious couples to have distinct fertility preferences and outcomes. Fertility preference hypotheses If ethnicity and religion are salient in shaping fertility preferences, interethnic and interreligious couples may have to compromise more in their desired fertility than intramarried couples. If interethnic and interreligious couples successfully communicate and compromise, they may experience lower levels of fertility preference discordance than intramarried couples. Contrarily, failed communication and a lack of compromise puts interethnic and interreligious couples at increased risk of discordance compared to intramarried couples. Given the ways fertility preferences may vary for interethnic and interreligious couples compared to those who are intramarried, I propose three competing hypotheses to explain fertility preferences in interethnic and interreligious marriages. First, interethnic and interreligious couples may experience lower levels of fertility preference concordance (Hypothesis 1a) than their intramarried counterparts. Due to the differing natalist orientations of 81 each partner’s ethnic and religious affiliation, interethnic and interreligious couples may face greater challenges in reconciling their fertility preferences than intramarried couples (Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Lehrer 1996). Normally, couples’ preferences converge (Yeatman and Sennott 2014). However, due to ethnicity’s and religion’s role in forming fertility norms, intermarried couples may be less willing to move their fertility preferences to match their partner’s, thus experiencing higher levels of conflict resulting in lower preference concordance (Lehrer 1996). Alternatively, interethnic and interreligious couples may experience higher levels of fertility preference concordance (Hypothesis 1b) relative to their intramarried counterparts. People marrying interethnically or interreligiously may select partners whose fertility goals match their own (Yeatman and Sennott 2014). Additionally, without shared ethnic and religious backgrounds to help mediate differences in fertility preferences, interethnic and interreligious couples may have to communicate more about their preferences. This increased communication, assuming it is without conflict, could increase the likelihood of preference concordance. If this is the case, interethnic and interreligious couples will exhibit a smaller preference gap and higher odds of concordance than intramarried couples. Furthermore, there may be no association (Hypothesis 1c) between interethnic and interreligious marriage and fertility preference concordance, meaning fertility preference concordance will be as likely among these couples as among intramarried couples. This may be because the gendered preference-gap between husbands and wives may be particularly salient and difficult to close, rather than ethnicity and religion playing key roles in preference concordance (Ibisomi and Odimegwu 2011). 82 Fertility outcomes hypotheses As being of different ethnic and religious backgrounds may shape fertility preferences, these preferences and compromises might, in turn, differentially shape fertility outcomes (Gyimah, Takyi, & Tenkorang, 2008; Ibisomi & Odimegwu, 2011; Lehrer, 1996). For example, some studies looking at religiously discordant couples find that they experience fertility negotiations differently than intrareligious couples and experience distinct fertility outcomes as a result (Gyimah et al. 2008; Lehrer 1996). However, if social identity is not salient in these compromises, fertility differences between interethnic and interreligious couples and their intramarried counterparts will not be significant (Ntoimo and Mutanda 2017). This may be because the partner who desires a more normative number of children will see their preferences win out due to support from their social networks (Voas 2003), regardless of the couple’s social identities. As social identities may be significant in determining what ideal fertility looks like, I propose two competing hypotheses. If ethnicity and religion are particularly salient in fertility outcomes, interethnic and interreligious couples will have higher or lower fertility (Hypothesis 2a) than their intermarried counterparts due to compromising (Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Lehrer 1996). Additionally, interethnic and interreligious marriages experience more scrutiny and are likely at increased risk of dissolution, which may affect their fertility compromises and outcomes. Interethnically and interreligiously married couples may have higher fertility in an attempt to legitimize the union and strain itself may lead to increased accidental pregnancies, potentially increasing fertility rates among interethnic and interreligious couples (Bean and Aiken 1976). Lower fertility, on the other hand, would minimize the costs of leaving the union in the event of a dissolution (Lehrer 1996). Thus, if different ethnic and 83 religious identities are salient in shaping fertility, interethnic and interreligious marriages will likely experience different fertility outcomes from their intramarried counterparts. Alternatively, the second hypothesis is one of no difference (Hypothesis 2b): interethnic and interreligious marriages will not differ in fertility outcomes from their intramarried counterparts. For example, there is evidence from Ghana and Nigeria that completed fertility was the same for both inter- and intraethnically married couples (Ntoimo and Mutanda 2017). This may be for several reasons, including the difficulty of accessing contraception which may lead to similar fertility rates regardless of preferences, as well as the gendered processes that allow men’s fertility preferences to win-out over his wife’s regardless of the ethnic and religious composition of the marriage (Ibisomi and Odimegwu 2011). Given the potential for ethnicity’s and religion’s salience in shaping fertility preferences and outcomes, I take a comprehensive approach using multiple indicators to estimate the association of interethnic and interreligious marriage on fertility preferences and outcomes. I look at ideal family size to see if it differs for men and women in inter- versus intramarriages. Using the difference between men’s and women’s ideal family size, I estimate preference concordance at the couple-level. I also compare couple-level concordance for the desire to have another child. To test if interethnic and interreligious marriages also influence fertility outcomes, I compare the completed fertility of inter- and intramarried women. 84 Study Context I begin with a brief introduction into ethnic and religious fertility-related differentials in Benin to understand how ethnicity and religion may be associated with fertility in this context. Ethnicity and fertility preferences and outcomes in Benin Appendix 4 Table 1 shows the ethnic differentials in fertility preferences and outcomes in Benin. Adja women with two or fewer children 12 want, for example, an average of 4.5 children, which is markedly higher than their Fon neighbors (4.0 children), who have the lowest desired number of children. Peuhl women, on the other hand, express a desire for 6.6 children, on average, which is the highest of any ethnic group. Similarly, there are differences between ethnic groups in women’s expression of a non-numeric 13 fertility preference. Only 3.4% of Adja women gave a non-numeric response in contrast to a full 20% of Dendi women. Beninese women experience high fertility regardless of their ethnicity. However, fertility levels do differ by ethnic background, although less markedly than for fertility preferences. Among women who are finishing their reproductive careers (40-49 years old 14 ), Fon women (the largest ethnic group) have the lowest average fertility levels at 6.2 children. These averages map on to those of women from the other southern ethnic groups — the Adja (6.4 children) and the Yoruba (6.3 children). However, women from northern ethnic groups have higher average fertility levels — reaching an average of 7.5 children among the Betamaribé. These distinctions have grown over time as southern and northern ethnic groups had similar completed fertility in 12 Only 2% of Beninese women state that their ideal family size is below two children so these women are unlikely to be reporting preferences that map on perfectly to their current parity. 13 This means the woman responded with “as many as possible,” “up to god,” or another non- numeric answer. 14 Only 29% of women gave birth within 5 years of turning 40, whereas 57% of women gave birth within 5 years of turning 35. 85 the early DHS surveys (7.3 children for southerners and 7.2 children for northerners in 1996). However, by 2018 southerners’ averages had dropped to 5.9 children as compared to 7.0 children for northerners. Religion and fertility preferences and outcomes in Benin Appendix 4 Table 1 shows the religious differentials in fertility preferences and outcomes in Benin. Among women with two or fewer children, Catholic women report the lowest ideal family size (4.2 children), followed by Protestants (4.5 children), other Christians (4.7 children), and others (4.7 children). Muslim (5.7 children) and Traditional (5.4 children) women, as well as those reporting no religion (5.5 children), have the highest ideal family size preferences. Similarly, Catholic women offer the lowest proportion of non-numeric fertility preferences of any religious group at 2.7%, whereas a full 10% of Muslim women report a non-numeric ideal family size. Beninese women experience high fertility rates across religious denominations. However, among women 40 years and older, there is more than a one-child difference between other religious women (5.8 children) and women who are not religiously affiliated (6.9 children). Of the three main religious traditions, Traditional and Muslim women have the highest fertility rate (6.8 children for both groups) and Catholic women have the lowest (5.9 children). Overall, women from predominantly southern religious groups have lower average fertility than women from northern religious groups. These religious differentials have grown over time. In the 1996 survey, women 40 years and older who were Traditional adherents had an average of 7.4 children, 7.5 children among Muslim women, and 7.2 children among Catholics. However, by the time of the 2018 survey, the fertility outcomes for Catholic women had decreased to 5.2 86 children, whereas it remained at higher levels for both Traditional women (6.9 children) and Muslim women (6.8 children). Sample All analyses include only men, women, and couples with full data on all fertility preferences and fertility histories. For example, men and women with non-numeric ideal family size preferences are not included (7.2% men; 6.1% women), as well as women who are sterilized or declared infecund (3.6%). After including only men and women who report full data for all other control variables, the fertility preference models have a sample size of 8,126. The completed fertility models for women aged 40-49 have a sample of 1,182. Measures Dependent variables Family size preferences. I measure both absolute preferences and couple-level concordance in ideal family size. The first key outcome variable is men’s and women’s individual ideal family sizes. Interviewers asked childless men and women: “If you could choose exactly the number of children to have in your lifetime, how many would that be?” Interviewers asked men and women with children “If you could return to the time when you did not yet have children and you could choose the exact number to have in your life, how many would you have?” The second key outcome variable is the gender difference in ideal family sizes between men and women. I measure this continuously by subtracting women’s ideal family sizes from men’s. Men in sub-Saharan Africa have higher ideal family size preferences than women, on average. However, the gender gap may look different between interethnically and interreligiously married men and women. Additionally, the ability to convert preferences into outcomes is not gender- 87 neutral in sub-Saharan Africa making it important to understand the average preference gap between women and their husbands. As a third key outcome, I code the absolute preference gap between couples continuously. I convert all negative values created by subtracting women’s ideal family sizes from men’s (8.3%) into their absolute value to reflect the absolute difference in the number of children couples want. As a fourth key outcome, I measure couple-level concordance as a binary using two types of concordance — couples who are concordant within zero or one child of each other and couples who are concordant when the child gap is expanded to higher levels — two or four children. I define concordance (=1) as couples reporting the same ideal family size preferences and discordance as those whose gap is larger than the defined concordance gap. Desire to have another child. The fifth key outcome is whether there is couple-level concordance on the desire to have another child. Interviewers asked men and women: “I would now like to ask you questions about the future. Would you like to have a/another child or would you prefer not to have a/another child? How long would you like to wait from now before the birth of a/another child?” I code couples who both report the desire to either have another child or to cease childbearing as concordant (=1) and couples where one partner wishes to cease childbearing while the other wishes to continue as discordant (=0). I code the following responses as wanting another child: wanting one within the next two years, wanting one but unsure of when, or being undecided. I code not wanting another child using the following responses: wanting another child but in more than two years or wanting no more children (McNamee 2009). As a robustness check, I also code the desire to stop childbearing using only couples who explicitly state the desire for no more children (Andro et al. 2002). 88 Number of children ever born. The sixth key outcome is the number of children ever born to the surveyed wife. Women are asked a combination of questions to understand their total parity including asking how many children reside with them, how many they have birthed who do not reside with them, and how many children they have lost. To confirm the total of these three questions, interviewers asked “I would like to ensure that I have fully understood you have had a total of ___ births in your life. Is that correct?” To robustly capture completed fertility, I measure fertility at ages 35-49, 40-49, and 45-49. I exclude women younger than 35 as they are unlikely to have reached their final parity meaning they are likely to have additional children in the future. Key independent variables Based on the robust literature on fertility preferences and outcomes in sub-Saharan Africa, in addition to controls for whether the couple is interethnic or interreligiously married, I also include a standard litany of control variables. Ethnicity of wife and husband. To examine ethnicity’s link to fertility, I control for the ethnicity of each spouse in their respective individual models. Religion of wife and husband. Similarly, to examine religion’s link to fertility, I control for the religion of each spouse in their respective individual models. Decade married. I measure how fertility preferences and outcomes are linked to time. To track this, I make use of data on women’s date of marriage to create a categorical indicator of the decade married ranging from pre-1980 to 2010 or later based on their response to the following question: “In which month and year did you start to live with your husband/partner?” Educational attainment. My main measure of educational attainment is a binary of having ever attended school due to overall low school attendance in Benin. Among women surveyed, 89 76% have never attended nor have 54% of men. Due to the large discrepancy in access to schooling for boys and girls, at the couple-level I measure educational discordance with a categorical variable coded as neither wife nor husband are educated (=0), wife and husband are educated (=1), husband is educated but wife is not (=2), and wife is educated but husband is not (=3). Due to parents prioritizing sending their sons to school over their daughters, few women (N=461 or 4.6%) have been to school when their husbands have not, meaning the results of this group should be interpreted with caution. However, because fertility preferences and outcomes are strongly correlated with schooling, particularly for women, these couples may have distinct preferences and outcomes. I also measure the absolute number of years of schooling attended individually for both husbands and wives as well as measuring education at the couple-level by including controls for his and her years of schooling. Age heterogamy. The average age gap between husbands and wives is largest in sub-Saharan Africa. Age heterogamy is correlated with earlier marital age for women, lower contraceptive usage, and higher parity (Barbieri and Hertrich 2005; Takyi 2001; Togunde 1998). African men overwhelmingly marry at later ages than their female counterparts (88% in the couples’ data), and, as a result, only 5% of women are older than their husbands. I code age homogamy (=0) as partners who are within five years of each other in age. When the husband is older, I code age heterogamy as couples where there is a five- to nine-year age gap between spouse(s) (=1) and an age gap of ten or more years (=2). Due to the limited number of women five or more years older than their husbands (N=86 or 0.9%), I only create one grouping for women who are older than their husbands by at least five years (=3). The results of this last category should, therefore, be interpreted with caution. 90 Age. In all individual models, I control for the individual ages of the husband or wife at the time of survey. Polygyny. As polygyny may shape the fertility preferences of men and women, and couples’ fertility outcomes, I include a control for polygynous marriage. To measure monogamy and polygyny, interviewers asked women: “Does your husband/partner have other wives or live with other women as though they were married?” I code women who do not report their husband/partner as having other wives as monogamously married and women who respond affirmatively as polygynously married. Duration of marriage. Prior work finds that women’s ideal number of children is greatly shaped by the amount of time she has been married (Bongaarts and Casterline 2012). Further, the longer she has been married, the longer she has been exposed to the risk of childbearing. To account for this, I code the duration of marriage continuously based on the number of years the wife reports being married. Number of living children. In the fertility preference analyses, I include a control for the wife’s number of living children. Women’s stated fertility preferences rarely dip below the number of children they currently have (Bongaarts and Casterline 2012) and women who have lost a child may be more likely to desire another birth. Location of residence. Fertility preferences and outcomes may be distinct based on whether the couple resides in an urban or rural location due to differences in space for large families, occupations, and the overall costs associated with raising children. However, urbanicity is not a static feature and may have changed since starting and/or completing childbearing. Due to the large north-south divide in Benin regarding cultures, religions, and resources, I also control for whether the couple resides in the north (=1) or the south (=0). 91 Household wealth. To measure household wealth, I use the wealth index score, which is a composite measure of a household living standard based on weighting assets and services available in the household, such as water sources and materials used to construct the home (Rutstein and Johnson 2004). It is not a measure of income but resources available to all household members (Rutstein and Johnson 2004). As wealth is measured at the time of the survey, it is not reflective of the wealth of couples at the time of their marriage nor is wealth independent of the number of children a family has as families with fewer children may be better able to accumulate wealth than families with more. Analytic Approach I examine the relationship between interethnic and interreligious marriage and fertility preferences and outcomes in several ways. I use ordinary least squares linear regression to model individual men’s and women’s ideal number of children, to examine the numeric difference between men’s and women’s preferences by relationship type, as well as to estimate the completed fertility of couples. In models estimating concordance, I begin with Kappa statistics to measure the level of agreement between couples using the agreement categories presented in Viera & Garrett (2005). I then use multivariate logistic regression models for indicators that estimate the likelihood of interethnic and interreligious couples’ preferences aligning compared to their intramarried counterparts: binaries of ideal number of children concordance and the desire to have (or not have) another child. All results are presented with four additive modeling strategies: 1) baseline models for interethnic and interreligious couples, 2) controls for characteristics that are unlikely to have changed since marriage, 3) marital characteristics, and 4) urbanicity, region, and wealth. 92 Results Table 4.1 offers descriptive statistics of key fertility variables of interest for interethnic couples and Table 4.2 offers the same for interreligious couples. Of note are a few key patterns — generally, men desire higher numbers of children than women both in terms of the average ideal family size and the frequency with which larger numbers are selected. Further, polygynous men report having more children than women, however, the number of children is similar for monogamous couples. Interethnically married men and women have, on average, lower ideal family size preferences and slightly lower parity levels than intraethnically married couples. Across genders and ethnic composition of marriage, reports for the desire to have another child are similar. There are no noticeable differences in ideal family size preferences by the religious composition of marriage except for the frequency of non-numeric responses: 4.8% of interreligiously married women report non-numeric preferences as compared to 6.1% of intrareligiously married women. Interreligiously married men and women report higher average parities at 40 years or older than their intrareligiously married counterparts. 93 Table 4.1: Means and percentages of fertility variables of interest by ethnic composition of marriage (Standard deviations are in parentheses) Men Women All Intra Inter All Intra Inter Average Ideal Family Size 1 8.21 8.43 6.52 5.34 5.43 4.69 (6.52) (6.64) (5.24) (2.35) (2.37) (2.05) Ideal Family Size Groupings 1 (%) 0 0.84 0.91 0.22 2.19 2.20 2.09 1 0.17 0.17 0.22 0.30 0.30 0.22 2 3.75 3.24 7.81 5.38 4.84 9.68 3 5.67 5.29 8.69 5.02 4.64 8.03 4 19.41 18.44 27.06 30.67 29.53 39.71 5 10.26 10.05 11.99 12.59 13.00 9.35 6+ 59.89 61.90 44.00 43.86 45.49 30.91 Non-numeric 7.42 7.28 8.08 5.72 5.47 6.63 Desire To Have Another Child 1 (%) 37.76 37.90 36.63 37.28 37.29 37.18 Parity (all ages) 1 6.35 6.53 5.00 3.88 3.95 3.31 (5.34) (5.43) (4.32) (2.33) (2.36) (1.96) Parity (39+) 2 9.64 9.84 7.87 6.43 6.51 5.68 (6.12) (6.19) (5.15) (2.41) (2.43) (2.05) Parity (monogamous & all ages) 3 4.29 4.38 3.70 3.61 3.69 3.12 (3.09) (3.14) (2.65) (2.27) (2.31) (3.12) Parity (monogamous & 35+) 4 6.66 6.75 5.94 6.36 6.49 5.39 (3.41) (3.45) (2.95) (2.49) (2.52) (2.06) Source: The Demographic and Health Surveys Program 1 N=8126 2 N=1182 3 N=5202 4 N=1143 94 Table 4.2: Means and percentages of fertility variables of interest by religious composition of marriage (Standard deviations are in parentheses) Men Women All Intra Inter All Intra Inter Average Ideal Family Size 1 8.21 8.29 8.01 5.34 5.38 5.25 (6.52) (6.63) (6.24) (2.35) (2.40) (2.22) Ideal Family Size Groupings 1 (%) 0 0.84 0.83 0.85 2.19 2.32 1.84 1 0.17 0.22 0.04 0.30 0.29 0.31 2 3.75 3.56 4.26 5.38 5.28 5.65 3 5.67 5.70 5.60 5.02 5.00 5.06 4 19.41 19.19 19.99 30.67 29.84 32.86 5 10.26 10.55 9.50 12.59 12.81 12.01 6+ 59.89 59.95 59.75 43.86 44.46 42.27 Non-numeric 7.42 7.42 7.43 5.72 6.06 4.84 Desire To Have Another Child 1 (%) 37.76 38.69 35.28 37.28 38.30 34.56 Parity (all ages) 1 6.35 6.41 6.22 3.88 3.86 3.92 (5.34) (5.44) (5.06) (2.33) (2.30) (2.41) Parity (39+) 2 9.64 9.65 9.60 6.43 6.28 6.82 (6.12) (6.32) (5.55) (2.41) (2.42) (2.32) Parity (monogamous & all ages) 3 4.29 4.27 4.32 3.61 3.59 3.66 (3.09) (3.04) (3.21) (2.27) (2.24) (2.35) Parity (monogamous & 35+) 4 6.66 6.51 7.05 6.36 6.20 6.78 (3.41) (3.34) (3.55) (2.49) (2.52) (2.37) Source: The Demographic and Health Surveys Program 1 N=8126 2 N=1182 3 N=5202 4 N=1143 Individual fertility preferences Men’s ideal family size preferences Men, on average, report an ideal family size of 8.2 children. There is a nearly two-child difference between intraethnically married men (8.4) and interethnically married men (6.5), as shown in Table 4.1. Intrareligiously married men report an average ideal family size of 8.3 versus 8.0 for interreligiously married men (Table 4.2). Among interethnic men, this gap is significant in a baseline model without controls. In Model 2 (Table 4.3), after introducing controls for men’s individual characteristics, interethnically married men continue to report significantly lower ideal family sizes. After the introduction of marital controls in Model 3 and full controls in Model 4, interethnically married men’s ideal 95 family sizes remain significantly smaller than intraethnically married men’s, but the effect size is smaller than in previous models. Contrarily, the gap in ideal family size preferences among inter- and intrareligiously married men does not reach significance in any of the models, including the baseline model (Table 4.4). In the individual models for both interethnic and interreligious ideal family sizes, Adja, Bariba, Dendi and Peuhl men all have significantly higher ideal family size preferences than their Fon counterparts. Compared to male Traditional adherents, men from all other religious backgrounds have significantly lower ideal family size preferences. Men who have attended school have significantly lower ideal family sizes and this is robust to when men’s years of education are included instead. The duration of marriage is non-significant. Compared to men married before 1980, all men in subsequent decades have significantly lower ideal family sizes. Increased age at survey, polygyny, and increased parity all serve to increase men’s ideal family size preferences. Compared to men in the poorest wealth quintile, men in the second wealth quintile report higher ideal family sizes, whereas men in the fourth and fifth quintiles — the rich and richest — have significantly lower ideal family size preferences. Rural men have significantly higher preferences than men in urban areas. The largest difference between the interethnic and interreligious models is that residing in the north is non-significant for interethnically married men whereas it significantly increases the ideal family size of interreligiously married men. 96 Table 4.3: Men’s ideal family size — Interethnic (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interethnic -1.91*** -1.09*** -0.76*** -0.47* (0.23) (0.21) (0.20) (0.21) Ethnicity Fon (ref) -- -- -- Adja 1.93*** 1.69*** 1.67*** (0.21) (0.20) (0.20) Bariba 1.79*** 1.60*** 0.90* (0.28) (0.27) (0.42) Betamaribé 1.36*** 1.39*** 0.61 (0.27) (0.26) (0.42) Dendi 4.04*** 3.84*** 3.31*** (0.44) (0.42) (0.51) Peuhl 3.44*** 3.16*** 2.28*** (0.35) (0.33) (0.45) Yoa/Lokpa 1.40*** 1.04** 0.31 (0.38) (0.36) (0.47) Yoruba -0.27 -0.38 -0.49* (0.24) (0.23) (0.23) Religion Traditional (ref) -- -- -- Muslim -1.76*** -1.20*** -0.97*** (0.28) (0.27) (0.27) Catholic -2.98*** -2.33*** -2.09*** (0.21) (0.21) (0.21) Protestant -2.99*** -2.25*** -2.06*** (0.32) (0.30) (0.30) Other Christian -2.48*** -1.80*** -1.70*** (0.23) (0.22) (0.22) Other -3.41*** -2.90*** -2.79*** (0.66) (0.62) (0.62) No religion -2.02*** -1.48*** -1.32*** (0.34) (0.32) (0.32) Educated -2.79*** -2.11*** -1.78*** (0.14) (0.14) (0.14) Duration of Marriage -0.03 -0.03 (0.02) (0.02) Decade of Marriage Pre-1980s (ref) -- -- 1980s -1.04** -1.02** (0.33) (0.33) 1990s -1.34*** -1.34*** (0.34) (0.33) 2000s -1.33*** -1.25** (0.36) (0.36) 2010s -0.97* -0.85* (0.42) (0.42) 97 Table 4.3 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Age at Survey 0.06*** 0.07*** (0.01) (0.01) Polygynous 3.15*** 3.11*** (0.14) (0.14) Number of Children 0.30*** 0.27*** (0.05) (0.05) Household Wealth Quintile Poorest (ref) -- Poor 0.39* (0.19) (0.19) Middle -0.31 (0.20) Rich -0.48* (0.21) Richest -0.58* (0.23) Northern 0.63 (0.35) Rural 0.91*** (0.15) Constant 8.43*** 10.58*** 7.00*** 6.05*** (0.08) (0.19) (0.51) (0.54) Obs. 8126 8126 8126 8126 R-squared 0.01 0.17 0.25 0.26 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 98 Table 4.4: Men’s ideal family size — Interreligious (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interreligious -0.29 -0.22 -0.17 -0.11 (0.16) (0.16) (0.15) (0.15) Religion Traditional (ref) -- -- -- Muslim -1.92*** -1.31*** -1.03*** (0.29) (0.27) (0.28) Catholic -3.08*** -2.40*** -2.12*** (0.22) (0.21) (0.21) Protestant -3.03*** -2.27*** -2.07*** (0.32) (0.30) (0.30) Other Christian -2.53*** -1.84*** -1.72*** (0.23) (0.22) (0.22) Other -3.32*** -2.83*** -2.74*** (0.66) (0.63) (0.62) No religion -1.98*** -1.44*** -1.29*** (0.34) (0.32) (0.32) Ethnicity Fon (ref) -- -- -- Adja 1.89*** 1.66*** 1.64*** (0.21) (0.20) (0.20) Bariba 1.84*** 1.63*** 0.86* (0.28) (0.27) (0.42) Betamaribé 1.42*** 1.43*** 0.58 (0.28) (0.26) (0.42) Dendi 3.93*** 3.76*** 3.22*** (0.44) (0.42) (0.51) Peuhl 3.46*** 3.16*** 2.22*** (0.35) (0.33) (0.45) Yoa/Lokpa 1.26** 0.93** 0.19 (0.38) (0.36) (0.47) Yoruba -0.34 -0.43 -0.52* (0.24) (0.23) (0.23) Educated -2.88*** -2.17*** -1.80*** (0.14) (0.14) (0.14) Duration of Marriage -0.03 -0.03 (0.02) (0.02) Decade of Marriage Pre-1980s (ref) -- -- 1980s -1.04** -1.02** (0.33) (0.33) 1990s -1.34*** -1.35*** (0.34) (0.33) 2000s -1.35*** -1.26** (0.36) (0.36) 2010s -0.97* -0.85* (0.42) (0.42) 99 Table 4.4 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Age at Survey 0.06*** 0.06*** (0.01) (0.01) Polygynous 3.18*** 3.12*** (0.14) (0.14) Number of Children 0.31*** 0.27*** (0.05) (0.05) Household Wealth Quintile Poorest (ref) Poor 0.38* (0.19) Middle -0.32 (0.20) Rich -0.49* (0.21) Richest -0.64** (0.23) Northern 0.68* (0.35) Rural 0.93*** (0.15) Constant 8.29*** 10.65*** 7.10*** 6.10*** (0.08) (0.20) (0.52) (0.55) Obs. 8126 8126 8126 8126 R-squared 0.00 0.17 0.25 0.26 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 Women’s ideal family size preferences On average, women have an ideal family size preference of 5.3 children (Table 4.1). Intraethnically married women have an average preference of 5.4 as compared to 4.7 for interethnically married women. This is a significant difference in the baseline model shown in Table 4.5. Intrareligiously married women have an average ideal family size of 5.4 as compared to the slightly lower 5.3 for interreligiously married women (Table 4.2). In the baseline model (Table 4.6), this difference is significant. In all models presented in Table 4.5, including those with full controls, interethnically married women have significantly lower ideal family sizes compared to their intraethnically married counterparts. However, the significance of the baseline model among interreligiously 100 married women is rendered non-significant once individual characteristics are included in Model 2 (Table 4.6). The addition of ethnicity alone renders this difference non-significant. The control variables for both interethnic and interreligious models work in the expected directions. Compared to Fon women, women from all other ethnic groups have significantly higher ideal family size preferences, all else being equal, barring the Dendi in interreligious models whose difference is non-significant. Women of all religions have lower ideal family size preferences than women who report being Traditional adherents. However, this loses significance for women who report no religion when full controls are included. In the interreligious full controls model, other religious backgrounds are also non-significant. Education significantly decreases women’s ideal family sizes and in models controlling for women’s years of education (not shown), women’s ideal family size decreases by approximately 0.07 children for every year of education she receives. The duration of marriage is not significant. Women married since the 2000s have significantly lower ideal family sizes than women married before the 1980s. Polygyny is associated with an increase in ideal family size, as is the number of living children a woman has. Only the richest wealth quintile is associated with a significant decrease in ideal family size as compared to the poorest households. Women residing in the north or rural areas express significantly higher ideal family sizes. 101 Table 4.5: Women’s ideal family size — Interethnic (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interethnic -0.74*** -0.51*** -0.42*** -0.32*** (0.08) (0.08) (0.08) (0.08) Ethnicity Fon (ref) -- -- -- Adja 0.46*** 0.44*** 0.43*** (0.08) (0.07) (0.07) Bariba 1.02*** 1.00*** 0.56*** (0.10) (0.10) (0.15) Betamaribé 1.39*** 1.34*** 0.87*** (0.10) (0.10) (0.15) Dendi 0.84*** 0.85*** 0.46* (0.16) (0.15) (0.19) Peuhl 2.27*** 2.18*** 1.69*** (0.12) (0.12) (0.16) Yoa/Lokpa 1.30*** 1.25*** 0.79*** (0.14) (0.14) (0.18) Yoruba 0.34*** 0.31*** 0.25** (0.09) (0.09) (0.09) Religion Traditional (ref) -- -- -- Muslim -0.49*** -0.41*** -0.36** (0.10) (0.10) (0.10) Catholic -0.67*** -0.58*** -0.51*** (0.08) (0.08) (0.08) Protestant -0.44*** -0.37** -0.33** (0.12) (0.12) (0.12) Other Christian -0.31*** -0.24** -0.19* (0.08) (0.08) (0.08) Other -0.56*** -0.47* -0.43* (0.19) (0.19) (0.19) No religion -0.28* -0.23* -0.20 (0.11) (0.11) (0.11) Educated -0.80*** -0.54*** -0.42*** (0.06) (0.06) (0.06) Duration of Marriage -0.01 -0.01 (0.01) (0.01) Decade of Marriage Pre-1980s (ref) -- -- 1980s -0.09 -0.08 (0.13) (0.13) 1990s -0.23* -0.23 (0.13) (0.13) 2000s -0.41** -0.39** (0.14) (0.14) 2010s -0.39* -0.37* (0.16) (0.16) 102 Table 4.5 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Age at Survey -0.02** -0.01 (0.01) (0.01) Polygynous 0.24*** 0.23*** (0.05) (0.05) Number of Children 0.24*** 0.23*** (0.02) (0.02) Household Wealth Quintile Poorest (ref) -- Poor 0.05 (0.07) Middle -0.07 (0.08) Rich -0.13 (0.08) Richest -0.23** (0.09) Northern 0.45*** (0.13) Rural 0.28*** (0.06) Constant 5.43*** 5.42*** 5.32*** 5.00*** (0.03) (0.07) (0.21) (0.22) Obs. 8126 8126 8126 8126 R-squared 0.01 0.14 0.17 0.18 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 103 Table 4.6: Women’s ideal family size — Interreligious (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interreligious -0.14* -0.12 -0.12 -0.11 (0.06) (0.06) (0.06) (0.06) Religion Traditional (ref) -- -- -- Muslim -0.55*** -0.46*** -0.40*** (0.11) (0.10) (0.11) Catholic -0.71*** -0.61*** -0.52*** (0.08) (0.08) (0.08) Protestant -0.43** -0.35** -0.30** (0.12) (0.12) (0.12) Other Christian -0.30** -0.23** -0.18* (0.08) (0.08) (0.08) Other -0.49* -0.40* -0.36 (0.19) (0.19) (0.19) No religion -0.24* -0.19 -0.16 (0.12) (0.12) (0.12) Ethnicity Fon (ref) -- -- -- Adja 0.43*** 0.41*** 0.41*** (0.08) (0.07) (0.07) Bariba 1.04*** 1.01*** 0.53** (0.10) (0.10) (0.15) Betamaribé 1.40*** 1.34*** 0.84*** (0.10) (0.10) (0.15) Dendi 0.76*** 0.78*** 0.37 (0.16) (0.15) (0.19) Peuhl 2.27*** 2.18*** 1.65*** (0.12) (0.12) (0.16) Yoa/Lokpa 1.28*** 1.22*** 0.74*** (0.14) (0.14) (0.18) Yoruba 0.28** 0.26** 0.21** (0.09) (0.09) (0.09) Educated -0.85*** -0.57*** -0.43* (0.06) (0.06) (0.06) Duration of Marriage -0.00 -0.01 (0.01) (0.01) Decade of Marriage Pre-1980s (ref) -- -- 1980s -0.09 -0.08 (0.13) (0.13) 1990s -0.23* -0.23 (0.13) (0.13) 2000s -0.41** -0.39** (0.14) (0.14) 2010s -0.39* -0.37* (0.16) (0.16) 104 Table 4.6 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Age at Survey -0.02** -0.01* (0.01) (0.01) Polygynous 0.25*** 0.23*** (0.05) (0.05) Number of Children 0.24*** 0.23*** (0.02) (0.02) Household Wealth Quintile Poorest (ref) -- Poor 0.05 (0.07) Middle -0.08 (0.08) Rich -0.15 (0.08) Richest -0.27** (0.09) Northern 0.49*** (0.13) Rural 0.29*** (0.06) Constant 5.38*** 5.45*** 5.37*** 5.02*** (0.03) (0.07) (0.21) (0.22) Obs. 8126 8126 8126 8126 R-squared 0.00 0.13 0.17 0.18 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 Couple concordance on fertility preferences Ideal family size gaps Intraethnic and intrareligious couples report slight agreement in their ideal family sizes (Kappa= 0.07), with 19% of couples reporting the same ideal family size. Interethnic couples also report slight agreement in their ideal family sizes (Kappa= 0.11) with 26% of couples reporting the same ideal family size, and 20% of interreligious couples agree for slight agreement levels (Kappa= 0.07). As the gender gap in fertility preferences between men and women is largest in sub-Saharan Africa, I first examine the average difference in ideal family size between men and women in Table 4.7. The gender gap in ideal family size preferences between men and women in 105 interethnic marriages (1.8 children) is significantly lower than for couples in intraethnic marriages (3.0 children) in baseline linear models, with men wanting more children than their wives on average. However, these effects disappear in Model 3. The unique addition of any of the variables in this model — the decade of marriage, polygyny, and duration of marriage — is sufficient to render it non-significant. The gender gap between interreligious couples is 2.8 children, which is very close to the gender gap for intrareligious couples, at 2.9 children, and is non-significant in baseline models. Table 4.7: Gender gap in ideal family size (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic -1.17*** -0.46* -0.23 -0.05 (0.22) (0.22) (0.21) (0.21) Interreligious -0.15 -0.11 -0.13 -0.11 (0.16) (0.15) (0.15) (0.15) Education Neither educated (ref) -- -- -- Both educated -3.23*** -2.29*** -1.66*** (0.18) (0.19) (0.21) Husband educated/Wife not -2.47*** -2.09*** -1.77*** (0.16) (0.16) (0.17) Wife educated/ Husband not -1.38*** -0.97** -0.71* (0.33) (0.33) (0.33) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.45** 0.33** 0.31* (0.16) (0.16) (0.16) Husband older >10 years 1.74*** 1.15*** 1.13*** (0.17) (0.17) (0.17) Wife older >5 years 1.65* 1.28 1.27 (0.76) (0.74) (0.74) Duration of Marriage 0.03 0.03* (0.02) (0.02) 106 Table 4.7 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Decade of Marriage Pre-1980s (ref) -- -- 1980s -1.04** -1.05** (0.35) (0.35) 1990s -1.26*** -1.29*** (0.35) (0.35) 2000s -1.00** -1.00*** (0.38) (0.38) 2010s -0.39 -0.46 (0.43) (0.44) Polygynous 3.21*** 3.19*** (0.14) (0.14) Number of Living Children 0.09 0.07 (0.05) (0.05) Household Wealth Quintile Poorest (ref) -- Poor 0.19 (0.20) Middle -0.42* (0.21) Rich -0.56** (0.22) Richest -0.59* (0.24) Northern 0.39** (0.15) Rural 0.62*** (0.15) Constant 3.00*** 2.91*** 3.67*** 2.73*** 2.21*** (0.07) (0.16) (0.15) (0.45) (0.49) Obs. 8126 8126 8126 8126 8126 R-squared 0.00 0.00 0.07 0.13 0.14 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 I also examine the absolute preference gap in couples to see if it varies for interethnic and interreligious couples versus their intramarried counterparts (Table 4.8). In baseline models, the absolute ideal family size gap between interethnic couples (2.8 children) is significantly lower than for intraethnic couples (4.1 children) but, similar to the gender gap, this also loses its significance in Model 4. While no single control variable renders it non-significant, the inclusion of polygyny with decade married, duration of marriage, or the number of living children renders 107 it non-significant. The absolute preference gap is non-significant among interreligiously married couples and their intrareligiously married counterparts. Table 4.8: Absolute ideal family size gap (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic -1.31*** -0.59** -0.34 -0.19 (0.20) (0.20) (0.19) (0.19) Interreligious -0.24 -0.20 -0.21 -0.19 (0.14) (0.14) (0.13) (0.13) Education Neither educated (ref) -- -- -- Both educated -3.28*** -2.29*** -1.64*** (0.16) (0.17) (0.19) Husband educated/Wife not -2.26*** -1.88*** -1.54*** (0.15) (0.14) (0.15) Wife educated/ Husband not -1.50*** -1.00** -0.72* (0.30) (0.29) (0.29) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.25 0.14 0.11 (0.15) (0.14) (0.14) Husband older >10 years 1.59*** 1.02*** 1.00*** (0.16) (0.15) (0.15) Wife older >5 years 1.45* 1.03 1.04 (0.68) (0.66) (0.66) Duration of Marriage 0.04** 0.05** (0.01) (0.01) Decade Married Pre-1980s (ref) -- -- 1980s -0.97** -0.99** (0.31) (0.31) 1990s -1.37*** -1.41*** (0.31) (0.31) 2000s -1.24*** -1.25*** (0.34) (0.34) 2010s -0.65 -0.75 (0.39) (0.39) Polygynous 2.80*** 2.79*** (0.13) (0.13) Number of Living Children 0.15** 0.13** (0.05) (0.05) 108 Table 4.8 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Household Wealth Quintile Poorest (ref) -- Poor 0.23 (0.18) Middle -0.45* (0.18) Rich -0.63** (0.19) Richest -0.58** (0.22) Northern 0.51*** (0.13) Rural 0.52*** (0.14) Constant 4.10*** 4.01*** 4.86*** 3.87*** 3.40*** (0.07) (0.14) (0.14) (0.40) (0.43) Obs. 8126 8126 8126 8126 8126 R-squared 0.01 0.00 0.08 0.15 0.16 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 In both the gender gap models and the absolute gap models, the control variables operate in the expected directions. Education decreases the gap and having a husband who is five or more years older than his spouse significantly increases the preference gap, particularly for age differences of a decade or more. Compared to couples married before the 1980s, all couples have smaller preference gaps, although this does not reach significance for those married in the 2010s. Polygyny significantly increases the gap, as does the number of children the couple has. Increasing wealth decreases the gap compared to the poorest wealth quintile and residing in the north or a rural location increases it. These results offer support for different findings among interethnic and interreligious couples. Interethnic couples have smaller preference gaps compared to their intraethnically married counterparts, supporting hypothesis (1b), higher levels of concordance. Contrarily, 109 interreligious couples offer support for hypothesis (1c) — there is no effect of different religious identities on the size of the preference gap. Likelihood of couple-level fertility preference concordance I use several measures of concordance to robustly capture whether interethnic and interreligious couples are more likely to agree in their preferences than intramarried couples. I begin by modeling low preference gaps, specifically desiring the exact same number of children or within one child of each other. However, Benin is a high fertility context with high ideal family sizes and large gender preference gaps meaning a one-child preference gap may not be a point of contention for couples (Bankole and Singh 1998; Ibisomi and Odimegwu 2011; Voas 2003). Additionally, fertility preferences are dynamic, fluctuating in response to economic and social circumstances (Johnson-Hanks 2004; Kodzi, Casterline, and Aglobitse 2010; Trinitapoli and Yeatman 2018; Yeatman et al. 2013). As a result, larger gaps of two or more children may be more reflective of low levels of couple communication in this context. I, therefore, also expand the definition of concordance in sensitivity analyses to reporting an ideal family size within two children of each other, as well as within four of each other. I also examine whether interethnic and interreligious couples are more likely to agree on their desire to have an additional child as compared to their intramarried counterparts. Relative ideal family size concordance Using logistic regression models (Table 4.9), I estimate the likelihood of exact preference concordance, meaning husbands and wives report the same numeric ideal family size. In a baseline model, interethnic couples are significantly more likely to agree about the number of children they desire than intraethnic couples. However, the addition of controls in Model 4, specifically polygyny, render this non-significant, meaning interethnic couples are as likely to 110 agree about their ideal number of children as intraethnic couples. Interreligious couples, on the other hand, are no more likely to report the same ideal family size than intrareligious couples in baseline models. Table 4.9: Likelihood of reporting the same ideal family size (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic 1.52*** 1.22* 1.16 1.13 (0.12) (0.10) (0.10) (0.10) Interreligious 1.03 1.02 1.03 1.01 (0.06) (0.06) (0.07) (0.07) Education Neither educated (ref) -- -- -- Both educated 2.84*** 2.29*** 1.69*** (0.21) (0.18) (0.15) Husband educated/Wife not 1.88*** 1.73*** 1.43*** (0.13) (0.12) (0.11) Wife educated/ Husband not 1.76*** 1.58** 1.37* (0.24) (0.22) (0.19) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.95 0.96 0.99 (0.06) (0.06) (0.07) Husband older >10 years 0.71*** 0.80** 0.83* (0.05) (0.06) (0.06) Wife older >5 years 0.76 0.84 0.81 (0.26) (0.29) (0.28) Duration of Marriage 0.98** 0.98** (0.01) (0.01) Decade Married Pre-1980s (ref) -- -- 1980s 0.73 0.73 (0.13) (0.13) 1990s 0.76 0.76 (0.13) (0.13) 2000s 0.68* 0.68* (0.12) (0.13) 2010s 0.50** 0.54** (0.10) (0.11) Polygynous 0.50*** 0.50*** (0.04) (0.04) (0.02) (0.02) 111 Table 4.9 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Number of Living Children 0.93** 0.93** (0.02) (0.02) Household Wealth Quintile Poorest (ref) -- Poor 1.00 (0.10) Middle 1.12 (0.11) Rich 1.21 (0.12) Richest 1.25* (0.13) Northern 0.59*** (0.04) Rural 0.87* (0.06) Constant 0.23*** 0.24*** 0.16*** 0.49** 0.62* (0.01) (0.01) (0.01) (0.10) (0.14) Obs. 8126 8126 8126 8126 8126 Pseudo R 2 0.00 0.00 0.04 0.06 0.07 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 Similar results are found for an expanded definition of concordance that allows couples to express concordance within a one-child gap (models not shown). At baseline, interethnic couples are significantly more likely to be concordant. However, the addition of wealth or urbanicity renders interethnic couples no more or less likely to be concordant than intraethnic couples. Interreligious couples are no more likely to be concordant than intrareligious couples at this level. In models with either definition of concordance (zero or one child), the association of the control variables operates in the expected directions. Education increases the likelihood of concordance, while an age gap of at least a decade and each year of marriage decrease the likelihood of concordance. All couples married since the 1980s have lower odds of concordance compared to couples married before the 1980s, although this only reaches significance for 112 couples married since the 2000s. Polygyny, increased parity, residing in the north, and living in rural areas all significantly decrease the likelihood of concordance. Overall, these results indicate that interethnic couples are somewhat more likely to be concordant in their fertility preferences than intraethnic couples, supporting hypothesis (1b), whereas I find no difference in concordance by the religious composition of the marriage, supporting my null hypothesis (1c). Sensitivity analysis of the likelihood of couple concordance To further explore fertility preference concordance among interethnic and interreligious couples, I expand the definition of concordance to a two- (Table 4.10) and four-child (Table 4.11) gap — which better reflect the average difference between men’s and women’s ideal family sizes. In baseline models, interethnic couples are significantly more likely to be concordant than intraethnic couples with the expanded definitions of concordance: within two or with four children of each other. Interreligious couples are no more likely to be concordant at two children. However, they are significantly more likely to be concordant at four. Unlike for lower levels of concordance, in models with full controls, interethnic couples remain significantly more likely to be concordant at two children and four children. Interreligious couples are not more likely to be more concordant at two children. At four children, the likelihood that they are more concordant becomes non-significant in Model 5 with full controls — specifically with the control for region of residence. Without controlling for whether the couple resides in the north or the south, interreligious couples remain more likely to be concordant at four children. 113 Table 4.10: Likelihood of reporting an ideal family size within two children of each other (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic 1.79*** 1.37*** 1.28** 1.20* (0.14) (0.11) (0.11) (0.10) Interreligious 1.09 1.09 1.10 1.07 (0.06) (0.06) (0.06) (0.06) Education Neither educated (ref) -- -- -- Both educated 4.50*** 3.43*** 2.33*** (0.32) (0.25) (0.19) Husband educated/Wife not 2.11*** 1.94*** 1.57*** (0.12) (0.11) (0.09) Wife educated/ Husband not 1.58*** 1.39** 1.17 (0.18) (0.16) (0.14) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.89* 0.92 0.95 (0.05) (0.05) (0.06) Husband older >10 years 0.61*** 0.73*** 0.75*** (0.04) (0.05) (0.05) Wife older >5 years 0.73 0.81 0.77 (0.19) (0.21) (0.20) Duration of Marriage 0.98** 0.98** (0.01) (0.01) Decade Married Pre-1980s (ref) -- -- 1980s 1.12 1.13 (0.14) (0.14) 1990s 1.08 1.12 (0.14) (0.14) 2000s 0.91 0.96 (0.13) (0.13) 2010s 0.71* 0.81 (0.11) (0.13) Polygynous 0.36*** 0.36*** (0.02) (0.02) Number of Living Children 0.92*** 0.93*** (0.02) (0.02) 114 Table 4.10 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Household Wealth Quintile Poorest (ref) -- Poor 0.96 (0.07) Middle 1.11 (0.08) Rich 1.21* (0.10) Richest 1.36** (0.13) Northern 0.60*** (0.03) Rural 0.76*** (0.04) Constant 1.30*** 1.35*** 0.97 2.38*** 3.38*** (0.03) (0.04) (0.05) (0.39) (0.62) Obs. 8126 8126 8126 8126 8126 Pseudo R 2 0.01 0.00 0.07 0.12 0.13 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 115 Table 4.11: Likelihood of reporting an ideal family size within four children of each other (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic 1.96*** 1.47*** 1.36** 1.25* (0.19) (0.15) (0.14) (0.13) Interreligious 1.14* 1.13* 1.15* 1.12 (0.07) (0.07) (0.07) (0.07) Education Neither educated (ref) -- -- -- Both educated 5.38*** 3.85*** 2.45*** (0.50) (0.38) (0.26) Husband educated/Wife not 2.34*** 2.13*** 1.70*** (0.15) (0.14) (0.12) Wife educated/ Husband not 1.60*** 1.39* 1.13 (0.20) (0.18) (0.15) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.87* 0.91 0.93 (0.06) (0.06) (0.07) Husband older >10 years 0.57*** 0.70*** 0.71*** (0.04) (0.05) (0.05) Wife older >5 years 0.72 0.80 0.75 (0.21) (0.24) (0.22) Duration of Marriage 0.99* 0.98* (0.01) (0.01) Decade Married Pre-1980s (ref) -- -- 1980s 1.07 1.09 (0.14) (0.14) 1990s 1.17 1.24 (0.15) (0.17) 2000s 0.98 1.06 (0.14) (0.16) 2010s 0.72 0.84 (0.13) (0.15) Polygynous 0.30*** 0.30*** (0.02) (0.02) Number of Living Children 0.91*** 0.92*** (0.02) (0.02) Household Wealth Quintile Poorest (ref) -- Poor 0.95 (0.07) Middle 1.35*** (0.11) Rich 1.32** (0.12) Richest 1.66*** (0.19) 116 Table 4.11 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Northern 0.62*** (0.04) Rural 0.74*** (0.05) Constant 2.74*** 2.82*** 2.11*** 5.85*** 7.88*** (0.19) (0.08) (0.12) (1.05) (1.58) Obs. 8126 8126 8126 8126 8126 Pseudo R 2 (0.00) 0.00 0.07 0.14 0.15 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 I also rerun the analyses of two-child concordance using interaction effects for monogamy and urbanicity in models with full controls (models not shown). Monogamous couples, both inter- and intraethnic, experience significantly higher predicted levels of concordance than their polygynously married counterparts. Both types of interethnic marriages have higher likelihoods of concordance compared to their intramarried counterparts although this does not reach significance. Both urban and rural interethnic couples have higher predicted levels of concordance than their intraethnic counterparts. Importantly, the difference in concordance levels between urban and rural interethnic couples is non-significant, whereas it is significant among intraethnic couples. I use this same strategy among interreligious couples. When modeling concordance as preferences within four children of each other, monogamous and polygynous interreligious couples are no more likely to be concordant that their intrareligious counterparts. However, there is a large gap between monogamous and polygynous couples of both types, with monogamous couples being significantly more likely to be concordant. Urban interreligious couples are more likely to be concordant within two or four children than urban intrareligious couples. This is no difference among rural couples. 117 The control variables operate similarly regardless of the relative gap between spouses. In the models with full controls, both partners being educated or just the husband being educated is significantly associated with increased concordance. Couples with an age gap of a decade or more between husband and wife are significantly less likely to be concordant, and the likelihood of concordance decreases with the duration of marriage. Being polygynously married lowers concordance, as does the number of living children a couple has. Compared to couples in the poorest wealth quintile, couples in the fourth and fifth quintiles — the rich and richest — are significantly more likely to be concordant. Further, couples residing in rural areas and the north of the country have lower odds of concordance. These findings offer support for higher levels of concordance among interethnic couples (1b), particularly at elevated child-gaps. The results among interreligious couples are less consistent. These findings do not offer robust support that interreligious couples are significantly more concordant than intrareligious couples, supporting hypothesis (1c). However, there is evidence that at larger child-gaps, interreligious couples experience higher odds of concordance, supporting hypothesis (1b). Further, urban interreligious couples may experience higher levels of concordance, also offering some support for hypothesis (1b). Desire to have another child Descriptively, approximately 38% of men and women in inter- and intraethnic marriages report wanting another birth within the next two years (Table 4.1). Using Kappa statistics, 67% of both interethnic (Kappa=0.30) and intraethnic (Kappa= 0.33) couples agree on their desire to have a child within the next two years, meaning they have fair agreement levels above chance. Similarly, 35% of interreligious couples report wanting another birth within two years, as compared to 38% for intrareligious couples (Table 4.2). In total 68% of interreligious and 66% 118 of intrareligious couples agree for a Kappa of 0.29 in both groups, indicating they experience fair agreement. These findings are consistent in baseline logistic models (tables not shown), as interethnic and interreligious couples are not significantly more likely to agree or disagree on the desire to have another child within two years than are intramarried couples. These results are further confirmed in a robustness check as they are no more likely to agree or disagree about the desire to stop childbearing. In a model with full controls, few control variables are significant. Both the husband and wife being educated and the husband being educated alone significantly increase the likelihood of agreement, as does being in the middle wealth quintile. However, being married in the 2000s, being polygynously married, or residing in the north all lower the odds of agreement. These results offer no support for ethnicity and religion as being salient in concordance between couples on the desire for a future birth, supporting hypothesis (1c). Completed fertility Women aged 40 and older report an average of 6.4 births, with intraethnically married women reporting 6.5 on average as compared to 5.7 for interethnically married women (Table 4.1). Intrareligiously married women aged 40 to 49 have, on average, 6.2 children compared to 6.8 for interreligiously married women (Table 4.2). In Table 4.12, at baseline, interethnically married women aged 40 to 49 have significantly lower completed fertility than women who are intraethnically married by nearly one full child. Interreligiously married women have significantly higher parity than intrareligiously married women at baseline. 119 Among interethnic couples, these findings are rendered non-significant in Model 4. Specifically, the unique inclusion of the decade of marriage, the duration of marriage, or the number of living children all render the difference in completed fertility non-significant. Contrarily, interreligiously married couples continue to have higher parity after full controls are added in Model 5, although the effect size is attenuated. When restricting to women aged 35-49 and 45-49 to capture various levels of childbearing completeness, the findings for both groups are robust. In Table 4.12, Model 5 with full controls, the duration of marriage, number of living children, and living in a rural area all significantly increase the total number of children ever born. Contrarily, both partners being educated or the husband alone being educated decrease the number of children ever born, as does being married starting in the 1980s, although being married in the 2010s does not reach significance. 120 Table 4.12: Completed fertility for women aged 40 to 49 (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic -0.83*** -0.50* -0.13 -0.11 (0.24) (0.23) (0.12) (0.12) Interreligious 0.54** 0.45** 0.19* 0.19* (0.15) (0.15) (0.08) (0.08) Education Neither educated (ref) -- -- -- Both educated -2.25*** -0.58*** -0.49*** (0.20) (0.11) (0.12) Husband educated/Wife not -0.55*** -0.29*** -0.27** (0.15) (0.08) (0.08) Wife educated/ Husband not -0.26 0.01 0.04 (0.37) (0.19) (0.20) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years -0.03 -0.17* -0.16 (0.16) (0.09) (0.09) Husband older >10 years 0.13 0.00 0.02 (0.17) (0.09) (0.09) Wife older >5 years -0.01 0.25 0.21 (0.48) (0.26) (0.26) Duration of Marriage 0.04*** 0.04*** (0.01) (0.01) Decade Married Pre-1980s (ref) -- -- 1980s -0.55*** -0.55*** (0.09) (0.10) 1990s -0.71*** -0.70*** (0.11) (0.11) 2000s -0.85*** -0.83*** (0.20) (0.20) 2010s -0.73 -0.56 (0.88) (0.88) Polygynous -0.01 -0.02 (0.07) (0.07) Number of Living Children 0.91*** 0.90*** (0.02) (0.02) Household Wealth Quintile Poorest (ref) -- Poor -0.03 (0.11) Middle -0.06 (0.11) Rich 0.17 (0.12) Richest -0.07 (0.13) 121 Table 4.12 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Northern 0.01 (0.08) Rural 0.19* (0.08) Constant 6.51*** 6.38*** 6.81*** 1.25*** 1.07*** (0.07) (0.08) (0.14) (0.26) (0.27) Obs. 1182 1182 1182 1182 1182 R-squared 0.01 0.01 0.12 0.76 0.76 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 I also looked at interaction effects for monogamy and urbanicity (models not shown). Monogamous marriages are characterized by lower completed fertility among interethnic couples, although this does not reach significance, whereas there is no difference for polygynously married couples. Among monogamous and polygynous marriages, there is no difference by religious composition of marriage, however, monogamous marriages have lower completed fertility than polygynous unions. Urban interethnic couples do not have significantly different fertility outcomes than urban intraethnic couples. However, rural interethnic couples do have lower fertility than their intraethnic rural counterparts, although it does not reach significance. In the interreligious models, rural interreligious couples have higher parity than rural intrareligious couples. Urban interreligious couples do have slightly higher parity than urban intrareligious couples although the effect size is smaller than between rural couples. Overall, these results offer mixed evidence of different fertility outcomes for interethnically married couples. The significantly lower fertility of interethnic couples is rendered non- significant by controlling for marital characteristics, especially the number of living children then have, suggesting that interethnic couples may be couples who experience lower levels of child mortality. These results lend themselves most to hypothesis (2b) — different ethnicity is 122 not salient in fertility outcomes among couples. Among interreligious couples, findings lend support for their having distinct fertility outcomes from intrareligiously married couples supporting hypothesis (2a). In this case, interreligious couples’ distinctive fertility is exhibited through higher than average fertility, particularly among rural couples. Discussion Social identities, such as ethnicity and religion, remain salient in shaping fertility preferences and outcomes in sub-Saharan Africa. This is due to social groups’ shared norms and ideals as well as the unequal distribution of resources among different populations (Agadjanian and Yabiku 2014; Takyi and Addai 2002; Takyi and Dodoo 2005; Weinreb 2001). However, much of our knowledge of the salience of social identities on fertility relies heavily on women’s fertility preferences and outcomes without accounting for their partner despite the important role men’s preferences play in shaping fertility outcomes (Andro et al. 2002; Gyimah et al. 2008; McNamee 2009; Tumlinson et al. 2013; Voas 2003). Further, men’s preferences and desires also differ by ethnic and religious affiliation (Gyimah et al. 2008; Speizer 1999). Resultantly, what constitutes an acceptable compromise between couples is heavily shaped by group norms and expectations — meaning studying solely women contributes to an incomplete understanding of how fertility preferences are enacted (Andro et al. 2002; Behrman et al. 2002; Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Isiugo-Abanihe 2014; Kritz and Makinwa-Adebusoye 1999; Speizer 1999). Of the studies that do measure fertility preferences and outcomes at the couple-level, the samples are predominantly comprised of intraethnic and intrareligious couples, as these are the most common spousal partnerships. As a result, what we know about couples’ fertility preferences and outcomes largely stems from people of the same social backgrounds. However, 123 past research suggests that interethnic and interreligious couples may experience unique fertility preferences and outcomes as fertility expectations are shaped by the couples’ social identities (Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Isiugo-Abanihe 2014; Lehrer 1996). However, some research finds no significant difference for couples of different social identities (Ntoimo and Mutanda 2017). In this study, I examined the salience of different ethnic and religious identities in fertility preferences, concordance, and outcomes. I did this by analyzing interethnic and interreligious couples to see if their fertility preferences and outcomes are distinct from their intramarried counterparts in Benin, West Africa. The results of this study suggest that intermarriage does not uniformly shape fertility preferences and outcomes — rather the results are distinct between interethnic couples and interreligious couples. This indicates that the dynamics in interethnic and interreligious marriages are not the same. At both the individual- and couple-level, men and women married interethnically in Benin have distinct fertility preferences that are lower, on average, than their intraethnically married counterparts, even after controls for educational attainment, wealth, and urbanicity are included. These robust results suggest that men and women who marry interethnically truly have distinctly lower fertility preferences and that being a man or a woman who chooses to marry someone of a different ethnicity correlates with other distinct preferences. Further, interethnic couples are more likely to be concordant in their fertility preferences than intraethnic couples, especially at higher levels of preference concordance (within two or more children of each other). The different ethnic backgrounds of the couples appear to be salient in shaping fertility preferences and compromises among interethnic couples in a way that renders them distinct from intraethnic 124 couples and more likely to be concordant (Crespin-Boucaud 2018; Ibisomi and Odimegwu 2011; Isiugo-Abanihe 2014). Interestingly, these distinct fertility preferences among interethnic couples do not appear to translate into distinct fertility outcomes, as they have similar parity levels as intraethnic couples once the number of living children is included in the model. This may indicate that interethnic couples experience lower levels of child mortality — perhaps due to reasons related to their education and wealth compared to intraethnic couples, especially in monogamous marriages and rural areas — and they, therefore, have fewer children overall. Further, because interethnic marriage is rapidly becoming more prevalent across marital cohorts, perhaps future interethnic couples will have lower completed fertility than intraethnic couples — matching their lower fertility preferences. However, the discordance between fertility preferences and outcomes may reflect higher levels of marital strain leading to higher than intended fertility (Bean and Aiken 1976). If this is true, the gap between fertility preferences and fertility outcomes will likely remain in future cohorts as it is a symptom of marital strain rather than a reflection of cohort preferences and access to contraception. Overall, these results suggest that there is something about the characteristics of interethnic marriages, rather than the couples being of different ethnic backgrounds, per se, that shapes their lower fertility outcomes. Contrarily to the interethnic couples, interreligious couples do not have significantly different fertility preferences from their intrareligiously married counterparts. Further, they do not experience different likelihoods of preference concordance than intrareligiously married couples except at the highest level measured (within four children of each other), although there is evidence that urban interreligious couples experience slightly higher concordance at two children. This suggests that interreligious couples are not more likely to be communicating 125 and/or compromising about fertility preferences than intrareligious couples, especially among rural interreligious couples. Despite not having distinct fertility preferences compared to intrareligious couples, interreligious couples do have higher completed fertility levels than their intrareligious counterparts, particularly among rural interreligious couples. This stands in contrast to research from other West African countries that finds that interreligious couples do not have distinct fertility outcomes, although this may be due to different definitions of interreligious marriage (Ntoimo and Mutanda 2017). This higher total fertility may be attributable to the marital strain experienced by couples of different religious identities. Marital strain can lead to higher levels of unintended pregnancy thereby raising the parity of interreligious couples despite them not having higher fertility preferences and they may also have more children than they want in an attempt to validate the union (Bean and Aiken 1976). Additionally, many interreligious marriages involve Traditional husbands who have significantly higher average fertility preferences. As a result, it could be that interreligious couples have more children because men’s fertility preferences are often prioritized in couples (Gyimah et al. 2008; Ibisomi and Odimegwu 2011), as are the preferences of the partner wanting more children in a high-fertility context like Benin (Voas 2003). Notably, neither interethnic nor interreligious couples have significantly lower fertility than their intramarried counterparts. This indicates that intermarried couples are not currently at the vanguard of lower fertility despite transgressing social norms in their partner selection. This is especially surprising for interethnic couples who are disproportionately in the socioeconomic categories that correlate with lower fertility. A potential explanation is the marital strain intermarried couples experience. Marital strain may cause couples to have more children than 126 they intend to both due to increasing accidental pregnancies, as well as due to having more than desired as a way to validate and cement the union (Bean and Aiken 1976). In the Beninese case, the couple may desire the approval of the extended family and may have more children as a result, even if they do not rely on their extended kin for monetary resources. If this strain is the cause of their higher fertility, rather than access to contraception or generational differences in preferences, it is unlikely that either interethnic and interreligious marriages will be at the forefront of fertility change. Overall, these results offer support for different hypotheses based on whether the couple is interethnic or interreligious, indicating that couples in these two types of intermarriages experience their fertility differently. Interethnic couples have distinctly lower preferences and greater odds of concordance than intraethnic couples, offering support that these couples experience greater levels of communication and are distinct from intraethnic couples. These results support hypothesis (1b) that these couples either select partners with similar fertility preferences and/or are more likely to communicate and compromise. Contrarily, interreligious couples do not experience distinct preferences or concordance, giving no evidence of selection into these relationships based on fertility preferences and/or higher levels of communication about preferences between spouses supporting hypothesis (1c) — my null hypothesis. Neither group lends itself to support hypothesis (1a) — lower preference concordance, suggesting that the differing natalist orientations of various ethnic and religious backgrounds do not seem to decrease concordance among interethnic or interreligious couples in Benin. The results of this study offer support for both of my fertility outcomes hypotheses. Interethnic couples do not have distinct fertility outcomes, mirroring another study in the region (Ntoimo and Mutanda 2017), despite their lower fertility preferences. This finding supports 127 hypothesis (2b) — my null hypothesis. In support of hypothesis (2a), interreligious couples do have different fertility outcomes from intrareligious couples, suggesting that they experience distinct fertility negotiations compared to their intrareligious counterparts (Gyimah et al. 2008; Ibisomi and Odimegwu 2011; Lehrer 1996). These findings, however, are not without their limitations. One key limitation is the flexibility of fertility preferences (Johnson-Hanks 2004; Kodzi, Johnson, et al. 2010; Sennott and Yeatman 2018; Trinitapoli and Yeatman 2018; Yeatman and Sennott 2014; Yeatman et al. 2013). Because fertility preferences in the DHS are measured at one moment in time, I am unable to gauge the consistency of preference gaps and levels of concordance. It is possible, for example, that men and women may experience preference concordance until some of their children age as husbands and wives have different financial responsibilities towards their children across the life course. Thus, their respective incomes may lead to disagreements about how many children they can support (Adida et al. 2018; Falen 2011). Couples may also fluctuate in their preference concordance as their social and economic circumstances change. Additionally, there is a temporal mismatch between measuring the fertility preferences of all couples and then comparing it to completed fertility, which involves a restricted age category. The completed fertility results among older couples may not be indicative of future completed fertility trends. For example, lower fertility preferences may not translate into lower parity because different cohorts have different fertility preferences and, importantly, older cohorts might have been less able to realize lower fertility preferences due to the unavailability of contraception. 128 Relatedly, measuring completed fertility offers a larger sample of interreligious couples 15 as their population has remained relatively stable in Benin throughout the survey waves. As a result, the interreligious completed fertility results are likely to be representative of their fertility experiences. However, interethnic couples 16 are largely concentrated among younger couples whose reproductive careers are not yet over. This may mean that older interethnic couples’ fertility careers are distinct from more recent interethnic couples in unknown ways. In studying interethnic and interreligious couples, this study finds that marriages in which couples cross important social group boundaries do result in unique fertility preferences and outcomes, although these differences are not the same depending on whether the couple is interethnic or interreligious. In the Beninese context, this may be due to the differential selection of couples into these marriages: interethnic couples are more likely to be educated, wealthier, and urban, which are characteristics that may result in higher levels of compromise and communication around preferences. Contrarily, interreligious couples experience less selection into their unions, as their characteristics appear to reflect those of the general population, meaning they do not exhibit higher frequencies of characteristics tied to a higher likelihood of communicating about fertility preferences. However, in contexts where there is greater selection into interreligious marriages than interethnic marriages, the results of this study may be reversed, with interreligious couples experiencing higher concordance. Similarly, in contexts where there is negative selection into intermarriages, intermarried couples would likely experience lower levels of concordance than their intramarried counterparts. Thus, in the Beninese context, these 15 40% (N=336) of couples with the wife aged 40+ are interreligious. They comprise 15% of all interreligious couples. 16 9% (N=112) of couples with the wife aged 40+ are interethnic. They comprise 12% of all interethnic couples. 129 findings are likely reflective of the compositional traits that comprise interethnic and interreligious marriages and render them more or less likely to communicate about fertility preferences. The consistent results of higher completed fertility compared to fertility preferences in this study speak to the difficulty many couples face globally in implementing their fertility preferences — whether due to familial pressure, marital strain, unintended pregnancy, the quality of communication between couples, and/or the negotiation between partners who desire different numbers of children. That interethnic couples do have significant preference differences and non-significant outcomes — and the reverse among interreligious couples — contributes to our understanding of how different social identities and backgrounds shape fertility and how social identities may be both a source of increased marital communication and increased marital strain. 130 CHAPTER 5 MARITAL INVESTMENT AND INTERETHNIC AND INTERRELIGIOUS MARRIAGES Researchers are interested in how partner selection and the characteristics of husbands and wives shape marital investment, meaning the couple’s level of commitment to their union and the effort they put into maintaining the marriage (Burton et al. 2010; Drefahl 2010; Gnoumou Thiombiano and Legrand 2014; Hohmann-Marriott and Amato 2008; Kashyap et al. 2015; Locoh and Thiriat 1995; Pearce-Morris and King 2012; Reniers 2003; Schwartz and Gonalons-Pons 2017; Smith et al. 2012; Zhang and Van Hook 2009). One way that researchers have measured marital investment in survey research is by asking about individuals’ engagement in marital destabilizing behaviors, including extramarital sex, which suggest a low level of commitment to the union (Previti and Amato 2004). Using marital destabilizing behaviors allows researchers to estimate marital investment and examine marital strain in intact unions rather than relying on measures such as union dissolution. A key question of interest is whether increasing shifts from away from social identity homogamy alters couples’ investment in their unions. Several studies find that couples who do not match on social identities have lower levels of investment in their marriages and lower quality unions compared to couples whose social identities do match (Burton et al. 2010; Chinitz and Brown 2001; Gnoumou Thiombiano 2017; Hohmann-Marriott and Amato 2008; Locoh and Thiriat 1995; Reniers 2003; Schramm et al. 2012; Smith et al. 2012; Zhang and Van Hook 2009). Lower investment is found in various global contexts for interethnic and interreligious couples, who face elevated stress relative to their intramarried peers (Burton et al. 2010; Gnoumou Thiombiano 2017; Hewitt, Baxter, and Western 2005; Hohmann-Marriott and Amato 2008; Locoh and Thiriat 1995; Reniers 2003; Smith et al. 2012; Zhang and Van Hook 2009). A key 131 reason for this stress across contexts, including West Africa, is the lack of familial support many interethnic and interreligious couples receive (Hohmann-Marriott and Amato 2008; Maïga and Baya 2011). For example, in West Africa, wives of different ethnic backgrounds are frequently not accepted as family by the husband’s extended kin (Maïga and Baya 2011). If the extended family does not accept the marriage, they may not consider it valid, meaning they would have no obligations towards the spouse and any children arising from the union (Awusabo-Asare 1990). This lack of acceptance may cause marital strain due to the importance of kinship, lowering relationship investment relative to those who are intramarried in the region. Marital strain and lowered relationship investment have significant implications for both the couple and their children (Hair et al. 2009; Robles et al. 2015; Slatcher 2010; Umberson et al. 2006). Research finds that marriages facing strain lead to more negative health outcomes for both partners across their life course, as well as for their children (Hair et al. 2009; Robles et al. 2015). While it is not the ethnic and religious compositions of these marriages, per se, that cause negative outcomes for families (Pearce-Morris and King 2012), the greater likelihood of marital strain in these unions increases their likelihood of negative outcomes (Hair et al. 2009). Measuring Marital Investment In developed countries, there is extensive research using multiple indicators — quality, cohesion, dissolution, stability — that seeks to understand the dyadic and dynamic nature of marriages (Bradbury, Fincham, and Beach 2000). These varied measures include how spouses interact with each other — measured in ways such as communication styles — as well as self-reported levels of happiness (Amato and Booth 1995; Brown, Manning, and Payne 2015; Hair et al. 2009; Umberson et al. 2006; Webster, Orbuch, and House 1995). Interpersonal violence and physical well-being, such as biomarkers of stress, are also commonly used (Bradbury et al. 2000; Robles 132 et al. 2015; Umberson et al. 2006). Divorce, or the dissolution of a union, is perhaps the most widely used measure over time, although it does not capture low marital investment in intact marriages, nor is it a perfect correlate of low marital quality (Amato 2010; Amato and Hohmann- Marriott 2007; Hewitt et al. 2005; Ono 1998; Previti and Amato 2003; Teachman 2002; Zhang and Van Hook 2009). This research also extends to how the spousal relationship affects the larger family unit, frequently measuring children’s experiences of family-life controlling for the characteristics of their parents’ marriage (Amato 2010; Hair et al. 2009; Webster et al. 1995). Compared to this robust literature in developed contexts, there is minimal research on the nature of marriages in low-income settings, including sub-Saharan Africa (Frye and Trinitapoli 2015; Takyi 2001). These varied indicators used in developed contexts are not easily measured in low-income settings due to the lack of surveys asking about men’s and women’s current marriages and their union histories (Frye and Trinitapoli 2015; Takyi 2001). Resultantly, our knowledge of the inner workings of marriages in low-income settings frequently relies on accounts of intimate partner violence (Decker et al. 2015; Vyas and Watts 2009), or, more commonly, union dissolution when this information is available (Clark and Brauner-Otto 2015; Gnoumou Thiombiano and Legrand 2014; Reniers 2003; Takyi 2001; Takyi and Gyimah 2007). As both indicators, intimate partner violence and divorce, are heavily tied to health and fertility outcomes, these are indicators captured in the Demographic and Health Surveys. However, what we can glean from these indicators and their relationship to marriage is minimal using these data. Their data on intimate partner violence are limited as they only collect these data in select countries 17 and there are few countries with multiple waves of data on the topic. 17 A module on intimate partner violence was added to the 2018 BDHS (INSAE and ICF International 2019). 133 Additionally, in countries such as Benin, they only include households that were not selected for men’s surveys, meaning there are no couple-level data available on intimate partner violence (INSAE and ICF International 2019). This limits our ability to measure changes over time, as well as how violence is shaped by the husband’s characteristics and the couple’s relative characteristics. Similarly, the ability to study divorce at the couple-level is tempered by the fact that the DHS does not collect data on full marital histories. While the DHS asks if people were formally in a union that ended in divorce, full marital histories would enable researchers to study the predictors of union dissolutions. As a result of this lack of data, in the majority of DHS, there are limited direct measures of couples’ non-fertility related marital experiences. Resultantly, measuring marital investment is difficult to do using existing data in sub- Saharan Africa. Marital investment is an important indicator due to its relation to health as well as marital stability and dissolution, among other outcomes (Gnoumou Thiombiano 2017; Previti and Amato 2004; Teachman 2002). Previous work attempts to parse out what contributes to a couple’s investment in their marriage because marital strain is not evenly spread across all couples. The social characteristics, such as religion and ethnicity, spouses bring into the marriage might shape their attachment to and investment in the couple-unit (Amato 2010; Bradbury et al. 2000; Umberson et al. 2006; Webster et al. 1995). People who enter into marriages with partners from different social groups, especially along the lines of ethnicity and religion, often face increased marital strain which may lower interethnically and interreligiously married people’s investment in the couple-unit (Burton et al. 2010; Hewitt et al. 2005; Hohmann-Marriott and Amato 2008; Kaplan and Herbst 2015; Zhang and Van Hook 2009). In focusing almost exclusively on intimate partner violence and marital intactness, we may overestimate marital investment and miss other locally salient indicators of disharmony in 134 relationships. Despite the limitations of data collected in sub-Saharan Africa, there are indirect measures of marital investment collected in the DHS, including the marital destabilizing behaviors of extramarital sex and polygyny (Bove and Valeggia 2009; Previti and Amato 2004). In the sub-Saharan context, both extramarital sex and polygyny often occur in lieu of divorce and introduce strain into the marriage (Agadjanian and Ezeh 2000; Bove and Valeggia 2009; Falen 2008a, 2011; Reniers 2003; Timæus and Reynar 1998). Therefore, these locally salient marital destabilizing behaviors can be used to grapple indirectly with marital investment in intact marriages. Faced with scrutiny and a lack of support from extended kin, men in interethnic and interreligious unions may be less invested in their marriages. As a result, they may be more likely to engage in marital destabilizing behaviors, extramarital sex and/or polygyny in the Beninese context, than intramarried men as an alternative to divorce. Similarly, a husband’s extended family may encourage him to take a second wife and/or engage in extramarital sex as a way for him to secure a wife of the same ethnic or religious background (Maïga and Baya 2011). Men in interethnic and interreligious marriages may also have an easier time engaging in these behaviors as extended family — both his own and his wife’s — may police his behaviors less. Alternatively, engaging in extramarital sex and polygyny may not be a reflection of current marital investment as extramarital affairs and polygyny are both common in Benin (Falen 2008a). In addition to potentially signaling of low marital investment, both extramarital sex and polygyny often create new marital problems. For example, wives in polygynous marriages experience lower marital satisfaction and poorer mental health (Bledsoe 1990; Bove and Valeggia 2009; Falen 2008a). Even when men do not formally take another wife, they often 135 maintain extramarital partners (Bledsoe 1990; Falen 2008a; Oluwaseyi, Odimegwu, and Adedini 2014; Orubuloye, Caldwell, and Caldwell 1997; Smith 2007). As is the case globally (Previti and Amato 2004), extramarital sex lowers trust and marital satisfaction (Smith 2007, 2010). Extramarital sex increases both spouses’ risk of sexually transmitted infections and it diverts resources — time, money, etc. — away from the primary household (Orubuloye et al. 1997; Smith 2007). Furthermore, in sub-Saharan Africa, extramarital pregnancies also risk creating co- wives (Falen 2008a; Orubuloye et al. 1997; Smith 2002, 2010). Extramarital sex and polygyny are thus two locally salient behaviors that may both indicate and cause a lack of investment in the couple-unit. Given that interethnic and interreligious couples may experience more stress in their marriages compared to their intramarried counterparts, I examine whether husbands in interethnic and interreligious unions are more likely to engage marital destabilizing behaviors, defined as extramarital sex and polygyny, than those in intramarriages. Study Context Extramarital sex and polygyny While extramarital sex occurs globally, in West and Central Africa extramarital sex and polygyny are intertwined practices (Falen 2011; Lacombe 1987). As Christianity, Western perspectives of the family, and general cultural sentiments have turned away from the practice of polygyny, informal long-term relationships have replaced what might have been polygynous marriages in another time (Falen 2008a, 2011; Lacombe 1987; Ware 1979). As a result, in West Africa, some men engage in long-term extramarital relationships as a type of informal polygyny. These informal long-term unions are frequently referred to as the “deuxième bureau,” meaning second office or post. While these unions often do not become official marriages, these are 136 partnerships that go beyond simply being a girlfriend or a mistress, with the children from these partnerships being considered legitimate (Lacombe 1987). These informal arrangements often strain the official marriage, in large part because the relationship between the official wife and the informal partner is not formalized in the way it would be under a polygynous union (Lacombe 1987). As a result of the informal nature of these relationships, many men do not consider themselves as being polygynously married, rendering polygyny undercounted (Falen 2011). The frequency of extramarital sex varies by religion and polygynous status in West Africa. Generally, monogamously married men — who are more frequently but not exclusively Christian — report higher levels of engaging in extramarital sex than their polygynously married counterparts (Mitsunaga et al. 2005; Oyediran et al. 2010; Vu et al. 2011). As a result, extramarital sex is more common among Christian men whereas polygyny is more common among Muslims. This is due to Islam permitting multiple wives versus Christianity’s emphasis on monogamy, among other reasons (Mitsunaga et al. 2005; Oyediran et al. 2010; Vu et al. 2011). However, these are by no means absolutes, as even within religious categories, the relative frequency of extramarital sex varies by region in ways that frequently overlap with ethnic clusterings (Mitsunaga et al. 2005). Men in this region frequently perceive extramarital sex, including casual encounters and longer-term partnerships, and polygyny as important displays of masculinity, wealth, and status (Falen 2008a, 2011; Lacombe 1987). As a result, men engaging in these behaviors may not always be a direct reflection of current marital unhappiness. However, the husband’s extramarital sex — both real and perceived — is a common source of conflict in marriages (Falen 2011). Additionally, men in Benin often use the threat of taking on additional wives as a way to exert 137 power over their current wives. Further, extramarital sex and polygyny are linked as extramarital pregnancies also risk creating formal co-wives (Falen 2008a; Orubuloye et al. 1997; Smith 2002, 2010) or children that the official wife will be expected to accept as legitimate children of her husband, and therefore share resources with her children (Lacombe 1987). This suggests that extramarital sex and polygyny reflect real sources of tension in Beninese marriages. It is important to acknowledge that polygyny is not exclusively recognized as harming family life. Many women in polygynous marriages enjoy having co-wives as sources of friendship, help managing the household and children, as well as being in a position of power over someone, in the case of first wives (Caldwell et al. 1991; Falen 2011; Ware 1979). Similarly to men who are unhappy in their marriage, unhappy women may also encourage their husband to take another wife (Meekers and Franklin 1995). Additionally, the higher-order wives in polygynous marriages are frequently divorced or widowed women and entry into a polygynous union can offer them the benefits of marriage they may not otherwise be able to find (Timæus and Reynar 1998). In these ways, formalized polygyny can offer benefits to all the wives involved. However, in the case of informal polygyny, relatively few benefits are available for the official wife, but the informal partners express benefiting from the informal nature of the relationship which does not restrict them to the formal rules, expectations, and hierarchies of marriage (Lacombe 1987). Ethnicity, religion, extramarital sex, and polygyny in Benin The frequency of extramarital sex and polygyny is not evenly spread across ethnic and religious groups in Benin. Appendix 5.A shows the reported levels of extramarital sex and polygyny in Benin by men’s ethnicity and Appendix 5.B shows the same by men’s religious identification. Eighteen percent of Beninese men report engaging in extramarital sex within the last year and this dips as low as 10% among Dendi men and is highest among Fon men at 21%. Among 138 religious groups, only 15% of Muslim men report engaging in extramarital sex, the lowest of any religious group, while Catholic men report the most at 23%. Overall, a lower proportion of northern men (17%) than southern men (20%) report engaging in extramarital sex. Thirty-nine percent of marriages in Benin are polygynous and this ranges from 33% of marriages with a Fon husband up to 49% of marriages with a Peuhl husband. Among religious groups, 28% of Protestant and 29% of Catholic men are polygynously married, and this reaches 50% for marriages with a Traditional husband. Overall, polygyny is more common in the north (44%) than the south (36%). Samples In the extramarital sex models, all couples and men are the units of analysis as extramarital sex is a risk in all marriages. After removing couples missing data for key control variables, the final sample size for these analyses is 9,828 couples and men. For the analysis of the transition into polygyny, I study only polygynously married couples and men, who comprise 37.5% of the total sample. This alters the sample population in several key ways: compared to monogamously married couples, polygynous couples are less educated and are more likely to have larger age gaps between the husband and wife. Additionally, they are more likely to be rural and reside in the north than in the south. In terms of wealth, they are more likely to be in the middle wealth quintiles than the poorest two quintiles. Appendix 5 Table 3 offers further information on how the polygynous sample varies from their monogamously married counterparts. The final polygynous sample includes 3,683 couples and men after couples missing data on control variables are excluded. 139 Measures Dependent variables Extramarital sex. The first outcome of interest is extramarital sex. Men were asked the number of times they estimate they had sexual relations in the past 12 months and their relation to that person. Choices included: spouse, cohabitating partner (coded as a spouse), girlfriend (non- cohabitating), occasional hookup partner, prostitute/client, other (must be given a label). The response is transformed into a variable indicating the number of partners other than their wife men had sex within the last 12 months. I code men who report engaging in sex only with their wife in the last 12 months as not having extramarital sex (=0) and I code men reporting at least one partner with someone who is not their wife as having extramarital sex (=1). I also include men who have missing data (i.e. “do not know”) as having extramarital sex. Men with missing data are likely those who engaged in extramarital sex but are unsure of whether it was in the last year or who do not recall the exact number of extramarital partners they had. Overall, there are only 326 men with missing data comprising 3% of men in these models. Although women also engage in extramarital sex (Oluwaseyi et al. 2014), I do not include them due to low reported levels of extramarital sex (<1%) among wives in Benin (author’s calculations). Polygyny. My second key outcome of interest is whether unions that are interethnic or interreligious have increased odds of transitioning into polygyny, meaning that a woman’s husband took a second wife after her, compared to intramarriages. To measure monogamy, polygyny, and a wife’s rank in polygynous unions, interviewers asked women: “Does your husband/partner have other wives or live with other women as though they were married?” If she responded affirmatively, the interviewer then asked: “Are you the first, second,…spouse?”. I code women who do not report their husband/partner as having other wives as monogamously married and women who respond affirmatively as polygynously married. Because wife rank is 140 essential for measuring the transition to polygyny, 144 women who report being unsure of their rank (2.8%) or have missing values (0.1%) are not included in this measure. To examine the transition to a polygynous union, I code first wives as (=1) and second or higher-order wives as (=0). Control variables I control for a robust set of variables standard in the sub-Saharan African marital literature. Ethnicity of husband. To examine ethnicity’s link to extramarital sex and polygyny, I control for the husband’s ethnicity in individual-level models. Religion of husband. Similarly, because extramarital sex and polygyny are linked to religion, I control for the husband’s religion in individual-level models. Decade married. To measure how extramarital sex and polygyny are linked to time, I utilize data on women’s date of marriage to create a categorical indicator of decade married ranging from pre-1980 to 2010 or later. This indicator is based on their response to the following question: “In which month and year did you start to live with your husband/partner?” Educational attainment. My main measure of educational attainment is a binary indicator of having ever attended school. This is due to overall low school attendance in Benin. Among women surveyed, 76% have never attended nor have 54% of men. Due to the large discrepancy in access to schooling for boys and girls, I measure the educational discordance of couples with a categorical variable. I code this variable as neither wife nor husband is educated (=0), wife and husband are educated (=1), husband is educated but wife is not (=2), and wife is educated but husband is not (=3). Due to parents prioritizing sending their sons to school over their daughters, few women (N=461 or 4.6%) have been to school when their husbands have not, meaning the results of this group should be interpreted with caution. 141 I also measure the absolute number of years of schooling attended individually for husbands and wives at the couple-level by including controls for his and her years of schooling. Age heterogamy. The average age gap between husbands and wives is largest in sub-Saharan Africa. Age heterogamy is correlated with earlier marital age for women, lower contraceptive usage, and higher parity (Barbieri and Hertrich 2005; Takyi 2001; Togunde 1998). African men overwhelmingly marry at later ages than their female counterparts (88% in the couples’ data). As a result, only 5% of women are older than their husbands. I code age homogamy (=0) as partners who are within five years of each other in age. When the husband is older, I code age heterogamy as couples where there is a five- to nine-year age gap between spouses (=1) and an age gap of ten or more years (=2). Due to limited numbers of women being five or more years older than their husbands (N=86 or 0.9%), I only create one grouping for women who are older than their husbands by at least five years (=3). Due to the small number of couples in this last category, these results should be interpreted with caution. Premarital pregnancy. I ask if men in unions where their wives experienced a premarital pregnancy are more likely to engage in extramarital sex or polygyny compared to husbands whose wives who did not experience a premarital pregnancy. In sub-Saharan Africa, a premarital pregnancy may lead couples to marry to have a “legitimate” birth. Further, unwed women who are already pregnant or mothers often have different marriage market experiences than women without children or visible pregnancies (Smith-Greenaway and Clark 2018). As a result, women with premarital pregnancies may be considered less attractive on the marriage market and their husbands may be less invested in their marriage. The survey measures the interval from marriage to first birth by comparing the date of marriage to the date of first birth. I code births that occurred within seven months of marriage as 142 premarital pregnancies (=1) while babies born anytime starting at eight months from marriage are coded as marital pregnancies (=0). The DHS does not inquire as to the identity of the male partner involved in the pregnancy, meaning the father of the child may or may not be the woman’s current spouse. Age. In the individual models, I control for the individual age of the husband at the time of survey. Duration of marriage. The longer a woman has been married, the longer she has been exposed to the risk of extramarital sex and her husband marrying additional wives. To account for this, I code the duration of marriage continuously based on the number of years the wife reports being married. In the polygyny models, I also include controls for the duration of time in a categorical fashion: (1) married fewer than five years (2) married fewer than ten years and (3) married for a decade or more. Number of living children. I include a control for the number of living children the wife has. Women who have relatively low parity may be at increased risk of her husband engaging in extramarital sex or selecting another wife. Location of residence. The likelihood of engaging in extramarital sex and polygyny may be distinct based on whether the couple resides in an urban or rural location due to differences in the availability of partners and the level of acceptance of polygyny. However, urbanicity is not a static feature and may have changed since marrying, meaning that it is a feature measured at the time of survey rather than at the time of extramarital sex and/or a transition into polygyny. Due to the large north-south divide that marks Benin regarding cultures, religions, and resources, I also control for whether the couple resides in the north (=1) or the (south=0). 143 Household wealth. To measure household wealth, I use the wealth index score, which is a composite measure of a household’s living standard based on weighting assets and services available in the household, such as water sources and materials used to construct the home (Rutstein and Johnson 2004). It is not a measure of income but resources available to all household members (Rutstein and Johnson 2004). As wealth is measured at the time of the survey, it is not reflective of the wealth of couples at the time of their marriage. Wealth is also not independent of the number of people living in the household: families with more wives may have more children and may be less able to accumulate wealth than families with fewer wives and/or children. Therefore, this measure should be interpreted cautiously, particularly in the polygyny models. In the extramarital sex models, but not the polygyny models, I also include the following controls: Polygyny. To measure monogamy and polygyny, interviewers asked women: “Does your husband/partner have other wives or live with other women as though they were married?” I code women who do not report their husband/partner having other wives as monogamously married (=0) and women who respond affirmatively as polygynously married (=1). Birth within the last year. When men’s wives have recently given birth and are observing a post-natal abstinence period, men are at increased risk of engaging in extramarital sex (Ali and Cleland 2001). To account for this, I code women who report no birth within the last year as (=0) and women who have given birth as (=1). 144 Analytic Approach Using logistic regression, I offer four modeling approaches to examine the likelihood of interethnically and interreligiously married men engaging in extramarital sex and transitioning to polygyny. For both couple- and individual-level models, I include baseline models for (1) interethnic marriages and (2) interreligious marriages, respectively, (3) controlling for characteristics that are likely to have remained constant since marriage, (4) controlling for marital features, and (5) full controls including urbanicity, region, and wealth. Using this modeling strategy, I capture the presumed temporal sequence of these variables vis-à-vis union formation. The temporal order of the variables is especially important in the polygyny models, as I do not have data on when an additional wife was added to the family unit and several of the variables of choice — such as wealth — may be a reflection of polygyny rather than a cause. Results Extramarital sex Descriptively, 18% of intraethnically married men and 17% of intrareligiously married men report engaging in extramarital sex within the last year, compared to 21% of interethnically and interreligiously married men. These differences are significant. Further supporting these findings is that, at baseline (Table 5.1), interethnically married men are 19% more likely to report engaging in extramarital sex within the last year than intraethnically married men and interreligiously married men are 20% more likely to report engaging in extramarital sex than intrareligiously married men. These baseline results indicate that men in interethnic and interreligious marriages may be less invested in their marriages. Among interethnic couples, however, the unique inclusion of education renders this non- significant regardless of whether it is coded categorically, as shown in Table 5.1, or with either 145 his or her years of education included. Similarly, the unique inclusion of a control for the duration of marriage, urbanicity, or current household wealth status also renders this non- significant. These findings are replicated in the men’s individual model (Table 5.2). In models including interaction effects (not shown), interethnic couples where the wife is educated but the husband is not have significantly lower predicted likelihoods of the husband engaging in extramarital sex than for intraethnic couples, but due to the low number of these couples, these results should be interpreted cautiously. Additionally, interethnic couples residing in the north have lower odds of the husband engaging in extramarital sex than intraethnic northern couples but there is no difference in the south. Contrary to the findings for interethnic couples, the baseline result for interreligious couples is robust to the inclusion of all variables in both the couple-level and individual-level models, as shown in Tables 5.1 and 5.3. This means that with full controls, interreligiously married men remain significantly more likely to engage in extramarital sex than intrareligiously married men. There are several significant interactions of note in the interreligious models that robustly support interreligiously married men having higher odds of engaging in extramarital sex than intrareligiously married men (models not shown). Educated men in interreligious marriages whose wife is not educated have significantly higher predicted likelihoods of engaging in extramarital sex than intrareligiously married men. There is no significance for other types of educational matching, indicating that women’s education is protective against extramarital sex in interreligious marriages. Additionally, while polygynous men have the same likelihood of engaging in extramarital sex, interreligious monogamously married men are significantly more likely to engage in extramarital sex within the last year than intrareligious monogamous men. Further, interreligiously married men in rural or northern areas are significantly more likely to 146 report extramarital sex than intrareligiously married rural or northern men. There is no difference for urban or southern men. In the full controls couple-level model (Table 5.1), men’s education is robustly tied to their reports of extramarital sex. Both partners being educated and only the husband being educated are both tied to increased odds of extramarital sex. When individual years of schooling are used for the husband and wife, only the husband’s education is significant. Compared to men married before the 1980s, only men married in the 2000s are significantly less likely to engage in extramarital sex. The more living children a woman reports, the lower the likelihood of her husband engaging in extramarital sex while the wife giving birth within the last year greatly increases the likelihood of the husband engaging in extramarital sex. Compared to the poorest couples, men from the middle and richer quintiles are significantly more likely to report extramarital sex. The age composition of the couple, duration of marriage, premarital pregnancy, polygyny, and urbanicity are all non-significant. In the individual-level models (Tables 5.2 and 5.3), compared to Fon men (the largest ethnic group), only Adja and Dendi men are significantly less likely to engage in extramarital sex. Compared to men of Traditional religious backgrounds, Muslim, Protestant, and other Christian men are significantly less likely to engage in extramarital sex. An increase in men’s education increases the husband’s likelihood of reporting extramarital sex and his age decreases it. His wife giving birth within the last year significantly increases his odds of engaging in extramarital sex while living in a rural area decreases his odds. Compared to the poorest men, men in the middle and higher quintiles (middle class to richest) are significantly more likely to engage in extramarital sex. Non-significant are the years of marriage, the number of living children his wife reports, and his residence in the north or the south. The results of both the individual- and 147 couple-level models highlight the importance of men’s increased capital — education and wealth — in his likelihood of engaging in extramarital sex. Overall, these results indicate that extramarital sex is not more (or less) common among interethnic couples compared to intraethnic couples. This indicates that interethnic marriage does not lower men’s marital investment as a result of ethnicity, per se. Although being interethnically married is associated with an increased likelihood of extramarital sex at baseline, the unique inclusion of controls for education (his or hers), urbanicity, or wealth is enough to eliminate the effect of interethnic marriage. This suggests that men in interethnic marriages may be more likely to report engaging in extramarital sex because of their high levels of human and economic capital, as measured by education and wealth, respectively. Further, the inclusion of a control for the duration of the marriage also renders it non-significant, indicating that this is also partly a story of the newness of these marriages relative to intraethnic marriages. On the other hand, these results robustly support the findings that interreligiously married men are significantly more likely to engage in extramarital sex than are intrareligiously married men. This indicates that men in interreligious marriages may be less invested in their marriages than are intrareligiously married men. 148 Table 5.1: Odds ratio of men reporting engaging in extramarital sex within last 12 months – Couples’ characteristics (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic 1.19* 1.03 1.03 0.97 (0.09) (0.08) (0.08) (0.08) Interreligious 1.20** 1.19** 1.19** 1.21** (0.25) (0.07) (0.07) (0.07) Education Neither educated (ref) -- -- -- Both educated 1.72*** 1.60*** 1.40*** (0.12) (0.11) (0.11) Husband educated/Wife not 1.48*** 1.46*** 1.37*** (0.09) (0.09) (0.09) Wife educated/ Husband not 1.19 1.15 1.09 (0.15) (0.14) (0.14) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.96 0.96 0.96 (0.06) (0.06) (0.06) Husband older >10 years 0.79*** 0.81** 0.80** (0.05) (0.05) (0.05) Wife older >5 years 0.80 0.84 0.85 (0.23) (0.24) (0.24) Premarital Pregnancy 1.07 1.11 1.11 (0.06) (0.06) (0.06) Decade Married Pre-1980s (ref) -- -- -- 1980s 1.18 1.00 0.99 (0.16) (0.14) (0.14) 1990s 1.64*** 1.11 1.11 (0.20) (0.15) (0.15) 2000s 1.56*** 0.91 0.88 (0.20) (0.14) (0.13) 2010s 1.86*** 0.93 0.88 (0.26) (0.16) (0.15) Duration of Marriage 0.99* 0.98* (0.01) (0.01) Number of Living Children 0.93*** 0.93** (0.02) (0.02) Polygynous 0.94 0.93 (0.05) (0.05) Birth within Last Year 1.36*** 1.37*** (0.08) (0.08) Rural 0.95 (0.06) Northern 1.12 (0.07) 149 Table 5.1 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Household Wealth Quintile Poorest (ref) -- Poor 1.20* (0.10) Middle 1.41*** (0.12) Rich 1.48*** (0.12) Richest 1.60*** (0.15) Constant 0.26*** 0.25*** 0.14*** 0.30*** 0.25*** (0.01) (0.01) (0.02) (0.05) (0.05) Obs. 9828 9828 9828 9828 9828 Pseudo R 2 0.00 0.00 0.02 0.03 0.03 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 150 Table 5.2: Odds ratio of men reporting engaging in extramarital sex within last 12 months – Men’s characteristics interethnic marriages (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interethnic 1.19* 1.13 1.11 1.04 (0.09) (0.09) (0.09) (0.08) Ethnicity Fon (ref) -- -- -- Adja 0.74*** 0.74*** 0.74*** (0.06) (0.06) (0.06) Bariba 1.25* 1.24* 1.19 (0.13) (0.13) (0.19) Betamaribé 0.81 0.79* 0.80 (0.09) (0.09) (0.14) Dendi 0.64* 0.64* 0.63* (0.11) (0.11) (0.13) Peuhl 0.84 0.83 0.89 (0.11) (0.11) (0.16) Yoa/Lokpa 0.92 0.95 0.92 (0.13) (0.14) (0.17) Yoruba 0.89 0.94 0.93 (0.08) (0.09) (0.09) Religion Traditional (ref) -- -- -- Muslim 0.82 0.77* 0.70** (0.09) (0.08) (0.08) Catholic 1.02 0.99 0.92 (0.08) (0.08) (0.07) Protestant 0.81 0.78* 0.74* (0.10) (0.10) (0.09) Other Christian 0.75** 0.69*** 0.67*** (0.07) (0.06) (0.06) Other 0.94 0.88 0.84 (0.22) (0.21) (0.20) No religion 1.03 0.95 0.92 (0.13) (0.12) (0.12) Educated 1.59*** 1.48*** 1.36*** (0.09) (0.08) (0.08) Age at Survey 0.99** 0.99*** (0.00) (0.00) Duration of Marriage 0.99 0.99 (0.01) (0.01) Number of Living Children 0.95** 0.95* (0.02) (0.02) Birth within Last Year 1.34*** 1.35*** (0.08) (0.08) Rural 0.93 (0.05) 151 Table 5.2 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Northern 1.13 (0.16) Household Wealth Quintile Poorest (ref) -- Poor 1.16 (0.09) Middle 1.35*** (0.11) Rich 1.41*** (0.12) Richest 1.55*** (0.14) Constant 0.26*** 0.25*** 0.52*** 0.47*** (0.01) (0.02) (0.07) (0.08) Obs. 9828 9828 9828 9828 Pseudo R 2 0.00 0.02 0.03 0.03 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 152 Table 5.3: Odds ratio of men reporting engaging in extramarital sex within last 12 months – Men’s characteristics interreligious marriages (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interreligious 1.20** 1.17** 1.15* 1.15* (0.07) (0.07) (0.07) (0.07) Ethnicity Fon (ref) Adja 0.75*** 0.75*** 0.75*** (0.06) (0.06) (0.06) Bariba 1.22* 1.22 1.18 (0.12) (0.12) (0.19) Betamaribé 0.79* 0.77* 0.78 (0.09) (0.08) (0.13) Dendi 0.65* 0.65* 0.64* (0.12) (0.12) (0.13) Peuhl 0.84 0.83 0.89 (0.11) (0.11) (0.16) Yoa/Lokpa 0.94 0.96 0.93 (0.13) (0.14) (0.17) Yoruba 0.90 0.94 0.94 (0.08) (0.09) (0.09) Religion Traditional (ref) -- -- -- Muslim 0.87 0.81 0.74** (0.09) (0.09) (0.08) Catholic 1.05 1.01 0.94 (0.08) (0.08) (0.08) Protestant 0.80 0.77* 0.73* (0.10) (0.10) (0.09) Other Christian 0.76** 0.71*** 0.69*** (0.07) (0.06) (0.06) Other 0.88 0.84 0.80 (0.21) (0.20) (0.19) No religion 0.98 0.92 0.88 (0.12) (0.12) (0.11) Educated 1.61*** 1.49*** 1.37*** (0.09) (0.08) (0.08) Age at Survey 0.99** 0.99*** (0.00) (0.00) Duration of Marriage 0.99 0.99 (0.01) (0.01) Number of Living Children 0.95** 0.95* (0.02) (0.02) Birth within Last Year 1.34*** 1.35*** (0.08) (0.08) Rural 0.93 (0.05) 153 Table 5.3 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Northern 1.13 (0.16) Household Wealth Quintile Poorest (ref) -- Poor 1.16 (0.09) Middle 1.36*** (0.11) Rich 1.42*** (0.12) Richest 1.56*** (0.14) Constant 0.25*** 0.23*** 0.49*** 0.44*** (0.01) (0.02) (0.07) (0.07) Obs. 9828 9828 9828 9828 Pseudo R 2 0.00 0.02 0.03 0.03 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 Polygyny I begin with a brief introduction into the sub-sample of polygynous couples used in these models, shown in Table 5.4. Polygyny is more common among intraethnic couples (39%) than interethnic couples (27%). Contrarily, the same percentage of intra- and interreligiously married couples are polygynous, at 37% and 38% respectively. This is confirmed in a baseline logistic model, shown in Appendix 5 Table 4, as interethnic couples are significantly less likely to be polygynously married, whereas there is no significant difference for interreligious couples. Overall, the trends seen among polygynous intra- and interethnically married women match their general population characteristics. For example, interethnically married polygynous women are educated at higher rates (27% versus 13%) and married nearly a year older (18.7 versus 17.9) than their intraethnically married counterparts in polygynous marriages. However, higher percentages of women in interethnic couples report being a first wife (52%) than their intraethnic counterparts (48%) and the reverse is seen for being second wives. Similarly, intra- and 154 interreligiously married polygynous women look akin to each other on key statistics as they do in the general population — they are equally educated (14%) and married at similar ages (18 years old), for example. Women in both types of marriages report similar levels of being a first wife, 72% for intrareligious women and 71% for interreligious women, however interreligious women report slightly higher rates of being a second wife (22%) than intrareligious women (20%). 155 Table 5.4: Means and percentages of characteristics of women in polygynous marriages by ethnic composition of marriage (Standard deviations are in parentheses) Wife’s Characteristics Intraethnic Interethnic Intrareligious Interreligious Polygynous (%) 38.73 28.23 37.49 37.51 Ethnicity (%) Fon 34.15 18.29 30.63 39.76 Adja 18.11 11.14 18.10 17.01 Bariba 13.39 8.86 13.16 12.76 Betamaribé 6.80 6.00 4.30 13.16 Dendi 4.06 16.57 6.25 0.30 Peuhl 10.53 8.57 12.60 3.96 Yoa/Lokpa 4.33 8.00 5.31 2.37 Yoruba 8.63 22.57 9.65 10.68 Educated (%) 12.82 28.66 14.06 14.34 Religion (%) Traditional 24.04 10.57 26.07 14.64 Muslim 29.18 47.43 38.82 6.53 Catholic 15.76 18.29 15.82 16.12 Protestant 4.21 4.86 3.29 7.22 Other Christian 16.34 12.00 13.54 15.62 Other 2.17 1.43 0.37 6.63 No religion 8.30 5.43 2.09 23.24 Age at Marriage 17.90 18.71 17.96 18.00 (3.88) (4.61) (3.86) (4.25) Remarried (%) 17.48 19.54 17.10 18.90 Premarital Pregnancy (%) 33.60 36.86 34.48 32.44 Age at Survey 32.24 32.11 32.25 32.20 (7.59) (7.84) (7.55) (7.75) Wife Rank (%) 1 51.65 48.29 72.25 71.02 2 38.81 43.14 19.75 21.65 3 6.98 7.14 5.16 4.95 4 1.47 1.14 1.31 1.19 >5 1.02 0.29 1.53 1.29 N 3378 307 2674 1011 Source: The Demographic and Health Surveys Program 156 Among these polygynous couples, I examine whether interethnic and interreligious marriages are more likely to transition into polygyny. Put another way, I test whether interethnically and interreligiously married women are more likely to be first wives in a polygynous union, meaning that their husbands took another wife after them. The results of these models (Table 5.5), including a baseline model and interaction terms, are non-significant. This indicates that interethnic and interreligious marriages are no more likely to transition into polygyny than are intraethnic and intrareligious marriages. These null results are consistent across all models for interethnic couples. However, in Models 3, 4, and 5, interreligious women are more likely to be higher-order wives, due to controlling for the couple’s age gap. This pattern holds when controlling for husbands’ and wives’ current individual ages rather than the categorical age gap. To further confirm these results, I also restrict the sample by the duration of marriage (models not shown). Recent marriages, defined as those that occurred within five years of the survey, are no more likely to have an interethnic or interreligious first wife. Nor is it more likely for marriages that occurred within nine years of the survey to have an interethnic or interreligious first wife. Marriages older than ten years are also not more or less likely to have a first wife who is in an interethnic or interreligious marriage. In the couple-level model with full controls (Table 5.5 Model 5), education is not significantly associated with wife rank, however, women married to men five or more years older are significantly less likely to be first wives. Compared to marriages occurring before the 1980s, only women married in the 2010s have lower odds of being a first wife. The duration of marriage and number of living children significantly increase the odds of being a first wife, 157 likely due to their relationship to time. Premarital pregnancy is not tied to increased or decreased odds, nor is urbanicity, region of residence, or household wealth. Table 5.5: Odds ratio of marriage transitioning to polygyny – Couples’ characteristics (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Interethnic 0.92 0.98 0.98 0.98 (0.11) (0.13) (0.14) (0.13) Interreligious 0.87 0.82* 0.83* 0.83* (0.06) (0.07) (0.07) (0.07) Education Neither educated (ref) -- -- -- Both educated 0.89 0.99 1.03 (0.12) (0.13) (0.15) Husband educated/Wife not 1.01 1.02 1.04 (0.09) (0.09) (0.10) Wife educated/ Husband not 0.72 0.77 0.80 (0.14) (0.15) (0.16) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 0.47*** 0.48*** 0.47*** (0.05) (0.05) (0.05) Husband older >10 years 0.12*** 0.12*** 0.12*** (0.01) (0.01) (0.01) Wife older >5 years 0.73 0.70 0.71 (0.32) (0.32) (0.32) Premarital Pregnancy 1.07 1.07 1.07 (0.08) (0.09) (0.09) Decade Married Pre-1980s (ref) -- -- -- 1980s 0.66** 0.92 0.90 (0.10) (0.15) (0.14) 1990s 0.48*** 1.02 0.99 (0.07) (0.17) (0.17) 2000s 0.31*** 0.86 0.82 (0.05) (0.16) (0.15) 2010s 0.15*** 0.60* 0.56* (0.03) (0.14) (0.14) Duration of Marriage 1.05*** 1.05*** (0.01) (0.01) Number of Living Children 1.10*** 1.10*** (0.03) (0.03) Rural 0.98 (0.09) Northern 1.17 (0.10) 158 Table 5.5 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Model 5 Household Wealth Quintile Poorest (ref) -- Poor 0.89 (0.10) Middle 1.01 (0.11) Rich 1.09 (0.13) Richest 0.93 (0.13) Constant 1.06* 1.10* 7.85*** 1.27 1.28 (0.04) (0.04) (1.25) (0.30) (0.34) Obs. 3683 3683 3683 3683 3683 Pseudo R 2 0.00 0.00 0.14 0.17 0.17 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 I also examine this outcome using the husband’s individual-level characteristics in Table 5.6 for interethnically married men and Table 5.7 for interreligiously married men. In the men’s individual model with full controls (Model 4), no ethnic or religious background is significant in predicting him taking another wife. His age significantly increases the likelihood that she will be a higher-order wife. An increase in the duration of the marriage and an increase in the number of living children increase the likelihood that the wife is the first wife. Education, urbanicity, region, and wealth are all non-significant. 159 Table 5.6: Odds of transition to polygyny — Men’s characteristics interethnic (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interethnic 0.92 0.86 1.10 1.05 (0.11) (0.11) (0.15) (0.15) Ethnicity Fon (ref) -- -- -- Adja 0.91 0.85 0.85 (0.09) (0.10) (0.10) Bariba 1.11 0.93 1.07 (0.16) (0.15) (0.29) Betamaribé 0.88 0.96 1.10 (0.13) (0.16) (0.30) Dendi 1.41 1.53 1.74 (0.29) (0.36) (0.55) Peuhl 1.15 0.80 0.93 (0.19) (0.15) (0.26) Yoa/Lokpa 1.19 1.05 1.20 (0.22) (0.22) (0.35) Yoruba 1.04 1.05 1.07 (0.14) (0.16) (0.16) Religion Traditional (ref) -- -- -- Muslim 1.08 0.98 0.96 (0.15) (0.16) (0.16) Catholic 1.14 1.04 1.02 (0.13) (0.13) (0.13) Protestant 1.23 0.93 0.91 (0.22) (0.19) (0.19) Other Christian 1.27* 1.18 1.17 (0.15) (0.16) (0.16) Other 1.34 1.10 1.08 (0.46) (0.44) (0.43) No religion 1.19 0.97 0.96 (0.20) (0.19) (0.19) Educated 1.17** 1.12 1.09 (0.09) (0.10) (0.09) Age at Survey 0.86** 0.86** (0.03) (0.03) Duration of Marriage 1.19*** 1.19*** (0.01) (0.01) Number of Living Children 1.17*** 1.17*** (0.03) (0.03) Rural 0.92 (0.08) Northern 0.88 (0.20) 160 Table 5.6 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Household Wealth Quintile Poorest (ref) -- Poor 0.90 (0.10) Middle 1.00 (0.12) Rich 1.07 (0.13) Richest 1.05 (0.15) Constant 1.06 0.89 23.74*** 26.60*** (0.04) (0.08) (5.33) (6.71) Obs. 3683 3683 3683 3683 Pseudo R 2 0.00 0.00 0.20 0.20 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 161 Table 5.7: Odds of transition to polygyny — Men’s characteristics interreligious (Page 1 of 2) Model 1 Model 2 Model 3 Model 4 Interreligious 0.87 0.89 0.85 0.85 (0.06) (0.07) (0.08) (0.08) Ethnicity Fon (ref) -- -- -- Adja 0.91 0.85 0.85 (0.09) (0.10) (0.10) Bariba 1.13 0.95 1.11 (0.16) (0.15) (0.30) Betamaribé 0.90 0.98 1.15 (0.13) (0.17) (0.32) Dendi 1.38 1.52 1.75 (0.29) (0.36) (0.55) Peuhl 1.16 0.80 0.95 (0.19) (0.15) (0.27) Yoa/Lokpa 1.16 1.06 1.23 (0.21) (0.22) (0.36) Yoruba 1.04 1.06 1.07 (0.14) (0.16) (0.16) Religion Traditional (ref) -- -- -- Muslim 1.04 0.95 0.93 (0.15) (0.15) (0.15) Catholic 1.14 1.04 1.02 (0.12) (0.13) (0.13) Protestant 1.23 0.95 0.93 (0.22) (0.20) (0.19) Other Christian 1.25* 1.16 1.15 (0.15) (0.16) (0.16) Other 1.41 1.19 1.16 (0.49) (0.47) (0.46) No religion 1.23 1.02 1.01 (0.21) (0.20) (0.20) Educated 1.16* 1.13 1.09 (0.09) (0.10) (0.09) Age at Survey 0.86** 0.86*** (0.03) (0.03) Duration of Marriage 1.19*** 1.19*** (0.01) (0.01) Number of Living Children 1.17*** 1.17*** (0.03) (0.03) Rural 0.91 (0.08) Northern 0.86 (0.19) 162 Table 5.7 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Model 4 Household Wealth Quintile Poorest (ref) -- Poor 0.88 (0.10) Middle 0.99 (0.12) Rich 1.06 (0.13) Richest 1.04 (0.15) Constant 1.10* 0.92 25.12*** 28.66*** (0.04) (0.09) (5.71) (7.32) Obs. 3683 3683 3683 3683 Pseudo R 2 0.00 0.00 0.20 0.20 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 These findings offer no evidence for either lower or higher relationship investment among interethnically married couples relative to their intraethnic counterparts nor does it offer evidence for a lack of investment among interreligious couples. However, it does offer limited evidence that women who enter into interreligious marriages are more likely to enter into a polygynous marriage compared to their intrareligious counterparts — meaning they are more likely to enter into a marriage where there is already at least one wife. Discussion This paper examined locally salient marital destabilizing behaviors, extramarital sex and polygyny, to see if men in interethnic and interreligious marriages are more likely to engage in behaviors that are both causes and outcomes of marital strain compared to their intramarried counterparts, indicating lower marital investment. Research indicates that interethnic and interreligious couples may be less invested in their marriages compared to their intramarried counterparts (Burton et al. 2010; Gnoumou Thiombiano 2017; Hewitt et al. 2005; Hohmann- Marriott and Amato 2008; Locoh and Thiriat 1995; Reniers 2003; Smith et al. 2012; Zhang and 163 Van Hook 2009). While this is not due to ethnicity or religion, per se, intermarried couples may face higher levels of social sanctions and less support from their families for their partner choice compared to people who select someone from their same social background (Hohmann-Marriott and Amato 2008; Maïga and Baya 2011). As a result of this higher marital strain, studies in various contexts outside of sub-Saharan Africa find that interethnic and interreligious marriages may face an increased risk of union dissolution (Burton et al. 2010; Chinitz and Brown 2001; Hohmann-Marriott and Amato 2008; Schramm et al. 2012; Smith et al. 2012; Zhang and Van Hook 2009). Although generally less is known about this topic in sub-Saharan Africa, studies find mixed evidence for whether these couples are at increased risk of dissolution in this region (Gnoumou Thiombiano 2017; Locoh and Thiriat 1995; Reniers 2003; Takyi 2001). However, divorce only captures marriages that formally dissolve and Benin is a low divorce context (Clark and Brauner-Otto 2015), meaning even low-quality unions in which the spouses are unhappy may not end in divorce in Benin. Extramarital sex and polygyny are locally salient marital destabilizing behaviors that are indirect indicators of low investment in the couple-unit from the husband. These two indicators are present in intact first marriages and have repercussions for the couple- and family-unit, particularly wives and children. In Beninese marriages (Falen 2011), men’s engagement in extramarital sex is a cause of conflict between spouses, lowering their trust and marital satisfaction, lowering overall marital investment (Falen 2011; Previti and Amato 2004; Smith 2007, 2010). Additionally, Beninese husbands use the threat of polygyny to exert power over their wives (Falen 2011), which is undoubtedly related to polygyny frequently resulting in lower levels of marital satisfaction and mental health among women (Bledsoe 1990; Bove and Valeggia 2009; Falen 2008a). In addition, both of these actions frequently result in the 164 household’s resources being diverted and split beyond the first/primary couple-unit, causing marital strain (Falen 2008a, 2011; Lacombe 1987; Orubuloye et al. 1997; Smith 2002, 2007, 2010), as well as putting women at increased risk of sexually transmitted infections (Smith 2007). In this paper, I looked at whether husbands in interethnic and interreligious marriages are less invested in their unions than intramarried husbands by measuring whether intermarried husbands are more likely to marital destabilizing behaviors in Benin. Using two marital destabilizing behaviors, extramarital sex and the transition to polygyny, this study found slight evidence of lower investment for interethnically married men. Men in interethnic marriages are significantly more likely to report engaging in extramarital sex within the last year. However, this increased likelihood is mitigated after including controls for couples’ characteristics, specifically educational attainment in both absolute terms and relative to each other. Apart from couples’ educational attainment, the unique inclusion of controls for the duration of marriage, the number of living children, polygyny, urbanicity, and current wealth status all respectively render the difference non-significant. Contrarily, there is no evidence that interethnic marriages are more likely to transition into polygyny than are intraethnic marriages. Overall, these findings suggest that interethnically married men are not less invested in their marriages than are intraethnically married men. Rather, due to the selection of educated, wealthy, and urban men into these marriages, interethnically married men may be better able to engage in extramarital sex. This may be due to their capital levels making them attractive partners and/or the expectation of men of high capital to engage in extramarital sex (Falen 2011; Lacombe 1987). Further, interethnic marriages are disproportionately new due to their recent increase in frequency. This relative newness of interethnic marriages compared to their intraethnic peers 165 contributes to their increased likelihood of reporting extramarital sex as younger men are significantly more likely to report engaging in extramarital sex than older men. Among interreligious marriages, there are robust indications of husbands’ lower marital investment compared to their intrareligiously married counterparts. In all models, interreligiously married men are around 20% more likely to report engaging in extramarital sex than their intrareligiously married counterparts. Although interreligiously married men are more likely to engage in extramarital sex, there is no indication that interreligious marriages transition into polygyny at higher rates than intrareligious marriages. There is, however, limited evidence that interreligious marriages are more likely to occur at higher wife ranks meaning the wife of a different religious background married a man who already had at least one wife. However, entering as a higher-order wife does not indicate a lack of marital investment in the interreligious couple-unit. Overall, these findings do suggest that men in interreligious marriages are less invested in the primary couple-unit and more likely to engage in marital destabilizing behaviors than are men in intrareligious marriages, but this is restricted to extramarital sex and potentially informal polygyny, rather than formalized polygyny. Part of this may be explained by the religious composition of interreligious marriages. There are low levels of intermarriage between Muslim men and non-Muslim women (12% of interreligious marriages) and Muslim men are among the least likely to report extramarital sex. Contrarily, there are high levels of interreligious marriages involving a male Traditional adherent (29% of interreligious marriages) or a Catholic man (23% of interreligious marriages), who report higher levels of extramarital sex than men of other religious groups. These religious differences corroborate similar findings in other parts of West Africa (Mitsunaga et al. 2005). However, the findings of higher levels of extramarital sex are 166 robust in the men’s individual model, which includes controls for men’s religion. Thus, the religious composition of men in these marriages cannot explain this finding fully. These results indicate that men in interreligious marriages may be more likely to engage in informal polygyny than men in intrareligious marriages, perhaps due to the majority of interreligious marriages (63%) involving one Catholic or Protestant partner — husband or wife — whose teachings do not permit formal polygyny (Falen 2008a). Most importantly, these robust findings may indicate that there is less investment both individually from men and extended family in interreligious marriages, opening up opportunities to engage in extramarital sex (Maïga and Baya 2011). This indicates that women in interreligious marriages may be at increased risk of experiencing negative health outcomes, including lower mental health and increased risks of sexually transmitted infections. Further, if her husband’s extramarital sex results in a pregnancy, the resources of her household will diminish, with negative implications for her and her children. The findings of this study suggest that being interethnically versus interreligiously married affects men’s marital investment differently, indicating that intermarriages as a whole do not equally experience marital strain. Rather, interethnic couples appear to be less likely to experience and/or be protected from marital strain related to their ethnic differences. This may be because their high levels of socioeconomic standing vis-à-vis the general population shields them from marital strain by allowing them to rely less on their families of origin and/or these couples may be more likely to select partners whose values agree with their own (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). Regardless, interethnic marriage does not appear to put women and children at a higher risk of negative health outcomes based on their husband’s behaviors compared to intraethnic marriage. 167 Contrarily, interreligious unions do appear to experience more marital strain than intrareligious couples. This marital strain may lead to negative outcomes for women in these marriages. While this study cannot explain why exactly interreligiously married men engage more in extramarital sex, it may be that their families disapprove of the union and they may be encouraged to engage in extramarital sex (Maïga and Baya 2011). Additionally, because they do not have the same levels of education and wealth as interethnically married men, they may rely more heavily on their family of origin, increasing marital strain and decreasing their investment in the couple-unit (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). Resultantly, although both interethnic and interreligious marriages are comprised of individuals who married someone with a different social identity, these couple types do not appear to experience the same levels of marital strain relative to each other. Due to the lack of direct indicators on marital investment in sub-Saharan Africa, I used two locally salient marital destabilizing behaviors, extramarital sex and polygyny, to indirectly capture marital investment. These serve as indicators of marital disharmony that makes sense in the Beninese context as both are a cause and consequence of marital problems (Falen 2008a, 2011; Previti and Amato 2004). Thus, these indirect indicators serve as a way to help address the lack of marital quality indicators in DHS data and sub-Saharan African literatures. However, these indicators are not without their limitations as they may understate the marital strain of interethnic and interreligious marriages. Both extramarital sex and polygyny are associated with status among men in Benin and are not uncommon (Falen 2008a, 2011), meaning these behaviors may not map on to men’s own perceived investment in their primary and/or first couple-unit. Due to the limited time frame around engagement in extramarital sex (within the last year) and the lack of questions about marital history, we do not know if husbands have ever 168 engaged in extramarital sex, only that they did so recently, nor do we know when the couple transitioned into polygyny. As a result, these two indicators may not be picking up on existing low investment in the marriage, especially if the wife does not know about her husband’s extramarital engagements, but the report of these behaviors is likely to cause future marital problems (Falen 2008a, 2011). Additionally, because extramarital sex and polygyny are relative status signs, these indicators may not equally correlate with marital investment across class boundaries. The poorest men are the least likely to engage in extramarital sex and polygyny, meaning these are not likely to be salient marital destabilizing behaviors for men of lower wealth backgrounds. Thus, using different indicators, such as intimate partner violence 18 , to measure marital investment may find that using extramarital sex and polygyny understate the existence of marital strain in interethnic and interreligious marriages. Further, extramarital sex and polygyny only capture indirect measures of men’s investment in the couple-unit, a reflection of the limited direct and indirect measures of women’s investment in the couple-unit. Across West Africa, women report very low levels of extramarital sex making women’s extramarital sex difficult to capture in a study (Oluwaseyi et al. 2014). However, because extramarital sex is normalized among men relative to women, extramarital sex among women may be a stronger correlate of low marital investment than the same behavior among men, both as a cause and a consequence of low marital investment. Due to the strong stigma against women engaging in extramarital sex, the marriages in which it occurs might be at higher risk of dissolution than when the man engages in it. If this is true, these couples are unlikely to be anywhere near as represented in intact first marriages as compared to couples where the 18 While men are affected by intimate partner violence, the current DHS data focus on women’s experiences rendering it an indicator of men’s behaviors akin to extramarital sex and polygyny. 169 husband engages in this behavior. Similarly, as polyandry is not practiced in Benin, developing indicators that focus on women’s expressions of marital destabilizing behaviors in intact marriages would greatly expand our understanding of the relative marital strain experienced in interethnic and interreligious marriages. One challenge of creating a measure of polygyny is that I measure monogamy and women’s wife rank based on a wife’s own reporting and wives may not know about informal spouses their husbands keep (Falen 2011). This means that there is a chance that a husband is courting another wife and/or has an informal partner that the surveyed wife is unaware of and therefore mistakenly reports herself as monogamously married. Additionally, because long-term extramarital partners are used in lieu of formal polygyny, men and women may not report these long-term relationships as polygynous marriages, even if they would qualify as such (Falen 2008a, 2011; Lacombe 1987; Ware 1979). Another limitation to measuring polygyny for interethnic marriages is that the transition to polygyny may occur after the survey due to the relative newness of these marriages. In the couples’ survey as a whole, only 9% of couples in the 1996 survey were interethnic compared to a full 20% in the 2018 survey, indicating that these marriages are very new and may not yet have had the time to transition into polygyny. This means that we may find in later surveys, once these couples have become more common and have been married for longer durations, that they do have different risks of transitioning into polygyny. Contrarily, interreligious marriages have been relatively stable at 31% of the couple’s survey in both 1996 and 2018. This indicates that although many of these marriages have had time to transition, they were no more likely to do so than intrareligious marriages. 170 Further, women who are opposed to their spouse engaging in extramarital sex or being married polygynously may end their marriage when their husband engages in these behaviors. This means that I am only capturing marriages that remained intact after experiencing extramarital sex and the transition to polygyny. There is also the possibility that interethnic and interreligious marriages are particularly unlikely to survive these experiences. As the DHS does not ask marital histories, I have no data on the reason for divorces, nor the characteristics of former couples, meaning I cannot determine how these transitions relate to union dissolution. However, Benin is a low-divorce context (Clark and Brauner-Otto 2015), which helps minimize the potential effects of limiting my analysis to intact first marriages. Despite the limitations posed by these indicators, I find that, overall, men in interethnic marriages do not appear to be less invested in the couple-unit than men who are in intraethnic marriages. In the Beninese context, women in interethnic marriages may not be at increased risk of experiencing behaviors and transitions that put her and her children’s health at risk. The overall lack of support for lower levels of marital investment for interethnic marriages is surprising given that, in other contexts, intermarriages of differing types are significantly less cohesive and more prone to end in divorce. However, the high levels of human and economic capital among interethnic couples may be protective of these unions as wealthier and higher educated unions are less likely to dissolve in other contexts (Amato 2010; Kaplan and Herbst 2015). Further research using other indicators of marital investment could explore how robust these results are in Benin as well as other sub-Saharan contexts. Contrarily, men in interreligious marriages do appear to engage in more marital destabilizing behaviors than intrareligiously married men, but only when using extramarital sex as an indicator. As a result of this increased risk of extramarital sex, women in interreligious marriages 171 may be at increased risk of negative physical and mental health outcomes, including increased exposure to HIV (Smith 2007) and their children may grow up with fewer resources. That interreligious couples do not differ from their intrareligious counterparts in their levels of human and economic capital may put them at increased risk of marital strain due to needing to rely on their extended kin networks more for support (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). Further explorations of the links between couples’ religious identifications, rather than men’s and women’s individual identification, and extramarital sex could be a fruitful avenue of research in understanding religious identification’s role in sexual behaviors. As people increasingly select partners along the lines of matching human and social capital rather than social identities, such as ethnicity and religion, intermarriages are becoming more common. This paper contributes to the minimal research available about whether interethnically and interreligiously married men are less invested in their marriages than their intramarried counterparts in West Africa. By using two locally salient indicators of men’s marital destabilizing behaviors that also potentially result in negative outcomes for their wives and children, I expand our understanding of the ways in which different types of intermarriages are differentially linked to lower marital investment among spouses and worse health outcomes for women and children. 172 CHAPTER 6 CONCLUSION Summary Marital formation patterns in sub-Saharan Africa are undergoing shifts due to large-scale structural changes — both social and economic (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Clark and Brauner-Otto 2015; Clark et al. 2017; Koski et al. 2017; Maïga and Baya 2011). These changes, including increasing educational attainment, — particularly for girls — urbanization, and individualization are correlated with changes in partner selection in West African cities (Calvès 2016a, 2016b; Calvès and Gnoumou Thiombiano 2014; Calvès and Marcoux 2007; Maïga and Baya 2011, 2014). Specifically, the decreasing role of the extended family in partner selection and the desire to find a partner who matches along the lines of educational attainment are altering who is considered an attractive potential spouse on the marriage market. These shifts in partner selection may increasingly open the doors to interethnic and interreligious marriages as social identity homogamy decreases in importance in favor of matching along the lines of human capital (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Maïga and Baya 2011). I build upon this marital literature by examining interethnic and interreligious marriage in Benin, West Africa. Using couples’ data collected by the Beninese Demographic and Health Surveys, I investigated the temporal and spatial distribution of interethnic and interreligious marriage, the characteristics associated with entry into these unions, as well as several different indicators of marital strain. The results presented in Chapter 3 indicate that interethnic marriage is correlated with changing economic and social structures in Benin that diversify social networks and correspond to prioritizing various forms of capital, economic and human, on the 173 marriage market more than shared ethnic backgrounds. Although ethnically homogenous marriages are still the overwhelming majority of pairings, interethnic marriages are rapidly increasing across marital cohorts, supporting research from capital cities that urbanization and education are reshaping ethnicity’s role in family life (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Shapiro and Tambashe 2017). Contrarily, interreligious marriages mirror intrareligious marriages’ average socioeconomic and demographic characteristics, with minimal changes in frequency across marital cohorts. Marrying across religious lines does not appear to correlate heavily with recent structural changes in Benin. This is in contrast to other studies that found religion’s salience in partner selection increased with large-scale structural changes (Crespin-Boucaud 2018; Maïga and Baya 2011). This suggests that, in Benin, religion may not be increasingly important in shaping social networks, even in urban areas. In Chapter 4, I examined the relationship between interethnic and interreligious marriage and fertility preferences and outcomes, and whether these couples report higher levels of preference concordance, indicating high levels of communication within intermarriages. I found that men and women in interethnic marriages have lower fertility preferences than intraethnically married men and women. They also experience significantly higher odds of concordance on their fertility preferences, indicating high levels of intra-couple communication (Andro et al. 2002; McNamee 2009). This is likely tied to their high levels of education which is correlated not only with lower fertility preferences, but a higher likelihood of communicating about fertility preferences and behaviors (Derose, Wu, and Dodoo 2010; Dodoo 1993; Ibisomi and Odimegwu 2011). However, these lower fertility preferences do not appear to translate into distinct fertility outcomes. Among interreligious couples, I found the reverse: although they do not have distinct fertility preferences or a higher likelihood of concordance, they experience significantly higher completed fertility 174 than intrareligious couples. This points to a higher likelihood of marital strain in interreligious marriages that may make compromise between spouses less successful and may make unintended pregnancies more likely (Bean and Aiken 1976; Lehrer 1996). These distinct outcomes for interethnic and interreligious marriages suggest that the high level of selection into interethnic marriages is protective against marital strain and encourages high levels of communication. Contrarily, these protective features are not present in interreligious marriages at higher levels than for the general population, which may make them more susceptible to marital strain. That neither type of intermarriage experiences lower fertility outcomes suggests that these are not couples currently at the vanguard of fertility declines, despite transgressing social norms in their partner selection. In Chapter 5, I measure two locally salient marital destabilizing behaviors that result in and cause low marital investment: men’s engagement in extramarital sex and the transition from monogamy to polygyny. Importantly, the presence of either is linked to lower mental and physical health for women, meaning that a higher prevalence in interethnic and interreligious marriages may have health implications for the family-unit (Bove and Valeggia 2009; Previti and Amato 2004). My results showed that men in interethnic marriages are more likely to engage in extramarital sex than intraethnically married men, but once socioeconomic characteristics are controlled for, there is no significant difference. In comparison, men in interreligious marriages are significantly more likely to report engaging in extramarital sex regardless of controls. This indicates that interreligious unions not only experience more strain, but the husband’s reduced investment in the relationship manifests in ways that put his wife in danger. There is no evidence that either type of intermarriage is at increased risk of transitioning from a monogamous marriage into a polygynous one. These results further support the findings from prior chapters: 175 the different selection into interethnic and interreligious marriages lead these two types of intermarriages to have distinct outcomes, with interreligious marriages exhibiting more signs of marital strain than interethnic marriages. In summation, interethnic and interreligious unions do not appear to be experienced in the same way in the Beninese context despite both involving a crossing of social identities. This may be explained by the differential selection that goes into these respective unions in Benin. Interethnic marriages do not exhibit greater signs of marital strain than intraethnic unions, both in terms of fertility concordance and outcomes, as well as extramarital sex and polygyny. Rather, these unions may be protected from higher strain due to their high levels of social and economic capital, which may buffer them from displeased extended kin and may offer better negotiating positions to wives. Contrarily, interreligious marriages exhibit characteristics that mirror their intrareligious counterparts, outside of their choice to marry interreligiously. However, these unions exhibit multiple indicators of strain, including increased completed fertility and increased odds that the husband has engaged in extramarital sex. That these couples are not highly selective may mean that they are more reliant on their extended kin-networks, who may not approve of the union, increasing marital strain and seemingly lowering the husband’s investment in these unions. Contributions to current research This research contributes to the literatures in demography, marriage, family, fertility, ethnicity, and religion, particularly in West Africa, a region of sub-Saharan Africa known for ethnicity’s and religion’s strong influence on family life (Caldwell 1996; Caldwell and Caldwell 1987). Specifically, I examined an understudied sub-set of relationships, interethnic and interreligious marriages. I used couples’ data to understand the trends and implications of these union types in 176 a context undergoing social and economic changes. I build on prior research showing that interethnic and interreligious marriages are tied to the shifting prioritization of social identity homogamy in favor of human capital homogamy on the marriage market and that these couples face distinct outcomes due to their different social identities. In contrast to most prior research, I give an analysis of one country to fully flesh out the spatial and temporal nature of these unions, as well as their characteristics and outcomes. In doing so, I offer a comprehensive analysis of interethnic and interreligious marriage in Benin, West Africa, a region and country where ethnicity and religion are highly salient in marriages (Crespin-Boucaud 2018). Importantly, this dissertation centers couples as the unit of analysis. Women are the most frequently used unit of analysis in demographic studies, particularly relating to marital quality and fertility preferences and outcomes. However, men and women negotiate throughout their marriages and centering both parties contributes to our understanding of both partners’ roles in marriage. Relatedly, most studies focusing on couples offer results based on intramarriages, as these are the most common type of union. My results highlight the importance of examining interethnic and interreligious marriages, as they are distinct in several ways compared to their intramarried counterparts, shaping the health of the whole family unit. The divergent results of interethnic and interreligious couples in this study suggest the need to further examine interethnic and interreligious unions separately. Because both of these marriages involve people crossing social groups, they are often studied at the same time (Crespin-Boucaud 2018; Maïga and Baya 2011; Ntoimo and Mutanda 2017). However, this study illustrates that ethnicity and religion do not operate in the same manner. Rather, the boundaries surrounding these social identities may respond differently to the same changes in the marriage market. In the Beninese context, due to the religious diversity of ethnic groups, 177 interreligious marriages are a common feature of the marital landscape (Falen 2011). Additionally, due to the syncretic nature of religion in Benin, marrying across religious boundaries may seem less like marrying across an identity category than marrying across ethnicities, which often correspond with languages and other social traditions. Interreligious marriage’s relatively stable frequency indicates that, in this context, interreligious unions may not be linked to structural changes that reshape social networks and increase the prioritization of achieved characteristics on the marriage market. Instead, the gender imbalance of religious identification makes interreligious marriage a necessity in a way that interethnic marriage is not. Despite being relatively common, interreligious marriages exhibit more indicators of low investment suggesting that these unions are at an increased risk of marital strain compared to couples who share a religious affiliation. Similarly, my data point to significantly higher levels of positive selection into interethnic unions than interreligious unions. This high level of selection into interethnic marriages may be due to their relative newness compared to interreligious marriages: early adaptors of social trends often experience high levels of selection. For example, early adopters of cohabitation are often distinct from couples who engage in this union type once the practice becomes widely diffused (Liefbroer and Dourleijn 2006). These high levels of positive selection into interethnic unions may be protective against marital strain despite being less common. These results point to the importance of context in understanding how specific social identities are tied to marriage market outcomes and how selection into these unions relates to the couples’ outcomes. Thus, further research should explore how different types of intermarriage correlate with broader social changes, particularly in contexts where one social identity is more salient than another in shaping spousal choice. 178 Limitations These studies are not without their limitations. Due to data limitations, all marriages analyzed are intact first unions. As a result, it is only possible to capture unions that have not dissolved, which may occur more frequently in interethnic and interreligious marriages (Locoh and Thiriat 1995; Maïga and Baya 2011). This means that my sample is affected by survivorship bias, especially among older unions. The findings presented are, therefore, based on couples who did not dissolve in the face of the additional pressure these union types are likely to incur. This would result in my sample being more cohesive than interethnic and interreligious marriages at large. To help mitigate this, I included robustness checks limiting my sample to recently married couples, and the results from this sub-population are similar to the larger sample, suggesting that the effect of dissolution may be minimal. Although the Demographic and Health Surveys are designed to be nationally representative and therefore generalizable, unfortunately, this is not true of the couples’ survey. While these three studies offer insight into the relationship between interethnic and interreligious marriage, broader social change, and marital strain in the Beninese context, they cannot be extrapolated to all interethnic and interreligious marriages or other contexts. These data are particularly likely to underrepresent marriages where the husband resides elsewhere due to migration and/or polygyny. Further, these data are unable to address the micro-level characteristics and interactions that influence partner selection, such as attractiveness. One of the largest limitations of studying social identities is that while we often think of these identities as constant, in reality, ones’ identity may change over time. Religious affiliation in sub-Saharan Africa is an especially mutable category, meaning that some of the interreligious marriages captured in these data may not have been interreligious at the time of marriage. 179 Additionally, some of the marriages identified as intrareligious may be the result of the conversion of one of the spouses upon marriage (Agadjanian 2017). Conversion to the same religion upon or after marriage would most likely serve to increase marital stability (Chiswick and Lehrer 1991; Lehrer 1996), which would render the remaining interreligious marriages more strained in comparison, as my results find. However, because religion plays a strong role in shaping norms and customs, currently intrareligious couples who come from different religious backgrounds may be at a higher risk of conflict at some point in their marriage. Contrarily, religious conversion after marriage may be a result of strain present elsewhere in the couple’s lives and a change in religious identity may also result in more strain than for couples who remain intrareligious or were already interreligious at the time of marriage (Agadjanian 2017). Resultantly, my findings that interreligious marriages experience more strain may not be due to the difficulty of being a couple from two different religious backgrounds per se, rather, the instability of a spouse’s identity within marriage may, in fact, increase marital strain and lower marital investment. Unfortunately, as the DHS does not ask religious histories, I was unable to determine if interreligious couples were interreligious at the time of marriage and I, therefore, treated them as though they were. This decision has particular implications for the third chapter, which attempts to predict the characteristics associated with entry into an interreligious marriage, as I may not be accurately capturing selection into interreligious marriages so much as the characteristics of couples who have different religious affiliations at some point in their marriage. However, this is a less severe limitation for the measures used in chapters four and five, which, minus completed fertility, are largely contemporaneous to the stated religious affiliations. 180 This limitation impacts the interreligious analyses more than the interethnic ones, as ethnicity is largely considered to be more static across time due to notions of biological ties (Brubaker 2009; Schraml 2014). If ethnic affiliation were to change, it would likely be older women who are in happy, long-term marriages changing their ethnic identity to match that of their husband and his family (Crespin-Boucaud 2018). The exclusion of these couples from the category of interethnic marriage would remove high cohesion marriages from the interethnic sample. However, as interethnic marriages exhibit signs of higher levels of communication and similar levels of male investment as marriages with intraethnic partners, their exclusion does not appear to have biased the results. That said there is no research, to my knowledge, that shows ethnic conversions occur among interethnically married women in sub-Saharan Africa, let alone in the Beninese context. Discussion and implications Benin is a perfect case study to examine interethnic and interreligious marriage in West Africa. It is highly ethnically and religiously diverse with both social identities playing a larger role in marital assortment than in many other sub-Saharan countries (Crespin-Boucaud 2018). In studying Benin, I offer further evidence that structural changes are tied to changing marital formation patterns and marital norms, including the salience of shared social identity on the marriage market (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011). Exploring interethnic and interreligious marriages in a single country allowed me to examine trends in the distribution of interethnic and interreligious marriages temporally and spatially, while also going further to investigate indicators of strain in these marriages. Using a single country context to examine both the changes in the likelihood of 181 these union types and the implications of these marriages is a perspective that has been missing from prior research. The results of this research indicate that the importance of shared ethnic identification on the marriage market is likely shifting in the face of broader structural changes that prioritize educational homogamy, supporting prior research finding that ethnicity’s salience in family life is in-flux (Calvès 2016a; Calvès and Gnoumou Thiombiano 2014; Crespin-Boucaud 2018; Maïga and Baya 2011; Shapiro and Tambashe 2017). On the other hand, my results offer no evidence that religion’s salience on the marriage market is correlated with broader structural changes, contrary to one study that found it to be increasingly salient (Crespin-Boucaud 2018). These differences may result from a combination of my using more expansive religious categories (Crespin-Boucaud 2018) and Benin being a context where religious boundaries are more frequently crossed than ethnic ones in marriage due, in part, to rural areas being more religiously than ethnically diverse (Falen 2011). Additionally, religion is syncretic in Benin, with people often practicing the traditions of multiple faith backgrounds at once, which may make different religious identities less of a barrier to marriage (Falen 2008a, 2011, 2016). As a result, ethnicity poses a greater barrier to marriage in Benin if for no reason other than the higher prevalence of ethnically homogenous social networks. Thus, in the Beninese context, the increase of ethnically diverse urban populations and schooling opportunities have likely opened up new social locations for networks to diversify ethnically, increasing the opportunity to select a partner of a different ethnic background. Further supporting this is that interethnic couples are strongly positively selected along the lines of human and economic capital. Interethnic couples are not randomly assorted, as interreligious couples appear to be. Rather, these are couples who are educated and have the 182 socioeconomic attributes that frequently correlate with increased individualization (Calvès and Marcoux 2007), increasing the likelihood that they selected their partner with minimal family influence (Calvès 2016b; Maïga and Baya 2011). Likely as a result of this positive selection into interethnic marriages, interethnic couples do not exhibit greater signs of marital strain than their intraethnically married counterparts. Contrarily, interreligious couples, whose characteristics are similar to intrareligious couples, do experience higher levels of marital strain than their intrareligiously married counterparts. Thus, high levels of capital appear to be protective against marital strain in intermarriages and intermarried couples without higher than average levels of resources may be at a higher risk of marital strain than their intramarried peers. Crossing social boundaries for marriage, therefore, does appear to increase the risk of negative outcomes for the couple-unit if they do not have the resources to mitigate some sources of stress, such as reliance on their extended families. To conclude, shared social identities still play an important role on the marriage market and in the relative strain couples experience in their marriage. Importantly, while intermarriages may experience more strain than intramarriages, increasing access to resources such as education may help diminish the association between intermarriage and higher marital strain. My research indicates that the link between social identity and marital partner selection is changing over time in West Africa. Additional research exploring under what conditions intermarriages exhibit higher marital strain will expand our understanding of how ethnicity and religion are linked to marriage and how marriage is, in turn, linked to fertility and the well-being of the family unit. 183 REFERENCES Acquah, Francis. 2011. “The Impact of African Traditional Religious Beliefs and Culture Values on Christian-Muslim Relations in Ghana from 1920 through the Present: A Case Study of Nkusukum-Ekumfi-Enyan Area of the Central Region.” University of Exeter. Addai, Isaac. 1996. “Ethnic Fertility Differentials in Ghana and Their Consequences.” University of Alberta. Addai, Isaac and Jelena Pokimica. 2010. “Ethnicity and Economic Well-Being: The Case of Ghana.” Social Indicators Research 99(3):487–510. Addai, Isaac and Frank Trovato. 1999. “Structural Assimilation and Ethnic Fertility in Ghana.” Journal of Comparative Family Studies 30(3):409–27. Adebayo, Akanmu G. 1997. “Contemporary Dimensions of Migration among Historically Migrant Nigerians: Fulani Pastoralists in SouthWestern Nigeria.” Journal of Asian and African Studies 32(1–2). Adida, Claire L. 2015. “Do African Voters Favor Coethnics? Evidence from a Survey Experiment in Benin.” Journal of Experimental Political Science 2(1):1–11. Adida, Claire L., Adam Chabi Bouko, Alex Verink, Ganz Chockalingam, and Jennifer Burney. 2018. “Pilot of a Mobile Money School Fee Payment System in Rural Benin.” PLoS ONE 13(6):1–14. Adongo, Philip B., James F. Phillips, and Fred N. Binka. 1998. “The Influence of Traditional Religion on Fertility Regulation among the Kassena-Nankana of Northern Ghana.” Studies in Family Planning 29(1):23-40. Adotevi, Stanislas Spero, Robin Law, and Dov Ronen. 2017. “Benin.” Encyclopædia Britannica. Agadjanian, Victor. 2017. “Women’s Schooling and Religious Mobility: Joining, Switching, and Quitting Church in a Christian Sub-Saharan Setting.” Sociology of Religion: A Quarterly Review 78(4):411–36. Agadjanian, Victor and Alex Chika Ezeh. 2000. “Polygyny, Gender Relations, and Reproduction in Ghana.” Journal of Comparative Family Studies 31(4):427–41. Agadjanian, Victor and Scott T. Yabiku. 2014. “Religious Affiliation and Fertility in a Sub- Saharan Context: Dynamic and Lifetime Perspectives.” Population Research and Policy Review 33(5):673–91. 184 Akintoye, Stephen. 2004. “Yoruba History: From Early Times to the 20th Century.” Pp. 3–30 in Understanding Yoruba Life & Culture, edited by N. S. Lawal, M. N. O. Sadiku, and P. A. Dopamu. Trenton, NJ: Africa World Press Inc. Alber, Erdmute, Tabea Häberlein, and Jeannett Martin. 2010. “Changing Webs of Kinship: Spotlights on West Africa.” Africa Spectrum 45(3):43–67. Ali, Mohamed M. and John Cleland. 2001. “The Link between Postnatal Abstinence and Extramarital Sex in Cote d’Ivoire.” Studies in Family Planning 32(3):214-219. Amadou Sanni, Mouftaou and Candide Mahouton Atodjinou. 2012. État et Dynamique Des Langues Nationales et de La Langue Française Au Bénin. Québec. Amato, Paul R. 2010. “Research on Divorce: Continuing Trends and New Developments.” Journal of Marriage and Family 72(3):650–66. Amato, Paul R. and Alan Booth. 1995. “Changes in Gender Role Attitudes and Perceived Marital Quality.” American Sociological Review 60(1):58–66. Amato, Paul R. and Bryndl Hohmann-Marriott. 2007. “A Comparison of High- and Low- Distress Marriages That End in Divorce.” Journal of Marriage and Family 69(3):621–38. Amoah-Boampong, Cyrelene. 2016. “Songhai Empire.” The Encyclopedia of Empire. Andro, Armelle, Véronique Hertrich, and Glenn D. Robertson. 2002. “Demand for Contraception in Sahelian Countries: Are Men’s and Women’s Expectations Converging?” Population (English Edition) 57(6):929–57. Anon. 2004. “Songhay.” Dictionary of Languages. Anon. 2017. “Fon.” Encyclopaedia Britannica. Attanasso, Marie-Odile. 2005. Determinents of Monetary Poverty among Female-Headed Households in Benin. Awusabo-Asare, Kofi. 1990. “Matriliny and the New Intestate Succession Law of Ghana.” Canadian Journal of African Studies / Revue Canadienne Des Études Africaines 24(1):1-16. Baba-Moussa, Abdel Rahamane. 2013. “Alphabétisation et Éducation En Langues Nationales Dans Les Politiques Globales de l’éducation Au Bénin : Valeurs, Principes d’actions et Stratégies d’acteur.” Cahiers de La Recherche Sur l’éducation et Les Savoirs (En Ligne). Bankole, Akinrinola and Susheela Singh. 1998. “Couples’ Fertility and Contraceptive Decision- Making in Developing Countries: Hearing the Man’s Voice.” International Family Planning Perspectives 24(1):15–24. 185 Barbieri, Magali and Véronique Hertrich. 2005. “Age Difference between Spouses and Contraceptive Practice in Sub-Saharan Africa.” Population 60(5):617–54. Bean, Frank D. and Linda H. Aiken. 1976. “Intermarriage and Unwanted Fertility in the United States.” Journal of Marriage and Family 38(1):61–72. Behrman, Jere, Hans-Peter Kohler, and Susan Watkins. 2002. “Social Networks and Changes in Contraceptive Use over Time: Evidence from a Longitudinal Study in Rural Kenya.” Demography 39(4):713–38. Bertrand-Dansereau, Anaïs and Shelley Clark. 2016. “Pragmatic Tradition or Romantic Aspiration? The Causes of Impulsive Marriage and Early Divorce among Women in Rural Malawi.” Demographic Research 35(1):47–80. Bialik, Kristen. 2017. Key Facts about Race and Marriage, 50 Years after Loving v. Virginia. Bierschenk, Thomas. 1996. “Peuls et État Colonial Dans Le Borgou Français/Nord-Dahomey (1895 - 1940).” Nomadic Peoples (38):99–124. Birkelund, Gunn Elisabeth and Johan Heldal. 2003. “Who Marries Whom? Educational Homogamy in Norway.” Demographic Research 8(1):1–30. Bledsoe, Caroline. 1990. “Transformations in Sub-Saharan African Marriage and Fertility.” The Annals of the American Academy of Political and Social Science 510:115–25. Boerma, J. Ties and A. Elisabeth Sommerfelt. 1993. “Demographic and Health Surveys (DHS): Contributions and Limitations.” World Health Statistics Quarterly 46(4):222–26. Bongaarts, John. 2006. “The Causes of Stalling Fertility Transitions.” Studies in Family Planning 37(1):1–16. Bongaarts, John. 2017. “Africa’s Unique Fertility Transition.” Population and Development Review 43:39–58. Bongaarts, John and John Casterline. 2012. “Fertility Transition: Is Sub-Saharan Africa Different?” Population and Development Review 38(SUPPL.1):153–68. Bongaarts, John and Susan Cotts Watkins. 1996. “Social Interactions and Contemporary Fertility Transitions.” Population and Development Review 22(4):639–82. Booth, Heather. 2010. “Ethnic Differentials in the Timing of Family Formation: A Case Study of the Complex Interaction between Ethnicity, Socioeconomic Level, and Marriage Market Pressure.” Demographic Research 23:153–90. Botha, Machaka Beatrice. 2008. “Patterns of Language Acquisition and Language Use among Members of Inter-Ethnic Marriages in Mansa Urban, Zambia.” University of Zambia. 186 Bove, Riley and Claudia Valeggia. 2009. “Polygyny and Women’s Health in Sub-Saharan Africa.” Social Science and Medicine 68(1):21–29. Bradbury, Thomas N., Frank D. Fincham, and Steven R. H. Beach. 2000. “Research on Nature and Determinants of Marital Satisfaction: A Decade in Review.” Journal of Marriage and the Family 62(November):964–80. Brown, Susan L., Wendy D. Manning, and Krista K. Payne. 2015. “Relationship Quality among Cohabiting versus Married Couples.” Journal of Family Issues (July):1–24. Brubaker, Rogers. 2009. “Ethnicity, Race, and Nationalism.” Annual Review of Sociology 35:21– 42. Bukari, Kaderi Noagah and Nicholaus Schareika. 2015. “Stereotypes, Prejudices and Exclusion of Fulani Pastoralists in Ghana.” Pastoralism 5(1):20. Bumpass, Larry L., James A. Sweet, and Andrew Cherlin. 1991. “The Role of Cohabitation in Declining Rates of Marriage.” Journal of Marriage and Family 53(4):913–27. Burton, Linda M., Eduardo Bonilla-Silva, Victor Ray, Rose Buckelew, and Elizabeth Hordge Freeman. 2010. “Critical Race Theories, Colorism, and the Decade’s Research on Families of Color.” Journal of Marriage and Family 72(3):440–59. Buzasi, Katalin. 2015. “Languages, Communication Potential and Generalized Trust in Sub- Saharan Africa: Evidence Based on the Afrobarometer Survey.” Social Science Research 49:141–55. Buzasi, Katalin. 2016. “Linguistic Situation in Twenty Sub-Saharan African Countries: A Survey-Based Approach.” African Studies 75(3):358–80. Caldwell, John C. 1996. “The Demographic Implications of West African Family Systems.” Journal of Comparative Family Studies 27(2):331–52. Caldwell, John C. 2001. “The Globalization of Fertility Behavior.” Population and Development Review 27(Supplement: Global Fertility Transition):93–115. Caldwell, John C. and Pat Caldwell. 1987. “The Cultural Context of High Fertility in Sub- Saharan Africa.” Population and Development Review 13(3):409–37. Caldwell, John C., I. O. Orubuloye, and Pat Caldwell. 1991. “The Destabilization of the Traditional Yoruba Sexual System.” Population and Development Review 17(2):229–62. Caldwell, John C., I. O. Orubuloye, and Pat Caldwell. 1992. “Fertility Decline in Africa: A New Type of Transition?” Population Council 18(2):211–42. 187 Calvès, Anne-Emmanuèle. 2016a. “First Union Formation in Urban Burkina Faso: Competing Relationship Transitions to Marriage or Cohabitation.” Demographic Research 34(1):421– 50. Calvès, Anne-Emmanuèle. 2016b. “Generational Control over Young Urban Couples in West Africa: Involvement of Elders in Union Formation and Infant Care in Ouagadougou.” Revue Quetelet/Quetelet Journal 4(1):57–72. Calvès, Anne-Emmanuèle and Bilampoa Gnoumou Thiombiano. 2014. “Quand Les Jeunes Célibataires Se Mettent En Couple: Scolarisation Féminine et Choix Des Partenaires a Ouagadougou.” in Le Marriage en Afrique. Pluralité des formes et des modèles matrimoniaux. Calvès, Anne-Emmanuèle and Richard Marcoux. 2007. “Présentation: Les Processus d’individualisation <<a l’africaine>>.” Sociologie et Sociétés 39(2):5–18. Calvès, Anne-Emmanuèle and Dominique Meekers. 1999. “The Advantages of Having Many Children for Women in Formal and Informal Unions in Cameroon.” Journal of Comparative Family Studies 30(4):617–39. Castle, Sarah. 2001. “‘The Tongue Is Venomous’: Perception, Verbalisation and Manipulation of Mortality and Fertility Regimes in Rural Mali.” Social Science and Medicine 52(12):1827– 41. Çelikaksoy, Aycan. 2016. “Household Formation Behavior: An Analysis of Relative Education and Exogamy for Descendants of Immigrants in Sweden.” Ethnicities 16(4):547–67. Chae, Sophia. 2016. “Forgotten Marriages? Measuring the Reliability of Marriage Histories.” Demographic Research 34(1):525–62. Chinitz, Joshua G. and Robert A. Brown. 2001. “Religious Homogamy, Marital Conflict, and Stability in Same-faith and Interfaith Jewish Marriages.” Journal for the Scientific Study of Religion 40(4):723–33. Chiswick, Carmel U. and Evelyn L. Lehrer. 1991. “Religious Intermarriage: An Economic Perspective.” Contemporary Jewry 12(1):21–34. Choi, Kate and Nancy Reichman. 2019. “The Health of Biracial Children in Two-Parent Families in the United States.” Demographic Research 41(8):197–230. Clark, Shelley and Sarah Brauner-Otto. 2015. “Divorce in Sub-Saharan Africa: Are Unions Becoming Less Stable?” Population and Development Review 41(4):583–605. Clark, Shelley, Alissa Koski, and Emily Smith-Greenaway. 2017. “Recent Trends in Premarital Fertility across Sub-Saharan Africa.” Studies in Family Planning 48(1):3–22. 188 Crespin-Boucaud, Juliette. 2018. Interethnic and Interfaith Marriages in Sub-Saharan Africa. PSE Working Paper: 2018–37. Croft, Trevor N., Aileen M. J. Marshall, and Courtney K. Allen. 2018. Guide to DHS Statistics. Davidson, Basil. 1988. West Africa before the Colonial Era. New York: Addison Wesley Longman. Debpuur, Cornelius and Bawah Agula Ayaga. 2002. “Are Reproductive Preferences Stable? Evidence from Rural Northern Ghana.” Genus 58(2):63–89. Decker, Michele R., Amanda D. Latimore, Suzumi Yasutake, Miriam Haviland, Saifuddin Ahmed, Robert W. Blum, Freya Sonenstein, and Nan Marie Astone. 2015. “Gender-Based Violence against Adolescent and Young Adult Women in Low- and Middle-Income Countries.” Journal of Adolescent Health 56(2):188–96. Decoret-Ahiha, Anne. 2005. “Benin.” Encylopedia of the World’s Minorities. Deose, Laurie F., F. Nii-Amoo Dodoo, and Vrushali Patil. 2002. “Fertility Desires and Perceptions of Power in Reproductive Conflict in Ghana.” Gender and Society 16(1):53–73. DeRose, Laurie F. and Alex C. Ezeh. 2005. “Men’s Influence on the Onset and Progress of Fertility Decline in Ghana, 1988-98.” Population Studies 59(2):197–210. Derose, Laurie F., Lijuan Wu, and F. Nii-Amoo Dodoo. 2010. “Inferring Gender-Power: Women’s Schooling and Relative Spousal Influence in Childbearing in Ghana.” Genus 66(2):69–91. Dickovick, J. Tyler. 2008. “Legacies of Leftism: Ideology, Ethnicity and Democracy in Benin, Ghana and Mali.” Third World Quarterly 29(6):1119–37. Dodoo, F. Nii-Amoo. 1993. “Insights into Spousal Differences in Reproductive Dis/Agreement.” Sociological Focus 26(3):257–70. Drefahl, Sven. 2010. “How Does the Age Gap between Partners Affect Their Survival?” Demography 47(2):313–26. Duflo, Esther and Christopher Udry. 2004. Intrahousehold Resource Allocation in Cote d’Ivoire: Social Norms, Separate Accounts and Consumption Choices. National Bureau of Economic Research. 10498. Cambridge, MA. Eberhard, David M., Gary F. Simons, and Charles D. Fennig (eds.). 2019. Ethnologue: Languages of the World. Twenty-second edition. Dallas, Texas: SIL International. Eloundou-Enyegue, Parfait M. and Sarah C. Giroux. 2012. “Fertility Transitions and Schooling: From Micro- to Macro-Level Associations.” Demography 49(4):1407–32. 189 Esteve, Albert, Christine R. Schwartz, Jan van Bavel, Iñaki Permanyer, Martin Klesment, and Joan García-Román. 2016. “The End of Hypergamy: Global Trends and Implications.” Population and Development Review 42(4):615–25. Ezeh, Alex C., Blessing U. Mberu, and Jacques O. Emina. 2009. “Stall in Fertility Decline in Eastern African Countries: Regional Analysis of Patterns, Determinants and Implications.” Philosophical Transactions of the Royal Society B: Biological Sciences 364(1532):2991– 3007. Fage, John D. and T. C. McCaskie. 2016. “Western Africa.” Encyclopaedia Britannica. Falen, Douglas J. 2008a. “Polygyny and Christian Marriage in Africa: The Case of Benin.” African Studies Review 51(2):51–74. Falen, Douglas J. 2008b. “The ‘Other’ Gender? Reflections on Fieldwork in Benin.” Men and Masculinities 11(2):164–73. Falen, Douglas J. 2011. Power & Paradox: Authority, Insecurity, and Creativity in Fon Gender Relations. Trenton, NJ: Africa World Press Inc. Falen, Douglas J. 2016. “Vodún, Spiritual Insecurity, and Religious Importation in Benin.” Journal of Religion in Africa 46:453–83. Falola, Toyin and Daniel Jean-Jacques. 2016. Africa: An Encyclopedia of Culture and Society [3 Volumes] : An Encyclopedia of Culture and Society. Santa Barbara, California: ABC-CLIO. Fearon, James D. 2003. “Ethnic and Cultural Diversity by Country.” Journal of Economic Growth 8(2):195–222. Ferree, Karen E. 2010. “The Social Origins of Electoral Volatility in Africa.” British Journal of Political Science 40(4):759–79. Feyisetan, Bamikale J. 1990. “Postpartum Sexual Abstinence, Breastfeeding, and Childspacing, among Yoruba Women in Urban Nigeria.” Social Biology 37(1–2):110–27. Feyisetan, Bamikale J. 2000. “Spousal Communication and Contraceptive Use among the Yoruba of Nigeria.” Population Research and Policy Review 19(1):29–45. Frye, Margaret and Jenny Trinitapoli. 2015. “Ideals as Anchors for Relationship Experiences.” American Sociological Review 80(3):496–525. Frye, Margaret and Daniela R. Urbina. 2019. “Fearing Such a Lady: University Expansion, Underemploymentm and the Hypergamy Ideal in Kampala, Uganda.” Journal of Family Issues 00(0):1–27. 190 Furtado, Delia. 2012. “Human Capital and Interethnic Marriage Decisions.” Economic Inquiry 50(1):82–93. Gage, Anastasia J. 1995. “Women’s Socioeconomic Position and Contraceptive Behavior in Togo.” Studies in Family Planning 26(5):264–77. Garenne, Michel and Veronique Joseph. 2002. “The Timing of the Fertility Transition in Sub- Saharan Africa.” World Development 30(10):1835–43. Garenne, Michel and Julien Zwang. 2006. “Premarital Fertility in Namibia: Trends, Factors and Consequences.” Journal of Biosocial Science 38(2):145–67. Gnoumou Thiombiano, Bilampoa. 2017. “Union Breakdown in West African Cities: The Cases of Ouagadougou and Lomé.” Demographic Research 37(1):101–28. Gnoumou Thiombiano, Bilampoa and K. Legrand. 2014. “Niveau et Facteurs de Ruptures Des Premières Unions Conjugales Au Burkina Faso.” African Population Studies 28(3):1432– 46. Goldscheider, Calvin and Peter R. Uhlenberg. 1969. “Minority Group Status and Fertility.” American Journal of Sociology 74(4):361–72. Gonalons-Pons, Pilar and Christine R. Schwartz. 2017. “Trends in Economic Homogamy: Changes in Assortative Mating or the Division of Labor in Marriage?” Demography 54(3):985–1005. Goody, Jack. 1971. “Class and Marriage in Africa and Eurasia.” American Journal of Sociology 76(4):585–603. Gordon, A. J. 2000. “Cultural Identity and Illness: Fulani Views.” Culture, Medicine and Psychiatry 24(3):297–330. Gorenburg, Dmitry. 2006. “Rethinking Interethnic Marriage in the Soviet Union.” Post-Soviet Affairs 22(2):145–65. Grace, Kathryn and Stuart Sweeney. 2016. “Ethnic Dimensions of Guatemala’s Stalled Transition: A Parity-Specific Analysis of Ladino and Indigenous Fertility Regimes.” Demography 53(1):117–37. Gubhaju, Bhakta, Eduard Jongstra, and Merewalesi Raikoti. 2014. “Below-Replacement Fertility of Ethnic Indians in Fiji: A Decomposition Analysis of the Components of Changes in the Total Fertility Rate.” Journal of Population Research 31(4):269–86. Guttmacher Institute. 2015. Contraceptive Use in Benin. 191 Gyimah, Stephen Obeng, Baffour Takyi, and Eric Yeboah Tenkorang. 2008. “Denominational Affiliation and Fertility Behaviour in an African Context: An Examination of Couple Data from Ghana.” Journal of Biosocial Science 40(3):445–58. Hahonou, Éric Komlavi. 2009. Slavery and Politics: Stigma, Decentralisation, and Political Representation in Niger and Benin. edited by B. Rossi. Liverpool. Hahonou, Éric Komlavi. 2015. “The Quest for Honor and Citizenship in Post-Slavery Borgu (Benin).” International Journal of African Historical Studies 48(2):3–4. Hair, Elizabeth C., Kristin Anderson Moore, Alena M. Hadley, Kelleen Kaye, Randal D. Day, and Dennis K. Orthner. 2009. “Parent Marital Quality and the Parent-Adolescent Relationship: Effects on Adolescent and Young Adult Health Outcomes.” Marriage and Family Review 45(2–3):218–48. Hallgren, Roland. 1993. “Some Aspects of the Relationship between Religion and Health among the Yoruba People in Souther-Western Nigeria.” Temenos: Nordic Journal of Comparative Religion 23:47. Hampshire, K. R. 2004. “Fulani.” Encylopedia of Medical Anthropology: Health and Illness in the World’s Cultures 351(21):2245–46. Hannemann, Tina and Hill Kulu. 2015. “Union Formation and Dissolution among Immigrants and Their Descendants in the United Kingdom.” Demographic Research 33(1):273–312. Hannemann, Tina, Hill Kulu, Leen Rahnu, Allan Puur, Mihaela Hărăguş, Ognjen Obućina, Amparo González-Ferrer, Karel Neels, Layla Van den Berg, Ariane Pailhé, Gina Potarca, and Laura Bernardi. 2018. “Co-Ethnic Marriage versus Intermarriage among Immigrants and Their Descendants: A Comparison across Seven European Countries Using Event- History Analysis.” Demographic Research 39(17):487–524. Hayford, Sarah R. and Victor Agadjanian. 2012. “From Desires to Behavior: Moderating Factors in a Fertility Transition.” Demographic Research 26(20):511–42. Heath, Elizabeth. 2008. “Fon.” Africana: The Encyclopedia of the African and African American Experience. Heaton, Tim B. 1986. “How Does Religion Influence Fertility?: The Case of Mormons.” Journal for the Scientific Study of Religion 25(2):248–58. Heaton, Tim B. 2011. “Does Religion Influence Fertility in Developing Countries.” Population Research and Policy Review 30(3):449–65. Hertrich, Véronique. 2017. “Trends in Age at Marriage and the Onset of Fertility Transition in Sub-Saharan Africa.” Population and Development Review 43:112–37. 192 Hewitt, Belinda, Janeen Baxter, and Mark Western. 2005. “Marriage Breakdown in Australia: The Social Correlates of Separation and Divorce.” Journal of Sociology 41(2):163–83. Hohmann-Marriott, Bryndl E. and Paul R. Amato. 2008. “Relationship Quality in Interethnic Marriages and Cohabitations.” Social Forces 87(2):825–55. Hollos, Marida and Ulla Larsen. 2008. “Motherhood in Sub-Saharan Africa: The Social Consequences of Infertility in an Urban Population in Northern Tanzania.” Culture, Health & Sexuality 10(2):159–73. Hosseini, Hatam and Amir Erfani. 2014. “Ethnic Differences in the Attitudes and Practice of Consanguineous Marriage among Kurds and Turks in Uromiyeh District, Iran.” Journal of Comparative Family Studies 45(3):389–403. Houngnikpo, Mathurin C. and Samuel Decalo. 2013. Historical Dictionary of Benin. 4th ed. Plymouth, United Kingdom: Scarecrow Press, Inc. Hu, Anning and Zhenchao Qian. 2016. “Does Higher Education Expansion Promote Educational Homogamy? Evidence from Married Couples of the Post-80s Generation in Shanghai, China.” Social Science Research 60:148–62. Ibisomi, Latifat and Clifford Odimegwu. 2011. “Understanding Resolution of Differential Fertility Preferences among Couples in Nigeria.” International Journal of Business and Social Science 2(4):98–105. ICF International. 2012. “Demographic and Health Surveys Sampling and Household Listing Manual.” MEASURE DHS, Calverton, Maryland, USA (September):1–98. INSAE and ICF International. 2019. Enquête d Émographique et de Santé Au Bénin 2017-2018. INSAE and ICF International. 2013. Enquête Démographique et de Santé Du Bénin 2011-2012: Rapport de Synthèse. Calverton, Maryland. Isiugo-Abanihe, Uche C. 2014. “Reproductive Motivation and Family-Size Preferences among Nigerian Men.” Studies in Family Planning 25(3):149–61. Johnson-Hanks, Jennifer. 2003. “Education, Ethnicity, and Reproductive Practice in Cameroon.” Population-E 58(2):153–79. Johnson-Hanks, Jennifer. 2004. “Uncertainty and the Second Space: Modern Birth Timing and the Dilemma of Education.” European Journal of Population 20(4):351–73. Johnson-Hanks, Jennifer. 2006. “On the Politics and Practice of Muslim Fertility : Comparative Evidence from West Africa.” Medical Anthropolgy Quarterly 20(1):12–30. 193 Kalmijn, Matthijs. 1998. “Intermarriage and Homogamy: Causes, Patterns, Trends.” Annual Review of Sociology 24(1):395–421. Kalmijn, Matthijts and Frank Van Tubergen. 2010. “A Comparitive Perspective on Intermarriage: Explaining Differences among National-Origin Groups in the United States.” Demography 47(2):459–79. Kaplan, Amit and Anat Herbst. 2015. “Stratified Patterns of Divorce: Earnings, Education, and Gender.” Demographic Research 32(1):949–82. Kashyap, Ridhi, Albert Esteve, and Joan García-Román. 2015. “Potential (Mis)Match? Marriage Markets amidst Sociodemographic Change in India, 2005–2050.” Demography 52(1):183– 208. Kehinde, Michael Olujimi. 2010. “Implications of Colonially Determined Boundaries in (West) Africa : The Yoruba of Nigeria and Benin in Perspective.” University of Durham. Killewald, Alexandra. 2016. “Money, Work, and Marital Stability: Assessing Change in the Gendered Determinants of Divorce.” American Sociological Review 81(4):696–719. Kodzi, Ivy A., John B. Casterline, and Peter Aglobitse. 2010. “The Time Dynamics of Individual Fertility Preferences among Rural Ghanaian Women.” Population Council 41(1):45–54. Kodzi, Ivy A., David R. Johnson, and John B. Casterline. 2010. “Examining the Predictive Value of Fertility Preferences among Ghanaian Women.” Demographic Research 22:965. Kohler, Hans-Peter, Jere R. Behrman, and Susan C. Watkins. 2001. “The Density of Social Networks and Fertility Decisions: Evidence from South Nyanza District, Kenya.” Demography 38(1):43–58. Koski, Alissa, Shelley Clark, and Arijit Nandi. 2017. “Has Child Marriage Declined in Sub- Saharan Africa? An Analysis of Trends in 31 Countries.” Population and Development Review 43(1):7–29. Kouta, Momar. 2016. “Benin Data Portal- The African Development Bank.” Kritz, Mary M. and Paulina Makinwa-Adebusoye. 1999. “Determinants of Women’s Decision- Making Authority in Nigeria: The Ethnic Dimension.” Sociological Forum 14(3):399–424. Kruse, Hanno and Clemens Kroneberg. 2019. “More Than a Sorting Machine: Ethnic Boundary Making in a Stratified School System.” American Journal of Sociology 125(2):431–84. Kwabena, Edward, Ameyaw Id, Eugene Budu, Francis Sambah, Linus Baatiema, Francis Appiah, Abdul-aziz Seidu Id, and Bright Opoku Ahinkorah. 2019. “Prevalence and Determinants of Unintended Pregnancy in Sub-Saharan Africa : A Multi- Country Analysis of Demographic and Health Surveys.” PLoS ONE 14(8):1–16. 194 Lacombe, Bernard. 1987. “Les Unions Informelles En Afrique Au Sud Du Sahara: L’exemple Du Deuxieme Bureau Congolais.” Genus 43(1):151–64. Lalmann, D. 2004. “Aja-Speaking Peoples: Aja, Fon, Ewe, Seventeenth and Eighteenth Centuries.” Encyclopedia of African History. Langer, Arnim and Ukoha Ukiwo. 2009. “Subjective Realities: Perceptions of Identity and Conflict in Ghana and Nigeria.” Journal of International Development 21:483–94. Lasee, Ashraf and Stan Becker. 1997. “Husband-Wife Communication about Family Planning and Contraceptive Use in Kenya.” International Family Planning Perspectives 23(1):15. Law, Robin and Paul E. Lovejoy. 1999. “Borgou in the Atlantic Slave Trade.” African Economic History 27(27):69–92. Lehrer, Evelyn L. 1996. “Religion as a Determinant of Marital Fertility.” Journal of Population Economics 9(2):173–96. Lesthaeghe, Ron. 2014. “The Second Demographic Transition: A Concise Overview of Its Development.” Proceedings of the National Academy of Sciences of the United States of America 111(51):18112–15. Leszczensky, Lars and Sebastian Pink. 2019. “What Drives Ethnic Homophily? A Relational Approach on How Ethnic Identification Moderates Preferences for Same-Ethnic Friends.” American Sociological Review 84(3):394–419. Liefbroer, Aart C. and Edith Dourleijn. 2006. “Unmarried Cohabitation and Union Stability: Testing the Role of Diffusion Using Data from 16 European Countries.” Demography 43(2):203–21. Lloyd, Cynthia B., Carol E. Kaufman, and Paul Hewett. 2000. “The Spread of Primary Schooling in Sub-Saharan Africa: Implications for Fertility Change.” Population and Development Review 26(3):483–515. Locoh, Thérèse and Marie-Paule Thiriat. 1995. “Divorce et Remariage Des Femmes En Afrique de l’Ouest. Le Cas Du Togo.” Population (French Edition) 50(1):61–93. Maas, Ineke and Richard Zijdeman. 2010. “Beyond the Local Marriage Market: The Influence of Modernization on Geographical Heterogamy.” Demographic Research 23:933–62. Maffioli, D., P. Sacco, and G. Gabrielli. 2007. “The Impact of Urbanization on the Traditional Family System in Sub-Saharan Africa.” Pp. 10–14 in Fifth African Population Conference 2007. Maïga, Abdoulaye and Banza Baya. 2011. “Au-delà des Normes de Formation du Couple : Quand Les Cultures S’épousent.” Pp. 1–23 in 6eme Conference Africaine sur la Population. 195 Maïga, Abdoulaye and Banza Baya. 2014. “La Coexistence Des Generations en Milieu Urbain au Burkina Faso: Entre Logique de Selection et Obligation de Parente.” Etude de La Population Africaine 28(2):1088–1102. Majekodunmi, Ayodele O., Akinyemi Fajinmi, Charles Dongkum, Alexandra P. M. Shaw, and Susan C. Welburn. 2014. “Pastoral Livelihoods of the Fulani on the Jos Plateau of Nigeria.” Pastoralism 4(1):20–36. Martin, Dribe, J. David Hacker, and Scalone Francesco. 2018. “Becoming American: Intermarriage during the Great Migration to the United States.” Journal of Interdisciplinary History 49(2):189–218. Mberu, Blessing U. and Holly E. Reed. 2014. “Understanding Subgroup Fertility Differentials in Nigeria.” Population Review 53(2):23–46. McLanahan, S. and D. Schwartz. 2002. “Life without Father: What Happens to the Children?” Contexts 1(1):35–44. McNamee, Catherine B. 2009. “Wanted and Unwanted Fertility in Bolivia: Does Ethnicity Matter?” International Perspectives on Sexual and Reproductive Health 35(4):166–75. McPherson, Miller, Lynn Smith-Lovin, and James M. Cook. 2001. “Birds of a Feather: Homophily in Social Networks.” Annual Review of Sociology 27(1):415–44. McQuilian, Kevin. 2004. “When Does Religion Influence Fertility?” Population (English Edition) 30(1):25–56. Meekers, Dominique. 1992. “The Process of Marriage in African Societies : A Multiple Indicator Approach.” Population and Development Review 18(1):61–78. Meekers, Dominique. 1995. “Freedom of Partner Choice in Togo.” Journal of Comparative Family Studies 26(2):163–78. Meekers, Dominique and Anne-Emmanuèle Calvès. 1997. “‘Main’ Girlfriends, Girlfriends, Marriage, and Money : The Social Context of HIV Risk Behaviour in Sub-Saharan Africa.” Health Transition Review 7(1997):361–75. Meekers, Dominique and Nadra Franklin. 1995. “Women’s Perceptions of Polygyny among the Kaguru of Tanzania.” Ethnology 34(4):315–29. Mensch, Barbara, Monica Grant, and Ann Blanc. 2006. “The Changing Context of Sexual Initiation in Sub-Saharan Africa.” Population and Development Review 32(4):699–727. Ministère de la Santé du Bénin/INSAE. 1996. “Benin: Enquête Démographique et de Santé.” 196 Mitsunaga, Tisha M., Antonia M. Powell, Nathan J. Heard, and Ulla M. Larsen. 2005. “Extramarital Sex among Nigerian Men: Polygyny and Other Risk Factors.” Journal of Acquired Immune Deficiency Syndromes 39(4):478–88. Moultrie, Tom A., Victoria Hosegood, Nuala Mcgrath, Caterina Hill, Kobus Herbst, and Marie Louise Newell. 2008. “Refining the Criteria for Stalled Fertility Declines: An Application to Rural KwaZulu-Natal, South Africa, 1990-2005.” Studies in Family Planning 39(1):39–48. Murphy, Caryle. 2015. Interfaith Marriage Is Common in U.S., Particularly among the Recently Wed. Murphy, Caryle. 2016. Religious Groups in Israel Keep to Themselves When It Comes to Marriages and Friendships. Nahmias, Petra. 2007. “The Importance of Ethnicity: Fertility and Ethnicity in West Africa.” Princeton Univertity: 1–43. Nordman, Christophe J., Anne-Sophie Robilliard, and François Roubaud. 2011. “Gender and Ethnic Earnings Gaps in Seven West African Cities.” Labour Economics 18:S132–45. Ntoimo, Lorretta Favour Chizomam and Nyasha Mutanda. 2017. “Homogamy and Heterogamy in Sub-Saharan Africa: Implications for Couples’ Fertility Behaviour.” Development Southern Africa 34(6):670–81. Nunn, Nathan. 2010. “Religious Conversion in Colonial Africa.” American Economic Review 100(2):147–52. Nuworsu, Anastasia. 2015. “Language Use in Interethnic Marriage Ceremonies in Greater Accra.” University of Ghana. Odimegwu, Clifford and Sunday A. Adedini. 2017. “The Role of Community Structure in Shaping African Fertility Pattern: Evidence from Demographic and Health Surveys.” Journal of Biosocial Science 49(S1):S46–61. Odimegwu, Clifford Obby and Assata Zerai. 1996. “Understanding the Proximate Determinants of Fertility of a Nigerian Ethnic Group.” Genus 52(3/4):67–87. Ojo, Olatunji. 2008. “Beyond Diversity: Women, Scarification, and Yoruba Identity.” History in Africa 35:347–74. Okeke, Chukwuma O., Christopher N. Ibenwa, and Gloria Tochukwu Okeke. 2017. “Conflicts between African Traditional Religion and Christianity in Eastern Nigeria: The Igbo Example.” SAGE Open 7(2). 197 Okello, Anna L., Ayodele O. Majekodunmi, Adamu Malala, Susan C. Welburn, and James Smith. 2014. “Identifying Motivators for State-Pastoralist Dialogue: Exploring the Relationships between Livestock Services, Self-Organisation and Conflict in Nigeria’s Pastoralist Fulani.” Pastoralism: Research, Policy and Practice 4(1):12. Okun, Barbara S. 2004. “Insight into Ethnic Flux : Marriage Patterns among Jews of Mixed Ancestry in Israel.” Demography 41(1):173–87. Oluwaseyi, Somefun, Clifford Odimegwu, and Sunday A. Adedini. 2014. “Gender Differentials in Extramarital Sexual Behaviour: Evidence from Countries in Sub-Saharan Africa.” Gender & Behaviour 12(4):6093–6113. Oni, Jacob. 2006. “Understanding Demogrpahic and Health Events through Yoruba Proverbs.” Pp. 195–226 in The Yoruba in Transition: History, Values, and Modernity, edited by T. Falola and A. Genova. Durham, North Carolina: Carolina Academic Press. Ono, Hiromi. 1998. “Husbands’ and Wives’ Resources and Marital Dissolution.” Journal of Marriage and the Family 60(3):674–89. Ononokpono, Dorothy Ngozi, Clifford Obby Odimegwu, Sunday Adepoju Adedini, and Eunice N. S. Imasiku. 2016. “Ethnic Diversity and Maternal Health Care in Nigeria.” Women’s Reproductive Health 3(1):45–59. Oppenheimer, Valerie Kincade. 1997. “Women’s Employment and the Gain to Marriage: The Specialization and Trading Model.” Annual Review of Sociology 23:431–53. Orubuloye, I. O., John C. Caldwell, and Pat Caldwell. 1997. “Perceived Male Sexual Needs and Male Sexual Behaviour in SouthWest Nigeria.” Social Science & Medicine 44(8):1195– 1207. Otite, Onigu. 1991. “Marriage and Family Systems in Nigeria.” International Journal of Sociology of the Family 21(2):15–54. Oyediran, Kola, Uche C. Isiugo-Abanihe, Bamikale J. Feyisetan, and Gbenga P. Ishola. 2010. “Prevalence of and Factors Associated with Extramarital Sex among Nigerian Men.” American Journal of Men’s Health 4(2):124–34. Pearce-Morris, Jennifer and Valarie King. 2012. “The Well-Being of Children Living with Interethnic Parents: Are They at a Disadvantage?” Journal of Family Issues 33(7):989–919. Posner, Daniel N. 2004. “Measuring Ethnic Fractionalization in Africa.” American Journal of Political Science 48(4):849–63. Prag, Ebbe. 2010. Entrepôt Politics: Political Struggles over the Dantokpa Marketplace in Cotonou, Benin. DIIS Working Paper (2010):03. 198 Previti, Denise and Paul R. Amato. 2003. “Why Stay Married? Rewards, Barriers, and Marital Stability.” Journal of Marriage and Family 65(3):561–73. Previti, Denise and Paul R. Amato. 2004. “Is Infidelity a Cause or a Consequence of Poor Marital Quality?” Journal of Social and Personal Relationships 21(2):217–30. Primativo, Giuseppina, Claudio Ottoni, Gianfranco Biondi, Sara Serafino, Cristina Mart, Maarten H. D. Larmuseau, Michele Scardi, Ronny Decorte, and Olga Rickards. 2017. “Bight of Benin: A Maternal Perspective of Four Beninese Populations and Their Genetic Implications on the American Populations of African Ancestry.” Annals of Human Genetics 81:78–90. Puur, Allan, Leen Rahnu, Luule Sakkeus, Martin Klesment, and Liili Abuladze. 2018. “The Formation of Ethnically Mixed Partnerships in Estonia: A Stalling Trend from a Two-Sides Perspective.” Demographic Research 38:1111–54. Qian, Zhenchao. 2005. “Breaking the Last Taboo: Interracial Marriage in America.” Contexts 4(4):33–37. Qian, Zhenchao, Jennifer E. Glick, and Christie D. Baston. 2015. “Crossing Boundaries: Nativity, Ethnicity, and Mate Selection.” International Electronic Journal of Elementary Education 7(3):301–18. Qian, Zhenchao and Daniel T. Lichter. 2007. “Social Boundaries and Marital Assimilation: Interpreting Trends in Racial and Ethnic Intermarriage.” American Sociological Review 72(1):68–94. Qian, Zhenchao and Daniel T. Lichter. 2011. “Changing Patterns of Interracial Marriage in a Multiracial Society.” Journal of Marriage and Family 73(5):1065–84. Qian, Zhenchao and Daniel T. Lichter. 2018. “Marriage Markets and Intermarriage: Exchange in First Marriages and Remarriages.” Demography 55(3):849–75. Reniers, Georges. 2003. “Divorce and Remarriage in Rural Malawi.” Demographic Research 9(SUPPL. 1):175–206. Robles, Theodore F., Richard B. Slatcher, Joseph M. Trombello, and Meghan M. McGinn. 2015. “Marital Quality and Health: A Meta-Analytic Review.” Psychological Bulletin 140(1):140–87. Rosenfeld, M. J. and R. J. Thomas. 2012. “Searching for a Mate: The Rise of the Internet as a Social Intermediary.” American Sociological Review 77(4):523–47. Rosenfeld, Michael J. and Byung-Soo Kim. 2005. “The Independence of Young Adults and the Rise of Interrracial and Same-Sex Unions.” American Sociological Review 70(39):541–62. 199 Rutstein, Shea Oscar and Kiersten Johnson. 2004. The DHS Wealth Index. Calverton, Maryland: ORC Macro. Safi, Mirna and Godfrey Rogers. 2008. “Intermarriage and Assimilation: Disparities in Levels of Exogamy among Immigrants in France.” Population (English Edition) 63(2):239–67. Salamone, Frank A. 1995. “Fulani.” Encylopedia of World Cultures 100–103. Sargent, Carolyn F. 1988. “Born to Die : Witchcraft and Infanticide in Bariba Culture.” Ethnology 27(1):79–95. Schraml, Carla. 2014. “How Is Ethnicity Experienced? Essentialist and Constructivist Notions of Ethnicity in Rwanda and Burundi.” Ethnicities 14(5):615–33. Schramm, David G., James P. Marshall, Victor W. Harris, and Thomas R. Lee. 2012. “Religiosity, Homogamy, and Marital Adjustment.” Journal of Family Issues 33(2):246–68. Schwartz, Christine R. and Pilar Gonalons-Pons. 2017. “Trends in Relative Earnings and Marital Dissolution: Are Wives Who Outearn Their Husbands Still More Likely to Divorce?” RSF: The Russell Sage Foundation Journal of the Social Sciences 2(4):218. Schwartz, Christine R. and Robert D. Mare. 2005. “Trends in Educational Assortative Marriage from 1940 to 2003.” Demography 42(4):621–46. Schwartz, Christine R., Zhen Zeng, and Yu Xie. 2016. “Marrying up by Marrying down: Status Exchange between Social Origin and Education in the United States.” Sociological Science 3:1003–27. Sennott, Christie and Sara Yeatman. 2018. “Conceptualizing Childbearing Ambivalence: A Social and Dynamic Perspective.” Journal of Marriage and Family 80(4):888–901. Shapiro, David and Basile O. Tambashe. 2017. “Fertility, Ethnicity, and Education in the Democratic Republic of Congo.” African Population Studies 31(1):3253–70. Slatcher, Richard B. 2010. “Marital Functioning and Physical Health: Implications for Social and Personality Psychology.” Social and Personality Psychology Compass 4(7):455–69. Sly, David F. 1970. “Minority-Group Status and Fertility: An Extension of Goldscheider and Uhlenberg.” American Journal of Sociology 76(3):443–59. Smith-Greenaway, Emily and Shelley Clark. 2018. “Women’s Marriage Behavior Following a Premarital Birth in Sub-Saharan Africa.” Journal of Marriage and Family 80(1):256–70. Smith, Daniel Jordan. 2001. “Romance, Parenthood, and Gender in a Modern African Society.” Ethnology 40(2):129–51. 200 Smith, Daniel Jordan. 2002. “‘Man No Be Wood’: Gender and Extramarital Sex in Contemporary Southeastern Nigeria.” The Ahfad Journal 19(2):4–23. Smith, Daniel Jordan. 2007. “Modern Marriage, Men’s Extramarital Sex, and HIV Risk in Southeastern Nigeria.” American Journal of Public Health 97(6):997–1005. Smith, Daniel Jordan. 2010. “Promiscuous Girls, Good Wives, and Cheating Husbands: Gender Inequality, Transitions to Marriage, and Infidelity in Southeastern Nigeria.” Anthropological Quarterly 83(1):123–52. Smith, Daniel Jordan. 2011. “Rural-to-Urban Migration, Kinship Networks, and Fertility among the Igbo in Nigeria.” African Population Studies 25(2):320–36. Smith, Sanne, Ineke Maas, and Frank Van Tubergen. 2012. “Irreconcilable Differences ? Ethnic Intermarriage and Divorce in the Netherlands, 1995-2008.” Social Science Research 41(5):1126–37. Smits, Jeroen, Wout Ultee, and Jan Lammers. 1998. “Educational Homogamy in 65 Countries: An Explanation of Differences in Openness Using Country-Level Explanatory Variables.” American Sociological Review 63(2):264–85. Speizer, Ilene. 1999. “Men, Marriage, and Ideal Family Size in Francophone Africa.” Journal of Comparative Family Studies 30(1):17–34. Sprehe, J. Timothy. 1974. “The World Fertility Survey: An International Program of Fertility Research.” Studies in Famly Planning 5(2):35–41. Stier, H. and Y. Shavit. 1994. “Age at Marriage, Sex-Ratios, and Ethnic Heterogamy.” European Sociological Review 10(1):79–87. Tabutin, Dominique and Bruno Schoumaker. 2004. “The Demography of Sub-Saharan Africa from the 1950s to the 2000s: A Survey of Changes and a Statisical Assessment.” Population (English Edition) 59(3):455. Takyi, Baffour K. 2001. “Marital Instability in an African Society: Exploring the Factors That Influence Divorce Processes in Ghana.” Sociological Focus 34(1):77–96. Takyi, Baffour K. and Isaac Addai. 2002. “Effects of Ethnicity on the Onset of Parenthood in a Developing Country: Some Evidence from the Early 1990s.” Journal of Comparative Family Studies 33(1):57–71. Takyi, Baffour K. and F. Nii-Amoo Dodoo. 2005. “Gender, Lineage, and Fertility-Related Outcomes in Ghana.” Journal of Marriage and Family 67(February):251–57. Takyi, Baffour K. and Stephen Obeng Gyimah. 2007. “Matrilineal Family Ties and Marital Dissolution in Ghana.” Journal of Family Issues 28(5):682–705. 201 Teachman, Jay D. 2002. “Stability across Cohorts in Divorce Risk Factors.” Demography 39(2):331–51. The World Bank. 2019. “Benin.” Country Profile. Timæus, Ian M. and Angela Reynar. 1998. “Polygynists and Their Wives in Sub-Saharan Africa: An Analysis of Five Demographic and Health Surveys.” Population Studies 52(2):145–62. Togunde, Oladimeji R. 1998. “A Social Structural Analysis of the Effects of Women’s Employment on Fertility in Urban Nigeria.” International Journal of Sociology of the Family 28(1):31–46. Tompkins, Barbara and Angela Kluge. 2002. Sociolinguistic Survey of the Aja Language Area. Vols. 2002–03. Torabi, Fatemeh and Angela Baschieri. 2010. “Ethnic Differences in Transition to First Marriage in Iran: The Role of Marriage Market, Women’s Socio-Economic Status, and Process of Development.” Demographic Research 22:29–62. Trinitapoli, Jenny and Sara Yeatman. 2018. “The Flexibility of Fertility Preferences in a Context of Uncertainty.” Population and Development Review 44(1):87–116. Tumlinson, Katherine, Ilene S. Speizer, Joshua T. Davis, Jean Christophe Fotso, Paul Kuria, and Linda H. Archer. 2013. “Partner Communication, Discordant Fertility Goals, and Contraceptive Use in Urban Kenya.” African Journal of Reproductive Health 17(3):79–90. Udogu, E. Ike. 1999. “The Issue of Ethnicity and Democratization in Africa: Toward the Millennium.” Journal of Black Studies 29(6):790–808. Umberson, Debra, Kristi Williams, Daniel A. Powers, Hui Liu, and Belinda Needham. 2006. “You Make Me Sick: Marital Quality and Health over the Life Course.” Journal of Health and Social Behavior 47(1):1–16. United Nations Children’s Fund. 2017. The State of the World’s Children 2017. United Nations Development Programme. 2018. Human Development Reports: Gender Inequality Index. Viera, Anthony J. and Joanne M. Garrett. 2005. “Understanding Interobserver Agreement: The Kappa Statistic.” Family Medicine 37(5):360–63. Voas, David. 2003. “Conflicting Preferences: A Reason Fertility Tends to Be Too High or Too Low.” Population and Development Review 29(4):627–46. Voas, David. 2009. “The Maintenance and Transformation of Ethnicity: Evidence on Mixed Partnerships in Britain.” Journal of Ethnic and Migration Studies 35(9):1497–1513. 202 Vu, Lung, Waimar Tun, Andrew Karlyn, Sylvia Adebajo, and Babatunde Ahonsi. 2011. “Attitudinal and Behavioral Factors Associated with Extramarital Sex among Nigerian Men: Findings from a National Survey.” International Journal of Sexual Health 23(4):258– 68. Vyas, Seema and Charlotte Watts. 2009. “How Does Economic Empowerment Affect Women’s Risk of Intimate Partner Violence in Low and Middle Income Countries? A Systematic Review of Published Evidence.” Journal of International Development 21:577–602. Wallerstein, I. 1960. “Ethnicity and National Integration in West Africa.” Cahiers d’Etudes Africaines 1(3):129–39. Ware, Helen. 1979. “Polygyny: Women’s Views in a Transitional Society, Nigeria 1975.” Journal of Marriage and Family 41(1):185–95. Watkins, Susan Cotts. 1987. “The Fertility Transition: Europe and the Third World Compared.” Sociological Forum 2(4):645–73. Watkins, Susan Cotts. 1991. From Provinces into Nations: Demographic Integration in Western Europe, 1870-1960. Webster, Pamela S., Terri L. Orbuch, and James S. House. 1995. “Effects of Childhood Family Background on Adult Marital Quality and Perceived Stability.” American Journal of Sociology 101(2):404–32. Weinreb, Alexander A. 2001. “First Politics, Then Culture: Accounting for Ethnic Differences in Demographic Behavior in Kenya.” Population Council 27(3):437–67. Yeatman, Sara and Christie Sennott. 2014. “The Relationship between Partners’ Family-Size Preferences in Southern Malawai.” Studies in Famly Planning 45(3):361–77. Yeatman, Sara, Christie Sennott, and Steven Culpepper. 2013. “Young Women’s Dynamic Family Size Preferences in the Context of Transitioning Fertility.” Demography 50(5):1715–37. Zhang, Yuanting and Jennifer Van Hook. 2009. “Marital Dissolution among Interracial Couples.” Journal of Marriage and Family 71(1):95-107. 203 APPENDIX CHAPTER 2 Appendix 2.A: Background information on ethnic groups Adja: The Adja people reside predominantly in southwestern Benin and southeastern Togo. They speak a Gbe language that is distinct from neighboring Gbe languages, such as Fon (Tompkins and Kluge 2002). They are considered one of the first indigenous ethnic groups in the region that is now modern-day Benin (Primativo et al. 2017). Traditional religious practices, such as Vodun, are most commonly reported among the Adja in Benin. Bariba: The Bariba are the largest ethnic group of Benin’s northern region and smaller populations also reside in neighboring countries including Nigeria. The Bariba arrived in Benin in the 13 th and 14 th centuries (Decoret-Ahiha 2005). The Bariba are known for their large kingdoms, including the Kingdom of Nikki, and for their role as slavers of ethnic groups such as the Betamaribé (Hahonou 2015; Houngnikpo and Decalo 2013). They have distinct family, economic, and gendered educational patterns compared to other ethnic groups in Benin (Maffioli, Sacco, and Gabrielli 2007). Although they are predominantly Muslim, their beliefs are syncretic with Traditional religions (Sargent 1988). Betamaribé: The Betamaribé people are located almost exclusively in Benin’s north- westernmost department. They are religiously diverse with near equal affiliations of Catholicism, Traditional beliefs, and reports of no religion, making them the only group for whom this response is common. Further, they are the only northern ethnic category for whom Islam is not a major religion, which is why, historically, they were raided as slaves by ethnic groups such as the Bariba (Houngnikpo and Decalo 2013). Dendi: The Dendi, a Songhay people, are concentrated in northeastern Benin bordering Niger, where they form a substantial portion of the population (Anon 2004). The Songhay 204 Empire thrived from the 1400s until colonized by the French in the 1800s (Amoah-Boampong 2016; Anon 2004) and they entered the region that is now Benin in the 16 th century (Decoret- Ahiha 2005). The Dendi language is a commonly used vehicular trade language across much of West Africa (Anon 2004; Bierschenk 1996). Fon: The Fon are the largest ethnic group in Benin, comprising about 40% of the Beninese population and their language, Fongbe, is the most widely spoken local language in Benin (Falen 2016). The Fon kingdom of Dahomey is renowned for their fight against colonization as well as their role in selling people to Europeans during the Atlantic slave trade (Anon 2017; Heath 2008; Lalmann 2004; Prag 2010). Although historically and linguistically related to the Adja people, the Fon language and cultural practices are distinct from those of the Adja (Heath 2008; Lalmann 2004; Tompkins and Kluge 2002). In modern Benin, the Fon ethnic group wields significant political power (Ferree 2010; Heath 2008; Houngnikpo and Decalo 2013). Peuhl: The Peuhl, referred to as the Fulani in English speaking countries 19 , are a group of traditionally nomadic cow-herders spanning most of West Africa (Davidson 1988; Salamone 1995). Many Peuhl groups are semi or completely nomadic, making exact population counts difficult, but their population is estimated to be around 15 million (Gordon 2000; Hampshire 2004; Salamone 1995). Although the Peuhl are historically nomadic, they currently have diverse settlement patterns: settled, seasonally nomadic, and nomadic (Salamone 1995). The Peuhl settled in most nations only after the 12 th century and are not considered an “indigenous” ethnic group in many nations (Adebayo 1997; Bukari and Schareika 2015; Okello et al. 2014; Salamone 19 This ethnic group goes by several names in West Africa. The term “Fulani” is of Hausa origin and is commonly used in English speaking countries (Hampshire 2004; Salamone 1995). 205 1995). In Benin, the Peuhl reside predominately in the northern regions and nearly 90 percent identify as Muslim. However, the Peuhl of Benin are known for having beliefs infused with local Traditional practices (Houngnikpo and Decalo 2013). Yoa/Lokpa: The Yoa/Lokpa are a small ethnic group residing in northwestern Benin. In the colonial period, many of the Yoa/Lokpa men migrated West to Togo and Ghana and upon return brought Islam back, where it became one of the main religions of this ethnic group starting in the 1960s (Alber, Häberlein, and Martin 2010). Yoruba: The Yoruba people are geographically concentrated in central and southern Benin and Nigeria. The term “Yoruba” refers to an ethnic group sharing cultural customs, origins, and norms, as well as a group of language dialects (Akintoye 2004; Kehinde 2010). The Yoruba have a history of empires in the modern Benin-Nigeria region, with several major empires falling apart or being split due to colonial rule (Akintoye 2004; Houngnikpo and Decalo 2013; Kehinde 2010). During colonization, the majority of Yorubaland fell under British rule, while a minority of the Yoruba kingdoms fell under French control (Akintoye 2004; Kehinde 2010). As a result, most Yoruba live in Nigeria but a smaller proportion of the Yoruba live in Benin, where they are a small but significant ethnic group. The Yoruba are an advantaged ethnic group both economically and politically. They are religiously diverse in Benin with Catholicism and Islam being common. Many Yoruba maintain strong ties to their traditional beliefs and culture, despite ties to Western religions (Houngnikpo and Decalo 2013). 206 Appendix 2.B: Background information on religions Traditional: Traditional beliefs in West Africa and Benin are diverse in their practices and frequently lack an established orthodoxy (Falen 2016; Falola and Jean-Jacques 2016). A key tenant that links Traditional beliefs throughout West Africa is the belief of ancestors and spirits/god playing active roles in the lives of the living (Acquah 2011; Falen 2016; Falola and Jean-Jacques 2016). Despite many people not identifying as Traditional adherents, the beliefs and practices associated with Traditional religious systems are widespread and have influenced Islam and Christianity in the region (Acquah 2011; Falen 2016; Falola and Jean-Jacques 2016; Okeke et al. 2017). The most famous of Traditional belief systems is Vodun, spelled Voodoo in English, the traditional belief system of the Fon people of Benin (Falen 2016; Falola and Jean- Jacques 2016). Islam: Islam was introduced to the region that is now Benin starting in the 11 th century (Falola and Jean-Jacques 2016). Islam largely spread along trans-Saharan trading routes throughout West Africa and was later spread through the formation of large Caliphates, with the Fulani Sokoto Caliphate shaping Islam in Benin (Falola and Jean-Jacques 2016). Islam in Benin is known for being particularly syncretic, with many people in predominantly Muslim communities also adhering to Traditional beliefs (Falola and Jean-Jacques 2016; Houngnikpo and Decalo 2013). Catholicism: Catholicism was brought to Benin by Europeans beginning in the 15 th century, beginning with the Portuguese, who first established the trans-Atlantic slave trade out of Benin, and later establishing a strong presence in the 19 th century with the French under colonization (Falen 2016; Falola and Jean-Jacques 2016). Catholicism was unable to make inroads in converting the population of Benin until the fall of the Dahomey Empire (Fage and 207 McCaskie 2016; Falola and Jean-Jacques 2016). The missionaries established schools, almost exclusively in the south, sowing the seeds for the north-south divide in Benin (Falola and Jean- Jacques 2016; Houngnikpo and Decalo 2013). Despite its long presence in Benin and its success in gaining adherents, Catholicism — and Christianity at large — is still referred to as “the white man’s religion” (Falen 2016). Protestantism: Protestantism is a relatively new addition to Benin’s religious landscape with Protestant missionaries and churches spilling over from British colonization in neighboring Nigeria (Houngnikpo and Decalo 2013). Methodists are the largest Protestant group at 5% of the Beninese population (INSAE and ICF International 2019). Other Christians (African Initiated Churches): Evidence of the syncretic nature of religion in Benin are “African Initiated Churches,” which blend Christian ideals with parts of Traditional beliefs and practices (Falen 2016; Falola and Jean-Jacques 2016). One common traditional practice incorporated into these church’s doctrines is the acceptance of polygyny (Falen 2008a), and these churches often gain members by offering protection to adherents from the powerful spirits prevalent in Traditional religions (Falen 2016). The Celestial Church of Christ, founded in Porto-Novo (the capital city) in 1947, is the largest African Initiated Church in Benin, comprising 8% of the population in 2018 (Falen 2016; INSAE and ICF International 2019). 208 Appendix 2 Figure 1: Beninese languages in the Niger-Congo language family Source: Author’s rendition based on Eberhard, David M., Gary F. Simons, and Charles D. Fennig (eds.). 2019. Ethnologue: Languages of the World. Twenty-second edition. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com. Niger-Congo Atlantic- Congo Atlantic Northern Senegambian Fula-Wolof Fula West Central Fulani Volta-Congo Benue-Congo Defoid Yoruboid Edekiri Yoruba North Gur Central Northern Oti-Volta Eastern Betamaribé Southern Grusi Eastern Yoa/Lokpa Bariba Bariba Kwa Left Bank Gbe Aja Adja Fon Fon 209 Appendix 2 Figure 2: Beninese language in the Nilo-Saharan language family Source: Author’s rendition based on Eberhard, David M., Gary F. Simons, and Charles D. Fennig (eds.). 2019. Ethnologue: Languages of the World. Twenty-second edition. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com. Nilo- Saharan Songhai Southern Dendi 210 Appendix 2 Figure 3: A map of Benin’s departments Notes: Uses data from BDHS 2006, 2011-2012, and 2017-2018 due to changes in Benin’s departments. The four northern departments are the Alibori (1), the Atacora (2), the Borgou (4), and the Donga (7). Map source: The Demographic and Health Surveys Program’s Spatial Data Repository. 1 2 3 4 5 6 7 9 10 11 12 -------- 8 Department 1.  Alibori 2.  Atacora 3.  Atlantique 4.  Borgou 5.  Collines 6.  Couffo 7.  Donga 8.  Littoral 9.  Mono 10. Ouémé 11. Plateau 12. Zou 211 Appendix 2 Figure 4: Beninese population pyramids Source: U.S. Census Bureau. International Programs. International Data Base. Revised: September 18, 2018. 212 APPENDIX CHAPTER 3 Appendix 3 Table 1: Pairings by ethnicity and gender of spouse by absolute number of pairings and percentage of all interethnic marriages (Page 1 of 2) Husband Wife Count % Wife Husband Count % Adja Bariba 9 0.5 Adja Bariba 4 0.2 -- Betamaribé 1 0.1 -- Betamaribé 1 0.1 -- Fon 183 10.2 -- Dendi 1 0.1 -- Other 3 0.2 -- Fon 229 12.8 -- Yoruba 25 1.4 -- Peuhl 2 0.1 -- Other 4 0.2 -- Yoruba 15 0.8 Bariba Adja 4 0.2 Bariba Adja 9 0.5 -- Betamaribé 9 0.5 -- Betamaribé 7 0.4 -- Dendi 45 2.5 -- Dendi 37 2.1 -- Fon 7 0.4 -- Fon 14 0.8 -- Peuhl 32 1.8 -- Peuhl 34 1.9 -- Other 60 3.4 -- Other 7 0.4 -- Yoa/Lokpa 16 0.9 -- Yoa/Lokpa 8 0.5 -- Yoruba 22 1.2 -- Yoruba 25 1.4 Betamaribé Adja 1 0.1 Betamaribé Adja 1 0.1 -- Bariba 7 0.4 -- Bariba 9 0.5 -- Dendi 1 0.1 -- Dendi 5 0.3 -- Fon 5 0.3 -- Fon 4 0.2 -- Peuhl 7 0.4 -- Peuhl 8 0.5 -- Other 12 0.7 -- Other 54 3.0 -- Yoa/Lokpa 15 0.8 -- Yoa/Lokpa 24 1.3 -- Yoruba 5 0.3 -- Yoruba 2 0.1 Dendi Adja 1 0.1 Dendi Bariba 45 2.5 -- Bariba 37 2.1 -- Betamaribé 1 0.1 -- Betamaribé 5 0.3 -- Fon 14 0.8 -- Fon 7 0.4 -- Peuhl 8 0.5 -- Peuhl 17 1.0 -- Other 44 2.5 -- Other 33 1.8 -- Yoa/Lokpa 71 4.0 -- Yoa/Lokpa 16 0.9 -- Yoruba 10 0.6 -- Yoruba 9 0.5 213 Appendix 3 Table 1 Continued (Page 2 of 2) Husband Wife Count % Wife Husband Count % Fon Adja 229 12.8 Fon Adja 183 10.2 -- Bariba 14 0.8 -- Bariba 7 0.4 -- Betamaribé 4 0.2 -- Betamaribé 5 0.3 -- Dendi 14 0.8 -- Dendi 7 0.4 -- Peuhl 7 0.4 -- Peuhl 6 0.3 -- Other 4 0.2 -- Other 9 0.5 -- Yoa/Lokpa 18 1.0 -- Yoa/Lokpa 15 0.8 -- Yoruba 193 10.8 -- Yoruba 170 9.5 Peuhl Adja 2 0.1 Peuhl Bariba 32 1.8 -- Bariba 34 1.9 -- Dendi 17 1.0 -- Betamaribé 8 0.5 -- Fon 7 0.4 -- Dendi 8 0.5 -- Betamaribé 7 0.4 -- Fon 6 0.3 -- Other 7 0.4 -- Other 6 0.3 -- Yoa/Lokpa 16 0.9 -- Yoa/Lokpa 9 0.5 -- Yoruba 20 1.1 -- Yoruba 30 1.7 Other Adja 4 0.2 Other Adja 3 0.2 -- Bariba 7 0.4 -- Bariba 60 3.4 -- Betamaribé 54 3.0 -- Betamaribé 12 0.7 -- Dendi 44 2.5 -- Dendi 33 1.8 -- Fon 9 0.5 -- Fon 4 0.2 -- Peuhl 7 0.4 -- Peuhl 6 0.3 -- Yoa/Lokpa 22 1.2 -- Yoa/Lokpa 52 2.9 -- Yoruba 9 0.5 -- Yoruba 8 0.5 Yoa/Lokpa Bariba 8 0.5 Yoa/Lokpa Bariba 16 0.9 -- Betamaribé 24 1.3 -- Betamaribé 15 0.8 -- Dendi 71 4.0 -- Dendi 16 0.9 -- Fon 15 0.8 -- Fon 18 1.0 -- Peuhl 16 0.9 -- Peuhl 9 0.5 -- Other 52 2.9 -- Other 22 1.2 -- Yoruba 12 0.7 -- Yoruba 9 0.5 Yoruba Adja 15 0.8 Yoruba Adja 25 1.4 -- Bariba 25 1.4 -- Bariba 22 1.2 -- Betamaribé 2 0.1 -- Betamaribé 5 0.3 -- Dendi 10 0.6 -- Dendi 9 0.5 -- Fon 170 9.5 -- Fon 193 10.8 -- Peuhl 20 1.1 -- Peuhl 30 1.7 -- Other 8 0.5 -- Other 9 0.5 -- Yoa/Lokpa 9 0.5 -- Yoa/Lokpa 12 0.7 Source: The Demographic and Health Surveys Program 214 Appendix 3 Table 2: Pairings by religion and gender of spouse by absolute number of pairings and percentage of all interreligious marriages (Page 1 of 2) Husband Wife Count % Wife Husband Count % Traditional Muslim 45 1.7 Traditional Muslim 62 2.3 -- Catholic 168 6.2 -- Catholic 122 4.5 -- Protestant 41 1.5 -- Protestant 14 0.5 -- Other Christian 208 7.7 -- Other Christian 41 1.5 -- Other 58 2.2 -- Other 14 0.5 -- None 262 9.7 -- None 94 3.5 Muslim Traditional 62 2.3 Muslim Traditional 45 1.7 -- Catholic 92 3.4 -- Catholic 90 3.3 -- Protestant 27 1.0 -- Protestant 14 0.6 -- Other Christian 35 1.3 -- Other Christian 18 0.7 -- Other 3 0.1 -- Other 3 0.1 -- None 96 3.6 -- None 28 1.0 Catholic Traditional 122 4.5 Catholic Traditional 168 6.2 -- Muslim 90 3.3 -- Muslim 92 3.4 -- Protestant 80 3.0 -- Protestant 54 2.0 -- Other Christian 182 6.8 -- Other Christian 98 3.6 -- Other 22 0.8 -- Other 15 0.6 -- None 91 3.4 -- None 87 3.2 Protestant Traditional 14 0.5 Protestant Traditional 41 1.5 -- Muslim 15 0.6 -- Muslim 27 1.0 -- Catholic 54 2.0 -- Catholic 80 3.0 -- Other Christian 166 6.2 -- Other Christian 118 4.4 -- Other 8 0.3 -- Other 5 0.2 -- None 8 0.3 -- None 33 1.2 Other Christian Traditional 41 1.5 Other Christian Traditional 208 7.7 -- Muslim 18 0.7 -- Muslim 35 1.3 -- Catholic 98 3.6 -- Catholic 182 6.8 -- Protestant 118 4.4 -- Protestant 166 6.2 -- Other 44 1.6 -- Other 35 1.3 -- None 17 0.6 -- None 80 3.0 215 Appendix 3 Table 2 Continued (Page 2 of 2) Husband Wife Count % Wife Husband Count % Other Traditional 14 0.5 Other Traditional 58 2.2 -- Muslim 3 0.1 -- Muslim 3 0.1 -- Catholic 15 0.6 -- Catholic 22 0.8 -- Protestant 5 0.2 -- Protestant 8 0.3 -- Other Christian 35 1.3 -- Other Christian 44 1.6 -- None 8 0.3 -- None 8 0.3 No religion Traditional 94 3.5 No religion Traditional 262 9.7 -- Muslim 28 1.0 -- Muslim 96 3.6 -- Catholic 87 3.2 -- Catholic 91 3.4 -- Protestant 33 1.2 -- Protestant 8 0.3 -- Other Christian 80 3.0 -- Other Christian 17 0.6 -- Other 8 0.3 -- Other 8 0.3 Source: The Demographic and Health Surveys Program 216 Appendix 3 Table 3: Individual characteristics and difference testing of recently married men and women in Benin by ethnic composition of marriage Married Women Married Men All Intra Inter All Intra Inter Ethnicity (%) Fon 41.22 43.70 25.47 *** 42.46 43.70 34.59 ** Adja 14.91 14.78 15.72 14.91 14.78 15.72 Bariba 11.40 11.51 10.69 11.23 11.51 9.43 Betamaribé 6.08 6.60 2.83 ** 6.13 6.60 3.14 * Dendi 4.54 3.52 11.01 *** 3.86 3.52 5.97 * Peuhl 7.41 7.49 6.92 7.37 7.49 6.60 Yoa/Lokpa 4.03 3.72 5.97 4.50 3.72 9.43 *** Yoruba 10.41 8.68 21.38 *** 9.55 8.68 15.09 *** Education Average years all 2.42 2.12 4.33 *** 4.27 3.92 6.51 *** (3.84) (3.55) (4.95) (5.29) (5.04) (6.24) Average years educated 6.50 6.17 7.78 *** 7.74 6.78 9.41 *** (3.63) (3.41) (4.14) (4.90) (4.72) (5.38) Educated (%) 37.28 34.38 55.66 *** 55.18 52.98 69.19 *** Literate (%) 28.98 25.91 47.49 *** 45.35 43.82 55.15 *** Religion (%) Traditional 14.61 16.52 2.52 *** 17.44 19.30 5.66 *** Muslim 25.92 24.31 36.16 *** 27.55 25.99 37.42 *** Catholic 25.58 23.81 36.79 *** 25.02 24.01 31.45 ** Protestant 6.34 6.55 5.03 5.83 5.85 5.66 Other Christian 19.92 20.63 15.41 * 17.52 17.81 15.72 Other 1.76 1.98 0.31 * 1.24 1.34 0.63 No religion 5.87 6.20 3.77 5.40 5.70 3.46 Age at Marriage 19.73 19.56 20.81 *** 23.74 23.65 24.28 * (4.02) (3.97) (4.14) (4.93) (4.92) (4.98) Premarital Pregnancy 40.19 39.48 44.64 --- --- --- N 2334 2016 318 2334 2016 318 Source: The Demographic and Health Surveys Program Note: Recently married is defined as within 5 years of each survey (1991-2018) Standard deviations are in parentheses Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 217 Appendix 3 Table 4: Individual characteristics and difference testing of recently married men and women in Benin by religious composition of marriage Married Women Married Men All Intra Inter All Intra Inter Religion (%) Traditional 14.61 15.09 13.45 17.44 15.09 23.10 *** Muslim 25.92 33.76 7.02 *** 27.55 33.76 12.57 *** Catholic 25.58 26.42 23.54 25.02 26.42 21.64 * Protestant 6.34 4.18 11.55 *** 5.83 4.18 9.80 *** Other Christian 19.92 18.12 24.27 ** 17.52 18.12 16.08 Other 1.76 0.48 4.82 *** 1.24 0.48 3.07 *** No religion 5.87 1.94 15.35 *** 5.40 1.94 13.74 *** Education Average years all 2.42 2.35 2.60 4.27 4.19 4.47 (3.84) (3.81) (3.92) (5.29) (5.25) (5.41) Average years educated 6.50 6.47 6.57 7.74 7.68 7.86 (3.63) (3.66) (3.57) (4.90) (4.86) (4.98) Educated (%) 3728 36.36 39.47 55.18 54.48 56.87 Literate (%) 28.98 28.05 31.26 45.35 45.56 44.82 Ethnicity (%) Fon 41.22 39.33 45.76 ** 42.46 41.58 44.59 Adja 14.91 13.94 17.25 * 14.91 13.15 19.15 *** Bariba 11.40 12.06 9.80 11.23 11.88 9.65 Betamaribé 6.08 3.58 12.13 *** 6.13 3.58 12.28 *** Dendi 4.54 6.36 0.14 *** 3.86 5.27 0.44 *** Peuhl 7.41 8.91 3.80 *** 7.37 8.97 3.51 *** Yoa/Lokpa 4.03 4.73 2.34 ** 4.50 5.39 2.34 ** Yoruba 10.41 11.09 8.77 9.55 10.18 8.04 Age at Marriage 19.73 19.64 19.98 23.74 23.67 23.90 (4.02) (3.85) (4.37) (4.93) (4.85) (5.14) Premarital Pregnancy 40.19 39.94 40.79 --- --- --- N 2334 1650 684 2334 1650 684 Source: The Demographic and Health Surveys Program Note: Recently married is defined as within 5 years of each survey (1991-2018) Standard deviations are in parentheses Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 218 Appendix 3 Table 5: Couple characteristics and difference testing of recent marriages in Benin by ethnic composition of marriage All Intra Inter Decade Married (%) 1990s 20.39 21.33 14.47 ** 2000s 43.14 43.20 42.77 2010s 36.46 35.47 42.77 * Marital Type (%) Polygynous (first marriage) 24.81 26.14 16.35 *** Polygynous (all marriages) 26.26 27.01 21.61 ** Interreligious marriage 29.31 29.17 30.19 Remarried 5.07 5.15 5.05 Education (%) Neither educated 37.92 40.23 23.27 *** Both educated 30.38 27.58 48.11 *** Husband educated/Wife not 24.81 25.40 21.07 Wife educated/Husband not 6.90 6.80 7.55 Age Gap (%) <5 years 36.46 35.37 43.40 ** Husband older 5-9 years 37.10 37.75 33.02 Husband older >10 years 25.79 21.19 23.27 Wife older >5 years 0.64 0.69 0.31 Household Wealth Quintile (%) Poorest 21.04 22.22 13.52 *** Poor 20.95 22.57 10.69 *** Middle 19.54 20.44 13.84 ** Rich 19.02 18.90 19.81 Richest 19.45 15.87 42.14 *** Urban (%) 38.69 34.87 62.89 *** N 2334 2016 318 Source: The Demographic and Health Surveys Program Note: Recently married is defined as within 5 years of each survey (1991-2018) Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 219 Appendix 3 Table 6: Couple characteristics and difference testing of recent marriages in Benin by religious composition of marriage All Intra Inter Decade Married (%) 1990s 20.39 19.88 21.64 2000s 43.14 44.97 38.74 ** 2010s 36.46 35.15 39.62 * Marital Type (%) Polygynous (first marriage) 24.81 24.30 26.02 Polygynous (all marriages) 26.26 25.56 27.88 Interethnic marriage 13.62 13.45 14.04 Remarried 5.07 4.46 6.50 ** Education (%) Neither educated 37.92 38.91 35.53 Both educated 30.38 29.76 31.87 Husband educated/Wife not 24.81 24.73 25.00 Wife educated/Husband not 6.90 6.61 7.60 Age Gap (%) <5 years 36.46 36.36 36.70 Husband older 5-9 years 37.10 37.21 36.84 Husband older >10 years 25.79 25.82 25.73 Wife older >5 years 0.64 0.61 0.73 Household Wealth Quintile (%) Poorest 21.04 21.15 20.76 Poor 20.95 20.61 21.78 Middle 19.54 20.42 17.40 Rich 19.02 19.33 18.27 Richest 19.45 18.48 21.78 Urban (%) 38.69 37.88 40.64 N 2334 1650 684 Source: The Demographic and Health Surveys Program Note: Recently married is defined as within 5 years of each survey (1991-2018) Significance tests used are pr-tests and t-tests *p < .05; **p < .01; ***p < .001 220 APPENDIX CHAPTER 4 Appendix 4 Table 1: Fertility preferences and outcomes by ethnicity of wife in Benin Ethnicity Ideal fertility size for women with two or fewer children Women giving non-numeric responses (%) Number of children ever born to women 40+ Number of living children for women 40+ Adja 4.50 (1.96) 3.38 6.35 (2.59) 5.31 (2.20) Bariba 5.06 (2.52) 8.35 6.83 (2.43) 5.79 (2.13) Betamaribé 5.38 (2.42) 5.53 7.54 (2.80) 6.00 (2.13) Dendi 5.03 (3.00) 20.03 6.61 (2.52) 5.44 (2.05) Fon 4.04 (1.63) 4.17 6.17 (2.74) 5.07 (2.03) Peuhl 6.58 (3.18) 11.86 7.48 (2.99) 6.18 (2.60) Yoa/Lokpa 4.93 (1.94) 13.30 6.63 (2.11) 5.50 (1.93) Yoruba 4.21 (1.67) 5.87 6.30 (2.36) 5.35 (1.98) Source: The Demographic and Health Surveys Program Standard deviations are in parentheses 221 Appendix 4 Table 2: Fertility preferences and outcomes by religion of wife in Benin Religion Ideal fertility size for women with two or fewer children Women giving non-numeric responses (%) Number of children ever born to women 40+ Number of living children for women 40+ Traditional 5.35 (2.39) 3.62 6.84 (2.57) 5.44 (2.09) Muslim 5.65 (2.76) 10.04 6.79 (2.58) 5.72 (2.24) Catholic 4.15 (1.65) 2.74 5.85 (2.49) 4.98 (2.04) Protestant 4.52 (1.61) 4.98 6.43 (2.21) 5.45 (1.96) Other Christian 4.68 (1.83) 3.82 6.45 (2.45) 5.35 (2.11) Other 4.74 (1.74) 4.11 5.82 (2.47) 4.98 (2.17) None 5.49 (2.36) 7.12 6.94 (2.58) 5.52 (2.12) Source: The Demographic and Health Surveys Program Standard deviations are in parentheses 222 APPENDIX CHAPTER 5 Appendix 5 Table 1: Extramarital sex and polygyny by ethnicity of husband Ethnicity Engaged in extramarital sex within last year (%) Polygynously married (%) Adja 17.72 45.41 Bariba 20.04 44.47 Betamaribé 16.41 38.22 Dendi 10.08 44.47 Fon 21.20 33.22 Peuhl 13.38 48.95 Yoa/Lokpa 16.06 45.21 Yoruba 17.49 34.27 Source: The Demographic and Health Surveys Program 223 Appendix 5 Table 2: Extramarital sex and polygyny by religion of husband Religion Engaged in extramarital sex within last year (%) Polygynously married (%) Traditional 18.77 50.06 Muslim 15.53 45.12 Catholic 22.50 29.44 Protestant 17.74 27.50 Other Christian 16.42 32.49 Other 19.44 34.46 None 22.31 37.11 Source: The Demographic and Health Surveys Program 224 Appendix 5 Table 3: Means and percentages of characteristics of women in general sample, monogamous marriages, and polygynous marriages (Standard deviations in parentheses) Wife’s Characteristics All Monogamous Polygynous Polygynous (%) 37.49 n/a 100.00 Ethnicity (%) Fon 39.57 43.43 33.13 Adja 14.97 13.27 17.80 Bariba 11.32 10.29 13.05 Betamaribé 6.86 6.93 6.73 Dendi 4.34 4.18 4.61 Peuhl 8.01 6.67 10.32 Yoa/Lokpa 4.05 3.78 4.50 Yoruba 10.88 11.44 9.93 Educated (%) 23.84 29.66 14.14 Religion (%) Traditional 16.86 13.22 22.93 Muslim 26.59 24.56 29.96 Catholic 23.28 27.71 15.90 Protestant 6.20 7.29 4.37 Other Christian 18.97 20.23 16.85 Other 1.74 1.53 2.09 No religion 6.37 5.45 7.90 Age at Marriage 18.31 18.52 17.97 (4.06) (4.09) (3.97) Remarried (%) 13.38 10.62 17.60 Premarital Pregnancy (%) 34.34 34.59 33.92 Age at Survey 30.98 (7.79) 30.23 (7.80) 32.34 (7.60) Wife Rank (%) No other wives 62.51 100.00 n/a 1 26.96 n/a 70.89 2 7.59 n/a 20.52 3 1.19 n/a 5.52 4 0.48 n/a 1.38 5+ 0.55 n/a 1.69 N 9282 6143 3685 Source: The Demographic and Health Surveys Program 225 Appendix 5 Table 4: Likelihood of being in a polygynous marriage (Page 1 of 2) Model 1 Model 2 Model 3 Interethnic 0.76*** 0.75*** 0.78** (0.06) (0.06) (0.06) Interreligious 1.03 1.03 1.05 (0.05) (0.05) (0.05) Education Neither educated (ref) -- -- -- Both educated 0.36*** 0.36*** 0.39*** (0.02) (0.02) (0.03) Husband educated/Wife not 0.69*** 0.69*** 0.70*** (0.04) (0.04) (0.04) Wife educated/ Husband not 0.74** 0.75** 0.76* (0.08) (0.08) (0.08) Age Gap <5 years (ref) -- -- -- Husband older 5-9 years 1.22*** 1.22** 1.21*** (0.07) (0.07) (0.07) Husband older >10 years 2.17*** 2.17*** 2.16*** (0.12) (0.12) (0.12) Wife older >5 years 1.12 1.11 1.12 (0.26) (0.26) (0.26) Premarital Pregnancy 1.05 1.08 1.08 (0.05) (0.05) (0.05) Decade Married Pre-1980s (ref) -- -- -- 1980s 0.88 0.96 0.96 (0.09) (0.10) (0.10) 1990s 0.65*** 0.80* 0.80* (0.06) (0.08) (0.08) 2000s 0.50*** 0.66*** 0.66*** (0.05) (0.08) (0.08) 2010s 0.41*** 0.58*** 0.58*** (0.05) (0.08) (0.08) Duration of Marriage 1.02*** 1.02*** (0.01) (0.01) Number of Living Children 0.97 0.97 (0.02) (0.02) Rural 1.37*** (0.07) Northern 1.14** (0.06) 226 Appendix 5 Table 4 Continued (Page 2 of 2) Model 1 Model 2 Model 3 Household Wealth Quintile Poorest (ref) -- Poor 0.98 (0.06) Middle 1.28*** (0.09) Rich 1.41*** (0.10) Richest 1.13 (0.09) Constant 0.93 0.63** 0.42*** (0.09) (0.09) (0.06) Obs. 9828 9828 9828 Pseudo R 2 0.07 0.07 0.07 Source: The Demographic and Health Surveys Program Standard errors are in parentheses *p < .05; **p < .01; ***p < .001 This appendix offers a brief introduction into the couples who are in polygynous marriages in Benin. In a logistic regression model with full couple-level controls, interethnically married couples are significantly less likely to be polygynous, however, interreligious couples are as likely to be polygynous as intrareligious couples. Couples where both spouses are educated have significantly lower odds of being polygynously married compared to couples where neither spouse is educated, as do couples where either spouse has been. Women who are married to men ten or more years older than they are have significantly higher odds of being polygynously married, whereas no other age gap is significant. Premarital pregnancy is not associated with polygyny and compared to couples married before 1980, couples in the 2000s and 2010s have significantly lower odds of being polygynous. The duration of marriage is not significantly associated with the likelihood of being polygynous, however, increasing numbers of living children decrease the likelihood that a couple will be polygynous. Rural couples are significantly more likely to be polygynous, as are northern couples. Compared to the lowest wealth quintile, couples in the middle and fourth (rich) quintiles are significantly more likely to be polygynous, 227 however, couples in the second (poor) and the fifth (the richest) quintiles are no more likely to be polygynous than couples in the lowest (poorest) quintile.

Abstract (if available)

Abstract Family formation patterns are in flux globally including assortative partner selection. Prior research notes the historical importance of shared social identities—specifically ethnicity and religion—in partner selection, with most couples sharing these social identities. Yet, in many contexts globally, large structural changes, such as increasing education, urbanization, and migration have created diverse sites for people of varying social groups to interact extensively, increasing the frequency of intermarriages along ethnic and religious lines. However, due to the importance of these social identities in value formation, marriages in which spouses’ social identities do not match often experience higher levels of marital strain. ❧ Despite sub-Saharan Africa’s changing socioeconomic structures, there is minimal research on the frequency of interethnic and interreligious marriages and the marital strain experienced by these couples in the region. In this dissertation, I analyze the trends in interethnic and interreligious first marriages in Benin, West Africa. I use couples’ data from five Demographic and Health Surveys (DHS) collected between 1996 and 2018 to analyze the temporal and spatial trends in interethnic and interreligious marriages and the characteristics that are associated with entry into these marriages. I then examine the correlation between these marital types and indicators of marital strain, including fertility preference discordance and men’s engagement in extramarital sex. ❧ Three empirical chapters provide an examination of interethnic and interreligious marriages in Benin. The first empirical chapter, Chapter 3, explores the changes in interethnic and interreligious marriage over time by comparing marital cohorts (pre-1980s to the 2010s), followed by a spatial exploration of the geographic spread of interethnic and interreligious marriages. I use binary and multivariate logistic models to explore the relationship between intermarriage and men’s and women’s individual characteristics, as well as their characteristics relative to each other. These analyses indicate that interethnic first marriages have increased rapidly and are tied to increasing education, urbanization, and wealth. The increase in interethnic marriage seems to correlate with structural changes affecting the marriage market. Specifically, interethnic marriages appear to correlate with higher levels of potential interethnic interactions and the prioritization of matching human capital over shared group identity. I found no such patterns for interreligious marriages: the socioeconomic and demographic characteristics of interreligious first marriages mirror those of the average intrareligious first marriage and these unions do not display consistent signs of increasing in frequency. My findings suggest that these two forms of intermarriage are not uniformly tied to larger socioeconomic changes and that they are distinct marital formation patterns. ❧ Chapter 4, the second empirical chapter, compares the fertility preferences and outcomes of interethnic and interreligious couples to their intraethnic and intrareligious counterparts. Ethnicity and religion both play important roles in shaping individuals’ fertility preferences and couples’ fertility outcomes. However, much of the literature on fertility is centered on women, and we know little about how different social identities within couples are linked to fertility preferences and completed fertility in sub-Saharan contexts. I find that men and women in interethnic marriages have significantly lower ideal family sizes than their intraethnic counterparts and that they are more likely to be concordant in their fertility preferences than intraethnically married couples. Contrarily, interreligious couples do not have distinct fertility preferences and they show only minimally increased concordance compared to intrareligious couples. The results for fertility outcomes among women ending their reproductive years indicate that interethnic couples do not have fewer children despite their lower preferences, however, interreligious couples have more children despite no difference in their preferences. These findings suggest that couples of different religious backgrounds experience fertility differently than intrareligious couples. Further, the differing outcomes between interethnic and interreligious couples indicate that interethnic and interreligious marriages are not uniformly tied to couples’ fertility preferences and outcomes although both involve partners of different social identities. ❧ The third empirical chapter, Chapter 5, examines the presence of marital destabilizing behaviors that indicate low marital investment in intact marriages. Specifically, I examine two locally salient marital destabilizing behaviors—extramarital sex and polygyny. I ask whether men in interethnic or interreligious marriages are less invested in their unions by estimating if they are more likely to engage in behaviors that are disruptive to the couple-unit and that cause negative health outcomes for their wives. Using logistic regression, I find that interethnically married men have higher odds of engaging in extramarital sex, however, when controlling for educational attainment or wealth, this difference is non-significant. This indicates that it is not being interethnically married, per se, that results in this higher likelihood of engaging in extramarital sex, rather interethnically married men are disproportionately in higher socioeconomic brackets where extramarital sex is more common. Importantly, interethnic marriages are no more likely to transition to polygyny. Interreligiously married men, on the other hand, are significantly more likely to engage in extramarital sex, although not polygyny, than intrareligiously married men even when controlling for educational attainment and wealth. This suggests that men in interreligious marriages are less invested in their unions as they are more likely to engage in marital destabilizing behaviors that are linked to negative outcomes for their wives and children. That these findings differ across the two types of intermarriage suggests that interethnic couples do not experience similar levels of marital strain as interreligious couples, despite both being marriages across social groups. This further supports interethnic and interreligious marriages as being distinct marital patterns with different social meanings. ❧ In sum, the results of this dissertation indicate that marriage markets are in flux in Benin, West Africa in ways that are shifting the salience of shared social identities in partner selection. One of the main findings of this dissertation is that selection into and the outcomes of being in an intermarriage are not the same for the average interethnic and interreligious couple, indicating that intermarriage is not a uniform experience across social groups. Interethnic marriage in Benin is correlated with recent structural changes, such as increased access to school and urbanization. This supports research in other regions of the world finding that social identity homogamy often decreases with structural changes, especially those that diversify networks. Although interethnic marriages are uncommon, their ties to higher levels of education and wealth appear to be protective against higher marital strain, both in terms of their relative fertility preferences and the husband’s engagement in marital destabilizing behaviors. This may be because interethnic couples, on average, have more resources to weather familial discontent and are more likely to exhibit the characteristics that correlate with more power for wives in the relationship and higher levels of communication at the couple-level. ❧ Contrarily, in Benin, the average interreligious marriage mirrors the average intrareligious marriage outside of their choice to marry interreligiously. These couples are neither richer, nor more educated, nor more urban—and there have been relatively stable high rates of interreligious couplings across survey waves. Yet, interreligious marriages appear to experience more marital strain regarding fertility and extramarital sex than their intrareligious counterparts. This suggests that interreligious marriages experience higher levels of strain in the Beninese context and, unlike the interethnic couples, they are less likely, on average, to have the resources to help buffer this increased marital strain. These findings tie into rich literatures showing the benefits of increased resources in staving off marital strain and indicate that although both types of intermarriage involve partners marrying across social groups, interethnic and interreligious marriages are distinct in their correlation with structural changes and their marital outcomes.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Diverging fertility patterns? Racial and educational differences in fertility behaviors and their implications for socioeconomic mobility

PDF

Marital quality, gender, and biomarkers of disease risk in the MIDUS cohort

PDF

The role of public policy in the decisions of parents and caregivers: an examination of work, fertility, and informal caregiving

PDF

Essays on the empirics of risk and time preferences in Indonesia

PDF

Essays on family and labor economics

PDF

What leads to a happy life? Subjective well-being in Alaska, China, and Australia

PDF

In visible families: gay fatherhood and the politics of family change

PDF

Global consequences of local information biases in complex networks

PDF

Impacts of caregiving on wellbeing among older adults and their spousal caregivers in the United States

PDF

Developing an agent-based simulation model to evaluate competition in private health care markets with an assessment of accountable care organizations

PDF

Acculturation team-based clinical program: pilot program to address acculturative stress and mental health in the Latino community

Asset Metadata

Creator Coritz, Alli (author)

Core Title Marital matching in West Africa: an examination of interethnic and interreligious first marriages in Benin

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Sociology

Publication Date 01/30/2022

Defense Date 12/11/2019

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag assortative marriage,Benin,demographic and health surveys,demography,Ethnicity,extramarital sex,Family,fertility,fertility preferences,interethnic marriage,interfaith marriage,intermarriage,interreligious marriage,marital sorting,marital strain,Marriage,OAI-PMH Harvest,Polygamy,polygyny,Religion,sub-Saharan Africa,West Africa

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Smith-Greenaway, Emily (committee chair), Casper, Lynne M. (committee member), Myers, Dowell (committee member)

Creator Email acoritz@asu.edu,acoritz@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-263010

Unique identifier UC11673462

Identifier etd-CoritzAlli-8136.pdf (filename),usctheses-c89-263010 (legacy record id)

Legacy Identifier etd-CoritzAlli-8136.pdf

Dmrecord 263010

Document Type Dissertation

Rights Coritz, Alli

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA