Child Marriage and Faith Affiliation in Sub-Saharan Africa: Stylized Facts and Heterogeneity

C hild marriage remains highly prevalent in Africa, especially in rural areas and among vulnerable socio-economic groups (UNFPA 2012; UNICEF 2014; Nguyen and Wodon 2015a). Physiological and social criteria suggest that most boys and girls aged 14 and younger are much too young for sexual, marital, and reproductive transitions, while many among the 15–17-year-olds are also too young (Dixon-Mueller 2008). Child marriage, defined here as marriage before the age of 18, is known to have negative impacts on health as well as educational attainment, labor force participation, intimate partner violence, and empowerment (for a review, see Parsons et al. 2015, in this issue). It therefore makes sense, as many countries have done, including in Africa, to legally restrict marriage before the age of 18. Yet despite such restrictions, many girls still marry early, suggesting that more needs to be done to eliminate the practice. If faith plays a role in the continuation of the practice, there should be a role for faith leaders and faith-based organizations in the fight against child marriage, as argued by Walker (2015) and Karam (2015) in this issue. But does faith really play an important role in the persistence of the practice of child marriage? This is a complex question to answer, and different types of data can be brought to bear. In order to provide an introductory discussion of the issue, this article has two main objectives. The first objective, which is the focus of the next section, is to provide an account of trends in child marriage in Africa and a brief analysis of the statistical association between child marriage and faith affiliation. This helps to establish broad stylized facts or general observations on the basis of household survey

C hild marriage remains highly prevalent in Africa, especially in rural areas and among vulnerable socio-economic groups (UNFPA 2012;UNICEF 2014;Nguyen and Wodon 2015a). Physiological and social criteria suggest that most boys and girls aged 14 and younger are much too young for sexual, marital, and reproductive transitions, while many among the 15-17-year-olds are also too young (Dixon-Mueller 2008). Child marriage, defined here as marriage before the age of 18, is known to have negative impacts on health as well as educational attainment, labor force participation, intimate partner violence, and empowerment (for a review, see Parsons et al. 2015, in this issue). It therefore makes sense, as many countries have done, including in Africa, to legally restrict marriage before the age of 18. Yet despite such restrictions, many girls still marry early, suggesting that more needs to be done to eliminate the practice. If faith plays a role in the continuation of the practice, there should be a role for faith leaders and faith-based organizations in the fight against child marriage, as argued by Walker (2015) and Karam (2015) in this issue. But does faith really play an important role in the persistence of the practice of child marriage?
This is a complex question to answer, and different types of data can be brought to bear. In order to provide an introductory discussion of the issue, this article has two main objectives. The first objective, which is the focus of the next section, is to provide an account of trends in child marriage in Africa and a brief analysis of the statistical association between child marriage and faith affiliation. This helps to establish broad stylized facts or general observations on the basis of household survey data as to whether the likelihood of marrying early differs by faith affiliation. Results suggest that it does, with higher child marriage among Muslim, Animist/traditional belief populations, and those without a faith affiliation in comparisons to Catholics and Protestants.
The second objective of the article, which is the focus of the following section, is to caution against broad generalizations about child marriage and faith affiliation, in order to show that beyond stylized facts derived from household survey data, there is also substantial heterogeneity within the adherents of any particular faith as to how the practice of child marriage is considered. The analysis focuses on Islamic communities. As pointed out by Lewis and Lockheed (2007), many Islamic communities throughout the world have managed to sharply reduce the prevalence of child marriage and have embraced girls' education and women's employment.
The heterogeneity of attitudes towards child marriage is illustrated with a case study for rural Muslim communities in Burkina Faso, based on previous work by Wodon (2015a, 2015b). The analysis suggests that child marriage does have deep socio-cultural and religious roots. Gender roles and social expectations, as well as prevailing conceptions about Islamic Law and the fear of pregnancy before marriage, all play a role in the persistence of the practice. But at the same time there are also important differences in the drivers of child marriage between communities, suggesting that policy and program responses to child marriage should take into account local conditionsincluding those pertaining to faith-if they are to be successful.

Stylized Facts: Data for Africa
Measures of child marriage are computed at the global and regional level by Nguyen and Wodon (2015a) in this issue. All measures are obtained from Demographic and Health Surveys. The incidence of child marriage is simply the share of girls born between 1985 and 1989 who married before the age of 18. The child marriage gap is a measure that takes into account how early girls marry-it is equal to the product of the incidence of child marriage times the number of years of early marriage for the girls who marry early, with the result expressed as a proportion of the appropriate minimum age of marriage (following common practice we use 18 years as the threshold for child marriage). For example, if 50 percent of girls marry early, and on average the girls who marry early marry at 16 (corresponding to 2 years of early marriage below the threshold of 18), the child marriage gap will be equal to 0.50 × (2/18) = 5.56 percent.
As discussed by Nguyen and Wodon (2015a), the region with the highest overall incidence of child marriage today (for girls born between 1985 and 1989) is South Asia where 45.4 percent of women born between those years were married below the age of 18. Sub-Saharan Africa is next, with 38.5 percent. Child marriage has been declining over time, but only slowly. Table 1 provides measures of child marriage for sub-Saharan African countries. The table suggests that at the country level, there is quite a bit of diversity in measures of child marriage. In Rwanda, only about five percent of girls born between 1985 and 1989 married early. By contrast, in Niger, almost three in four girls continue to marry before 18. The incidence of child marriage and the child marriage gap have been reduced in most but not all countries over time, but even in the majority of countries with gains, the gains achieved over the last 30 years have been limited.
Is there a relationship across African surveys between faith affiliation and the number of years of early marriage-the expected number of years a girl is likely to marry before the age of 18? This is not necessarily clear from the data in Table 1 since for example in some cases majority Muslim countries have higher measures of child marriage than majority Christian countries, while in other cases the comparison is reversed. In order to look at the relationship between child marriage and faith affiliation in a systematic way, it is necessary to rely on household level data and to use regression analysis to assess the impact of faith affiliation on the likelihood of child marriage controlling for a range of other household and community characteristics.
In work conducted on the relationship between child marriage and education, Nguyen and Wodon (2015b) look at the correlates or determinants of early marriage. The factors affecting child marriage taken into account in the regression analysis include the contemporaneous and past incidence of child marriage at the local level (in the primary sampling unit of the survey where households are located). These variables have, as expected, a strong positive impact on the likelihood that a girl will marry early and the number of years of early marriage. In addition, having a child before marriage is associated with a reduction in the number of years of early marriage. This is as expected, given that girls who have a child before marriage tend to be older when actually marrying as compared to those marrying without a previous delivery. Being an orphan does not seem to have a statistically significant effect on the number of years of early marriage (this may seem a bit surprising given that child marriage has often been presented as  1955-1959 1985-1989 1955-1959 1985-1989 Benin 2006  Source: Adapted from data processed by Nguyen and Wodon (2015a). Note: a, b, and c indicate another (closest) period is used due to lack of data for the period considered.
regina gemignani and quentin wodon the review of faith & international affairs | 43 being in part motivated by the need to protect orphan girls by integrating them into new families through early marriage).
More importantly for the question at hand in this article, there are also substantial differences in the number of years of early marriage according to the religious affiliation of the girl even after controlling for the other correlates of child marriage already mentioned. Because the analysis is conducted with Demographic and Health Surveys, only broad faith affiliations can be taken into account, which limits the richness of the analysis (as discussed for example by Olivier and Wodon 2015, in the case of the relationship between faith affiliation and reproductive health behaviors). Still, Nguyen and Wodon (2015b) find that, even after controlling for other variables, Muslims, those who declare having no religion, and those with traditional or Animist beliefs all tend to have higher measures of child marriage than the reference category of Catholics (the differences between Catholics and Protestants are small). In other words, there is evidence on the basis of household survey data of a stylized fact according to which the likelihood of marrying early and the number of years of child marriage do differ according to faith affiliation, even if other factors may play a larger role in the decision for girls (or their parents) to marry early. This of course does not imply that the relationship between faith affiliation and child marriage is the same across countries, or even within countries, as illustrated by a case study for Burkina Faso in the next section.

Heterogeneity: The Case of Burkina Faso 1
The fact that girls from some faith traditions are more likely than others to marry early than girls from other traditions does not imply that there is not a lot of heterogeneity in the factors that drive child marriage within a particular tradition. There is clearly not only heterogeneity between countries, as suggested by Table 1, but also within countries, as discussed by Wodon (2015a, 2015b) in a case study for three Muslim and rural communities in Burkina Faso based on mostly qualitative data from focus groups and in-depth interviews. The authors suggest that the gendered construction of men's and women's separate spheres of activity has a significant effect on decisions to marry early, since girls are identified with their future roles as mothers and wives and not as providers and leaders in the home or community. From this gendered division of labor, it is not surprising that many girls marry early and as a result of that decision follow a different education and labor force participation path than is the case for boys. Some girls themselves, in the face of pervasive stereotypes and limited options for education and future employment opportunities, may show low motivation or interest in school, and choose to marry early instead.
At the same time, the case study by Wodon (2015a, 2015b) suggested some important differences in the role played by faith in child marriage between communities, so that it is useful to summarize the key results here. In all three communities where data were collected the respondents were Muslim (data were collected in Tenkodogo District, Djibo District, and Bobo Dialousso District). Study participants were asked how child marriage is viewed within the Islamic faith. As given in Table 2, a majority of respondents in the first two districts, which are located further away from urban centers, agreed or totally agreed that early marriage was a very important practice for the religious community. By contrast, in Bobo Dioulasso District, that proportion was at less than a third. When asked more specifically about the nature of child marriage from the point of view of their faith, 56.9 percent of respondents in Tenkodogo District and 38.5 percent in Djibo District said that the practice is obligatory or recommended by Islamic law. By contrast again, in Bobo Dioulasso District the share of those holding those views was much lower, at 16.9 percent.
In Tenkodogo and Djibo districts, many respondents who said that early marriage is important to the Islamic faith also described prescriptions about women's role in the home, which in turn were believed to support the practice of early marriage and the fact that girls would then curtail their education. Quotes from Djobo District are illustrative of these perceptions: Early marriage allows a daughter to have the blessings of parents for respecting their will. She will also have a home and will be in accordance with the religious requirements that say a woman without a husband and home is not a good Muslim. If she prays, her prayers will be answered. She will be an example to her sisters and the girls in the community; Religion always teaches us to prepare our girls for marriage, instilling in them values that will help them to become good wives and good mothers and to be wives and women who obey the men and their spouses. Religion focuses on the place of women in the home so as to take good care of the children … [By contrast] school makes the girl rebellious with regard to men and may delay her marriage.
Similar perceptions were observed in Tenkodogo District, the other of the three districts with substantial support for early marriage for girls: "During the celebration of weddings in the mosque, religious leaders insist on the benefits of marriage and invite parents to quickly prepare their children for this new life that strengthens their faith and brings them closer to God"; "I know that among Christians, the girls go to school. They complete lengthy studies and work. But among us Muslims, what we prepare the girl for is marriage. The rest is unimportant"; "I'm not against girls attending school, but marriage is still the objective. It is the event that marks the end of the studies of a girl because our religion places marriage above all." A few respondents in the two districts went further to describe religious beliefs about women's subordination to men, the role of child marriage in this respect, and the danger of schooling. A respondent in Djibo District explained that "[The school] wants the woman to be the equal of the man. Remember, God created man first … According to us Muslims, the school wants to inverse God's order by giving women the same opportunities as men." Another respondent in Tenkodogo District stated: The place of women in society, if we follow Islamic prescription, is such that we are not interested in sending her to school. We speak of the woman in terms of obedience and submission, which makes us believe that her place is in the home.
Although this was less common in the interviews, some respondents in the two districts supporting child marriage stated that the age of marriage is fixed according to Islam (or that it is linked to menarche) so that it is a religious requirement for a girl to be married at this specific time. Quotes from Djibo District again illustrate those views: "I am against the marriage age of 17 because that is much older than 12 to 14 years, the age fixed by the Qu'ran and by our tradition"; "It is said that a girl must be in her marital home by a certain age, 14 years. If this does not happen, it is as though the parents do not respect the prescriptions of the Qur'an"; "If a girl begins menstruation, it means that she is a woman; she must marry if she is a good Muslim." By contrast, attitudes towards child marriage and schooling were very different in Bobo Dioulasso District, an area where the population is also Muslim, but closer to urban centers (Bobo Dioulasso is the second largest city in Burkina Faso, and the area visited for the fieldwork, while rural, was not located too far away from the city). The population in Bobo Dioulasso is much more supportive of girls' education, and thereby willing to delay the age of marriage for girls beyond 18. As a number of parents in Bobo Dioulasso District explained: "Before parents said that to educate a girl is a useless investment … but now with the many sensitizations on the radio and everywhere, we are starting to educate our girls"; "Some parents are beginning to educate their daughters because they say that they will have work when they complete their studies, and they can help their parents and help with the education of their brothers and sisters"; "If a woman goes to school and finds work outside the home like teaching, it is a good thing for the household"; "Before many people preferred to find husbands for their daughters rather than send them to school. But things have evolved; many girls are now in school and they are the pride of their parents." This does not mean that child marriage does not take place in Bobo Dioulasso District, but that it is less of a norm in that district in comparison with the other two districts. As given in Table 3, a majority of respondents in Bobo Dioulasso District suggested that girls should marry after 18 years of age, while very few were of that opinion in the other two districts. In practice, most girls still marry before 18 years of age even in Bobo Dioulasso District (the average age at marriage in that District is 16.2 years), but the shifts in public perceptions suggest that progress could be achieved relatively rapidly. The same cannot be said of the two other districts, suggesting that within a single country, there can be substantial heterogeneity in the perceptions of the relationship between faith, child marriage, and schooling for girls between different rural communities. Source: Gemignani and Wodon (2015a, 2015b) based on fieldwork data.
child marriage and faith affiliation in sub-saharan africa

Conclusion
Child marriage is declining in sub-Saharan Africa, but only slowly. Part of the reason for this is that the practice is deeply rooted culturally, and it is seen as recommended or even mandated by their faith in some communities. There are systematic differences in the likelihood of early marriage and the expected years of early marriage between faith affiliations according to data from Demographic and Health Surveys. However, at the same time, as illustrated by a case study for Burkina Faso, there is also a lot of heterogeneity in practices and beliefs between communities who share the same faith. This implies that policies and programs aiming to eliminate or curb child marriage must adapt to local circumstances -including those related to the specific understanding by local communities of what their faith has to say about child marriage. The role of local religious leaders therefore may matter greatly for the elimination of child marriage since these leaders are almost universally viewed as the most influential people in the lives of the communities and their spiritual and moral guidance has a profound effect on daily life.