1In countries severely affected by the AIDS epidemic, religious leaders may show a certain pragmatism in adapting to the realities of the disease, sometimes changing their doctrines and their teachings in order to protect those who are ill and limit the spread of the disease. Should a person with an HIV-infected spouse abandon his or her partner to avoid contamination, or stay to provide care? Men and women facing such dilemmas may turn to their religious leader for advice. Drawing on survey data and in-depth interviews, Jenny Trinitapoli examines both the perceptions of lay women living in rural Malawi, and the content of sermons on marriage and sexuality given by religious leaders of various denominations. In the context of the HIV-AIDS epidemic, religious doctrines can be used either to justify or to condemn divorce, depending on the health status of the persons involved, with the dual aim of maintaining social cohesion within the community and limiting the spread of HIV.
2Studies of religion since the onset of AIDS in Africa (around 1990) provide evidence of an increased emphasis on sexual behaviour in religious spaces across the region. HIV prevention has become central to religious life, with multiple studies confirming the active role of religious leaders in promoting AIDS-related behaviour change (Green, 2003; Krakauer and Newbery, 2007; Pfeiffer, 2004). In Mozambique, Agadjanian and Menjívar (2008) identify mainline Christian congregations as spaces that facilitate informal communication about AIDS and related issues. In rural Malawi in 2004, over 70% of religious leaders were preaching explicitly about AIDS on a regular basis and a similar proportion reported preaching regularly about sexual morality (Trinitapoli, 2011).
3In contrast to these abundant estimates of quantity, the nature of religious messages about sex and AIDS has been less fully explored. In sub-Saharan Africa, religious messages about marriage and family life pre-date AIDS, but the epidemic may have altered the ways in which these topics are engaged from religious perspectives. Given the high levels of uncertainty due to AIDS and other longstanding problems like food shortages, political instability, and spiritual insecurity, contemporary sub-Saharan Africa provides a clear example of “unsettled times” – historical periods in which ritual practice, doctrine, and ideological commitments can take on heightened salience. Swidler (1986) argues that during unsettled (or less-settled) times, the desire to enforce a new ethos can drive ideology, reflecting on the example of how Calvin reformulated Protestant doctrine in order to advance a new ethos of austerity and self-control. Extending this logic to the case of Africa’s AIDS epidemic, we expect to find doctrine being organized by an identifiable ethos that aims to address some of the challenges the epidemic has brought about with regard to family life.
4To examine how AIDS factors into religious teachings about family life, a case study approach is used, examining religious leaders in rural Malawi during 2004 and 2005 – the height of the country’s generalized AIDS epidemic. The first section of this article shows how religious leaders integrate HIV and AIDS into religious teachings on family life to illustrate how religion has been responsive to this demographic event. The second section extends recent research showing that Malawians increasingly use divorce as a strategy to prevent HIV infection and explores how religious ideas are invoked both to justify and to oppose divorce in the Malawian context. Arguing that religious teachings on marriage and divorce are the result of a combination of individual agency and institutional action, the third section examines some implications for advancing the study of religion (and culture more broadly) within demography.
I – Religion, family life, and divorce in Malawi
5Marriage in southern Africa has long been described as a “fluid” arrangement (Bledsoe and Cohen, 1993; Lesthaeghe, 1989). Historically, marriage and divorce have been social arrangements rather than legal ones, governed by customary law (Chimango, 1977; Wanda, 1988). Recognized marriages can take a variety of forms: civil, customary (chinkhoswe), religious, “by repute” or de facto; but civil marriages are uncommon. A household survey from 2000 estimated that the vast majority of marriages (approximately 77%) are chinkhoswe marriages (i.e. officiated by marriage mediators known as ankhoswe), with about 23% of couples having religiously officiated marriages (Malawi Law Commission, 2006).
6As with marriage, divorce has been almost entirely within the purview of local, traditional courts rather than the formal legal system (Wanda, 1988). Divorce is not new in Malawi; divorce rates have been high (i.e. between one-half and one-third of marriages) since the 1940s at least (Mitchell 1956). However, many scholars agree that the rules surrounding marriage and divorce have changed in response to AIDS. A historical comparison of the rhetoric surrounding marriage and divorce in Malawi during the 1940s and the 1990s documents approximately equivalent rates of divorce between the two time periods but identifies an important rhetorical shift from economic issues to questions of sexual impropriety (Kaler, 2001).
7More recent research suggests that divorce has become both more common and more acceptable in the minds of most Malawians, specifically when infidelity is involved (Reniers, 2008; Smith and Watkins, 2005). Not only has divorce risen generally, but the association between suspected infidelity and divorce in Malawi has increased since the visible onset of AIDS, suggesting that individuals (women in particular) increasingly sanction their spouses’ sexual indiscretions with divorce (Reniers, 2006). The existing literature on marriage and divorce in the context of AIDS focuses primarily on individuals, leaving questions about the role of local institutions in promoting, constraining, and otherwise managing the many changes occurring in response to the epidemic.
8Malawi is both highly religious and religiously diverse. Nearly 98% of rural Malawians affiliate with some church or mosque, and more than 60% report attending religious services at least once a week (Trinitapoli, 2009). Religious groups are the most common type of voluntary association in Malawi, and they play a particularly important role in women’s lives (Yeatman and Trinitapoli, 2008). Malawi’s religious diversity can be summarized by five broad denominational categories: Catholic, Protestant (i.e. Presbyterian, Anglican, Baptist), Muslim, African Independent Churches (AIC),  and Pentecostal or Revivalist traditions, including new mission Protestant denominations (i.e. Seventh Day Adventist, Church of Christ, Jehovah’s Witnesses). Muslims are heavily concentrated in the southern region, and mission Protestants in the northern region (home to the Livingstonia Mission).
9In this context, Christian and Muslim leaders may reinterpret doctrine in ways that advance the pragmatic goal of curbing the spread of AIDS by enforcing strict norms regarding sexual behaviour – particularly faithfulness within marriage. In fact, the epidemic provides an unparalleled opportunity for Christian and Muslim leaders to emphasize religious ideals concerning marriage and to advance a new ethos of faithfulness. To preserve their communities, religious leaders may also leverage their authority in ways that reinforce social cohesion in times of hardship. The historically high and rising levels of divorce in Malawi (Reniers, 2008), combined with widespread illness, food shortages, and high levels of early-adult mortality due to AIDS (de Waal and Whiteside, 2003) create a context of material and spiritual insecurity (Ashforth, 2005, 2011). Frequent death, divorce, and witchcraft accusations disrupt social relationships, threatening the very essence of a religious community. In light of this reality, the ethos of faithfulness may extend beyond the sexual realm to emphasize obligations that maintain critical relationships of social reciprocity to keep the congregation intact and growing.
10In primarily agrarian societies like Malawi, marriage organizes family and economic life by establishing a division of labour along traditional gender lines. Ethnographic data from Malawi places a good marriage at the forefront of “the good life” for both women and men. Women prioritize marriage to a faithful husband who is a good provider; men also desire faithfulness from their partners, expecting essential domestic support (i.e. caring for children, washing, cooking, fetching firewood) in return for their material contributions to the family (LeVine et al., 1994; Weisner et al., 1997). According to Chaves (1994), religious authority structures attempt to enforce order by controlling access to desired religious goods, such as marriage, and providing a supernatural basis for legitimizing control over these goods.
11Expanding religious authority is often described as an imposing force that structures the lives of individuals by limiting freedoms and setting rules (Bartholomew, 1981; Berger, 1999). But religion can also be viewed as a cultural resource that individuals draw upon in order to navigate critical challenges they face in their personal lives (Stark and Finke, 2000; Verter, 2003). In the face of AIDS, individuals may use religion both to navigate safe paths to marriage and to maintain stability in their families. Spronk (2009) documents an expanding pre-marital counselling programme among middle-class Kenyans that is based on Christian ideologies of modern personhood, complete with a lake retreat, an intense focus on faithfulness, and a curriculum featuring “best practices” for healthy communication.
12Evidence from Nigeria further illustrates the important role of religion in regulating sex within marriage. Christianity, in particular, provides a set of ideals for faithful marriage and practical mechanisms for enforcing them. For Nigerian men, religious settings both facilitate confessions of cheating and serve to enforce discretion in infidelity, if not actual faithfulness (Smith, 2009). Women appeal to Christian values to convince their husbands to end extramarital affairs; they often reach out to extended family members to intervene and admonish their husband’s infidelities and, in some cases, expose their husband’s affairs to their pastors or to their entire congregation. In other words, religious values, leaders, and communities have become increasingly relevant to how women navigate infidelities in their relationships and to the way in which men respond.
13Seen this way, religious teachings on family life are part of a bi-directional process, fuelled by both a demand for leadership on complex issues and a supply of religious leaders, eager to promote their understandings of what constitutes a good life on earth and in heaven. The official repertoire for HIV prevention focusing on Abstinence, Being faithful to one partner, and using Condoms (ABC) has expanded to include careful partner selection, HIV testing, divorce, and the active regulation of marriage (Reniers, 2008; Watkins, 2004). The following analysis seeks to establish that religious leaders in Malawi have responded to the epidemic by creating scripts for entering into marriage and by revising doctrines on divorce. It shows how this advice is structured by the pragmatics and organization of religious authority to achieve a broader set of goals that involves, but is not limited to, curbing the spread of HIV.
II – Setting and data
14HIV prevalence in Malawi is the eighth highest in the world, with an estimated 11% of adults currently infected. As elsewhere in sub-Saharan Africa, HIV prevalence rose during the 1980s and 1990s before stabilizing in the late 1990s, when new infections began to decline. In recent years, HIV prevalence has remained stable, thanks to increased access to anti-retroviral treatment (ART) and behaviour change (White et al., 2007). The data analysed here were collected in 2005, approximately one year before the rapid scale-up of free ART in Malawi. In this period, HIV-prevalence, AIDS-related mortality, and general anxiety about AIDS were close to their peak (Jahn et al., 2008; Watkins, 2004), but incidence (i.e. new infections) had already begun to decline (Figure 1).
AIDS trends in Malawi (1990-2012)
AIDS trends in Malawi (1990-2012)Note: Dotted lines represent timing of MLSFH survey rounds; solid line MRP data collection.
15A mixed-methods approach is used, drawing primarily upon survey and qualitative data from the Malawi Religion Project (MRP) to provide an account of how and why attitudes about marriage and divorce are relevant for understanding AIDS in this context. To triangulate and augment the data from religious leaders, three waves of the Malawi Longitudinal Study of Families and Health (MLSFH, 2001-2006) are also used to provide a parallel set of analyses from the perspective of lay women. 
16For the Malawi Religion Project (MRP) – a sister project to the MLSFH – data were collected at a single point in time (2005). The MRP is a mixed-methods study designed to assess how local religious organizations and their “moral communities” influence responses to the AIDS epidemic. The MRP sample was created using a hypernetwork sampling approach (McPherson, 1982). The MRP research team identified every congregation attended by respondents participating in wave 3 of the MLSFH (N = 200) and conducted both a survey and a semistructured in-depth interview with the primary leader from each congregation (N = 187; response rate = 93.5%). Leaders were asked to report on various characteristics of their congregation and offer their opinions about theological and AIDS-related issues. In the semi-structured interviews, congregation leaders talked at length about their formal and informal responsibilities, about the problems their communities face, and about AIDS. Interviews were digitally recorded and transcribed.
17One potential concern with the MRP data is that religious leaders may be motivated to present themselves as good leaders, claiming that, of course, they talk about AIDS, counsel members, and provide material assistance. In order to assess this potential bias, the MRP conducted between one and three interviews with ever-married lay women (randomly selected from MLSFH-3 respondents) from each congregation to which more than four respondents (religious leaders) belonged (N = 110 congregations). Religious-leader data was paired with the lay women data in order to assess over-reporting of behaviours that may be socially desirable in this context – activities like preaching about AIDS, caring for orphans, or sponsoring micro-credit endeavours. But the lay women/religious leader comparisons revealed high levels of consistency between leaders and members on the topic of AIDS.
18Finally, participant observation data are used from 116 religious services in 85 Malawian congregations, collected in 2004. A census of churches and mosques was conducted in the MLSFH’s field sites in Balaka and Rumphi and at least one observation was collected from each one. Trained Malawian research assistants observed religious services and wrote “sermon reports” on each congregation’s organizational structure, the service itself (with particular focus on the message of each sermon), and the congregation’s other activities. 
1 – Measures and methods
19Attitudes about the acceptability of divorce under a variety of circumstances were assessed based on binary (yes/no) responses to survey questions beginning with the prompt: “Do you think it is proper for a wife to leave her husband if…” We focus specifically on the responses to questions about the acceptability of divorce under the two circumstances that directly relate to the epidemic: infidelity and suspected HIV infection, examining particular combinations of attitudes and changes in these over time. Points of consensus and divergence between the two groups (i.e. religious leaders and lay women) and by religious tradition are highlighted. Based on the five categories used to capture the broad religious categories, the religious composition of the sample of married or ever married women is shown in Figure 2. Over the period 2001-2006, there are small increases among the Pentecostal/Revivalist and AIC categories, a slight decline in the Protestant sector, and relative stability of Catholics and Muslims (Cox, 2001; Isichei, 1995). At each wave, fewer than 10 women who reported “no religious affiliation” were excluded from the analyses presented here.
Religious composition of the ever-married lay women, 2001-2006
Religious composition of the ever-married lay women, 2001-2006
20The denominational composition of the religious-leader sample (i.e. MRP) corresponds closely to that of the female population, reflecting differences in average congregation size: Catholic (N = 21) and Muslim (N = 22) congregations tend to be large, while the Pentecostal and Revivalist congregations are smaller (N = 66) and the Independent (N = 38) and Mission Protestants (N = 40) somewhere in-between in terms of size and proportional representation in the MRP sample. 
21The MLSFH and MRP survey data are useful for providing a general overview of AIDS-related activities and attitudes. However, the idea that religious teachings are sensitive to demographic realities is rooted in the qualitative evidence from in-depth interviews with lay people and religious leaders and from the participant observation data collected from weekly religious services. The in-depth interview and sermon report data were coded according to a specified set of themes related to the role of religious organizations in addressing AIDS, including (but not limited to) formal prevention messages, family issues, informal counselling, condoms, marriage, divorce, HIV testing and counselling, healing and miracles, outreach activities, and caregiving endeavours. During the coding process, several other salient themes emerged, such as congregational discipline, current problems in the congregation and community, recent congregational scandals, and networks with other religious leaders and local non-governmental organizations (NGOs). 
22The qualitative analyses draw heavily upon references to marriage and family life, in particular as these themes overlap with actions, attitudes, and doctrines about HIV and AIDS both in terms of prevention for the uninfected and caregiving for people living with HIV and AIDS. For each theme of interest, coded passages were sorted and analysed by region and by religious tradition. The analyses of the qualitative data by religious tradition revealed few substantive denominational differences. 
23The results are presented in three parts. First, the qualitative data are used to show how religious leaders regulate marriage and family life in the context of AIDS and how lay persons are using religion as a resource for navigating the dangers of the epidemic. The second part identifies and describes a consensus on the acceptability of divorce – held both by lay persons and by religious leaders across religious traditions – in light of AIDS. The third part explains how these new views on divorce are rooted in local epidemiological understandings of the epidemic and argues that these beliefs and teachings about marriage and divorce represent a new ethos of faithfulness that structured changes to religious life in the era of AIDS, pre-ART.
III – Results
1 – Religious authority over marriage
24General references to AIDS – including sexual behaviour, suffering, and caregiving – are commonplace in religious services across Malawi (Trinitapoli and Weinreb, 2012; Trinitapoli, 2011). Religious leaders are leveraging the AIDS epidemic in ways that assert and expand religious authority over marriage in two distinct ways: first, by encouraging religiously sanctioned marriages over traditional ones, and second, by monitoring the sexual behaviour of members and actively intervening in marriages troubled by infidelity.
25The promotion of religious marriage is anchored in strategies that resemble what population health scholars recognize as partner selection or sero-sorting processes (Reniers, 2008). According to the pastor of a Presbyterian congregation in Mchinji district, “if boys and girls have fallen in love, the older members encourage them to go for blood test before kulowana [marriage] so that they can be assured their spouse is free of AIDS”. Similarly, the leader of an African Independent Church that allows polygamy advises HIV testing and counselling (HTC) before adding another wife to the family, along with several other recommendations for maintaining peace in a polygamous household:
In our congregation you are allowed to have mitala [two or more wives], but it should not be a habit to have many wives. And before marrying another wife, there should be a procedure to go for a blood test at the hospital. If the doctor finds that both the husband and the wife are okay, then they can marry. 
27Importantly, however, expanding religious authority over marriage is not limited to testing. Some leaders complained about the poor provision of HTC in rural areas, while others explained that the opportunity costs involved with travelling to a hospital or HTC facility (i.e. at least one full day’s work in 2005, when these data were collected) made testing unfeasible for most people. These leaders often emphasized the importance of a background check on possible spouses – especially those who are not from the community – as an alternate strategy for partner selection. In describing the congregation’s response to HIV in their village, the chairman  of one Catholic congregation spoke first about marriage:
We do teach each other nowadays there is a disease of AIDS, so playing with someone else’s wife is not good. For boys this is not allowed. If they want to marry a girl, they should go through the proper channels and see. If that girl does not have any other boy lovers, we will encourage him to marry that girl.
29Religious leaders’ emphasis on going through the proper channels is distinct from, but consistent with, insights from scholars who emphasize the important role of gossip in partner selection for avoiding HIV (Kaler, 2004; Watkins, 2004). “Because there is so much gossip in the villages, those who are considering a particular partner may already know a great deal about that person, and, if not, they search for information by asking others. Those who do not consult local knowledge are considered foolish” (Watkins, 2004, p. 689). The gossip networks Watkins emphasizes are comprised of friends and acquaintances who provide information that is essential for safely navigating the terrain of sexual relationships in a context of high HIV prevalence.
30Religious leaders are well positioned to facilitate the transmission of local knowledge, especially as they actively seek to expand their authority over family life relative to the traditional authorities – chiefs and traditional marriage mediators, known as ankhoswe. Whether one relies on fellow villagers, ankhoswe, religious leaders, or some combination of these for gathering background information on a potential partner, the general purpose is the same: to obtain both a sexual and health history and assess the level of risk involved with this new relationship. In fact, many religious leaders maintain that good counsel is better than being tested. Knowing that a potential partner is currently negative is important, but knowing the character of a potential partner – his or her likelihood of remaining faithful – is essential for remaining disease-free and cannot be determined by any clinical test. While they certainly are not the only source of such information (and may not be the best either), religious leaders and their networks offer an additional source of such counsel, and the qualitative data analysed here suggest that in many instances they are indeed being utilized as marriage brokers, and not just mere officiators.
31In addition to promoting religiously sanctioned marriage, AIDS has given religious leaders a platform for actively regulating it. Religious messages about faithfulness within marriage are even more prevalent than explicit messages about AIDS. Like many others, this Anglican pastor in Balaka district preached both against unfaithfulness (directed at men) and about how Christian women can and should respond to this problem:
The Bible tells us that when a Christian is in problems he/she has to kneel down and pray for Christ Jesus to intervene. These problems are like when a husband is not faithful to his wife. The wife has to pray in order for her husband to change the behaviour. Instead of doing this, most of the women seek help from traditional healers to give them some kind of love potion. This is the way through which Satan is destroying most of the families. Whenever we discover that our families are being destroyed by Satan, we have to seek God’s help. Believe you me, when you go to the unfaithful traditional healer for a love potion, the husband dies from the potion itself.
33Beyond emphasizing the power of prayer to change an unfaithful spouse, this pastor urges women in his congregation to bring “family problems” (i.e. unfaithfulness) to him so that he can intervene. In this way he elevates both God’s and his own authority over a competing alternative – love potions from a traditional healer.
34Analysis of the interviews with lay women reveals that religious leaders also regularly provide members with private counsel on AIDS-related issues. To illustrate, Agnes, a Muslim woman from Southern Malawi, discusses taking her AIDS-related worries to her sheikh, emphasizing the uncertainty involved with having a husband who travels often and is not trustworthy:
I first heard about AIDS on a radio and when I ever went to collect water with my friends they would say things about it. And I am very worried with this disease just because I can die with the disease anytime just because of my husband’s behaviour… I always chat about AIDS with my friends, and one of them is also worried of getting it from her spouse who is in Mangochi [a nearby town] but his behaviour is not good… I went to the sheikh when I found my husband having sex with another woman. My husband asked for forgiveness from me, and the sheikh told me to forgive him.
Our sheikh always preaches [about] the dangers of the disease. He also advises us to take care of the people who are suffering from the disease; we need to visit him or her so that she must not lose hope. She must have the idea that her fellow Muslims are still taking care of her. We go there to give support. But we are only concerned that the person will just stay [live] for a short time and die because of the disease. We always pray for the sick to have at least a longer life. Anyone can have AIDS. A sheikh, boy, girl, pastor, church elders, Christians. So we [Muslim women in her village] call the sheikh to a secret place and tell him to continue preaching about AIDS. So when we meet either during the week or on Friday he preaches about AIDS.
38This sheikh’s own report of his approach to counselling couples in instances of infidelity hinges on repentance. When he perceives true repentance, he counsels couples to stay together, but an unrepentant adulterer is a double-threat: first to his wife’s wellbeing and second to the leader’s authority. Like many other leaders in Malawi, this sheikh responds to requests from women in his congregation to keep teaching about AIDS in the mosque so that their husbands hear. Thus women like Agnes are using (male) religious leaders as a strategic resource to maintain their marriages and protect themselves from AIDS.
39The interview data from both lay women and from clergy themselves show that beyond formal messages and informal advice, Malawian religious leaders are taking on disciplinary roles in their communities. Such action is dually legitimized by biblical teachings on fidelity within marriage and the current threat of AIDS. A clear example of this comes from Patricia, who attends a congregation in which the leader reports taking an active role in monitoring the fidelity of members. Patricia explains:
Some from our congregation committed adultery and they were suspended. These people went to a funeral service. There was chimeto [hair shaving after death]. Their friends saw them committing adultery and reported them at the church. When they [the church leaders] asked them they were just quiet. Then the church suspended them… After a year, it’s when they called them back… They told them, “We have called you back, but if we hear from people that you have such a wrong for the second time we wont allow you again in your church, because you paint the church black. You paint the congregation black, doing wrong things regularly.”
41This instance exemplifies an approach to disciplining unfaithful members that appeared frequently in the MRP data – suspension until a public repentance, at which time the member is welcomed back. In addition, it shows that villagers call upon religious leaders to intervene in cases of adultery (as recommended by the Anglican pastor quoted above), thereby acknowledging religious authority over this sphere of social and family life.
2 – Divorce acceptability
Divorce in cases of infidelity
42In addition to expanding their authority over family life by promoting and regulating marriage in the context of AIDS, Malawian religious leaders tend to articulate a particular position on divorce. The MRP survey data provide a helpful overview of the AIDS-related messages religious leaders report giving and the prevention strategies they promote. Particularly relevant for understanding the role of religion in facilitating what Reniers (2008) calls “negative selection” (i.e. divorce of an adulterous spouse) is the fact that nearly one third of religious leaders (32%) have encouraged a member to leave a spouse in order to avoid contracting HIV (Trinitapoli, 2011). 
43Table 1 summarizes the attitudes of lay women about the acceptability of divorce, based on a set of questions about the acceptability of a wife leaving her husband under a variety of circumstances. That the acceptability of divorce in cases of infidelity was high and rising between 1999 and 2001 has been previously documented by Smith and Watkins (2005). Here, we see acceptance levels continued to rise (to 85% in 2006) as the epidemic progressed.
Views of acceptable reasons for a woman to divorce her husband
Views of acceptable reasons for a woman to divorce her husband
44Malawian religious leaders’ perspectives on divorce have not been documented previously in empirical research. Table 2 shows that a majority of religious leaders support divorce in the case of infidelity. Infidelity is by far the most acceptable reason for an individual to divorce a spouse – about 30 percentage points more than partner violence, the next-most-acceptable circumstance. Some religious leaders, like this AIC pastor from Rumphi district, tolerate divorce under such circumstances. “If she catches him, the Bible allows it. If she catches him, you ask him in the house: ‘Do you want to give this disease to me?’” Rural Malawians emphasize the need to catch an adulterous spouse “red-handed,” and the qualitative data shows that religious leaders tend to agree. Stories from Malawian villages involve elaborate schemes of individuals pretending to go out of town and sneaking back to catch a cheating spouse. Of course, as others (e.g. Kaler, 2001; Schatz, 2005; Watkins, 2004) have noted, one can simply leave or send a spouse packing – “take your mat and go”. But a divorce needs to be seen as legitimate in the eyes of others in one’s community and by the ankhoswe in order for Malawian divorcees to be eligible for remarriage in the eyes of their communities. When discussing issues of infidelity, religious leaders use phrases like “in the house” to emphasize the importance of first confronting a spouse privately – refraining from making accusations in public places where neighbours and other villagers would overhear.
Malawian religious leaders’ views of acceptable reasons for a woman to divorce her husband
Malawian religious leaders’ views of acceptable reasons for a woman to divorce her husband
45This AIC leader in patrilocal Rumphi advised his member, Stella, in ways that provide a clear example of what “encouragement to divorce” actually looks like, capturing some of the complexities of divorce in this context. Stella (a faithful church member) confronted her husband (who rarely attends) about his infidelities several times in the past before bringing the issue to her pastor. The pastor’s first action was to visit their house and talk to Stella’s husband about the dangers of AIDS. After perceiving no change in his behaviour, Stella went back to her pastor. She explained that she intended to leave her husband but was reluctant to move back to her parents’ compound as if she were a child. The pastor encouraged her to leave her husband but not to remarry hastily for financial reasons; the congregation could assist her in the short-term and she need not move back to her parents’ home. Unlike in the past, he explained, there are many single mothers (i.e. widows and divorcees) in their community now. Other religious leaders virtually mandate divorce in the case of infidelity, saying things like: “She should leave him; he could kill her” or “Yes. She must [leave him] if he is moving around with other women carelessly. Nowadays it is dangerous”.
Divorce of an infected spouse and the window of opportunity
46In contrast to the acceptance of and religious support for divorce under circumstances of unfaithfulness, the data on divorcing an infected spouse is more difficult to interpret. First, among lay women, acceptance of divorcing an infected spouse rose between 1999 and 2001 (12 percentage points) but then declined in subsequent years – to 29% in 2004 and to 24% in 2006 (see Table 1). Second, divorcing a spouse with AIDS receives the lowest level of support among religious leaders (Table 2) – more than 10 percentage points below leaving a spouse because of sexual dissatisfaction. These rather anomalous findings contradict both conventional wisdom and linear-type predictions about liberalizing attitudes to divorce in the area of AIDS, pre-ART. The qualitative data provide some clarity on this apparent contradiction.
47The in-depth interview data from religious leaders show that the primary reason given for opposing AIDS-related divorce is the mandate to care for the sick. One leader who was emphatically opposed to divorce under suspicion of AIDS emphasized the wife’s role in caring for her husband: “No! It is not appropriate! Who are you going to leave him with? Who would take care of him?” Similarly, a Church of Christ pastor emphasized that the only acceptable reason for a woman to leave her husband was infidelity, in which case “she must divorce him”. When asked, “what if she suspects he has HIV?” he replied, “No. Not to leave as to divorce but they have to be on separation [abstain from sex], if possible”. The religious leaders who invoked the caregiving mandate addressed husbands and wives almost evenly. The words of one sheikh concisely summarized the sentiment of religious leaders who opposed divorce over suspicion of AIDS: “He must continue to stay together because they were staying together, and she has catched AIDS while with him. If he drops her, who is going to take care of her? He must care for her”.
48A different Church of Christ pastor recounted having counselled three men who were concerned that their wives might have AIDS and were considering divorce. The case he described in detail referred to one whose wife had experienced sudden weight loss, rashes, and a general decline in health – all characteristic of AIDS. Basing his counsel on obligations to provide care and support, even in sickness, he summarized: “I talked with him, and he did not leave her. They are still members of the church”.
49A number of leaders insisted that suspicion of AIDS was insufficient grounds for divorce, but that a confirmed infection was a legitimate justification. “Do not divorce your spouse based on suspicion, but first go for a blood test.” Leaders offering such an explanation said they encourage both parties to get tested and to receive counselling to decide how to proceed. Several leaders expressed their concern that accepting divorce under AIDS suspicion could lead to an outbreak of unfounded AIDS accusations as grounds for divorce, which would wreak havoc on the community: “No. The husband is not allowed to leave the woman because there is not strong evidence. It is just a rumour”. This emphasis on proof: proof of infidelity (catching the spouse red-handed) and proof of HIV infection (through HTC) relates directly to community cohesion in a context where accusations and rumours pose a multitude of dangers (Ashforth, 2003).
50Given the reality of AIDS, religious leaders are demonstrating flexibility and pragmatism on the issue of divorce. This does not reflect a general liberalization of attitudes towards divorce but is specific to AIDS-related concerns and closely circumscribed. What emerges is a new doctrine – a “window of opportunity” – for a religiously legitimate divorce: after known infidelity but before infection. Tabulated by religious tradition, the MRP survey data provide evidence that the window-of-opportunity position is one of consensus across religious traditions (see Figure 3). With the exception of Catholic religious leaders, the majority of which oppose divorce under both circumstances (and, incidentally, under all others), the window of opportunity position is the modal position among religious leaders.
Religious leaders’ positions on divorce acceptability by religious tradition
Religious leaders’ positions on divorce acceptability by religious tradition
51The lay perspective from ever-married women provides some evidence that the window of opportunity position is reflected in the general population, across all five religious traditions (Figure 4). Among Catholic, Protestant, and Pentecostal women, we see a steady increase in the window of opportunity position over the three waves. Among Muslim women the prevalence of this view jumps between 2004 and 2006 and is attributable to a decline in the proportion of women who oppose divorce for both reasons. Among AIC women, the window of opportunity view is solidified earlier (between 2001 and 2004), and the acceptability of divorcing an infected spouse declines over this same period of time.
Attitudes towards divorce among ever-married women in Malawi by religious tradition, 2001-2006
Attitudes towards divorce among ever-married women in Malawi by religious tradition, 2001-2006
3 – Local epidemiology and the window of opportunity
52On its face, this consensus position on divorce appears contradictory: leave an unfaithful spouse, stay with an infected spouse. While the former can be easily understood as a prevention strategy, the later clearly places the partner at risk for infection. A closer examination of local epidemiological beliefs shows that this position is not as paradoxical as it first appears, given that Malawian religious leaders are balancing their local view of the epidemiological terrain among other salient concerns: asserting religious authority, maintaining community cohesion, and protecting individuals from infection.
53Contrary to epidemiological data suggesting that the likelihood of transmission for each act of unprotected sex with an infected partner is relatively low (Boily et al., 2009), most Malawians believe that HIV is easily transmitted through sexual intercourse. Indeed, once is considered enough. MLSFH data collected in 2004 reveals that 94% of survey respondents believed that if you have sex once with an infected person you were certain or highly likely to be infected yourself.  Local understanding of the epidemiology of HIV amounts to this: if one spouse is infected, so is the other. In the words of one Catholic leader: “If your wife is becoming thin, you yourself shall be thin soon. What you should do is live together as husband and wife. God will judge”.
54A person who is known to be unfaithful, or even suspected of being so, is considered dangerous. But once a person has AIDS, the situation is different. Divorce in this circumstance is not tolerated; abandoning a sick spouse in his or her hour of need meets strong disapproval. A few leaders expressed concern that sick individuals who are abandoned by their spouses might turn to suicide, but almost all of those who elaborated on their opposition to divorcing an infected spouse explained that the very essence of Christianity and Islam prohibits the abandonment of a sick person.
55The caregiving mandate was by far the most common basis for opposing AIDS-related divorce. However, the second most prevalent rationale can be described as a local public health initiative to contain the infection. Condom use remains generally low across the sub-continent and is especially rare within stable unions (Tavory and Swidler, 2009). Remarriage following divorce or the death of a spouse is common (Reniers, 2003), and abstinence is widely considered impractical, if not impossible, for adults; even religious leaders agree with this (Trinitapoli and Weinreb, 2012). A sizable minority of religious leaders in the MRP sample articulated the importance of making sure positive-positive unions stay intact: no divorce because “This is exactly how the disease spreads!” According to one Church of Christ pastor in Mchinji:
That [divorce] is not good because when one in the family has got this disease automatically the other one has it too. If we can accept it [divorce], it means now both people will spread a disease to other people. If we can say these things [accept divorce in this circumstance] it means we are not thinking properly because what you say to your friend, the same thing is following you and when one is sick, this is the time for his partner to show his/her love.
57An AIC leader from Balaka expressed a similar sentiment:
If a person wants to divorce a wife when she has AIDS. But if a wife is in such status [HIV positive] that means he too has the virus. So, if he divorces her and marries somewhere he is spreading the disease. So, we encourage these people that if you have this disease you should just stay. Be lovely in your heart. Because to us who are well, that people have diseases we do advise them to stay together.
59Accurate and inaccurate, local understandings about HIV risk and transmission colour the advice religious leaders give their members. In a context where access to HIV testing is scarce and treatment unavailable, tolerance of divorce depends on perceptions of the HIV status of both husband and wife, the potential for spread of HIV (through re-marriage and subsequent childbearing), and Christian and Muslim obligations to provide care for loved ones in their hour of need. Given the direct trade-off between people abandoning ill spouses and risking infection, this tension is a very real one. The Church of Christ pastor who recommended that couples stay together while abstaining (if possible) situates his advice within this contradiction, acknowledging that life-long abstinence is unrealistic and recognizing that African pronatalism and the semiotic meaning of condoms make their consistent use unlikely.
60Religious teachings on marriage and family life in the era of AIDS are complex. This complexity is the result of religious leaders’ struggle to balance competing mandates of abandoning neither one’s children (through death) nor ailing spouses (through divorce). Given all the trade-offs and pressures leaders face, their view is, in essence, that the major opportunity to stop the spread of HIV is to strongly discourage infidelity and to allow – even encourage – divorce of an unfaithful spouse. Not all go this far because for some (Catholics, in particular), divorce directly conflicts with long-standing doctrine. But despite these constraints, many have come to accept or even demand divorce when confronted with infidelity. On the other hand, the desire for community cohesion leads religious leaders to two other positions: first, that members should not divorce a spouse who is ill with AIDS, and second, that no one should start unfounded rumours (disruptive to community cohesion in a place where accusations of things like witchcraft constantly threaten community life) about another’s possible illness.
61This Pentecostal pastor from Malawi’s Northern district articulated the window of opportunity message clearly: “Yes. She has to do that, before the husband transmits diseases to her. But she has to do that only if she has enough evidence that her husband had sex with this woman. She has to do that before she gets any disease!” Others, from a variety of traditions, described their approach in similar terms, saying things like: “The Bible says that if a woman or man is caught in an adulterous act, he or she should be divorced”, while maintaining that the divorce must happen before infection occurs.
62Religious leaders are using their authority to chastise and discipline unfaithful spouses at the point where they are putting their families at risk. They strongly advise divorce in cases of incorrigible adultery. This view – that one can or should divorce at the point of infidelity but not on the basis of pure rumour and not after the partner is ill – is designed to serve several purposes: to reinforce sexual morality and the moral authority of the church over marriage, to protect individuals biologically while also protecting the religious leaders’ social claims on their members, and to protect community cohesion by insisting that people care for the ill and avoid spreading disruptive rumours.
63The “window of opportunity” doctrine is rooted in local understandings of the epidemiology and the theodicy of AIDS and hinges on an ethos of faithfulness – sexual faithfulness within marriage and faithfulness in death. The Biblical verse “Be faithful even to the point of death, and I will give you the crown of life” (Revelation 2:10) has traditionally been interpreted as emphasizing faithfulness to God, but the sermon report data show Malawian religious leaders reinterpreting this passage to esteem the onerous task of caregiving, which is disproportionately borne by women (Chimwaza and Watkins, 2004). In the context of AIDS, faithfulness has become the principle by which religious leaders are organizing both congregational and social life more broadly in Malawi. Importantly, the collective dimension of this doctrine has a dual purpose: to limit the spread of disease and to manage the caregiving burden within the community.
IV – Discussion
64This inquiry into religious teachings on family life moves us away from what would be considered traditional territory for the fields of demography and population health and towards questions of meaning in relation to demographic events. The findings highlight a drawback of the current state of knowledge on religion – indeed, culture more broadly – and demographic processes. Considerable emphasis has been placed on religion as an independent variable, predicting outcomes like marriage, divorce, sexual behaviour, fertility, mortality, and health. However, these relationships are reciprocal ones: religion is also influenced by these demographic processes. In examining religious responses to AIDS in Malawi, it is clear that religious teachings about family life, particularly marriage and divorce, reflect a new reality. While teachings about sexual morality in the pre-AIDS era were rooted in ideas about self-discipline and individual morality (Ross, 1995), in the era of AIDS these are couched in broader narratives about family and community obligations.
65The contours of religious life in Africa can and should be viewed as both cause and consequence of the broader changes underway in the region. On one hand, religious teachings on sexual behaviour, marriage, divorce, and caregiving influence how people navigate the AIDS epidemic in sub-Saharan Africa (Trinitapoli, 2009; Watkins, 2004). At the same time, the evidence presented here suggests that religious beliefs and practices are reflective of the broader demographic and epidemiological patterns evolving in the region. Both the survey and in-depth interview data from Malawi show that religious leaders are adapting doctrines and practices surrounding marriage and divorce using a pragmatic approach that aims to assert religious authority, maintain stability in communities, and curb the spread of HIV within their local epidemiological understandings of the virus. The survey data from lay women suggest that a similar perspective has also taken root in the general population – whether as a reflection of the teachings or as a consequence of them, we can’t say for sure.
66In Malawi, religious messages and doctrines reflect broad cultural values that include, but are not limited to, the physical health of individuals. These messages are out of step with the western biomedical approaches to HIV prevention that focus on individual-centred understanding of risk and prevention. In discussing the rules surrounding AIDS and divorce, religious leaders tend to emphasize community wellbeing, broadly defined, over the health of any particular individual. For example, by encouraging presumably positive-positive unions to stay intact with the goal of preventing the further spread of the disease in the community, religious leaders may be sacrificing the health of some individuals (i.e. actually encouraging an uninfected person to stay with an infected partner) as they seek to prevent sick spouses from being abandoned and to prevent the further spread of the disease through remarriage, which is inevitable in this context. Here, I have focused on the meaning behind these religious messages and on their intended functions, however it is important to note that the actual epidemiological consequences of these doctrines and teachings remain unknown.
67The example of religiously based health advice in Malawi offers four key insights into how population processes shape social life generally and religion in particular. First, this case from Malawi stands in contrast to the convention within both demography and the sociology of religion of estimating the associations between religious identities, beliefs, or practices and various outcomes. The emerging doctrine of acceptable divorce in the context of AIDS highlights the importance of looking beyond individual-level religiosity to understand religious change and religious trends. Malawi’s generalized AIDS epidemic offers an example of some conditions under which population processes may invoke religious change. This angle on the relationship between demographic processes and cultural change is a critical one (Bachrach, 2014), yet it goes undetected by studies that focus exclusively on the religious sphere and ignore the population processes that surround it.
68Second, rather than understanding religion as a force that shapes family life by providing sets of directives, ideals, and doctrines to follow, the Malawian example points to religion as a resource that individuals draw upon in strategic ways to navigate their personal lives. Women like Agnes rely on their religious leaders both for general advice and for private counsel on their AIDS-related concerns. Those privy to their neighbour’s adultery take the issue to their religious leader to handle. Religious leaders in Malawi are eagerly responding to the demand for moral and practical guidance and leadership on these issues. This insight from Malawi is supported by parallel findings from Nigeria (Smith, 2009) and Zanzibar (Stiles, 2003), and it serves as a caution against over-institutionalized perspectives on religion that assume a top-down model of influence.
69Third, these findings challenge facile assumptions about what religious leaders believe about marriage and divorce (i.e. marriage good, divorce bad) and underscore the importance of considering the nuanced theological and practical positions that religious leaders “on the ground” actually hold. The development of a religiously legitimate path for dissolving relationships that put individuals at risk for contracting HIV is a novel and pragmatic innovation. It is a direct response to AIDS. Hesitations about this doctrinal turn are not rooted in religious ideas about the sanctity of marriage, interpretations of biblical texts, or established doctrines; they too stem from pragmatic concerns about the chaos that could result from false AIDS-accusations being used as grounds for divorce. From the qualitative data leveraged here, this task of juggling competing practical concerns (e.g., HIV prevention and social cohesion) is surprisingly void of particularist doctrine or theology. The emerging doctrinal solution is rooted in a broad ethos of faithfulness and reflects earthly goals of curbing the impact of AIDS and promoting social cohesion. The formulation of this doctrine might be described in evolutionary terms as an “adaptive trait” that could ensure the reproductive success of the congregation by minimizing deaths, supporting marriage and childbearing, and maintaining social cohesion. 
70Fourth, the Malawian example highlights the ways in which religious leaders balance the goal of curbing the spread of HIV with a broader set of interests and responsibilities. The AIDS epidemic has provided an environment in which religious leaders are well-positioned to expand their authority over family life by making themselves central to the processes by which individuals access and maintain fundamental aspects of a “good life” in this society: access to a safe and stable marriage and to a religiously legitimate script for exiting from a dangerous one. This example of the reinterpretation of religious doctrine in the interest of promoting a broader ethos may be applicable to understanding changing beliefs and practices in other contexts characterized by high levels of uncertainty.
71Despite its contributions, this study suffers from several limitations. Obviously, data from religious leaders over time would be optimal for documenting changing attitudes throughout the course of the AIDS epidemic, but such data simply do not exist. The argument that religious teachings are dynamic and responsive to demographic realities further implies that future research must examine how these changes unfold over longer swaths of history. It is unknown, for example, whether the consensus positions identified here will persist or evolve further in the face of new HIV-related technologies, such as widespread access to anti-retroviral therapy, which has the capacity to transform HIV from a death sentence into a manageable, chronic illness. Malawi’s high levels of religious diversity make it an ideal setting for identifying consensus across religious traditions; however, whether these teachings are unique to Malawi or also present in epidemiologically similar but religiously distinct contexts (i.e. Zambia, Mozambique) remains unknown.
72To summarize, in the face of AIDS, Malawian religious leaders have been promoting a particular set of marital scripts, and lay Malawians have been invoking religion, an existing cultural resource, to address their AIDS-related concerns. Doctrines about marriage and divorce are being organized around an ethos of faithfulness to promote family and community stability, the basis upon which religious authority rests. The AIDS epidemic’s impact on religious teachings and authority in Malawi provides just one example of how the interplay between demographic processes and religious change is a reciprocal one. This insight extends beyond the problems of AIDS and beyond sub-Saharan Africa. Demographers and scholars of social change will know where to look for evidence that population dynamics influence religious life across a wide variety of settings.
AcknowledgementsThree anonymous reviewers provided fair and lucid critiques of this work, and the Population editors gave strong editorial guidance. I’m grateful to them for seeing the merit of this unconventionally demographic paper. Earlier versions of this paper were foisted upon a lot of readers; Jimi Adams, Lauren Bachan, Michelle Frisco, Sarah Hayford, Molly Martin, Georges Reniers, Emily Smith-Greenaway, Ann Swidler, Susan Watkins, Alex Weinreb, and Sara Yeatman commented on previous iterations. The MRP data were collected by a wonderful team of 20 young Malawians, including Chikondi Singano and Fanizo George, who are no longer with us. RIP.
The Malawi Religion Project was supported by a grant from NICHD (R01-HD050142-01). The Malawi Longitudinal Study of Families and Health (MLSFH) has been supported by the National Institute of Child Health and Development (grant numbers R03 HD058976, R21 HD050652, R01 HD044228, R01 HD053781), the National Institute on Aging (grant number P30 AG12836), the Boettner Center for Pensions and Retirement Security at the University of Pennsylvania, and the National Institute of Child Health and Development Population Research Infrastructure Program (grant number R24 HD-044964), all at the University of Pennsylvania. The MLSFH has also been supported by for pilot funding received through the Penn Center for AIDS Research (CFAR), supported by NIAID AI 045008, and the Penn Institute on Aging.
University of Chicago, United States. Correspondence: Jenny Trinitapoli, University of Chicago, United States, email: firstname.lastname@example.org
The AIC category represents a wide range of congregations that have broken off from a mission church, usually over issues related to polygamy or other cultural practices. “Traditional” and “new” Mission Protestant churches are distinguished according to the era when missionaries first began work in Malawi. Traditional Mission Protestant churches (i.e., Anglican, Presbyterian) arrived during the latter half of the nineteenth century, while New Mission Protestant churches (i.e., Church of Christ, Seventh Day Adventist, Jehovah’s Witnesses) arrived in the early twentieth century.
We focus on women’s attitudes about divorce for two reasons. First, women’s perceptions of HIV-risk and their related opinions are closely tied to family life, whereas those of men are more heavily informed by their own behavior and concerns about their extra-marital partners (Kengeya-Kayondo et al., 1999; Smith and Watkins, 2005). Second, the MLSFH began with a random sample of ever-married women and their partners (who do not constitute a random sample). Because of the study design, generalizing about men’s views from these data would be unwise.
Variability in the sermon report data could reflect differences in the actual services, but they may also reflect the ability or idiosyncrasies of the report-taker. Given that all the report-takers were residents of the study sites, were selected from a large pool of trained interviewers, had similar levels of education, and received the exact same training about how to conduct the sermon reports, there is no reason to believe that interviewer effects in the sermon report data would be any greater than interviewer effects on any other type of data.
Details about the MRP data and sample can be found in Adams and Trinitapoli (2009) and Trinitapoli (2011) and are not reproduced here.
Two research assistants coded a total of 10 randomly selected interviews (five each) to provide confirmation of these themes. Across the three coders, intercoder reliability was 0.80 overall and 0.89 for the AIDS-related themes.
The observed lack of denominational differences has been noted elsewhere (see Trinitapoli, 2011). Although this is a ripe area for future research, it is not explored in detail here.
It is reasonable to be skeptical about leaders’ claims that they advise HIV testing: the MRP data were collected just after the Malawian government began to strongly promote HTC in conjunction with efforts to expand access across the country. Public health efforts to scale-up HTC and encourage voluntary uptake have targeted trusted community leaders, including religious leaders, in service of this goal.
Chairmen and church secretaries were frequently key informants for the MRP. This was especially true for Catholic congregations. The shortage of priests in Malawi and elsewhere in the world is dramatic. In the Southern diocese of Malawi, one priest is in charge of 80 parishes. Each of these is run by a chairman/chairlady and visited by the priest about once a year.
This message is more common than the promotion of condoms (26% in informal counseling), though less prevalent than encouraging members to get tested for HIV (66%).
Analyses not shown, but these data are publicly available from the MLSFH website: http://www.malawi.pop.upenn.edu/
This hearkens back to both Durkheimian functionalism and more recent evolutionary adaptations of it, such as Wilson (2002).