Sources of behavioural data In 1998 the Malawi Diffusion and Ideational Change Project (MDICP) interviewed 1541 ever-married women of childbearing age (15-49) and 1065 men (husbands of those women who were currently married) in three culturally and demographically distinct rural areas of Malawi: Balaka in the south; Mchinji in the central region, and Rumphi in the north. The sampling procedures are described in detail elsewhere (www.pop.upenn.edu). Interviewing took place in a language with which the respondent was comfortable, which in most cases meant the
major local language of the region. A follow-up survey (MDICP2) conducted in 2001 sought more detailed information on sexual partnerships and marriages and on sexual activity outside marriage. Respondents were asked the calendar years of the beginning and, if applicable, the end of each of their marriages, and then provided more detailed information on their current or most recent marriage, their previous marriage, and their first marriage. We concentrate in the empirical analysis on the experience of women, since they are the focus of the paper.
The reason some marriages ended is not known in cases where a woman had been married more than three times: a woman with four marriages, for example, did not provide detailed information on her second marriage. Nevertheless, this convention resulted in very little missing data since few women reported more than three marriages. Multiple marriages are most common in the south and least common in the north.
Our interpretations of the empirical statistics derived from the MDICP2 survey are informed and indeed enriched by various types of qualitative data collected as part of the overall project. A few of the best interviewers from the first round of the MDICP were asked to keep observational journals in which they recorded conversations about AIDS that they had either overhead or participated in, although they were not to instigate such conversations nor to add to their journals until they could be alone. The journal keepers quoted here all live in the southern region of Malawi. The journals, written in English although the vast majority of conversations would have been conducted in a local language, shed light on current areas of investigation and suggest new avenues of research. When quoting from the journals we employ pseudonyms for the writers and have obscured the names of people they mention. We have edited the text only very lightly, either to enhance its readability where the meaning may be unclear, or to correct those gross errors of English grammar that, by misrepresenting the native linguistic skills of the conversationalists, may downgrade the substance of their conversations.
What is marriage? What is divorce?
The MDICP2 survey questionnaire implicitly defined marriage to include cohabitation, the English version of the first question on women’s current marital status taking the form, “Are you now married or living with a man, or are you now widowed, divorced, or are you no longer living together?”. This formulation, and the use of the local-language equivalent of “husband” to describe a man married to or living with a female respondent, appeared to cause respondents no difficulty. Nevertheless, what respondents understood by “marriage” appears to be quite diverse, and whether or not short marriages and cohabiting unions were under-reported, or even over-reported, cannot be known with certainty. It is conventional wisdom that rather short periods of marriage or cohabitation may be under-reported in sample surveys and, indeed, some surveys seek information only on those unions that endure for some minimum duration.
Insights can be gained from both the empirical and the qualitative data. Of the 927 marriages reported in the MDICP2, seven marriages were reported to have lasted one week (recorded as seven days), and one to have lasted three weeks (recorded as 21 days). (All of these marriages ended in divorce.) Moreover, a further 61 marriages were reported to have lasted only a matter of months, and twelve, indeed, were reported to have lasted for exactly one month. Very short marriages are mentioned also in the observational journals.
For example:
After some months, he married a certain woman who he stayed with for 3 months and he divorced her because she was a CCAP member and her parents refused [to let] her to join the new Salvation Army and another reason was that before he became a pastor, he was working for the ministry of agriculture as a field assistant. When he became a pastor, he was told to stop working and depend on preaching God’s words.
When his wife’s parents heard that, they just told their daughter to divorce.
In referring to the marriages of a less respectable woman, whom he calls a prostitute, another recorder describes what sounds like a series of short affairs with commercial elements:
This woman is our neighbor in our village and this is the only thing that she does to earn her living and even her family depends on her. She has been doing this since 1990. She has been married about six times but the marriages take only three to four months and she divorces because she is used to moving around and she feels bound to move around if she has a husband at home, that’s why she divorces.
It is natural to wonder whether rural Malawians define marriage or even cohabitation somewhat differently than we, who tend to view the latter as the culmination of a process whereby one partner spends an increasing proportion of nights at the other’s residence, and leaves an increasing number of personal possessions there — first a toothbrush, then some clothes, and so on. More relevantly, many respondents may also have defined marriage differently from their own elites, as evidenced by the following extract from one of the project’s journal keepers. She is describing a radio programme in which “some important people” — a Muslim cleric, an Anglican cleric, and a number of NGO officials — were asked their views on whether someone with AIDS should be permitted to marry. The clerics had similar views of marriage; below is that of the Muslim:
… it needs a long process for one to get married, [an individual’s] parents are concerned, his or her relatives are also concerned, the nkhoswes (councillors) are concerned, the sheikhs and pastors at the mosques and churches are concerned, the Government through the Courts and the District Commissioners also are concerned if you want to register your marriage.
Just as marriage may be formally defined, so may divorce. A man separated from his wife told another writer:
We have not yet divorced because the marriage was at our Roman Catholic Church and we are waiting for our church elders to discuss the matter and after that we will go to the court of law for a final divorce.
In contrast, divorce may also be dramatically simple. Of a married woman found having sex with a third party, the same writer records:
The woman told her husband that she was going to the garden to fetch some vegetables there, and when her husband saw that she was late, he decided to follow her and he caught her in action with the pastor. The incident was heard in the village so that the pastor is not going to his church since this happened. And the wife who was living Chitengwa was chased away marking their divorce.
In the following journal extract the woman whose conversation has been recorded appears to define a sexual union as a marriage if the man lives with her, even for a short time; by extension, she calls the end of such a union a “divorce”:
But Abiti A_ said that she is not married but she has a boyfriend who gave her a newly born.
And in addition to that Abiti A_ complained that two men came to her house to propose her for the marriage. The first man stayed with her as a marriage for three weeks and then she divorced him. The second man stayed with her for one week and she divorced him. After a few months came another man but not for marriage. He was just her sexual partner6 and he made her pregnant which she bore a newborn daughter.
The apparent ease with which many people move into and out of marriages is reflected in confusing stories, such as the following, in which an individual moves from one spouse to another and then returns to the previous one. Such stories appear not only in the journals but, in skeletal form, in interviewer jottings on MDICP2 schedules:
E_ asked Abiti A_ to explain how a woman can get married to two men at the same time. Abiti A_ explained that M_ was first married to a certain man from Blantyre called Mr. C_. She gave birth to a child with that man and their marriage ended. After some months, she then got remarried to a certain man from Chapoila village and she went to South Africa (Jubeck) with him. While they were there the woman was divorced and she came back home. She had given birth to two children with her second husband. After some time she went back to South Africa alone for the job and she was working there. When she met with her [second] husband who had divorced her, they remarried again and after some time, M_ came back home and she went to Blantyre to see her first husband who she gave birth with to one child and agreed to remarry again. The man from Blantyre came to his wife’s home and employed piece workers to dig in his wife’s land and then he went back to Blantyre where he was working. After some days some properties came from South Africa including money from her [second] husband sending her that she should be going back to South Africa.
The diversity of marriage is evident. It may be of the most recognizably old-fashioned kind — not that we feature such marriages in our selection of journal extracts — or a brief affair of a matter of a few weeks. In analyzing the marriage histories we take as marriages those that were reported to us as such. In addition, since it is clearly inappropriate to distinguish between divorce and separation, we take “divorce” to have occurred when a marriage is said to have ended but neither spouse has died.
We recognize that that there may be slippage in either direction. It is possible that some short, possibly early, marriages may have gone unreported, especially by older respondents, although such unreporting could not have been common since ages at first marriage were rather low. On the other hand, some reported marital unions may correspond merely with the loss of virginity, while others may have involved only brief or token co-residence. It may be useful to think of a continuum ranging from a one-off sexual encounter — a “hit and run” in journal terms — to a union approved by local notables and blessed by a pastor or imam. Somewhere along this continuum lies the point at which a particular respondent says “this is a marriage”, or “this is not a marriage”, and the precise point may well differ from person to person. It is thus important to bear in mind the reason that we are interested in marriage in the first place, which is that it is a proxy for sexual behaviour. We expect women to be more sexually active within marriage than outside it, but to have fewer sexual partners.
Perhaps the MDICP2 respondents would appreciate such a functional definition of marriage if it were couched in the appropriate language.
Reasons for divorce
The MDICP2 the questionnaire did not seek the reason a particular marriage ended in divorce.
As a result, we cannot tell in individual cases which partner instigated the separation, or why. For a general picture we therefore turn once again to the observational journals, in which various reasons are mentioned. One is conflict over religious affiliation, as appeared in the first journal extract we quoted. Another is apparent infertility, in this case of the husband:
She stayed with him for three years and divorced the marriage because Mr M_ was a barren man. He was not bearing children with her therefore our grandparents were not happy about that and they told my Aunt to divorce her husband.
Another is drunkenness:
I_ said that her husband told her that she should stop drinking and if he will once hear that she is still drinking beer, she will be divorced.
Like anywhere else in the world where divorce occurs, a common catalyst is infidelity. Such infidelity may be merely suspected or it may be proved, as in the journal extract in which an adulterous woman was driven away from her husband’s compound. Infidelity of a husband may not be a sufficient cause for divorce on its own since it may carry a financial penalty for the wife: the example below suggests that, although the man implies he was the instigator of the divorces, the man’s wives had taken matters into their own hands by seeking new partners who would be better providers:
K_ told me that he got married two times but divorced these two marriages and the reasons are that he doesn’t care for the wife properly and the wives do go out of his hands because his system of not sleeping at home when he receives his pay and instead he sleeps with bar girls at the rest houses until the money runs out.
The importance of infidelity as a reason for divorce is highlighted not just in the journals but by the responses to some general questions posed by the MDICP2. Respondents were asked whether they thought it was proper for a wife to leave her husband under the following five circumstances (in the order posed): if he did not support her and the children financially, if he beat her frequently, if he was sexually unfaithful, if she thought he might have AIDS, and if he did not allow her to use family planning. Table 2 presents the proportions of women approving various justifications for divorce, ranked in order of the level of approval.
Of the five hypothetical situations, the most commonly approved justification for divorce was a husband’s sexual infidelity, with approval ranging between 78 per cent in the south to 66 per cent in the north. Wife beating followed fairly closely behind, but the remaining three justifications gained less approval. Most significantly for the present paper, between two-fifths (south) and one-quarter (centre and north) of women approved of a wife’s leaving her husband if she thought he might be infected with AIDS.
The five sets of proportions are clearly differentiated by region, with greatest approval of divorce for one of the stated reasons in the south, and least in the north. In contrast, the proportions are not clearly differentiated by current marital status, number of times married, or, with two exceptions, current age (not shown). As a result, it is not possible to infer from these data that the expressed opinions either derive from the respondents’ own experience or influenced that experience.
There are insights into change in the approval of various grounds for divorce by matching respondents in the original 1998 survey and the 2001 follow-up survey.
The proportion of women reporting that it was proper for a wife to leave her husband if he was unfaithful rose between 1998 and 2001 from 68 per cent overall — they do not distinguish by region — to 76 per cent. Thirteen per cent of women changed their answer from “Yes” to “No” over the three-year period, but fully 21 per cent moved the other way, from believing divorce was unjustified in such circumstances to believing it was justified. A similar hardening in attitudes is apparent from responses to the companion question that explicitly mentioned AIDS. Although women who reported that a woman was justified in leaving her husband if she believed he was infected with AIDS remain in the minority — 18 per cent in 1998 and 29 per cent in 2001 — the proportional increase over the three-year period is marked. Of the women who changed their minds between the surveys, two out of three evinced a new acceptance of divorce in the case of a husband who was probably AIDS-infected. In contrast, the proportion of women reporting wife-beating as sufficient grounds for divorce did not increase between the surveys.
On the basis of this contrast, some argue that tolerance of divorce in general did not increase between the surveys, but tolerance of divorce to rid oneself of an unfaithful husband, and as a strategy to protect oneself from infection with HIV.
The journals, too, make it plain that divorce in the case of a straying husband is not merely a reaction to the evidence of his infidelity, but may be a strategy to avoid being infected oneself.
In the following extract a friend of the writer has contracted gonorrhoea from her husband, who had slept with another woman:
My friend told her husband that it was better to divorce because he will get the AIDS disease and she will be infected. She also said that this time she was lucky because she received treatment and she was sure that she will recover. But if it will be AIDS, she will not receive any treatment that will help her to recover, therefore it is good to divorce their marriage before it was too late.
Divorce as a means of protecting oneself from contracting AIDS from an unfaithful spouse is a strategy not just of women, but of men. For example:
One day in November, 2001 I went to a bread baking factory to buy some bread there, and I met some men on the factory gate who were also waiting to buy some bread, we were about five men and while there at Portuguese bakery at Chirimba in Blantyre, there came a certain woman who was looking unhealthy (sick) and one man whom I don’t know his name said that woman could be suffering from AIDS, he further said that this disease is claiming many lives of people, he said that he divorced his wife two years ago because he was afraid that she could infect him with AIDS. He said that this decision came into his mind when he realized that his wife had started having sex with her extra-marital partner while they were under post-partum abstinence. He said that he instantly divorced her the same day he noted that, and that he is not thinking of getting remarried because he is afraid that he might face the same problem and get AIDS. And another man said that these days of AIDS, when your wife is sexually misbehaving you better just divorce her and live alone without any sexual partner the whole rest part of your life. The gate was then opened for us so that we entered into the factory before the other men said their ideas on that.
These men do not see just divorce as a protection from AIDS, but divorce followed by subsequent abstinence. Certainly, they say they are apprehensive of remarrying.
Our final journal extract concerns a case in which a man excites disapproval for divorcing and remarrying too often:
A certain man who is rich in Dowa District failed to attend his daughter’s funeral at his home, fearing that his relatives were going to accuse him about his system of divorcing women frequently. The relatives said they were ready to accuse him of this because nowadays there is AIDS and if he keeps on changing women he might get it. To avoid all this they said they want to talk with him to change his system so that he should be safe from getting AIDS, as the changing of women can be one way that he might catch AIDS. The man has maize mills and cars. He is married to a fifth wife currently.
As his critics recognize, this man’s behaviour may be risky. Divorce can be protective only if the individual who divorces a suspect spouse does not form sexual relationships with other suspect individuals, whether these be affairs or marriages.
Without another empirical investigation it is impossible to quantify the extent to which some divorces in rural Malawi now occur not just because of a spouse’s infidelity but because of the fear that such infidelity poses a risk of infection with HIV. Certainly, the journals give the impression that neither the fear, nor such divorce, is rare, while the quantitative data derived from the MDICP2 indicate the increasing acceptability of this justification for divorce. One of the questions we investigate later in the paper is the extent to which the strategy of divorcing a spouse who may be infected may be successful.
Levels of divorce and remarriage Table 3 presents the marital-status composition of women in the three surveyed regions, according to age, as reported in the MDICP2. In each regional sample, around nine out of ten women are married, and divorced or separated women are about twice as common as widows.8 As was to be expected, widows are more common among older than younger women, but an age pattern among separated and divorced women is less clear.
Since the static picture created by such statistics reflects only dimly the dynamics of divorce because divorced women disappear from the divorced population when they remarry, we next estimated life-table survival functions of remaining in a first marriage. The underlying life tables were calculated in the conventional way, censoring on widowhood or, if an individual was still married to her first husband at the time of the survey, at interview. First marriages with unknown start or stop dates, numbering 57 in the south, four in the central region and a mere one case in the north, were perforce excluded from the analysis.
The calendar years delimiting each marriage, although sought, were certainly not easily obtained. In many cases they were calculated by interviewers from reported ages at the event in question, and since a degree of error is to be expected more attention should be paid to the broad features of the following analysis than the fine detail.
Divorce is indeed common in rural Malawi, and divorce patterns are clearly region-specific. First marriages are least stable in the south and most stable in the north, which mirrors the regional gradient in approval of divorce (Table 2). As many as 26 per cent of first marriages in the south end in divorce within five years, compared with 19 per cent in the central region and 14 per cent in the north. By the tenth anniversary these proportions rise to 40, 30 and 25 per cent respectively, and by the fifteenth to 49, 38 and 31 per cent.
Table 4 shows the proportions divorcing by the time of selected anniversaries not just overall for each region but by marriage cohort. Not only is this a fairer way to examine the time to divorce, given that only women who have been married for a considerable period can contribute to the survival functions at the long durations, but we were curious whether any recent rise in divorce — possibly as a response to HIV/AIDS — was detectable in the MDICP2 data. An apparent recent rise is detectable in the south where the proportion of women divorcing within five years rose from 24 per cent among women who first married between 1982 and 1991 to 36 per cent of women who first married between 1992 and 2001, and also in the north, with comparable figures of 14 per cent and 21 per cent, but in each case the numbers are small. No such trend is evident in the central region.
Table 5 reveals that just as women in the south are the swiftest to divorce — or be divorced — so are they the swiftest to remarry; likewise women in the north, the slowest to divorce, are the slowest to remarry. No trends were evident when separate life tables were estimated according to the calendar period during which the first marriage ended. Indeed, remarriage rates appear to have been rather constant over recent decades..
To sum up, just as the regions differ according to the timing of first marriage, so do they differ according to their patterns of divorce and remarriage. Of the three regions, marriage comes earliest in the south, divorce is the most likely, and remarriage is the swiftest. In contrast, marriage comes latest in the north, divorce is the least likely, and remarriage is the slowest.
Extramarital sexual activity
In the detailed marriage-history section of the MDICP2 questionnaire women were asked whether they had had a boyfriend while they were married to their current or most recent husband, their previous husband, or their first husband, or had had sex with anyone else other than that husband. We first tabulate the responses using each marriage as the unit of observation and categorizing according to whether the marriage in question ended in divorce or the husband’s death, or was still current, in order to see whether affairs were reported more often when the marriage in question had ended in divorce (Table 6). Overall, the question elicited few positive reports: affairs were reported in not quite four per cent of marriages in the south and not quite two per cent of marriages elsewhere. Women were indeed more likely to report that they had been unfaithful to an ex-husband than to their current husband, but the differences are slight and may result merely from less complete reporting of more recent than of distant affairs.
Taking respondents as the unit of observation, tabulating according to the number of times married, and designating respondents according to whether they reported ever having been unfaithful produced a clearer gradient, with reports of infidelity increasing with the number of marriages. Nevertheless, even in the south, no more than eleven per cent of women who had been married three or more times reported ever having sex with someone else while they were married to their husband.
Women were considerably less reluctant to report that their husbands had been unfaithful to them. Depending on the region, between one-fifth and one-quarter of women volunteered that they knew that their husbands had had sexual relations with other women, and around three in ten said that they either knew or suspected. Women were more likely to report that divorced husbands had been unfaithful than either dead or current husbands, and less likely to affirm that they had “probably not” been unfaithful.
The general picture, then, is that married men have more extramarital sexual partners than married women. This is the impression we gain also from the journals. In the following example, one writer overhears a discussion at a hospital about a young woman who is clearly very ill, people think with AIDS:
Some people who were near the three women who were speaking about the patient were also talking about the same story. Some people were saying that faithfulness is very important only to those who are lucky that their husbands were born faithful. But it is very rare to have a husband who is faithful and someone said that there is no man who just depends on his wife alone. All men have other sexual partners and those who claim to be faithful have only one other sexual partner but many of them have several sexual partners in addition to their wives. But always women are stupid people because they are usually faithful in their marriages because they fear their husbands will beat them when their husband hears that they have the extramarital status. She said that there are other women who have other sexual partners apart from their husbands but they are not many and they don’t have several sexual partners as men do. Most women have [only] one other sexual partner at a time. Women who have several sexual partners are those who are not married because they fear nobody and they need much help from all their partners.
This raises the important additional question of women’s sexual behaviour after one marriage has ended but before the next one has begun. In the MDICP2, women were asked how many sexual partners they had had after their most recent marriage, after their previous marriage and after their first marriage. If they had remarried since the marriage in question they were instructed to include their next husband if they had had sex with him before they married. We used this question to tabulate the proportions of women who reported any such extramarital sexual partner. The baseline population is a mix of divorcees, widows, and women in a second or subsequent marriage.
The proportions of women who report ever having had an extramarital sexual partner ranges from a high of nearly one-quarter in the south down to a low of nearly ten per cent in the central region. At least in the north and the south it appears that women who have been married only once were more likely to have an extramarital partner after that marriage ended than women who had remarried, but the lower panel of the table reveals that the proportions are based on very few women. The reason for this is swift remarriage: the table must perforce be based on women whose first marriage is no longer intact, but these women are likely to have moved on to a second or subsequent marriage.
An additional interpretational difficulty is introduced by the questionnaire’s including a subsequent spouse among a woman’s extramarital partners. We have no way of knowing whether interviewers and respondents heeded this direction, but if they did then from the point of
view of gaining an understanding of women’s sexual behaviour between marriages, the number of extramarital partners will be over-reported. After all, if a woman reported that she had one sexual partner after her first marriage ended but subsequently married this man, it is simplest to think of her as having not an extramarital partner but a second marriage that started somewhat earlier than she had originally reported.
In an attempt to circumvent this problem we restricted the baseline population to current divorcees and widows. Of those who reported a post-marital affair, most reported only one, but two (of 13) in the south reported two affairs as did one (of four) in the centre. In the north, two women (of nine) reported multiple affairs: one divorcee of 14 years reported three affairs; and one of six years reported five affairs.
Overall, around one in five divorcees and widows in the south and north report such an affair, and one in ten women in the central region. Any clear trend according to the length of time for which women had been divorced or widowed is obscured by various selection factors. On the one hand, one might expect the proportions to fall with marriage duration (as they do in the south and the centre), since embarking on a post-marital affair may be a way of finding a new husband; as a result, those at the longer post-marital durations may be selected for not having such affairs. On the other hand, one might expect the proportions to rise with marriage duration (as they do in the north) since the longer the duration, the longer the period of “exposure to risk”, in a demographic sense, of an affair. An overriding problem is that of small numbers. Given swift remarriage when divorce is common (as in the south) or comparatively low divorce rates (as in the north), there are simply very few unmarried women.
The conversations recorded by the journal writers reveal no hesitation among conversational partners in discussing their own or other people’s sexual adventures: either people were chatting and the writer overheard them, or they were chatting with a friendly individual — our journal keeper — who did not take notes in front of them. Under-reporting of extramarital sexual activity in the MDICP2, however, is a very real possibility: respondents were faced with a stranger with an interview schedule and, given the physical situations in which interviews were conducted, privacy could not always be guaranteed. Nevertheless, the pictures painted by the quantitative and the qualitative data are in broad agreement. Married women have fewer affairs than married men, and fewer affairs than divorced or widowed women. Some regional differences emerge, although less clearly than in the case of divorce and remarriage. Women’s reports of their own infidelities are most common in the south and least common in the north.
Women’s reports of their husbands’ infidelities, on the other hand, are not clearly distinguished according to region. Affairs between marriages are reported most commonly by women in the south, but least commonly by women in the central region.
