Polygyny and Fertility

The population of women in polygynous marriages consists of two groups: women who originally married a monogomist and who later acquired a cowife or cowives, and women who married a husband who already had a wife or wives. For the most part, the first group will be first or senior wives at the time of the survey, while the second group will mainly be junior (second, third, or fourth) wives. Since the associations between polygyny and fertility may be quite different for the two groups, we examine the results for first wives in polygynous marriages separately from junior wives, in all countries except Ghana which did not collect information on wife's rank.

The top panel of table 7.1 shows total marital fertility rates calculated for all currently married women by type of marriage, for the 3 years prior to the survey.^[3] In each country, marital fertility rates for all women in polygynous marriages combined are lower than for women in monogamous marriages, although the differences range from 0.16 in Lesotho to 1.8 in Kenya. When the polygynous group is divided by rank, there is no uniform pattern across countries in fertility rates. First wives have lower fertility than more junior wives in Kenya and Lesotho. However, the rate for senior wives in Lesotho is based on a very small number of cases. The differences between junior and senior wives are small and most likely not significant in Cameroon and the Ivory Coast. Senior wives have higher fertility than junior wives in Senegal. In fact, all of the fairly small fertility differential by marriage type in Senegal is due to the lower fertility of junior wives.

The differentials in marital fertility in table 7.1 must be the result of differences in one or more of the factors that have come to be known as the proximate determinants of fertility: fecundity, coital frequency, the length of the postpartum nonsusceptible period, fetal loss, and contraceptive use. One explanation for fertility differentials by type of marriage may be differences in female fecundity because of selection into polygynous marriages of sterile or subfecund women (Pison, 1986; Lesthaeghe, 1984; Clignet, 1969). This selection may occur in two ways. The sterility of a wife in a monogamous marriage is often an important factor precipitating the husband's marriage to a second wife. Husbands are also more likely to divorce subfecund or sterile wives. When these women remarry, they are more likely to become second or third wives in a polygynous household. Thus, we would anticipate that both senior and junior wives may be less fecund than wives in monogamous marriages, ceteris paribus.

In order to determine how much of the fertility differentials by marriage type are due to higher sterility of wives in polygynous marriages, we recalculated the total marital fertility rates for the 3-year period before the survey excluding women who had never had a live birth by the beginning of this observation period. The results are shown in the bottom panel of table 7.1. In the process of eliminating childless women, we also exclude a few, mainly younger, women who are newly married and thus have not had adequate time to have a child by the beginning of the 3-year period. However, during the period of observation itself, they generally experience relatively high fertility. Exclusion of these women accounts for the lower fertility rates, as in the case of Ghana, in the bottom panel of table 7.2 when compared with rates in the top panel.

Elimination of childless women from the calculation reduces the gap between the fertility of wives in monogamous and polygynous marriages in all countries except Lesotho. However, in some countries the change is only slight. In the Ivory Coast and Kenya, the difference by marriage type remains large: slightly more than one child, on average. A smaller difference also remains in Lesotho and Cameroon, but there is no longer any difference in Ghana and Senegal. In Cameroon and Kenya, senior cowives have the lowest fertility. In fact, in Cameroon the fertility of junior wives is about the same as that of wives in monogamous marriages. The situation is reversed in Ivory Coast and Senegal where senior cowives have higher fertility than junior cowives.

The classical hypothesis about fertility differences between wives in monogamous and polygynous marriages is that they are caused by differences in coital frequency (Muhsam, 1956). The usual argument is that if the husband's frequency of intercourse remains the same or increases only slightly if he marries another wife, then the coital frequency of each wife will decline as the number of wives increases. However, Garenne and van de Walle (1985) point out that models of coital frequency and fertility suggest that even substantial reductions in the frequency of intercourse may have a relatively minor impact on fertility. Furthermore, in societies in which women are frequently pregnant or abstinent, it may be relatively easy for each wife to be as sexually active as a wife in a monogamous marriage since cowives are likely to be abstinent at different times.

Data on coital frequency are difficult to collect and may be highly unreliable when available. In several of the surveys included in this analysis one or more questions were asked about whether a woman was sexually active at the time of the survey or about coital frequency. We will return to these data later on in the chapter.

The fertility differentials between women in monogamous and polygynous marriages may also be due to differences in postpartum variables and contraceptive use. Table 7.2 shows summary statistics for the distributions of

several of these variables, by marriage type for each country. In Ghana, all women in polygynous marriages are again grouped together since no information is available on cowives' status or rank. We have also grouped all wives in polygynous marriages together for Lesotho, in this and all subsequent tables, since there are relatively few women in polygynous marriages and very few senior cowives.

The first three columns of table 7.2 show that the wife's age, her duration in her current marriage, and her husband's age, at the time of interview, all differ consistently by marriage type. Wives in monogamous marriages and junior cowives in polygynous marriages are, on average, very close in age in all countries in which rank can be determined. The results for Ghana and Lesotho are at least consistent with this pattern. Senior cowives are considerably older than the other two groups. Similarly, senior cowives have also been married for longer average durations than junior cowives and wives in monogamous marriages. The latter two groups have roughly equivalent durations of marriage.

The reason for these two related patterns is the operation of the marriage system itself. Since all marriages may potentially become polygynous, many wives in monogamous marriages are those whose marriages are not yet polygynous: that is, they tend to be younger and at shorter marriage durations. Junior cowives in polygynous marriages are also more recently married, by definition, than senior cowives and are, therefore, younger and at shorter marriage durations. In fact, junior cowives at any given age are more likely to have shorter marriage durations in the current marriage than wives in monogamous marriages, because they are more likely to have been divorced or widowed than wives in monogamous marriages.^[4] Thus, one possible explanation for the lower fertility of senior cowives in Kenya and Cameroon may be the longer average marriage duration of these women. We have eliminated effects on the fertility differentials by marriage type of differences in primary sterility by excluding women who never had a live birth from the bottom of table 7.2. However, women at longer marriage durations may have lower coital frequency.

In the third column of table 7.2, we compare the mean ages of husbands to whom wives of each union type are married. However, substantial proportions of the sample did not report a husband's age in Cameroon and the Ivory Coast. In Senegal, a question on husband's age was not included in the individual questionnaire. To obtain husband's age information, we matched husbands and wives using information on the household schedules. Unfortunately, we were unable to match all currently married women with husbands. Even when matches were made, information on the husband's age was not always available. The proportion of wives for whom we have no husband's age information is shown in Appendix J. In Cameroon and the Ivory Coast, this information is more frequently missing for wives in polygy-

nous marriages. Thus, the means shown in table 7.2 for husband's age are averaged only across wives for whom husband's age information is available. These means are standardized for the wife's age since wives in different types of marriage have different age distributions. The standardization procedure used is analogous to calculating a total fertility rate. A separate mean husband's age was calculated for each 5-year wife's age group. These means were then averaged again across all wife's age groups, which has the effect of giving each wife's age group equal weight. The resulting mean husbands' ages reflect primarily the age difference between husbands and wives at marriage since wife's age is held constant.

The results indicate that when wife's age is held constant, wives in monogamous marriages and senior cowives in polygynous marriages are married to men of roughly the same mean age. This is not surprising, since the women in both groups are generally the first wives that these men married. Junior cowives, on the other hand, are at any given age married to much older husbands. The means for Ghana and Lesotho show that when all wives in polygynous marriages are grouped together, they are, on average, married to older men. But we can see from the means for other countries that the entire effect is most likely due to the fact that junior co-wives have older husbands.

Recent research on a population in Senegal by Garenne and van de Walle (1985) and on nineteenth-century American Mormons by Bean and Mineau (1986) suggests that the fact that husbands of junior cowives are older may be part of the reason for lower fertility among these wives. Two possible explanations for these effects of husband's age may be reduced coital frequency and reduced fecundity among older husbands.

In the next four columns of table 7.2, we examine differences in breast-feeding, amenorrhea, and abstinence by marriage type. Each median is calculated using life table methods^[5] and exposure in the open birth interval (the interval since last live birth) and the last closed birth interval, which occurred during a 3-year observation period. The sample of birth intervals is the same one used for multivariate analysis and was described in the preceding section. Because of the sample on which they are based, the medians pertain to a period of observation shortly before interview.

Many observers argue that polygyny may be associated with fertility in part because it allows wives to breastfeed and to observe the postpartum sexual taboo for a longer time after giving birth, since polygynists have other legitimate sexual partners while monogamists do not. Wives in polygynous marriages may also breastfeed and abstain from sex for longer periods because they come from more traditional backgrounds than wives in monogamous marriages. This longer median duration of breastfeeding for senior cowives is at least consistent with the fertility differentials by marriage type and rank observed for Kenya in table 7.2. Thus, one of the hypotheses we will

test in the next section is that longer breastfeeding for senior cowives accounts for their lower fertility relative to other women.

The median duration of breastfeeding is generally shortest for women in monogamous marriages, longer for senior cowives in polygynous marriages, and longest for junior cowives, although the differences among these three groups are very small in the Ivory Coast and Senegal. The exception to this pattern is Kenya, where senior cowives have the longest median duration of breastfeeding. The median duration of full breastfeeding shows no strong association with marriage type, although in most countries the medians for wives of polygynists are slightly higher.

The median lengths of postpartum amenorrhea vary relatively little by marriage type within each country. It is encouraging to note that for the most part these variations follow the same patterns as those in the median length of breastfeeding. For example, in Kenya, the median length of postpartum amenorrhea is highest for senior cowives, as in the median length of breastfeeding. Questions on postpartum amenorrhea and postpartum abstinence were not included in the Senegal questionnaire.

The median durations of postpartum abstinence vary markedly from country to country, with high median durations reported in Cameroon and the Ivory Coast, and very low median durations reported in Kenya. The relatively long durations of postpartum abstinence in Lesotho are less the result of strict observance of a postpartum sexual taboo than of the frequent absence of substantial portions of the male population who leave home to work in South Africa. Postpartum abstinence does not appear to differ significantly by marriage type and rank in the Ivory Coast and Kenya, where fertility differentials are the greatest. In Cameroon and Ghana, women in polygynous marriages report abstaining 2 to 3 months longer than wives in monogamous marriages. In Lesotho, however, wives in monogamous marriages abstain for about a month longer, on average, than those in polygynous marriages.

The last three columns of table 7.2 indicate the frequency of contraceptive use and the proportion of wives coresiding with their husbands by type of marriage and rank. Since contraceptive use in the open and last closed birth intervals is generally higher among wives in monogamous marriages in most countries, differential contraceptive use can not be an explanation of the lower fertility among wives in polygynous marriages. In fact, differentials in contraceptive use in some countries such as Kenya may tend to reduce the size of fertility differentials by marriage type. However, even in Kenya, contraceptive use is not common and the differentials in use by marriage type are not large.

In many African societies, separate residence of spouses does not necessarily imply the lack of sexual relations. However, spouses who are not coresident may have a lower frequency of intercourse than spouses who core-

side. As the figures in table 7.2 show, the vast majority of women coreside with their husbands, except in Lesotho where male labor migration to South Africa makes coresidence of spouses more difficult. The proportion of wives coresiding with husbands is also low in Ghana primarily because separate residence of spouses is traditional in several Ghanaian cultures. No information on coresidence of spouses is available for Cameroon or Senegal. In the Ivory Coast, junior cowives are the least likely to coreside with their husbands. The situation is reversed in Kenya, where senior cowives are least likely to live with their husbands.

Thus, differentials in several of the variables shown in table 7.2 may explain the fertility differentials by marriage type and rank shown in table 7.1. Marriage duration and husband's age all vary consistently with marriage type. Differences in the length of breastfeeding may be important in Cameroon, Ghana, and Kenya, as may be many differences in postpartum abstinence in Cameroon. The patterns of contraceptive use suggest that fertility differentials by marriage type might actually be larger if there were no contraceptive use. Finally, the lower proportions of senior cowives in Kenya and junior cowives in the Ivory Coast who coreside with their husbands may be associated with lower fertility for these two groups. We have not investigated differentials in one other proximate determinant of fertility—fetal loss—because of apparently severe problems of underreporting. However, there is no reason to believe that fetal loss rates would differ sufficiently by type of marriage to contribute to the fertility differentials between women in monogamous and polygynous marriages.