Until recently, fertility and family planning research in developing countries, as well as policy and program formulation, has generally relied on data collected from women. Increasingly, however, attention is being paid to including men. The reasons for the new interest in men are not hard to find: First, information that has become available from surveys conducted over the past decade suggests that men and women do not necessarily have similar fertility attitudes and goals.1 Second, the scope of fertility and family planning research has expanded to include such broader reproductive health issues as sexually transmitted diseases, on which data from both men and women are needed.2

Although women bear children and most modern contraceptives are female methods, childrearing has an impact on men's lives too. This impact may be felt financially, if men accept the responsibility of supporting their children, and in a range of other ways, including through the health and well-being of their wives and children. Often, a man's social status is also affected when he becomes a father.3

The male partner may play an important role in decision-making regarding contraceptive use and the timing and number of a couple's births. In some countries or among some social groups, the male partner has greater influence than his spouse.4 In Ghana, the wife's attitude toward contraception is wife.5

On the other hand, the perception that men will necessarily have more influence on reproductive decisions because they typically control the family's assets, are accepted as the household head and are older may be an exaggeration. The actual situation is likely to depend on other factors and to vary over time and by location. For instance, among the Yoruba of Nigeria, the fertility desires of both marriage partners are important predictors of the couple's fertility. However, whereas the husband's desire is dominant in predicting the couple's behavior when the number of living children is small, the wife's desire becomes more important as the number of children grows.6 In Taiwan, when spouses disagree over whether to have another child, the wife tends to prevail.7

Efforts to promote family planning in developing countries have often been criticized for their exclusion of men. The consequence of the female-only approach has been that some men view family planning with suspicion, regarding it as being aimed at undermining their authority in the family. For instance, men in Nigeria typically believe that contraception makes it easy for their wives to engage in extramarital sexual relationships.8 While men's attitudes toward family planning are generally positive, some studies show that men believe that they should be in control of whether and when a couple uses contraceptives.9

Failure to involve men in family planning programs can have serious implications. Even when women are educated and motivated to practice contraception, they may not do so because of opposition from their husbands. Individuals interviewed in urban Sudan believed that the male partner decides if a couple will use contraceptives and, if they do, supplies the method.10 In light of these findings, some researchers question the validity of the estimates of unmet need derived from information collected only from women.11

This article examines the reproductive preferences and behavior of married men and their wives in 18 developing countries. The main objective is to understand the role of husbands in reproductive decision-making by focusing on their preferences concerning family size and having additional children, whether they are aware of family planning and what role they play in contraceptive use. We examine these phenomena by comparing the responses of husbands with those of their wives in order to bring out the similarities and differences. For instance, it is important to know whether husbands and wives are equally knowledgeable about contraceptive methods, and how their family-size goals compare. In the case of disagreement about fertility desires, we attempt to identify whose view carries the greater weight in terms of contraceptive use. Although relatives and friends may play important roles in couples' reproductive decision-making, we do not examine their contributions.

Many studies that have examined differences in reproductive preferences by sex take the aggregate-level approach and compare men as a group with women as a group.12 This approach tends to understate the extent and role of disagreement in fertility preferences, particularly in populations where the norm is to have many children. When small numbers of individuals begin to prefer and actually have smaller families, the impact of these innovations is often submerged, since the overall average is likely to reflect the prevailing norm of the society.

On the other hand, comparing the individuals who make up a couple, and treating couples as the unit of analysis, adds a different perspective and enhances our understanding of reproductive preferences and decision-making. The couple-level approach involves direct comparison of the preferences of the two individuals who matter most in reproductive decision-making. Furthermore, such differences are more real in terms of their impact on reproductive outcome (e.g., contraceptive use or subsequent fertility). An important limitation of the couple-level approach, however, is that it does not cover the preferences and behavior of men and women who are not in a legal or cohabiting union.

Methodology

Data

Data in this analysis are from national surveys conducted as part of
the Demographic and Health Survey (DHS) program between 1990 and 1996 in 18 developing countries--13 in Sub-Saharan Africa, two each in North Africa and Asia, and one in Latin America. The DHS is the only major series of cross-national surveys of reproductive behavior in developing countries that cover both men and women; our analyses include all but two countries for which data on men and women were available.* The dominance of Sub-Saharan African countries among those in which men have been surveyed probably reflects a focus on HIV and AIDS, and the need to incorporate men in any analysis of this topic; in other countries, the motivation for including men was to explain husband-wife agreement and communication in matters related to fertility decision-making.

The female data for the Sub-Saharan African countries, Brazil and Morocco are based on samples of all women aged 15-49, while the data for Bangladesh, Egypt and Pakistan are for ever-married women of the same age range. The male samples are less uniform (Table 1). All men, regardless of marital status, were surveyed in 13 countries, but only the husbands of interviewed women were included in the remaining five. The age range of the male samples also varies, with some countries having no age limits, some having a lower limit and others having both lower and upper limits. To obtain the couple data used in this study, we combined data from the separate interviews of husbands and wives in all 18 surveys.

The male questionnaire is similar in structure to the female questionnaire, but is shorter. Men are asked about their background characteristics, fertility experiences, contraceptive knowledge and use, marriage, sexual behavior and reproductive preferences. In recent DHS questionnaires, the fertility section includes detailed questions on the number of children ever born and the number surviving and deceased by sex. (In some of the earlier surveys, men were asked only to report their total number of children or their number of living children by sex.)

In the section on contraception, the respondent is asked to name all the methods that he knows. After he has done so, the interviewer reads a description of each modern and traditional method that was not mentioned and asks the respondent whether he knows any of these and whether he has ever used each of the methods that he has heard of. Also, detailed questions are asked about current method use and, among nonusers, intention to use.

The section on fertility preferences includes questions about respondents' ideal number of children (in some cases by sex); whether they intend to have any more children and, if so, the preferred timing of the births; and their own and their partners' attitudes toward family planning. As much as possible, the questions in the male questionnaire are worded the same way as in the female questionnaire.

The data are limited to the extent that the DHS, like other large-scale national surveys, uses structured interviews that do not probe deeply into most topics and usually do not include open-ended questions. In some countries, even standard questions were not included because of cultural preferences or concerns. These limitations prevent deeper coverage of some of the issues examined here. Furthermore, the lack of uniformity in the ages of male respondents leads to some bias in overall comparisons. Nevertheless, by allowing comparison across 18 countries, the DHS data may help our understanding of reproductive preferences and decision-making among couples.

Analytic Methods

We adopted the DHS definition of a couple as consisting of a man and a woman who are legally married or who are living together in a consensual union. In countries where polygyny is widely practiced (most of which are in Sub-Saharan Africa), this implies that the sample includes some men who have more than one wife. (We categorize a union as polygynous or monogamous on the basis of the husband's report of his number of wives.) For our analysis, the number of couples in a polygynous household is the same as the number of wives, and the information for the male partner is the same for each couple. However, since most of the questions asked of husbands that relate to their wives did not require a polygynous man to be wife-specific in his answer, we do not know to which wife or wives the responses refer. Unfortunately, nothing can be done to correct this problem. Thus, in cases where this ambiguity may affect the results, we either carry out the analysis by type of marriage or include monogamous couples only.

To examine couples' ideal number of children, contraceptive knowledge and modern method use, we constructed measures of these variables combining husbands' and wives' reporting. For instance, the measure of modern method use is a three-category variable showing the proportions of couples in which only the husband reports use, both spouses report use and only the wife reports use. This measure allows us both to show the level of agreement between spouses and to measure use separately for husbands and wives.

The analysis of the effects of fertility intentions on couples' contraceptive behavior is restricted to fecund monogamous couples in which the wife was not pregnant at the time of the survey. (Couples are considered fecund if neither spouse reports that either partner is infecund.) Polygynous couples are excluded because men's answers to the question about fertility intentions are not wife-specific.

A couple is defined as using a modern method of family planning if the wife reports current use of any method or (because women may underreport use of male methods, especially condoms) if the husband reports condom use. We used logistic regression analysis to examine the effects of fertility intentions on couples' use of modern contraceptives; the results are presented as predicted proportions, to facilitate a comparison of the effects before and after controlling for other variables.†13

Results

Characteristics of Husbands and Wives

In some settings, the difference in the ages of husband and wife is a determinant of whether the spouses have similar reproductive preferences.14 In all 18 countries, husbands are typically older than their wives; the median age difference ranges from 2.7 years in Brazil to 12.2 years in Senegal (Table 2). Generally, the gap is widest in Sub-Saharan Africa.

Polygyny is evident in the 16 countries for which data are available on type of marriage. However, while polygyny is very common in Sub-Saharan Africa, its prevalence is negligible in other regions. On average, 23% of husbands and 29% of wives are in polygynous unions in Sub-Saharan Africa (not shown), but wide variations exist within the region: Polygyny is most prevalent in West African countries, which are predominantly Muslim.ý The relatively high prevalence of polygyny may account for the larger age gap between spouses in these countries, since women in societies where polygyny is common tend to marry at younger ages than their counterparts in societies where the practice is less prevalent.15

In every country, at least 80% of husbands are currently working. A substantial proportion of wives work, although wide variations exist between countries. More than 50% of wives are currently working in 10 countries, but in North Africa and Asia, the proportions are only 15-22%.

The literacy level among men varies from 10% in Senegal to 77% in Zimbabwe. In every country except Brazil, wives are considerably less likely than their husbands to be literate. The proportion of wives who can read without difficulty ranges from 3% in Niger to 76% in Brazil. In nine countries, the proportion literate among husbands is at least two times that for wives.

The number of years of schooling that men and women receive remains very low in many developing countries. The proportion of husbands with seven or more years of schooling ranges from 5% in Burkina Faso to 67% in Zimbabwe; it is greater than 20% in 13 countries. Wives tend to spend fewer years in school than their husbands in most of these countries; the proportion of wives with seven or more years of education exceeds 20% in only six countries.

Childbearing Goals

Although conventional wisdom suggests that men desire more children than women in developing countries, data that permit empirical, cross-cultural studies to verify this claim have only recently become available. A review of earlier studies, however, does not support this notion.16 And a recent study that examined male and female preferences using DHS data showed that except in some countries in West Africa, the family-size preferences of men and women are quite similar.17 However, this conclusion was based on aggregate-level results, which may conceal disagreements at the couple level.

Desired number of children. One commonly used measure of reproductive preferences is the number of children that a respondent would like to have if he or she could choose. The meaning and utility of this measure have been debated since the 1960s, when researchers argued that responses may be no more than polite answers to meaningless inquiries.18 The problems usually identified with the measure include inconsistency (which depends on how stable reproductive preferences are, as well as on whether questions are meaningful to the respondent) and rationalization (whereby the respondent simply gives a number that is equal to or greater than the number of children she or he already has, rather than the number she or he would prefer to have).

Nonetheless, the question is still asked in fertility surveys because it is simple and allows comparison with earlier survey data. Although the measure may be unstable over time at the individual level, it is remarkably consistent at the aggregate level.19 Therefore, some researchers consider it a good indicator of societal norms about family size, and believe that changes in this indicator over time reflect changes in attitudes that may affect behavior.20

Husbands tend to want a larger family than their wives in many countries, especially in Sub-Saharan Africa (Table 3).§ The proportion of couples in which the husband desires at least two children more than his wife ranges from 17% in Pakistan to 49% in Niger. By contrast, the corresponding estimate for wives ranges from 11% in Bangladesh and Egypt to 23% in Malawi and Niger. The gender differential is more pronounced in Sub-Saharan Africa than in other regions and is larger in West Africa than in East Africa.

If we define agreement to mean that a husband and wife reported either the same desired number of children or a one-child difference,** the proportion of couples in which spouses agree ranges from 27% in Niger to 70% in Bangladesh. Agreement is higher among couples in countries outside Sub-Saharan Africa than in that region. While more than 60% of couples in the five North African, Asian and Latin American countries included here are in agreement, fewer than half agree in eight of the 13 Sub-Saharan African countries; in five of these eight countries, only 30-40% of husbands and wives desire the same number of children. (The 68% shown for Pakistan is probably too high, because 64% of men and 60% of women gave nonnumeric responses, which are assigned a value of six children and thus may lead to false agreement. The actual level of agreement probably resembles levels in Sub-Saharan Africa.)

On average, married men want a large number of children in many of these countries. The mean number of children desired by husbands ranges from 2.9 in Brazil to 11.5 in Niger; it exceeds five in 11 of the Sub-Saharan African countries. Husband's desired family size tends to be higher in West Africa than in East Africa. On average, husbands in all of the other countries except Pakistan want fewer than five children. Wife's average preferred family size shows a similar range across countries and similar regional patterns.

The difference in spouses' mean desired family size is substantial--one child or more--in seven Sub-Saharan African countries, five of which are in West Africa. No gender difference in mean desired number of children is discernible in countries from other regions.

Fertility intentions. Another prominent measure of reproductive preferences is whether or not the respondent intends to have another child. This measure is a robust predictor of contraceptive and fertility behavior at both the aggregate and the couple levels;21 it also has become an indispensable ingredient in the estimation of unmet need for family planning.22

Using DHS data to compare husbands' and wives' responses on this issue among all couples is somewhat problematic: Since the surveys do not ask polygynous husbands whether they intend to have more children with each wife, the results probably understate the extent of agreement between spouses in polygynous unions. For example, if a man with two wives says he wants more children, he may want them with both wives or only with one. If he wants them only with one, a comparison of his survey response and that wife's will accurately show whether they agree. However, if the wife with whom he wants no more children also wants no more (i.e., their intentions agree), the couple will nevertheless be counted as disagreeing because his response did not reflect his intention with this wife.

Despite the possibility of bias in the direction of disagreement, husbands and wives show a high degree of agreement about their fertility intentions (Table 4). The proportion of couples in which both spouses agree on this measure (i.e., either both want more or both want no more) ranges from 74% in Kenya to 90% in Niger.

Couples' type of agreement--whether both want no more children or both want more--differs by region. Throughout Sub-Saharan Africa, husbands and wives whose fertility intentions agree generally want more children. Of all couples in agreement, the proportion who want more children ranges from 53% in Kenya to 99% in Niger. This supports the finding that a high proportion of both husbands and wives want a large family in Sub-Saharan Africa.

Outside Sub-Saharan Africa, spouses whose intentions agree typically want no more children. The only exception is Pakistan, where findings resemble those for Sub-Saharan Africa. Of all couples in agreement in the remaining four countries, the proportion who want to stop childbearing ranges from 51% in Morocco to 79% in Brazil. This regional differential is not surprising, since desired family size and actual fertility are lower and declining in most developing countries outside Sub-Saharan Africa.

In all settings, though, some couples have disparate fertility intentions. Overall, the lowest level of disagreement is in Niger (10%), and the highest is in Kenya (26%). In most countries, when couples disagree about future childbearing, the wife wants no more children but the husband wants more. (The exceptions are Bangladesh and Malawi, where both types of disagreement are about equally likely.) Among all couples experiencing disagreement, the proportion in which the wife wants no more children but the husband wants more is 60% or greater in 13 countries.

One important implication of disagreement about desired family size and fertility intentions relates to how it affects contraceptive use. However, even when spouses agree, potential areas of conflict remain. For instance, in many countries, the DHS asks men and women who say they intend to have another child how long they would like to wait before doing so; responses indicate that husbands and wives may disagree as to whether they want to have the next birth soon (within two years) or they want to postpone it.

Among couples in which both spouses want more children, the proportion who disagree as to the timing of the next birth ranges from 21% in Brazil to 40% in Burkina Faso and Uganda. This implies that in the majority of couples, the spouses agree either to have another child soon or to wait. Among those in disagreement, there is some evidence of a gender difference concerning which partner wants the child sooner than the other. In six of the nine Sub-Saharan African countries for which data are available, husbands want the next child sooner than their wives; this difference is five or more percentage points in four of these countries. Therefore, at least in Sub-Saharan Africa, husbands not only want a larger family than their wives, but also want the next child sooner than their spouses.

Type of Marriage and Reproductive Goals

Although the causal path is unclear, polygyny may be associated with men's reproductive preferences: Men may either have more than one wife because they want many children or want many children because they have more than one wife.23 The fertility preferences of wives in polygynous unions are less clear: On the one hand, wives may want many children in order to compete favorably with co-wives in terms of childbearing and status in the household. On the other hand, this desire may be curtailed by the fact that women in polygynous unions shoulder greater responsibilities in rearing their children than do those in monogamous marriages.24

Some differences exist in the desired family size of marital partners by type of union. In almost all countries for which we can classify couples by type of union, the proportion in which the husband's family-size preference exceeds the wife's by two or more children is higher for polygynous than for monogamous unions. In Sub-Saharan Africa, the proportion ranges from 21% in Kenya and Zimbabwe to 48% in Niger among monogamous couples, and from 33% in Malawi to 57% in Mali among polygynous couples.

For Sub-Saharan Africa overall, the proportion in which the husband's family-size preference exceeds the wife's by two or more children averages 32% for monogamous couples and 47% for polygynous couples. By contrast, 47% of couples in monogamous unions and 36% in polygynous unions in Sub-Saharan Africa agree about family size. The differences are probably associated with factors that are believed to be more prevalent in monogamous unions, such as conjugal closeness and spousal communication.25

We found only a small difference in the joint distribution of fertility intentions of monogamous and polygynous couples. First, the proportion who agree either to have more children or to stop childbearing is slightly higher among monogamous couples (83% in Sub-Saharan Africa overall) than among their polygynous counterparts (75%). Second, the implied higher disagreement among polygynous couples is manifested in both types of disagreement (i.e., the husband wants to stop childbearing while the wife wants more children and vice versa). Among monogamous couples, on average, 5% of husbands and 12% of wives want to stop childbearing while the other spouse wants more children. Similarly, among polygynous couples, on average, 8% of husbands and 17% of wives want to stop childbearing, disagreeing with their spouse. Assuming that agreement is understated among polygynous couples because of a lack of appropriate data, monogamous and polygynous couples appear to differ very little in levels of spousal agreement on fertility preferences.

Contraceptive Knowledge and Behavior

A substantial proportion of married men know of at least one method of family planning, but in some countries, only a small proportion of those who know of a method are practicing contraception. However, husbands often report greater method use than their wives.26 This disparity may stem from several factors: multiple sexual partners, differential reporting of condom use by husbands and wives, differences in perception of rhythm among marital partners, and underreporting of method use because of the presence of other adults during wives' interviews.27 The difference may also be related to the type of method used, the frequency of use or the reference period: For example, a man who used the condom once with his wife a week before the survey may report that he and his wife currently use condoms, while if his wife has forgotten about that incident or recalls several acts of unprotected intercourse they have had since then, she may report that they use no method.28 The DHS data permit us to examine the contraceptive knowledge and practice of husbands and their wives across countries.

Knowledge of modern methods of family planning is generally high among both husbands and wives, but variations by country are substantial. The proportion of husbands who know at least one modern method ranges from 57% in Burkina Faso to 100% in Brazil (Table 5). Knowledge of modern methods is lower among husbands in West Africa than elsewhere. Similarly, wives exhibit a high level of knowledge of modern methods: from 57% in Cameroon to 100% in Bangladesh, Brazil and Egypt. The difference between the proportion of husbands and wives who know a modern method is generally small within a country. It is less than five percentage points in 12 countries, 6-8 points in five countries and 16 points in one country.

Every respondent who reports knowing at least one method of family planning and who is not currently pregnant (or whose spouse is not pregnant) is asked if she or he is "doing something now or using any method with any partner to delay or avoid pregnancy." If the respondent reports using more than one method, the interviewer is instructed to record the most effective of the methods mentioned. Responses vary considerably across countries for both husbands and wives. The proportions of men and women reporting modern method use are lowest in Niger and Central African Republic, and are highest in Brazil. Except in Kenya and Zimbabwe, fewer than 20% of husbands and 13% of wives in Sub-Saharan Africa reportedly use modern methods. On the other hand, use is substantial outside Sub-Saharan Africa, except in Pakistan.

In most of the countries included here, a higher proportion of husbands than wives report using a modern method. The estimated level of modern contraceptive use varies, sometimes substantially, depending on whose reporting is taken into account. For instance, in Ghana, the level of use according to husbands' reports (i.e., whether only he reported use or both he and his spouse reported use) is 18%; according to wives' reports, it is 13%. However, if the proportion is based on couples in which both spouses report use, the level is 10%, and if it is based on those in which either spouse reports use (i.e., combining the data for husbands only, wives only and both), it is 21%.

Similarly, in Bangladesh, the level of modern method use is 44% according to husbands' reports, 39% according to wives' responses, 36% as measured by both spouses' replies and 47% if at least one partner reports use. In Central African Republic, the use of modern methods among couples when either spouse reports use (5%) is roughly five times the level when both spouses report use (1%) and more than double when wives only report use (2%).

While differences between husbands and wives in the reporting of current use of modern contraceptives can be substantial, particularly in Sub-Saharan Africa, it is not clear why these differences exist. Polygyny is not likely to explain much of the discrepancy. First, we see the difference even in countries where polygyny does not exist or is uncommon. Second, an analysis of men's and women's reporting of contraceptive use by type of union reveals little or no difference between monogamous and polygynous couples (not shown). For instance, in Sub-Saharan Africa, on average, only husbands reported modern contraceptive use in 5% of monogamous and 6% of polygynous couples. Similarly, the proportion of couples in which only the wife reported use of modern methods was 2% for monogamous and 3% for polygynous couples.

Is condom use the source of the differences in husbands' and wives' reporting of modern contraceptive use? If this were the case, we would expect substantially larger proportions of husbands than wives to report condom use. When we examine the proportion of couples in which only the husband reports condom use, we find that this may be an important explanation for the discrepancy. The proportion ranges from less than 1% in Egypt, Morocco and Niger to 6% in Kenya (Table 5); the average is 3% for Sub-Saharan Africa. The results are also very similar for monogamous and polygynous couples in most countries (not shown).

Furthermore, among couples in which only the husband reports use of a modern method, the proportion of use that is due to husbands' reporting of condom use ranges from 8% in Morocco to 84% in Malawi. This proportion is more than one-half in eight countries and close to one-half in three others. The only countries where this proportion is low (21% or less) are Bangladesh, Egypt and Morocco. Thus, although other factors may contribute to the observed differences between husbands' and wives' reporting of modern method use, the role of differential reporting of condom use seems to be very important. Whatever the explanation, these discrepancies also emphasize the importance of obtaining information from both men and women when measuring contraceptive prevalence.

Fertility Intentions and Contraceptive Use

Assuming that individuals who want no more children or wish to postpone childbearing practice contraception, it is easy to see why contraceptive use will be high when both spouses want to stop or postpone childbearing and low when both want to have another child. However, whether and to what extent contraceptives are used in situations where spouses disagree about their intentions is the question. Do couples use contraceptives more when the husband wants more children and the wife does not or when the wife wants more and the husband does not? Where men are favored in terms of access to household and community resources and recognition, do they also have a greater influence on contraceptive use?

The data largely confirm the predicted direction of method use among couples who are in agreement. Except in Côte d'Ivoire, use of modern methods is highest when spouses agree to stop childbearing (Table 6). The proportion of couples in this category who are using modern methods ranges from 14% in Côte d'Ivoire to 87% in Brazil; it exceeds 20% in 16 countries. On the other hand, in most countries, modern method use is lowest among couples who agree to have more children. The level of use for this group of couples ranges from 2% in Cameroon to 66% in Brazil; it is less than 10% in 11 countries and above 20% in only six. Apparently, these couples are using contraceptives to postpone the next birth.

When spouses' fertility intentions disagree, there is no clear pattern with respect to the direction of contraceptive use. In 10 countries, modern contraceptive use is higher when only the husband wants to stop childbearing. In the other eight countries, however, it is higher when only the wife wants to stop having children. The difference in use according to which spouse wants no more children is generally trivial--less than eight percentage points in 12 countries. For the remaining countries, the difference ranges from nine percentage points in Malawi to 22 percentage points in Egypt, and the level of use is higher when only the husband wants no more children in five of these countries, all in Sub-Saharan Africa (Burkina Faso, Côte d'Ivoire, Kenya, Malawi and Senegal).

These findings suggest that although there is no systematic difference, husbands' preferences have a greater influence than wives' preferences have on contraceptive use in Sub-Saharan Africa. Is this conclusion valid after the effects of other variables are taken into account? Data for 14 countries allow us to conduct a logistic regression analysis controlling for the effects of both spouses' age and education, residence and the number of living children. The results indicate that in nine countries, couples' joint fertility intentions are a significant predictor of current use of modern methods, and the pattern and the direction of the relationship remain largely similar to the unadjusted results.

Use of modern methods is highest when both spouses want to stop childbearing and lowest when they want to have more children. Only in Malawi are couples more likely to be using modern methods when the husband wants no more children and the wife wants more. On the other hand, use of modern contraceptives seems to be higher among couples in Egypt, Kenya, Mali, Morocco, Pakistan and Uganda when the wife wants no more children and the husband wants more. In Bangladesh and Brazil, levels of use in the two groups of couples are very similar. Thus, in the majority of these countries, when the effects of other variables are taken into account, the wife's preference appears to be dominant in determining whether or not the couple use modern contraceptives. However, this conclusion should be interpreted with caution, since the analysis is based on only 14 countries.

Discussion

In many countries covered in this study, both husbands and wives want a large family. While this is especially true in Sub-Saharan Africa, evidence from Ghana, Kenya and Zimbabwe suggests that the norms that support the large family are not uniform even within that region. At the aggregate level, husbands in Sub-Saharan Africa are more likely than their wives to want a large family; no difference is discernible between the number of children desired by husbands and wives in other regions.

At the couple level, though, differences between the family-size preferences of husbands and wives may be larger than suggested by the aggregate-level measure. In all 18 countries, there is a substantial discrepancy between the preferences of spouses: In about two-thirds of couples, husbands and wives differ by one child or more in the family size they consider ideal. Even when disagreement is defined as a difference of two or more children between partners' reported ideal family size, it remains high: between 30% of couples (in Bangladesh) and 73% (in Niger). Our findings also show that in most countries, husbands want a larger family than their wives. Thus, husbands and wives differ in their fertility goals, although the magnitude of the difference and its importance for behavior vary across countries and regions.

Spouses for the most part agree about whether they want more children. More than 70% of couples are in agreement on this subject, with little variation by country. However, in 10-26% of couples, partners disagree; usually, the husband wants more children but the wife does not. Furthermore, when marital partners agree to have another child, they may differ about whether they want the child within two years or later. This type of disagreement occurs in 21-40% of these couples, and more often husbands want to have the next child sooner than their wives do.

The findings from these two indicators of reproductive preferences have implications for fertility and family planning behavior. First, they show that decline in family-size preferences, which is a necessary precursor of decline in actual fertility, tends to occur first among wives. Furthermore, the results indicate that married women probably have a better understanding of the benefit of spacing their children and the danger associated with having births in quick succession than their husbands have. Therefore, contraceptive use either to space births or to limit family size is likely to be initiated by wives rather than their husbands. But success in achieving a smaller family will depend on how responsive husbands' fertility preferences are to the changes in their spouses' preferences and on the influence of husbands' preferences on couples' reproductive behavior.

Contraceptive knowledge is high among husbands and wives in the 18 countries, and only small differences are observed between the reporting of marital partners. On the other hand, husbands are more likely than wives to report modern method use. While differential reporting of condom use between marital partners is an important source of this discrepancy, it does not appear to be the sole cause. This finding highlights a potential problem associated with the conventional measure of contraceptive prevalence based only on women's reports. Estimates of contraceptive use may vary widely depending on whether the researcher relies solely on data obtained from women or takes into account husbands' reporting of use. For instance, in Ghana, 18% of husbands report modern method use, compared with 13% of wives, a difference of 28%. Studies of fertility and family planning will benefit from adopting measures of contraceptive use that are based on the reporting of both partners.

Our findings support the claim that reproductive intentions are important predictors of contraceptive behavior. Before controlling for other variables, the joint fertility intentions of the spouses significantly determine whether the couple will use modern methods of family planning in 15 of the 18 countries for which we have information. After controlling for the effects of spouses' characteristics, joint fertility intentions remain a significant predictor of use in nine of the 14 countries for which information exists. As expected, contraceptive use is more likely when couples want to stop childbearing and less likely when they intend to have more children, even in countries where overall use is low.

The level of contraceptive use when one spouse wants more children but the other does not is not as easy to predict. Overall, the results of the multivariate analysis show that the wife's preference exerts a stronger influence on the couple's contraceptive behavior in six of the nine countries where the variable significantly predicts contraceptive behavior.

Although this study included only a few countries outside Sub-Saharan Africa, the results show that Sub-Saharan Africa and the rest of the developing world may differ in important ways. Among the countries covered, there is more agreement between spouses on the ideal number of children in Asia, Latin America and North Africa than in Sub-Saharan Africa. And although levels of agreement as to fertility intentions are similar across all regions, couples in Asia, Latin America and North Africa are more likely to agree to stop childbearing than to have another child, whereas the reverse is the case for couples in Sub-Saharan Africa.

In terms of contraceptive use, inconsistency in spousal reporting is evident in all 18 countries studied; the proportions of couples in which only husbands or only wives report use are very similar across countries. However, because use of modern methods is lower in most of the Sub-Saharan African countries, these differences imply a greater relative bias in measurement of contraceptive prevalence there than in other regions.

Our findings suggest that more work is needed on gender differences in reproductive preferences and behavior. The investigation should be extended to unmarried men and women, particularly never-married men and their sexual partners, and to more countries, especially in regions other than Sub-Saharan Africa. In addition, survey questions should be more specific, especially on matters that involve respondents and one or more of their partners. Attempts to examine such issues as why marital partners report contraceptive use differently will benefit from information gathered in in-depth surveys of men's and women's attitudes, preferences and behavior regarding contraceptive use and childbearing. Longitudinal data are required if we are to gain a better understanding of the effects of reproductive attitudes and preferences on contraceptive use and fertility.