This contribution studies the variation in desired family size and excess fertility in four East African countries by analyzing the combined impact of wealth, education, religious affiliation, and place of residence. The findings show an enormous heterogeneity in Kenya. Wealthy and higher educated people have fertility desires close to replacement level, regardless of religion, while poor, uneducated people, particularly those in Muslim communities, have virtually uncontrolled fertility. Rwanda is at the other extreme: poor, uneducated people have the same desired fertility as their wealthy, educated compatriots, regardless of their religion—a case of “poverty Malthusianism.”. The potential for family planning is high in both countries as more than 50% of the women having 5 children or more would have preferred to stop at 4 or less. Tanzania and Uganda have an intermediate position in desired family size and a lower potential for family planning. Generally, the main factor that sustains higher fertility is poverty exacerbated by religious norms among the poor only.
The demographic developments in sub-Saharan Africa continue to puzzle population experts. After the turn of the millennium the stalling of the fertility decline in many countries gave new food for thought [
This paper investigates the levels and determinants of excess fertility in East Africa, using data from the latest Demographic and Health Surveys (DHSs). It links up with the debate on variations at the national and the community levels and aims at identifying the community contexts in these countries that shape both actual and desired fertility and then determine excess fertility and at finding possible target groups for future family planning programs.
To provide a better understanding of the regional background a condensed picture of the socioeconomic and demographic position of the countries concerned is presented in Section
The four countries under study are members of the East African Community (EAC). The countries differ in many ways, but they all have still a predominantly rural-based society and share a modest position on the UNDP Human Development Index list [
Thus, although the structural socioeconomic changes that are traditionally linked to changes in demographic regime are occurring in all four countries, the pace and the extent of change are not uniform. The fertility transition is on its way with a moderate to substantial decline [
Total fertility rate and desired family size for all women (year of survey).
Indicator | Kenya | Tanzania | Uganda | Rwanda |
---|---|---|---|---|
Total fertility rate | 4.6 (2008/9) | 5.4 (2010) | 6.2 (2011) | 4.6 (2010) |
Total fertility rate (previous DHS) | 4.9 (2003) | 6.1 (2004/05) | 6.7 (2006) | 5.8 (2005) |
Desired family size | 3.8 (2008/9) | 4.9 (2010) | 4.8 (2011) | 3.3 (2010) |
Desired family size (previous DHS) | 3.9 (2003) | 5.0 (2004/05) | 5.0 (2006) | 4.3 (2005) |
Source: DHSs reports.
Looking at the desired fertility (Table
The family planning programs vary in duration, government support, approach, and objectives. The oldest program is that of Kenya. It received more real support from the government during the 1980s and 1990s than the programs in other countries. The program was among the best in sub-Saharan Africa [
According to the dominant economic interpretation of fertility behavior, the main driving forces that reduce the desired fertility are structural socioeconomic transformations in societies, which lead to increasing expected costs for and diminishing benefits from children [
In this explanatory framework, fertility is seen as the outcome of a couple’s rational and conscious decision-making. Irrespective of the type of society, couples balance the costs of having children against the expected economic, social, and psychological gains they could obtain through them. The wide application of this framework in numerous studies of fertility change, however, did not stop the debate on the driving forces of fertility decline. The first issue is whether the cost-benefit approach is equally valid in every phase of the demographic transition [
Particularly in the early to mid phases of the transition, couples are expected to conform to their community’s norms and attitudes concerning reproductive behavior, because familial and social networks are important in a context where formal social security systems are absent and trust in state institutions is weak. People rely on their familial and social networks for access to resources and support in the case of need and will therefore not deviate strongly from the shared values and norms of their community [
Several socioeconomic factors changing with socioeconomic development or family planning campaign affect the demand for children (and then excess fertility). Educational level that slows down traditional beliefs seems to be the most important. A substantial amount of empirical literature demonstrates that there is a strong negative correlation between educational level and fertility preferences and behavior [
Regardless of the indicator used, studies show that the desired family size is negatively associated with economic position: couples with a low position desire and have more children than those with a higher position [
Finally, the level of child mortality plays an important role to determine the desired family size and the decision to use contraception to achieve the desired one, through the mechanisms of insurance and replacement mechanisms [
The phenomenon of excess fertility is directly related to lower fertility preferences, because a lower desired fertility can be achieved only by limiting the number of births. This requires adaptations in sexual behavior and access to and trust in contraceptive means. Unless these conditions are fulfilled, the ultimate family size will probably surpass the desired one.
According to the classical demographic transition theory, the desired fertility generally declines and excess fertility increases in the early and mid phases of the transition, following an inverted U-shaped curve [
In the mid and late transitional phases, reproductive change spreads rapidly as more population groups profit from socioeconomic development and the diffusion of new ideas through social interaction processes reinforces rather than inhibits demographic change. Traditional sociocultural factors gradually lose their influence, leading to less resistance to having a limited number of offspring and a drop in the social costs of contraception. Hence, excess fertility peaks before it starts to diminish substantially. In those phases, fertility behavior is more consistent with the expectations of most demographic and economic theories of fertility [
Excess fertility could be conceived as unmet need for contraception. Apart from limited access to services, four constraints influence the use of effective modern family planning methods: insufficient knowledge, fear of social disapproval, fear of side effects, and perception of husband’s opposition [
Most African societies are marked by gender inequities and a patriarchal social system [
The data for this research are drawn from the Kenya 2008/9, Tanzania 2005, Rwanda 2010, and Uganda 2011 Demographic and Health Survey data sets [
This study is focused on excess fertility. Excess fertility is a situation in which actual fertility exceeds desired fertility. Because of this definition, women without children were removed from this analysis. Actual fertility is the number of living, not ever born, children a woman had at the moment of interview. Desired fertility is obtained from the question on the ideal number of children, a question that aims to measure fertility preferences of the population. Excess fertility is also called unwanted fertility [
The dependent variable is a binary variable with value zero if the actual number of children is less than or equal to the desired number and one if otherwise. As the outcome variable is binary, having excess fertility or not, we will be using the logistic regression model that converts the outcome into a logit function, providing the probability/odds of being with excess fertility or not.
In this research, excess fertility is assessed as a function of various socioeconomic and sociocultural factors that shape both the desired and actual fertility from which excess fertility is drawn. The logistic regression establishes a linear relationship between the logarithm of
The logistic regression model is written as follows:
To obtain data about the desired number of children, two questions were used in the DHS questionnaire. The one for women with children was: “If you could go back to the time you did not have any children and could choose exactly the number of children to have in your whole life, how many would that be?” Although responses to these questions should be integer values, between 1.1 and 7.3 per cent of respondents gave nonnumerical answers, such as “It depends on God,” “As many as I can support,” or “As many as my husband wishes.” These respondents were removed from the data set, although it is acknowledged that this exclusion may have distorted the results. However, considering the limited number of excluded respondents, the bias is acceptable.
These questions about the ideal number of children are aimed at measuring the reproductive norms in the population and providing a quantitative basis for assessing variation in desired and actual fertility. The desired fertility also enables the calculation of excess fertility through the comparison of desired and actual fertility. Despite this important objective, responses to the question about the ideal number of children, as an indicator of the desired fertility, have been criticized regarding their validity and reliability [
The first criticism relates to the survey questions themselves and the answers they elicit. The answers could be misleading, as they reflect unformed, ephemeral views that change during the life course, and the effects of child mortality risks are not explicitly taken into account in the questions. Respondents presumably do not include possible child deaths in their ideal family size and may need to bear additional children in order to achieve that size. Thus, the total number of births in a marriage may exceed the desired familysize without any child being unwanted [
The second criticism is that there could be a rationalization of the desired family size to the actual family size. Despite the likelihood that some rationalization occurs, many respondents report ideal sizes that are lower or higher than their actual number of surviving children, also in East Africa (Tables
Percentage of women with two children or more whose desired fertility is greater than, equal to, or less than their actual fertility.
Country | Category of desired fertility | Number of living children | |||||||
---|---|---|---|---|---|---|---|---|---|
2 | 3 | 4 | 5 | 6 | 7 | 8+ | Total | ||
Greater | 72.7 | 59.3 | 33.1 | 29.7 | 25.9 | 22.8 | 18.3 | 47.7 | |
Kenya | Equal | 24.7 | 24.0 | 41.5 | 16.5 | 21.2 | 11.1 | 13.9 | 24.8 |
Less | 2.6 | 16.7 | 25.4 | 53.8 | 52.9 | 66.3 | 67.8 | 27.5 | |
|
|||||||||
Greater | 96.2 | 92.0 | 72.1 | 59.9 | 42.9 | 42.8 | 26.1 | 77.3 | |
Tanzania | Equal | 3.7 | 6.4 | 21.6 | 19.0 | 29.0 | 13.0 | 15.0 | 10.8 |
Less | 0.2 | 1.6 | 6.3 | 21.2 | 28.2 | 44.3 | 58.9 | 11.9 | |
|
|||||||||
Greater | 93.2 | 87.3 | 59.5 | 55.4 | 32.0 | 31.5 | 15.4 | 60.8 | |
Uganda | Equal | 5.7 | 6.3 | 32.0 | 13.0 | 26.0 | 8.1 | 7.7 | 14.2 |
Less | 1.0 | 6.4 | 8.5 | 31.6 | 42.0 | 60.4 | 76.9 | 25.0 | |
|
|||||||||
Greater | 82.3 | 55.8 | 23.6 | 16.6 | 11.0 | 10.6 | 6.1 | 41.5 | |
Rwanda | Equal | 17.0 | 33.6 | 39.2 | 11.7 | 10.7 | 4.0 | 6.5 | 22.1 |
Less | 0.7 | 10.6 | 37.2 | 71.7 | 78.3 | 85.4 | 87.4 | 36.4 |
Source: computation from the datasets.
At parity 8 and more, 59 to 87 per cent of the women declared to be with excess fertility. Even if there is rationalization in some cases, this does not apply to the majority of the respondents who had a high number of surviving children. The correlation coefficients between the number of surviving children and the ideal number show a weak relationship; the values range from 0.32 for Rwanda to 0.34 for Kenya, 0.43 for Tanzania, and 0.31 for Uganda.
The third criticism is that responses from women alone may not describe the real norms that guide fertility decisions. The attitudes of other family members, especially husbands or partners, may exert a great influence on reproductive decisions, particularly in sub-Saharan Africa where stronger unbalanced gender relations prevail. In practice, however, the importance of this criticism is doubtful. The evidence from surveys in which both husbands and wives were interviewed suggests that there is no radical difference between the views of the two sexes [
Excess fertility was modeled as a function of various socioeconomic and sociocultural variables that define communities which are the focus point of this research. We have restricted the analysis to socioeconomic and cultural variables, because they are exogenous to the dependent variable. Demographic factors, like mortality, have been excluded from the analysis since they operate through the components of excess fertility, namely, actual and desired fertility. Only the number of living children is used as it is an important variable of control. Also, in order to avoid the problem of endogeneity, such proximate determinants as knowledge of contraception and approval of contraception were not included in the analysis.
We shall note that socioeconomic and sociocultural factors operate differently. While the first group leads to desired lower fertility, the second is expected to reflect a pronatalist attitude. To check the effect of communities on the excess fertility, an interaction effect of the religious affiliation with educational level was included. For more details about independent variables and their categories, see Table
Percentage distribution of respondents by background characteristics.
Variable | Categories | Kenya | Tanzania | Uganda | Rwanda | ||||
---|---|---|---|---|---|---|---|---|---|
% |
|
% |
|
% |
|
% |
| ||
Woman’s education | No education | 15.0 | 653 | 28.9 | 1742 | 19.3 | 941 | 19.6 | 1242 |
Inc. primary | 29.3 | 1278 | 16.5 | 992 | 46.0 | 2237 | 54.5 | 3458 | |
Comp. primary | 26.5 | 1155 | 45.4 | 2732 | 11.2 | 545 | 14.7 | 930 | |
Secondary and + | 29.3 | 1279 | 9.2 | 556 | 23.5 | 1145 | 11.2 | 707 | |
|
|||||||||
Woman's occupation | Cultivator | 26.5 | 1157 | 70.6 | 4253 | 52.9 | 2575 | 76.5 | 4846 |
Craftswoman | 4.9 | 215 | 11.4 | 685 | — | — | 4.1 | 259 | |
Works in services | 30.9 | 1350 | 4.4 | 265 | 27.5 | 1339 | 9.2 | 584 | |
Others | 37.6 | 1643 | 13.6 | 819 | 19.6 | 954 | 10.2 | 648 | |
|
|||||||||
Woman’s religion | Catholic | 19.4 | 848 | 23.3 | 1403 | 43.8 | 2134 | 42.4 | 2685 |
Protestant | 62.8 | 2740 | 23.6 | 1423 | 41.3 | 2012 | 54.2 | 3432 | |
Muslim | 13.7 | 598 | 42.6 | 2563 | 13.8 | 671 | 1.6 | 99 | |
Others | 4.1 | 179 | 10.5 | 633 | 1.0 | 51 | 1.9 | 121 | |
|
|||||||||
Partner’s education | No education | 11.2 | 491 | 19.4 | 1167 | 11.3 | 551 | 19.5 | 1235 |
Inc. primary | 18.7 | 818 | 18.9 | 1136 | 36.7 | 1787 | 50.9 | 3225 | |
Comp. primary | 28.2 | 1233 | 48.7 | 2935 | 15.8 | 767 | 16.2 | 1028 | |
Secondary and + | 41.8 | 1823 | 13.0 | 784 | 36.2 | 1080 | 13.4 | 849 | |
|
|||||||||
Wealth index | Poorest | 20.2 | 882 | 18.7 | 1129 | 23.0 | 1120 | 19.5 | 1237 |
Poorer | 16.4 | 715 | 19.4 | 1170 | 18.7 | 911 | 19.9 | 1264 | |
Middle | 18.0 | 785 | 19.2 | 1159 | 17.1 | 831 | 19.7 | 1248 | |
Richer | 19.4 | 847 | 23.0 | 1386 | 16.4 | 797 | 20.3 | 1289 | |
Richest | 26.0 | 1136 | 19.6 | 1178 | 24.8 | 1209 | 20.5 | 1299 | |
|
|||||||||
Type of residence | Urban | 27.8 | 1213 | 20.2 | 1218 | 24.2 | 1180 | 14.8 | 937 |
Rural | 72.2 | 3152 | 79.8 | 4804 | 75.8 | 3688 | 85.2 | 5400 | |
|
|||||||||
Partner desire for children | Same | 51.0 | 2228 | 39.0 | 2347 | 28.3 | 1376 | 58.0 | 3675 |
More | 18.6 | 813 | 32.4 | 1948 | 28.6 | 1390 | 10.5 | 663 | |
Fewer | 6.3 | 273 | 4.5 | 273 | 8.8 | 426 | 17.9 | 1136 | |
Don’t know | 24.1 | 1051 | 24.1 | 1454 | 34.4 | 1676 | 13.6 | 863 | |
|
|||||||||
No living children: mean (StD) | 3.4 (2.09) | 3.6 (2.26) | 4.03 (2.34) | 3.5 (2.0) | |||||
|
|||||||||
Total | 100 | 4365 | 100 | 6022 | 100 | 4868 | 100 | 6337 |
Source: computation from the datasets.
As stated early, this study is focused on excess fertility which is defined in reference to desired and actual fertility. To understand the existence of excess fertility, our analysis starts with descriptive statistics on desired fertility that show the variability of demand for children between and within countries. Descriptive statistics are presented with respect to the two important socioeconomic and sociocultural factors that define communities of interest, mainly educational level and religion.
According to Table
Mean ideal number of children by country according to religion, education, and wealth.
National level | Kenya | Tanzania | Uganda | Rwanda | ||||
---|---|---|---|---|---|---|---|---|
4.3 | 5.7 | 5.3 | 3.6 | |||||
Christians | Muslims | Christians | Muslims | Christians | Muslims | Christians | Muslims | |
Educational level | ||||||||
No education | 5.9 | 8.3 | 5.8 | 7.0 | 6.8 | 6.5 | 4.0 | 4.0 |
Inc. primary | 4.1 | 5.6 | 5.4 | 6.1 | 5.3 | 5.3 | 3.6 | 3.2 |
Comp. primary | 3.7 | 5.1 | 4.9 | 5.1 | 4.7 | 5.1 | 3.5 | 3.1 |
Secondary and + | 3.2 | 3.9 | 3.6 | 5.7 | 4.2 | 4.4 | 3.5 | 3.5 |
Wealth index | ||||||||
Poorest | 4.9 | 8.3 | 5.8 | 6.3 | 6.3 | 6.0 | 3.6 | 3.4 |
Poorer | 4.0 | 7.4 | 5.5 | 6.3 | 5.1 | 5.7 | 3.7 | 3.6 |
Middle | 4.0 | 5.9 | 5.3 | 6.4 | 5.2 | 5.4 | 3.6 | 2.6 |
Richer | 3.6 | 5.4 | 4.9 | 6.2 | 5.3 | 5.1 | 3.8 | 4.0 |
Richest | 3.1 | 4.5 | 3.9 | 5.1 | 4.3 | 4.7 | 3.5 | 3.2 |
|
||||||||
Total Religion | 3.8 | 6.7 | 5.1 | 6.0 | 5.3 | 5.2 | 3.6 | 3.4 |
Source: computation from the datasets.
Kenya exhibits the largest diversity of fertility preferences according to either religious communities or socioeconomic ones. Rwanda has the lowest diversity. In Kenya, Muslims desire nearly twice the number of children desired by Christians (6.7 versus 3.8) whereas in Rwanda they desire slightly less than Christians, but the difference is small (3.4 versus 3.6). The homogeneity between religious communities is also a feature of Uganda, yet with higher preferences than in Rwanda: 5.2 children for Muslims and 5.3 for Christians. Tanzania occupies an intermediate position: Muslims prefer one more child than Christians, respectively, 6 and 5 children.
The differences in fertility preferences appear also when considering educational or economic communities. With regard to education, irrespective of religion, the differences are clear and straightforward in all countries. The desired fertility decreases progressively with the level of education. However, there are large variations across religious communities. The gap between women with no education and those who have reached the secondary school or more between Christians and Muslims is striking only in Kenya where the gap is 2.7 for Christians and 4.4 for Muslims. In all other countries the gap is very limited: less than one child.
With reference to economic status measured with wealth index of households, the pattern is similar to education, but the differences are less pronounced, except for Muslims in Kenya where the lowest quintile desires 3.8 more children than the highest. Everywhere else, the gap is lower than 2.0 children.
Considering the variations between religious communities in the same socioeconomic stratum, Table
The differentiation in excess fertility is demonstrated through a graph that indicates the level of excess and a regression analysis which identify the determinants.
Figure
Percentage of women having excess fertility according to number of living children (source: Table
These results indicate that a very substantial number of Rwandan and Kenyan women would want to limit their childbearing earlier than women in the other countries. In Kenya and Rwanda, we even find a category that would have preferred to stop at three children.
The results of the binary logistic regression are shown as log-odds ratios in Table
Log odds of being in excess fertility from the binary logistic regression.
Variable | Kenya | Tanzania | Uganda | Rwanda |
---|---|---|---|---|
Constant | −8.236*** | −10.138*** | −9.610*** | −9.511*** |
No living children | 1.809*** | 2.064*** | 1.715*** | 3.029*** |
No living children squared |
−0.096*** | −0.100*** | −0.073*** | −0.193*** |
Residence (ref. rural) | ||||
Urban | −0.078 | 0.153 | 0.316** | −0.020 |
Woman’s religion (ref. catholic) | ||||
Protestant | −0.124 | −0.180 | 0.441** | −0.036 |
Muslim | −2.352*** | −1.182*** | 0.085 | −0.649 |
Others | −1.374*** | −0.782*** | 0.555 | 0.155 |
Woman education (ref. no education) | ||||
Inc. primary | 0.481 | 0.075 | 0.756*** | 0.145 |
Com. primary | 1.053*** | 0.162 | 1.173*** | 0.331 |
Secondary and + | 0.697* | 0.279 | 1.404*** | 0.790*** |
Husband education (ref. no education) | ||||
Inc. primary | 0.799*** | 0.270* | 0.875*** | 0.074 |
Com. primary | 0.898*** | 0.292* | 0.796*** | 0.232* |
Secondary and + | 0.629*** | 0.140 | 0.950*** | −0.083 |
Woman occupation (ref. cultivator) | ||||
Craftswoman | −0.932*** | 0.049 | (—) | 0.121 |
Worker in service | −0.126 | 0.590** | 0.170 | −0.019 |
Others | −0.299** | −0.126 | −0.146 | −0.303** |
Wealth index |
||||
Poor | 0.475*** | 0.273 | 0.133 | −0.190 |
Middle | 0.361** | 0.331** | 0.090 | −0.080 |
Richer | 0.548*** | 0.656*** | 0.001 | −0.378*** |
Richest | 0.531** | 1.125*** | 0.042 | −0.180 |
Husband desire for children (ref. same) | ||||
More | 0.852*** | 0.261** | 0.434*** | −0.017 |
Fewer | 0.303 | 0.172 | −0.379** | −0.133 |
DK | 0.212* | 0.146 | 0.115 | −0.174 |
|
||||
Interaction effects | ||||
Protestant/education (ref. no education) | ||||
Inc. primary | 0.169 | −0.078 | −0.418 | −0.033 |
Com. primary | −0.391 | 0.289 | −0.737** | −0.433 |
Secondary and + | 0.324 | 0.855 | −0.599* | −0.532* |
Muslim/education (ref. no education) | ||||
Inc. primary | 1.327** | 0.463 | −0.579 | 0.776 |
Com. primary | 0.793 | 0.923*** | −0.523 | 1.247 |
Secondary and + | 2.194*** | −0.132 | −0.840* | −1.352 |
Other religions/education (ref. no education) | ||||
Inc. primary | 0.464 | 0.168 | −0.725 | 0.524 |
Com. primary | −1.100 | −0.025 | 0.188 | 0.745 |
Secondary and + | 1.819 | (—) | (—) | −0.679 |
|
||||
Pseudo- |
0.513 | 0.493 | 0.527 | 0.631 |
|
4365 | 6022 | 4868 | 6337 |
(—) indicates that there are insufficient cases in this category to calculate a coefficient.
Looking at the other factors, again it turns out that the heterogeneity in the awareness of excess fertility is most prominent in Kenya. The effects of education of the woman and her partner are highly positive, as is the effect of increasing wealth. The higher socioeconomic classes in this country not only want fewer children, but are also more aware of having too many. Muslims provide a special case again. Given the actual number of children they have, few of the uneducated indicated that they would have preferred fewer children (−2.35). However, this is offset by the interaction effect of the higher educated Muslims (+2.19), meaning that this group is just as aware of excess fertility as higher educated Catholics or Protestants.
In Tanzania, the role of education is less pronounced and the women’s educational level is not even significant. The effect of increasing wealth is stronger than it is in Kenya. Even the better educated Muslims are less aware of excess fertility than their Christian counterparts. Tanzania is the country in which religious affiliation has an impact on fertility preferences for all socioeconomic strata. Uganda resembles Kenya in the role of education, of both women and their partners, but the effects of wealth and religion are completely different; the poor are more aware of having too many children than the Ugandan middle and richer classes even if the results are not significant. Just as in Rwanda, this could be interpreted as a form of “poverty Malthusianism.” Women are aware that they do not have the means to support large families. Religion does not play a significant role in this awareness.
A striking outcome for all countries, except Uganda, is that the effect of the urban/rural divide disappears after controlling levels of education, occupational structure, wealth, and religious composition. The urban/rural dichotomy is a container concept that captures the socioeconomic and sociocultural differentiation of the population rather than being an explanatory factor in itself. Table
Parity (mean number of children) at which excess fertility starts to be dominant (>50%) within women for highest and lowest socioeconomic classes by religion.
Country | Religion | |||||
---|---|---|---|---|---|---|
Catholic | Protestant | Muslim | ||||
Highest category | Lowest category | Highest category | Lowest category | Highest category | Lowest category | |
Kenya | 5 | 8 | 5 | 8 | 6 | UF* |
Tanzania | 5 | 8 | 5 | 9 | 7 | UF* |
Uganda | 6 | 9 | 6 | 8 | 7 | 9 |
Rwanda | 4 | 5 | 5 | 5 | 5 | 5 |
Source: computation from the results of the model in Table
It confirms the conclusion that in Kenya and Tanzania there is a large contrast between the wealthiest and best educated part of the population on the one hand, and the poor, uneducated part on the other. In these countries, even at parity 12 or higher, fewer than 50% of the poor uneducated Muslim women are estimated to be in excess fertility. Given the fact that they also want many children, this means that the preferences of these women are near to uncontrolled fertility. Uganda does not show any signs of religious differentials but Kenya and Tanzania exhibit important social-economic contrasts. In Rwanda, having more than four or five children is problematic in all strata of society.
There are remarkable differences in desired and excess fertility between the four East African countries and between certain communities in those countries. The differences are largest in Kenya and smallest in Rwanda, while Tanzania and Uganda occupy intermediate positions. Our outcomes also contribute to the understanding that both socioeconomic and sociocultural factors should be taken into consideration when studying the fertility behavior and that relations that hold for one community in a country are of no or less importance for another. New attitudes to desired family size diffuse along different paths within the various communities in a specific national context.
The effect of education—which is known to be one of the most important determinants of change in fertility attitudes and behavior—differs widely between the religious communities in a country. This difference probably relates to the minority status of these communities, and perhaps also to the strong social cohesion and control within them. The large communities of poor/uneducated Muslims in Kenya and Tanzania in particular seem to differ in reproductive norms from the Christian communities. The proportion of Muslims in Rwanda is very small, and the difference between religions communities is absent. The variation within religious subgroups undermines the role ascribed to religion. To get a clear picture of the significance of religion, a more precise categorization of religious communities would be preferable in future research.
The finding that the variation in desired family size and excess fertility according to religion within the socioeconomic strata in every country is limited supports the conclusion that the main factor that sustains the higher demand for children in sub-Saharan Africa is poverty rather than religious norms. However, the most remarkable outcome of our analysis is the consensus among Rwandan population groups regarding desired fertility and excess fertility. This country shows homogeneity in desired family size for the various communities with only a few determinants displaying some significant impact. Contrary to the other East African countries, the awareness of being in excess fertility is found among all Rwandan communities. This limited differentiation can be understood if the current land problems are taken into account. The desire for fewer children might be a result of population pressure on the land and a lack of labor opportunities outside agriculture, which negate any current or future benefits from children’s work. Instead, children are seen as a burden in terms of extra mouths to feed and extra outlay on school fees, clothes, and health care. These findings confirm the assumption that the desired fertility is the outcome of parents’ assessment of the costs and benefits of their offspring, besides the future opportunities of their children. Land problems would also explain why in Rwanda excess fertility is higher among agrarian and poorer communities than it is among similar communities in neighboring countries. It indicates that Rwanda is likely to be undergoing a type of demographic transition that in Latin America and Asia is called “poverty Malthusianism.” Its fertility decline does not depend on socioeconomic development, but is induced by poverty and a lack of income-generating activities.
The last implication of our findings is that low desired fertility is a necessary but not a sufficient condition to bring down actual fertility; unmet need is also a major determinant. Rwanda provides a good example. The sensitizing campaign brought about a sharp decrease in desired family size to 3.3 in 2010. The expansion of reproductive health services was therefore welcomed by the population, leading to a very large drop in actual fertility to 4.6 in 2010. Yet excess fertility is still prominent and the potential for further reduction by improving access to services is still high. In other countries the situation is more complex.
In Kenya, there is a need to orient the campaign towards specific communities that are still in a pretransitional phase of fertility transition. Without a reduction of desired fertility, family planning programs will not be effective. In Tanzania, the high desired fertility and low access to family planning are both determinants of the high fertility. Therefore, family planning programs should focus on both aspects, that is, they should aim at reducing higher desired fertility and at meeting the need for family planning. Family planning programs should be more oriented towards certain groups, such as rural Muslims and poorer people.
For details see Table
The authors declare that there is no conflict of interests regarding the publication of this paper.