Ngalinda, Innocent: Age at First Birth, Fertility, and Contraception in Tanzania


Chapter 2. Analytical Framework and Literature Review

2.1. Introduction

The aim of this chapter is to give an introduction of other scholars’ work on the subject matter. The discussion will concentrate on findings based on sub-Saharan Africa. The main focus of this chapter will be the current state of knowledge on age at first motherhood and the fertility change in sub-Saharan Africa. The debate in the literature regarding the onset of the fertility transition in this region is the central issue addressed. The determinants of African fertility are outlined. The discussion focuses on the proximate determinants of fertility putting more emphasis on age at first birth and contraception. This is followed by a review of socio-economic and demographic determinants of fertility and their applicability in sub-Saharan Africa.

2.2. Conceptual Framework

As Davis and Blake (1956) pointed out that cultural, social, and economic settings impinge on fertility through the intermediate fertility variables. There could be biological or behavioural factors that determine exposure to sexual intercourse and hence to child bearing. The relative importance of each variable may differ from one society to another. In particular, fertility is directly determined by intermediate variables. The intermediate fertility variables that will be examined in this study are duration variables measured in terms of the time elapsed before a particular event occurred. The beginning of child bearing which is measured by the age of a woman at the birth of her first child is the dependent variable.

Although age at first sexual intercourse is the first step in the child bearing process, it is closely associated with age at marriage in most traditional set-ups. Hence age at first marriage is always considered to represent the beginning of the exposure to the risk of child bearing. However, in recent days child bearing is not just confined within marriage but is also taking place outside marriage due to prolonged delay in the entry


into unions. In addition, child bearing is now a common phenomenon among adolescents as a result of adolescent promiscuity.

In the past most African societies considered virginity to be essential for the first marriage and premarital pregnancy was a social embarrassment among most ethnic groups in Africa. Today however, this is being accepted as an inevitable consequence of the modernisation process even among the conservative nomadic communities. In the past for example, among the Wanyambo of Tanzania, a girl who became pregnant before marriage was required to confess, and the man responsible was forced to marry her. Today the man responsible is only required to pay a bride price to her parents and to accept supporting the child financially.

However many societies that traditionally condemned premarital sexual intercourse have become much more tolerant nowadays. For example the Baganda in Uganda changed their attitude towards premarital sexual activities. Whereas parents used to control the sexual behaviour of their adolescent children by having them living in the parental home, nowadays they indirectly encourage premarital sexuality by building separate houses for the sexually active adolescents (Ntozi and Lubega, 1990).

The rise of age at first marriage in Tanzania in the recent past as seen in the first chapter can be attributed to several factors, among them are the changing educational policies which are now geared towards boosting women’s education. Economic changes reflected by rising standards of living and stimulation of women’s employment are beginning to change people’s attitude towards early marriage.

As contraception is not widespread in Tanzania, fertility could, among other factors, be mostly determined by age at sexual intercourse as well as the duration of exposure to the risk of pregnancy, age at first marriage, frequency of intercourse, and age at first birth. Behavioural factors that may be modified by the level of education, religion, place of residence may also play a key role in determining fertility. Studies have hypothesised that women who start child bearing at an early age especially in the non-contraception societies are likely to end up with higher completed family size than their counterparts who start at later age provided other fertility depressing factors do not intervene.


Age at first birth as the dependent variable is influenced by a number of factors that could be social, economic, cultural and demographic. Among these are the background determinants and the proximate determinants of age at first birth. Social background has a moderate yet significant effect upon the timing of the first birth, but it seems that education and pre-marital sexual experiences also have a strong influence on age at first birth. Higher education provides women with status or opportunities that reduce the importance of early child bearing. As the use of contraception becomes more prevalent, age at first intercourse will decrease while the age at first birth will vary independently of age at first marriage, and fertility may be lowered.

In this thesis the analytical framework will constitute independent variables which operates through intermediate variables or proximate determinants to influence dependent variables. This study therefore deals with three models. The first model treats age at first birth as a dependent variable (proximate determinants of this model have already been discussed above). The second model is on fertility or rather the number of children ever born which is taken to be a dependent variable (the proximate determinants of the second model will be discussed in the next chapter). Contraceptive use similarly is treated as a dependent variable in the third model.

There are two main proximate determinants of the use of contraception: motivation to control fertility and the cost of regulation. Both operate through a set of socio-economic and demographic variables to affect the use of fertility regulation. At any given point, motivation is regarded as a function of the interaction between the supply of children (actual number of surviving children) and the demand for children (number of children desired). The cost of regulation includes economic costs (money and time), social costs (the outcome of transgressing social norms favouring child bearing), and health and psychological costs (the consequences of experimenting with something new that may be risky or unpleasant).

Factors affecting motivation to use contraceptives include:

  1. the desire for children, which is the most direct measure of the motivation for use;


  2. demographic status, which is measured by age and the number of living children. Older women and women with more than five children are likely to be more motivated to use contraception than younger women with low parity.
  3. reproductive knowledge, which is measured by accurate knowledge of the ovulation cycle;
  4. costs and benefits of children, measures related to the costs and benefits of children including education and place of residence;
  5. family life values;
  6. exposure to family planning, information, education, and communication (IEC).

However in this study, family life values and the exposure to family planning and IEC will not be examined.

Factors affecting the cost of contraceptive use include:

  1. economic costs, which is measured by the accessibility of the source and the time taken to reach it, and economic circumstances of the household;
  2. normative and psychological costs, which are measured by a woman’s own approval of family planning, the number of modern methods of which she approves, her perception that husband and religion approve family planning, and exposure to IEC through various media;
  3. social costs, which are measured by spousal communication about fertility and agreement about fertility goals, the influence of relatives on the decision to seek family planning information, and place of residence.

In lieu of these factors the Tanzanian fertility can be analysed by using a simple framework (see Figure 3 ). Age at first birth is directly linked with fertility as women who enter early into motherhood and are sexually active for a long period in their life have higher fertility than their counterparts who delay their first births. This is well covered by findings discussed in the introductory chapter.


Figure 2: A Conceptual Framework to study Tanzania fertility


POR = Place of Residence

AFM = Age at First Marriage

AFB = Age at First Birth

AFSI = Age at First Sexual Intercourse

CMS = Current Marital Status

Poly = Polygyny

The above Analytical framework can be simplified as follows:

Figure 3: A simplified framework for the study of fertility in Tanzania

The duration of the exposure to the risk of becoming pregnant is conventionally defined as the interval between the entry into the first union and the first live birth. To base this definition on first union or first marriage does not seem to make sense in a modern societal context anymore. Since the number of premarital sexual activities and births increases, marriage or rather first union does not signify the entry into the risk of pregnancy any longer as explained earlier. Whether the entry into sexual relations coincides with marriage is not significant here because we are concerned with the interval between the first sexual intercourse and the first live birth. Prolonging age at


entrance into sexual union will certainly result in lower birth rates. However, early age at first birth will result in high birth rates, other things being equal.

The socio-economic and demographic factors are perceived as determining the timing of birth and the duration of the exposure to the risk of becoming pregnant for those who have entered into sexual relations, which in turn directly influence the age at first birth. Therefore both factors influence fertility. These factors also influence the use or non-use of contraception that in turn influences the birth interval. Early entry into sexual relations with contraception can prolong the birth of the first child. Contraception on the other hand has a negative effect on childbirth hence reduces fertility.

2.3. Literature Review

2.3.1. Age at first motherhood

In the demographic literature, age at marriage has long been regarded as one of the proximate determinants of fertility (Davis and Blake, 1956; Bongaarts, 1982). However the empirical evidence for the effects of age at marriage on fertility is inconsistent (Durch, 1980; van de Walle and Foster, 1990). As a result, the effect of age at marriage on fertility in developing countries remains mostly speculative.

There are two schools of thought on the effect of age at marriage on fertility. Some argue that it has a major impact on fertility because the female reproductive span of life is determined by age at marriage. While this may be true when age-specific fecundability is invariant for changes in age at marriage, others argue that the contribution of age at marriage to fertility may be limited. They assume that a couple, who marries later, will compensate this delay by reducing the birth interval. This latter effect is likely to be of much less importance when age and duration of marriage interact to influence fecundability.

In most situations, age at marriage may have no effect on fertility. First, if women start having children no matter when they marry, then the effect of age at marriage on fertility may be limited. Second, if fertility is controlled within marriage by using


contraceptives or other means, age at marriage may not have much of an effect on fertility because couples may decide how many children they would like to have regardless of the age at marriage. Third, if the level and pattern of fecundability depends upon marriage duration and little on age, fertility is likely to be affected little by age at marriage because a woman, whether she marries early or late, will have the same fertility experience. Although fecundity is related to the age of a woman, it does not depend on age at marriage.

In Tanzania many traditional values and social practices concerning procreation have undergone changes during the course of modernisation. There is evidence that premarital sexual behaviour is on the increase. It is a fact that adolescents in Tanzania engage in early sexuality and subsequently have early births according to the 1991/92 and 1996 TDHS data. It was found that almost half of the women aged 15-19 years had sexual intercourse, 21 percent of all respondents in this age group had at least one birth (Bureau of Statistics, 1993 and 1997). At the same time large proportions of women choose to remain single but bear children. Therefore, it seems that in modern times the concept of marriage being universal and births occurring only in marriage is a misleading concept. Blanc and Rutenberg (1990) found that it was 12-67 percent of women sub-Saharan Africa experienced intercourse one or more years prior to their first union or marriage). This suggests that age at first marital union is insufficient to capture all sexual activities.

A growing number of studies suggest that family wellbeing is conditioned by how soon child bearing begins and how rapidly it proceeds. The complete family sizes appear to be strongly influenced by age at first motherhood. Bumpass (1978) in a study on age and marital status at first birth in the USA came up with a conclusion that both a young age at first birth and premarital first conception might be associated with rapid subsequent fertility. Therefore, he associated adolescent motherhood with rapid subsequent fertility.

The timing of marriage and child bearing appears more recently to be undergoing changes towards the direction of longer delays as revealed by a study done in India (Boom and Reddy, 1986). This implies that rapid rise in age at first birth appears likely for younger cohorts. The Demographic and Health Survey data also reveal that in 14 out


of the 19 sub-Saharan countries studied, the average ages at first birth for women aged 45-49 exceed those of women aged 40-44 (Westoff et al., 1994). In this study, it was found that the average age at first birth for women aged 35-39 is lower than that for women aged 40-44. Thus there seems to be a decline in age at first birth among younger mothers. However, it is possible that this decline could partly be due to the omission of first births or the misplacement in time of first births in the maternity histories reported by older cohorts of women.

Gaisie (1984) in a study on the proximate determinants of age at first birth in Ghana found that low age at first birth persists in the country. A cohort analysis of the mean age at first birth in Nigeria using World Fertility Survey data revealed a declining age at first birth. The study also shows that the major proximate determinants of age at first birth are age at marriage, age at first sexual relations and age at menarche. In the analysis of the Kenyan fertility survey, Konogolo (1985) found that women in Kenya start child bearing early as well. But it is interesting to note that in those provinces, where child bearing started early, fertility was above average. This shows that there is an inverse relationship between age at first birth and fertility.

Women who start child bearing at early ages are likely to have lower levels of education. They are likely to be rural residents or urban poor. Studies have shown that these women would adhere more to traditional patterns of birth spacing that results in long birth intervals and thus reduced fertility (Trussell and Reinis, 1989). Denied access to good medical and nutritional facilities might lead to problems of infecundity, especially of very young mothers. Infecundity would subsequently depress their reproductive potential.

In many countries, education, particularly women’s education, has been demonstrated to have a significant effect on fertility. Education brings in a new outlook on life as well as skills for taking advantage of new opportunities. A rise in the level of women’s education leads to a rise in age at first birth and consequently to a decline in fertility. Studies done in Latin America have shown that education is probably the most important socio-economic variable associated with greater occupational differentiation and social mobility both of which can affect nuptiality and the reproductive behaviour in various ways (Weinberger et al., 1989).


Women with higher educational levels are more likely to break with traditional patterns including early marriage and child bearing. Education indirectly influences age at first birth, and change in the traditional work role. Women with gainful employment may be more likely to postpone marriage and even child bearing within marriage. Maxwell (1987) in a study of 5,000 US women observed a positive relationship between education and age at first birth.

With regard to education, Gaisie (1984) found that the median age at first birth for women with secondary or tertiary education was 25 years compared to 19 years for the middle and primary school leavers. Similar studies in Kenya by Konogolo (1985) confirmed that post-primary schooling (especially of 9 or more years) has a strong effect in postponing the onset of fertility often by 3 to 4 years.

One of the most important variables for marriage and child bearing can be the influence of religious beliefs and practices, which therefore became important background variables for age at first birth. Studies done in India indicate that Hindus marry and bear children at younger ages than non-Hindus (Bloom and Reddy, 1986). In Tanzania, religion influences age at first marriage and, of course, age at first birth in religions such as the Islam that encourages early marriage. This will ultimately mean early age at first birth. The main reason for this encouragement (for Moslems to marry early) is the emphasis the Islam puts on premarital virginity. However religion can also influence the level of contraceptive use and therefore has an effect on age at first and subsequent births. Some religions like Catholicism have negative attitudes towards the use of modern contraceptives while others for example, Protestantism, have a more liberal stand. In societies where traditional norms and values are fading away, Catholics are likely to have low ages at first birth and short intervals between subsequent births. Thus fertility might be high for Catholics. This statement was confirmed by a study carried out in Sierra Leone by Gage (1986), who noted that Catholics had a lower age at first birth than Moslems.

Place of residence is a useful measure or indicator of the degree of change from traditional or rural behaviour to a modern or urban behaviour. Significant rural-urban differences in marriage and fertility timing are partly the result of the greater impact of


education on age at first marriage, and the incidence of cohabitation and first birth in urban areas in comparison with those in rural areas (Laurie, 1986). Urban marriage, cohabitation and first birth distribution appears to be more dispersed than the rural distributions. Urban women have greater heterogeneity in their marriage and fertility patterns.

Ethnicity is associated with age at first birth, as one of the main functions of culture is to maintain the biological continuity of the members of the society. This fact is supported by Ohadike’s argument (1979) that although natural fertility variations are primarily determined by biological process, it might be modified by socio-cultural factors. Every cultural group has its own socio-cultural ideologies of biological functions and their social accommodation (socio-continuities). These ideologies are comprised of the norms, beliefs and values as well as the practices that are likely to affect positively or negatively the reproductive performance of a given society. It is likely that each cultural group may stress certain aspects in their reproductive institutions. These may form peculiar elements that may serve to explain fertility differentials and levels to a greater or lesser extent from the fertility levels of other cultural groups.

In the traditional settings of most Tanzanian societies teenagers are encouraged to marry due to pressure from parents, peer groups and the society as a whole. These societal measures are being changed in order to conform to the government’s policy for controlled population growth. It must however be emphasised that the timing of marriage and childbirth still differs among the various ethnic groups in Tanzania. For instance, the inhabitants of the Lake regions and Southern zones marry early and start child bearing at an early age. A point to note here is that the mean number of children ever born to women who are now 40-49 years old is almost the same as in other parts of the country (Bureau of Statistics, 1997). This means that starting child bearing earlier than others does not necessarily lead to a greater number of children than those who start late.

Other things being equal, an earlier age at marriage for women whose first sexual intercourse proceed follow the first marriage, implies either a higher level of life time fertility or a longer period of exposure to the risk of unwanted child bearing once the


desired family size has been attained. The way in which initial fertility is postponed is by delaying marriage and not by using contraception to delay the first pregnancy. Early marriage may result in the delay of a first birth due to sub-fecundity that may be caused by coital frequency, as a woman might not be biologically mature. On the other hand, early marriage may be selective of more fecund women if premarital conception leads to marriage or where premarital conception is a precondition of marriage.

In their study, Marini and Hodson (1981) found that age at first marriage has a causal effect on the spacing of the second birth for those who conceived their first child within marriage. The timing of the first birth however appears to have a casual effect on the spacing between the first and the second birth. Their study also confirmed that as the ages at first marriage and first birth increase the incidence of high fertility after the first marriage (measured by the proportion of having a birth within a given period after marriage) decreases.

In many societies, age at first marriage is practically synonymous with age of entry into sexual relations and thus a major determinant of family formation. In a study carried out in Kenya, Bumpass and Mburugu (1977) noted that there could be a tendency among those with sexual experience for the more fecund to get ‘caught’ by pregnancy and for the less fecund to ‘survive’ the risk of pregnancy. This means that the more fecund women will marry early and be exposed to a higher risk of unwanted pregnancy than their less fecund counterparts who marry somewhat later.

Bumpass and Mburugu (1977) also observed that when marriage and child bearing began early, each subsequent stage of decision making is reached at a less mature age than for women who begin motherhood later. For example, it was noted that women who had their second child in their teenage years had to contracept or to consider the possibility of a third child in a markedly different setting of perceived alternatives than women who reached this stage in their mid-twenties. It is therefore clear that if anything is learned during adolescence that leads women to believe that one or two children could already fulfil their social and personal goals of motherhood, it will not be too late to affect their fertility. Hence, there is a need for sexual education in schools in order to inform girls as soon as possible about family planning measures etc.


2.3.2. Fertility Changes in Sub-Saharan Africa

The study of fertility in sub-Saharan Africa has been an area of interest for at least the past two decades. Since the early 1980s serious efforts were made to try to understand fertility trends and differentials in this part of the developing world. Surveys such as the Contraceptive Prevalence Surveys (CPS), World Fertility Surveys (WFS), and Demographic and Health Surveys (DHS) have made a major contribution to the study of fertility in this region.<9> The major reason for this increasing interest is the evidence of fertility decline observed in different African countries. Although a decline in fertility had not been anticipated for the near future. It has been documented that fertility decline has begun in at least three countries: Kenya, Botswana, and Zimbabwe (Arnold and Blanc, 1990; van de Walle and Foster, 1990; Cross et al., 1991; Freedman and Blanc, 1991; Robinson, 1992; Cohen, 1993; Brass and Jolly, 1993; Rutenberg and Diamond, 1993; Cleland et al., 1994) and parts of some other countries, for example Southwest Nigeria (Caldwell et al., 1992a; Cohen, 1993) and Northern Sudan (Cleland et al., 1994). Compared to Europe or North America it is argued that the African fertility transition is markedly different in certain important aspects to that experienced in pre-modern Western countries (Caldwell et al., 1992a; Robinson, 1992). We will pursue the key arguments of the debate on fertility decline in sub-Saharan Africa in the following paragraphs.

Caldwell and his colleagues, who attempted to explain the factors that influenced high fertility in sub-Saharan Africa, developed a first important concept. The ‘wealth-flow’ theory (Caldwell, 1976) is among Caldwell's early writings on the theory of fertility decline. The theory asserts that fertility decreases only when there is a change in economic relations. In other words, the level of fertility depends on whether children are a financial asset or a burden to their parents. In industrial societies, children are known to be a financial burden, as education, food, clothing, and entertainment etc cost parents


a lot. In contrast to industrialised regions, children in traditional societies are generally considered an asset. They help working on the land and are involved in the collection of firewood and water (Caldwell, 1976). They also provide security for their parents when the latter become older and infirm. According to Caldwell, since wealth flows in Africa have continued to be in favour of parents, fertility is not likely to decline. However, one can argue otherwise. Van de Walle and Foster put it this way: ‘that children often provide some security for their parents later in life is not in doubt; however, the proposition that parents of large families are better off than those with few children has not been adequately tested’ (van de Walle and Foster, 1990: 32).

In the late 1970s Caldwell (1977) further argued that high fertility is economically rational in traditional African economies where land is held by the lineage. An increasing number of direct family members provide the best form of investment available to control the land and its products. This rationality might not be justified any longer. Nowadays there is a tendency that many young people take non-agricultural jobs. This is not only because of modernisation but also because of the increasing number of foreign investors plus the fact of many offspring. The size of the land that a family holds decreases with time. For instance, the increasing scarcity of land in Northern Tanzania meant that landholdings were broken up so that most sons could inherit land. In consequence, land litigation among kinsmen increased, and the value of land in the highlands rose by 700 percent (Maro, 1974, cited in DeLancey, 1990). However, Caldwell (1977) makes the point that even when children take jobs in non-agricultural sectors, the family ties lead them to transfer remittance money to their families. Even though a family does not have much land, having many children increases the chance for parents to do well with the help of their children’s remittances. Although the traditional occupational sectors might not be the same as in the past anymore, Caldwell’s argument of the rationality of having many children might still hold for the present.

Caldwell and Caldwell (1987) have reviewed ways in which cultural and religious factors in sub-Saharan Africa affect the supply of and the demand for contraception and consequently sustain high fertility. They characterised ancestor worship and the horrors of infertility as fundamental characteristics of the African reproductive system. This point of view, however, has been questioned. It seems there is no direct evidence that


ancestor cults in Africa are still uniformly important to the extent of having a strong impact on fertility. Moreover, childlessness does not have as serious an implication as suggested by the Caldwells (van de Walle and Foster, 1990). In Africa too, life is undergoing modernisation. Old traditional cults and conventions are not as widespread anymore. Effects of modernisation in Africa can be seen in the increasing demand for modern methods of birth control, such as female and male sterilisation (Coeytaux, 1988; Rutenberg et al., 1991; Westoff and Ochoa, 1991; Robinson 1992).

In matrimonial societies, the traditional family structure gives husbands the power of decision making regarding reproduction, while placing most of the economic burden for raising children on mothers (Caldwell and Caldwell, 1987). Women are also responsible for the agricultural production (Boserup, 1985; Frank and McNicoll, 1987). These characteristics have been outlined as the major factors influencing high fertility in most of sub-Saharan African matriarchal countries. The argument is that, since husbands receive the advantages of the status and the prestige from paternity as heads of the household, while not having to bear any economic burden, they opt for a great number of children. As Page (1988) suggests that lowering fertility must be in the interest of those making decision about fertility, it implies that either mothers should be the decision-makers in reproduction or husbands should be economically responsible for their children. There is some evidence that the second option is occurring in Africa particularly in connection with the cost of educating children (Oppong, 1987).

Nevertheless it can be concluded that both the family structure and its consequences for power and decision-making, and the fact that children provide old-age security for their parents are the major socio-cultural factors causing high fertility in sub-Saharan Africa. It has been demonstrated, however, that fertility can decline without changing these conditions. In rural Kenya, for example, fertility declined in the 1980s even though the traditional concepts of family structure, decision making, and old-age security were still the same (Dow et al., 1994). Although Dow found out that fertility in these regions declined, he does not state reasons for the decline.

It seems that until the late 1980s demographers as well as other social scientists were convinced that fertility would remain high in sub-Saharan Africa. Experts assumed that a decline would not be experienced until the forces that support high fertility in the


region were weakened. For instance, Kenya has had one of the highest fertility rates in the world and was believed to have rather poor prospects for a fertility decline (Frank and McNicoll, 1987). However, this country on the whole is currently experiencing a considerable fertility decline. According to the experience of pre-modern societies, fertility in Africa was not supposed to decline, at least not in the late 20th century. The apparent fertility decline in sub-Saharan Africa, therefore, is a new experience, which in a way is a challenge to demographers.

Caldwell et al. (1992a) examined the features of the countries in sub-Saharan Africa where the fertility transition began. They concluded that Africa has a different type of fertility decline than that experienced elsewhere in the world. The common characteristics of Kenya, Botswana, and Zimbabwe include the following: First, they are the only countries in the region that have reached infant mortality rates below 70 per 1,000 live births. Second, they have unusually high levels of education compared to others. And third, they are unique in their high levels of contraceptive practice, ranging between 27-44 percent among married women. However, the predominant feature that marks these fertility declines in these African countries as a possibly new type of demographic transition is the similarity in contraceptive use and fertility decline at all ages (Caldwell et al., 1992a).

With regard to Western countries Knodel (1977) argued that during the initial stages of fertility decline, one should anticipate a larger relative decline in marital fertility rates at older ages since most couples want at least some children during the early years of marriage. As the decline of fertility progresses, this differential between young and old women is reduced (Knodel, 1977). This feature has been widely applicable to pre-modern Western countries.

When analysing the fertility transition in Kenya, Robinson (1992) has concluded that there is a general perception in Kenya that large families are a growing economic burden. This has led to a positive attitude towards family planning and a decline in the desired number of children. Further, the increase in contraceptive prevalence, particularly in modern methods, and the overall increase in the demand for contraception has contributed to the fertility transition in Kenya. In order to make better predictions about other countries in sub-Saharan Africa, Robinson wrote:


‘They [African countries] should continue effects at rapid socio-economic transformation, stressing health, education, and social infrastructure. They should not be obsessed with cultural 'barriers' to demographic change since in Africa, as elsewhere, these seem to have an ability to change and accommodate even while persisting at a formal level. Above all, it seems important to supply family planning services which stress approaches consistent with the tradition of birth-spacing’ (Robinson, 1992: 457).

The study of African fertility trends, however, has been controversial. Due to the absence of a reliable registration system of births, the major sources of data used are not vital statistics but censuses and surveys in which women were asked to report their children retrospectively. These data contain errors due to memory lapse, misreporting and omissions. It has been reported that women omit some of their children and/or misreport their own birth dates or their children's birth dates, both of which affect fertility estimations (United Nations, 1983). The changes in fertility observed during the analysis can be due to these problems and do not necessarily reflect an actual change in fertility. These problems have been minimised in recent surveys, particularly in the Demographic and Health Surveys (Arnold, 1991). However, as argued by Thomas and Muvandi (1994) for the case of Botswana and Zimbabwe, the use of different surveys to study fertility trends in a country may create biases due to different sample compositions. Therefore the study of fertility trends in Africa is not a straightforward exercise.

In Tanzania the average number of children a woman is expected to bear during her child bearing period - Total Fertility Rate (TFR) - was estimated to be 6.6 in 1967, and it had gone down to 5.8 in 1996 (Bureau of Statistics, 1997). If the fertility trend for Tanzania is closely scrutinised, it can be observed that between 1967 and 1996, almost 30 years, the TFR has only declined by one child, a marginal decline.

However, the 1991/92 Tanzania Demographic and Health Survey (TDHS) indicated that by the time Tanzanian women had completed their child bearing years, they had given birth to an average of seven children. The mean number of children ever born to a woman rises steadily with age from 0.3 children at age group 15-19 to about 7.29 children in age group 45-49. It is therefore evident that longer periods of exposure to child bearing lead to higher fertility. However some studies contradict this, for example


data collected in North America and India showed a different trend. It was found that the average number of live births to a woman, whose marriage remained intact beyond age 50, who were not using contraception, and under the same conditions, tended to differ in the average number of children ever born. Women of subgroup of the Hutterites, living in Midwestern North America were found to have an average of more than 9 live births per woman compared to Indian women (6 to 7 live births per woman on average), although the Hutterites marry at an average age of over 21 and Indian women, on average, marry several years younger (UNFPA, 1993).

2.3.3. Proximate Determinants of Fertility

It may seem superfluous to state that a birth is the result of the exposure to intercourse, the successful conception, gestation and parturition. In particular, fertility is directly determined only by a few variables: the intermediate or proximate variables. Any change in individual fertility must occur through an alteration of one, or a combination of several of these variables. The indirect determinants include socio-economic, cultural and environmental variables. The proximate variables provide a link between social, cultural and economic factors on the one hand, and the physiological process which ultimately determines fertility on the other. As noted by Freedman, ‘the proximate variables stand between fertility and all other proceeding variables. They immediately determine fertility, and all other variables act through combinations of them’ (Freedman, 1986:773).

Davis and Blake (1956) set the stage for rapid advance by formulating a framework in which live births were viewed as the consequences of intermediate factors or more clearly basic biosocial mechanisms affecting the exposure to intercourse, the exposure to the risk of conception, successful gestation, and delivery. Bongaarts (1978 and 1982) has named eight intermediate fertility variables:

  1. The proportion of married women,
  2. contraception,
  3. induced abortion,
  4. lactation infecundability,
  5. frequency of intercourse,
  6. sterility,


  7. spontaneous intrauterine mortality,
  8. and the duration of viability of the ova and sperm.

Five of these have been identified to be of particular relevance for fertility levels and trends in Africa (Page, 1988):

Fertility will be lowered as a result of delaying exposure to intercourse, e.g. through later marriage for those who experience their sexual intercourse in marriages. Other forms of consensual unions, where f.e. the partners do not live together, limits the risk of being exposed to sexual intercourse. Employing sexual practises that reduce the risk of becoming pregnant, such as sexual disruption, can also lower fertility. The use of contraception including abortion, postpartum infecundibility and non-susceptibility due to breastfeeding, and sexual abstinence may on the other hand lower fertility for those who are sexually active.

While the indirect determinants of fertility have relevance for policy makers since they provide mechanisms susceptible to be influenced by official policy, the change in these variables does not necessarily change fertility levels. The direct determinants, on the other hand, influence fertility directly. The change in one or more of these variables changes fertility unless another variable offsets the effect. The direct determinants in a general sense are biological and/or behavioural in nature. Proportion of Women engaged in Sexual Relations

The frequency of sexual intercourse is the underlying variable of interest, but information on this is rarely available. The proportion of women of reproductive age that is regularly engaged in sexual intercourse is believed to be the major determinant of high fertility in sub-Saharan Africa since contraceptive prevalence is still very low (Mturi and Hinde, 1994). Various measures of marital status are used as proxies for this


concept. A few studies have been able to employ direct data on coital frequency (Brunborg, 1983), but so far these attempts have been limited to only a few developed countries. The only sub-group of women whom we can assume to be sexually active is the currently married women.

In the past, marriage was thought to be universal (Bongaarts et al., 1984; Page, 1988), and postponement of first marriage has been outlined as the main determinant of fertility decline observed (Cleland et al., 1994). But marital dissolution through divorce or widowhood is a common phenomenon (Blanc and Rutenberg, 1990).

In studying recent trends in age at first marriage using data from 14 regions in Tanzania, van de Walle observed that the proportion of women never marrying decreases progressively along the age distribution. This influenced him to conclude that ‘the Tanzanian data suggest the old norm of universal female marriage may be changing’ (van de Walle, 1993: 146).

Furthermore, the definition of marriage is problematic in Africa. Usually marriage in African societies is ‘a process’. There is some ambiguity in determining exactly when a couple is getting married (van de Walle, 1993). This implies that the magnitude of the proportion of married women will depend on the indicator of marriage used. The use of the data of the proportion of married women is misleading because there is a rise in premarital sexuality and child bearing in sub-Saharan Africa (Meekers, 1994), which waters down the use of the ‘proportion of married women’ variable in the study of proximate determinants of fertility. Contraception

Any deliberate practice undertaken to reduce the risk of conception by sexually active women (and their male partners) is considered as contraception. The tool used to prevent or reduce the frequency of conception is known as contraceptive. Contraceptive use has been described as the most important proximate determinant of fertility (Sherris et al., 1985; Mauldin and Segal, 1988). Robey and his colleagues have shown that differences in the levels of contraceptive use explain 92 percent of the variation in fertility among the 50 countries they studied (Robey et al., 1992). This implies that where contraceptive use is widespread, fertility is low. It is therefore essential to study


the extent of the use of contraception in order to make sensible statements about the current and future fertility rates in a society.

Contraceptive prevalence is lower in sub-Saharan Africa than in other parts of the world. The contraceptive prevalence rates estimated in all African countries were less than 15 percent in 1990 except in Zimbabwe, Botswana and Kenya (Rutenberg et al., 1991; Robey et al., 1992). Moreover the reason given for using contraceptives in many African societies is birth spacing rather than limiting the number of children (Bertrand et al., 1993). It can therefore be argued that low contraceptive prevalence is partly responsible for the high fertility levels in sub-Saharan Africa except in Central African countries with a low contraceptive prevalence rate and low fertility due to pathological sterility. However, the higher rates of contraception to be anticipated in Africa are likely to reduce fertility. Indeed the recent DHS conducted in sub-Saharan Africa has shown an increase in the contraceptive prevalence rate in various countries. For instance in Tanzania, the second and third phases of the DHS show that the contraceptive prevalence rate has doubled from 10 (1991/92) to 12 (1994) percent in less than 3 years (Weinstein et al., 1995).

Figure 4 presents the percentage of currently married women (15 to 49 years) using any contraceptive method between 1986 and 1995. Of the countries in Eastern and Southern Africa, South Africa has the highest contraceptive prevalence rate (50 percent) followed by Zimbabwe (48 percent), Tanzania has the lowest of all countries.


Figure 4: Percentage of currently married women (15-49 years) using a contraceptive method in Eastern and Southern African

Source: Demographic and Health Survey Newsletter, 1997 Post-partum Infecundability

The primary cause of prolonged post-partum infecundability is breastfeeding, which results in lactational ammenorrhea. It is known that breastfeeding has an influence on fertility by lengthening the period of postpartum infecundability (Bongaarts and Potter, 1983). In societies where breastfeeding is generally prolonged and universal, and contraceptive use is rare, the primary determinant of birth interval length is the duration of breastfeeding. Breastfeeding leads to the release of prolactin which inhibits the release of gonadotrophin (the hormone which initiates the resumption of the menstrual cycle) (van Ginneken, 1978; McNeilley, 1993). The longer and the more intensive breastfeeding is, the greater the release of prolactin and therefore the greater the contraceptive effect of breastfeeding. Full breastfeeding, where the infant has no other source of food, suppresses it almost totally, whereas less intense and frequent suckling suppresses it partially. Thus ovulation can resume while a woman is still breastfeeding. However, it has been noted that a woman who has stopped breastfeeding is more likely


to became pregnant once ovulation returns compared with a woman still breastfeeding, due to a reduction of fecundability of breastfeeding women (Guz and Hobcraft, 1991).

Lactational infecundability arises after a pregnancy when a woman is unable to conceive until the normal pattern of ovulation and menstruation is restored. Bongaarts and Potter (1983) have observed the period of postpartum amenorrhea to be 1.5 months following the delivery in the absence of any lactation. The average duration of postpartum amenorrhea increases in proportion to the average length of breastfeeding, lasting for about 60 to 70 percent of the duration of breastfeeding where breastfeeding lasts two years and more. For the longest breastfeeding duration’s observed in practice, amenorrhea periods of up to two years occur (Bongaarts and Potter, 1983).

Long and intensive breastfeeding is evidently universal throughout sub-Saharan Africa. However, breastfeeding duration varies between countries and particularly within countries. The mean duration of breastfeeding is about 19 months in Lesotho, 18 months in Ghana, 16.5 months in Kenya and Sudan. The corresponding duration of postpartum amenorrhea are 13 months in Lesotho, 12 months in Ghana and 11 months in Sudan and Kenya (Bongaarts et al., 1984). A median duration of breastfeeding of 22.7 months in Bamako, the capital of Mali, corresponded to a length of amenorrhea of 15.2 months. For a period of breastfeeding of 18.5 months in Bobo-Dioulasso (Senegal), amenorrhea lasted 13 months (van de Walle and Omideyi, 1988). The mean duration of breastfeeding observed in Kibaha, Tanzania was between 18 and 19 months and the mean duration of amenorrhea was between 7 and 10 months for different educational groups (Komba and Kamuzora, 1988).

The general observation is that the duration of breastfeeding declines with development. In particular, breastfeeding declines with urbanisation and education (Lesthaeghe et al., 1981c). Therefore breastfeeding is still a potential factor in reducing fertility in sub-Saharan Africa.

The postpartum non-susceptible period is usually defined for each woman according to whichever period is longer, that of postpartum amenorrhea or that of postpartum sexual abstinence (Lesthaeghe et al., 1981b). In many African cultures, the resumption of intercourse is linked with weaning. Breastfeeding and sex are considered to be


incompatible since sperms are believed to poison the mother's milk. Therefore prolonged durations of postpartum abstinence are observed in sub-Saharan Africa. For the countries cited by Page (1988), the duration ranges from 12.4 months (Ghana) to 18.2 months (Benin) with the exception of Kenya which had a duration of 4.1 months. The Yoruba of Nigeria have recorded a duration of about 3 years which means that the sexual taboo lasted longer than the breastfeeding period (Caldwell and Caldwell, 1977). Erosion of the practice of postpartum abstinence has been observed in many areas of sub-Saharan Africa. Bongaarts et al. (1984) noted that in Tanzania in the 1970s, the period of postpartum abstinence rarely exceeded 6 months. However in Kibaha, Tanzania, the period recorded ranged from 8.4 months for women with at least 9 years of schooling to 10.6 months for illiterate women (Komba and Kamuzora, 1988). The most notable observation is that the period of postpartum sexual abstinence is becoming shorter, especially in East Africa, and this is likely to raise fertility. However, the demographic role of abstinence is much reduced by the relative stability of lactation. Induced Abortion

Data on induced abortion, a practice that deliberately interrupts the normal course of gestation, are very rare in Africa. This is due to the fact that in most African countries, induced abortion is illegal unless performed to save the mother's life. It is therefore difficult to assess the effects of induced abortion on fertility in this part of the world. It has been observed, however, that abortion is in fact not uncommon, particularly in urban areas, and that the number of cases presented at hospitals for abortion is increasing though it is restricted to young and unmarried women (Coeytaux, 1988; Justesen et al., 1992). This topic will be discussed in Chapter 4. Sterility

Sterility, whether primary or secondary, has been known to affect fertility particularly in areas where there is high incidence of sterility. In Gabon for example, the key determinant of the exceptionally low fertility (TFR of 4.1) was noted to be widespread pathological sterility (Bongaarts and Frank, 1988). If sterility is reduced, fertility is likely to rise - this is the trend in countries where the prevalence of sterility is high. However, sterility seems to be relatively lower in East and West Africa compared with Central Africa. Bongaarts et al. (1984) have noted that the highest levels of infertility


are found in Central Africa where over a large area more than 20 percent of women aged 45-49 are childless. The percentage of women aged 45-49, who are childless is 12-20 percent and 3-12 percent in East and West Africa respectively. A more recent estimation procedure developed by Larsen and Menken (1991) has shown that prevalence of sterility in Kenya is relatively low compared with other sub-Saharan African countries included in their analysis (Ghana, Lesotho, Cameroon, Sudan). This implies that sterility is still low in East Africa and its impact on fertility is small. This topic will be discussed in Chapter 4.

2.3.4. Determinants of Fertility

The World Fertility Survey (1977) has produced a list of explanatory variables in a simple framework for fertility analysis including:

These variables affect fertility indirectly through the proximate determinants explained in the previous section. In this study background variables will include education, religion, place of residence, and polygyny.<10>

The spread of education and literacy among women is believed to be fundamental to changes in the reproductive behaviour. The effect of women's education on fertility in less developed countries is found to be curvilinear, i.e. fertility tends to rise first with education and then decreases sharply once a certain level of education is attained (Cochrane, 1979). The argument is that education is positively associated with improved health, lower levels of infertility, abandonment of traditional constraints upon


sexual behaviour and the practice of breastfeeding, all of which are known to raise fertility levels. As the educational level increases, marriage tends to be postponed which causes a negative effect on fertility and counteracts the initial effect of fertility increase. Moreover, educated women desire relatively fewer children. They have high contraceptive prevalence and a high chance of working outside their homes. All of these factors are known to lower fertility levels (Cochrane, 1979). However, there is also a possibility of the reverse causation which is less documented, i.e. the initiation of child bearing causing the termination of education (Cochrane, 1979). While analysing the relationship between fertility and the level of education in sub-Saharan Africa, Cohen (1993) has shown that fertility is either curvilinearly or negatively related with education but does not appear very responsive to few years of education.

Generally, fertility is higher for women residing in rural areas compared with those residing in urban areas. Higher levels of education, occupation, a more modern environment, aspirations for higher levels of living are among the factors which can cause fertility among urban women to be lower than among rural women (Stolnitz, 1983 cited in Bulatao, 1984). Also, it is assumed that urban women have a better knowledge of and access to modern contraception than women in rural areas (Cohen, 1993). A recent demonstration has shown that rural fertility is substantially higher than urban fertility in every African country included in the analysis (Cohen, 1993).

Education and occupation of the husband (or the partner) can be used to measure the socio-economic status of a family, and is also an indicator of the quality of child rearing if more appropriate variables such as income are not available. The basic assumption is that the higher the educational level of the husband and his occupational status, the higher the income of the family. This leads to improved living conditions. Bulatao and Lee have argued that,

‘In principle, whether children are net producers or net consumers, higher income or greater wealth should make them more affordable and therefore increase demand for surviving children, with a subsequent increase in the number of desired births. However, income increases may lead to a demand for higher quality children rather than a large number’ (Bulatao and Lee, 1983a: 767-768).


In various societies religion has been found to have an impact on fertility. Lucas (1980) has argued that Moslems often have higher fertility than non-Moslems, and Catholics frequently a higher fertility than other Christians. While the Catholic doctrine is argued to be pronatalist by favouring large families and rejecting all efficient methods of birth control, Moslems have high fertility because of early and universal marriage. Also the majority of Moslems live in traditional agricultural societies where children are economically useful and levels of education are relatively low (Lucas, 1980). However as a society develops, the fertility differentials by religious groups are expected to become smaller.

It is generally argued that polygyny enhances child spacing in most African societies (Schoenmaeckers et al., 1981). For instance female abstinence can be maintained more easily in a society that practises polygyny. The Tanzanian experience has shown that pregnancies are more frequent among women living in monogamous unions than those in polygamous unions (Henin, 1979). On the contrary, it has been argued that polygyny is negatively associated with contraception (Caldwell and Caldwell, 1981) and women married to the same man can compete to bear children particularly in societies where the status of a woman depends on the number of surviving children. Therefore the effect of polygyny on fertility can be in either direction.

<9> In addition to the efforts of specific institutions and individuals, the following works are among the examples which show that there is a concern to understand fertility: the work done by the World Bank (Cochrane and Farid, 1989; Acsadi et al., 1990; van de Walle and Foster, 1990), the Ife Conference (van de Walle and Ebigbola, 1987), and the papers and books written under the auspices of the Panel on the Population Dynamics of sub-Saharan Africa of the National Research Council (NRC) Committee on Population (Bertrand et al., 1993; Blesdoe and Cohen, 1993; Brass and Jolly, 1993; Foote et al., 1993; National Research Council, 1993).

Later on we will speak of Protestants or Catholics, who live in polygamous marriages/unions. Although the religion and polygyny seem to be incompatible, it is a common practice for many people to identify themselves with a certain religion while not following the doctrine. In this respect one normally finds Christians in polygamous unions while the doctrine insists on monogamous unions.

[Titelseite] [Abkürzungsverzeichnis] [1] [2] [3] [4] [5] [6] [7] [8] [Bibliographie] [Anhang] [Selbständigkeitserklärung]

© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.

DiDi DTD Version 1.1
a subset from ETD-ML Version 1.1
Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML - Version erstellt am:
Mon Jan 17 19:31:05 2000