Understanding why particular patterns of adolescent fertility emerge in Africa is no easy task. Most basic is the task of identifying our subject population: Who is an "adolescent"? What is a "youth," a "teen," or a ''schoolgirl''? The studies we rely on are frequently inconsistent in their terminologies. And we ourselves are not perfectly consistent.
Studies of adolescents usually use a five-year age range, 15 to 19, as their primary source of data. The problem with restricting the analysis to this range—or even with including 15-and 19-year-olds in the same age group—is that the difference between old and young teens may be critical for the health of young women and their infants and for the kinds of options and repercussions they face. A 13-year-old who becomes pregnant may come from a family of white collar civil servants. She may have just reached puberty and may be contemplating secondary school. By contrast, an 18-year-old living a hundred miles away in a remote village may be married and carrying her second child. In locales where adult work and marriage begin early, a realistic analysis must include individuals as young as 12 or even 10. In sharp contrast are the areas, usually urban, where education ends late and childbearing begins late; here, an analysis that excludes 20-to 24-year-olds misses considerable information.
Even within the same geographical area and the same social class, a young man and a young woman who are both described as adolescents may
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa Appendix Difficulties in Analyzing Adolescent Fertility DEFINITION OF ADOLESCENT Understanding why particular patterns of adolescent fertility emerge in Africa is no easy task. Most basic is the task of identifying our subject population: Who is an "adolescent"? What is a "youth," a "teen," or a ''schoolgirl''? The studies we rely on are frequently inconsistent in their terminologies. And we ourselves are not perfectly consistent. Studies of adolescents usually use a five-year age range, 15 to 19, as their primary source of data. The problem with restricting the analysis to this range—or even with including 15-and 19-year-olds in the same age group—is that the difference between old and young teens may be critical for the health of young women and their infants and for the kinds of options and repercussions they face. A 13-year-old who becomes pregnant may come from a family of white collar civil servants. She may have just reached puberty and may be contemplating secondary school. By contrast, an 18-year-old living a hundred miles away in a remote village may be married and carrying her second child. In locales where adult work and marriage begin early, a realistic analysis must include individuals as young as 12 or even 10. In sharp contrast are the areas, usually urban, where education ends late and childbearing begins late; here, an analysis that excludes 20-to 24-year-olds misses considerable information. Even within the same geographical area and the same social class, a young man and a young woman who are both described as adolescents may
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa actually be very different ages. In many areas, women from 15 to 19 are considered mature enough to be wives and parents; indeed, a childless woman of 18 may be making trips to the local healer, desperate to cure her infertility. Conversely, few African societies consider men of 15 to 19 mature enough for the responsibilities of married life or parenthood. A definition of adolescents as 15-to 19-year-olds thus includes many females who are considered "adults" and many males who are considered "children." Because we need to draw a line somewhere, however, we use the term adolescent in a relative sense to define young people in the stage of life between childhood and adulthood, however their own societies define those terms for males and for females. We use the term adolescence flexibly, allowing it to stretch into the early twenties in areas where late secondary school completion is possible and where the average age at first marriage has increased. The term also at times includes children as young as 10, especially where labor force participation is at issue. By contrast, "teenagers" refers to a group defined in fixed chronological years—people aged 13 to 19—though in most cases the age group we use for quantitative analysis is 15 to 19. Otherwise, age groups are specified. Besides the problems that we face in simply defining our population, analysis is difficult because adolescence is a very different experience for males and females, and the two life cycles, when subjected to new pressures, do not necessarily change in synchrony. Further, adolescence often encompasses several life transitions that occur at much the same time so that it is difficult to attribute causality to one event or another. For example, educational opportunities may induce young women to postpone marriage and fertility; but it is also true that low fertility facilitates education. Although cause and consequence may seem on the surface easier to untangle in the realm of health, analogous problems arise once we delve below the surface. Are the high risks of morbidity and mortality to be interpreted as "natural" consequences of the mother's physical immaturity, or do many of these risks stem from the condemnation that makes young women forgo prenatal care or attempt abortion? DEFINITION OF MARRIAGE The key transitions surrounding adolescence pose some of the most significant problems of definitional ambiguities. In Africa, many teens who report themselves as unmarried acknowledge sexual experience and childbirth. A major stumbling block to understanding this anomaly is the frequent difficulty in deciding whether a woman is married and, if so, when her marriage began. Two fundamental features of marriage in contemporary Africa demonstrate that the definitional issue runs deep. First, marriage in many African societies is often a fluid process that
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa may be drawn out over months or even years. This processual nature of marriage creates immense difficulties for identifying a marriage and dating its inception and for establishing convincing causal connections between marriage and fertility. For most young women, sexual relations become more frequent after marriage begins, so that pregnancy becomes more likely. Yet if marriage is a process, pregnancy and childbirth may be formative events in creating a marriage: They may help legitimize a developing conjugal relationship. Hence, does an adolescent become pregnant because she is married, or is her marriage consolidated by her pregnancy? Second, not only is it unclear when a marriage begins, but also marriage can take many different forms, even within the same ethnic group. Partners and their families may contract a marriage within "customary" procedures, the church, Islamic law, civil statutory codes, or more than one of these. Furthermore, although formal polygyny appears to be diminishing in many urban areas, it is thriving in de facto forms, some of them new (Dinan, 1983; Mann, 1985; Karanja, 1987). A man who is technically monogamous may have "outside" wives, with whom he has informal unions. Which type of union is regarded as legitimate, and thus whether a woman is considered married, can vary with the context. Taken together, these complications make it difficult to draw reliable inferences about the association between marital status and fertility. What they do suggest is that increases in premarital sexual experience and childbearing in such contexts stem not so much from changes in actual rates of childbearing, or even from increases in sexual experience during the adolescent years; rather, they stem from the possibility that young women in certain kinds of unions or stages in the conjugal process are now less inclined to claim that they are married. The consequences are no less real, however. THE DATA: DEFINITIONS AND DEFICIENCIES Although demographic patterns in Africa are drawing increasing attention, the task of investigating their causes and consequences has been hampered by problems of data. Time-series data are extremely thin in Africa, so that assessing change over long spans of time is quite difficult. For the demographic past in areas such as Europe and Japan, the extraordinarily rich quantitative and cultural record has allowed hypotheses to be tested far more systematically than has been possible for the African historical experience. The situation has been improving rapidly, however. Nationally representative data on marriage and childbearing patterns in sub-Saharan Africa in the past two decades have come largely from the World Fertility Surveys (WFS) and their successors, the Demographic and Health Surveys (DHS). Together with national censuses, these surveys provide the most compre
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa hensive cross-sectional demographic data on adolescent fertility in sub-Saharan Africa. At the time we write, we have in hand data from the DHS obtained between 1986 and 1990 from random household samples of women aged 15–49 for 11 sub-Saharan African countries: Botswana, Burundi, Ghana, Kenya, Liberia, Mali, Nigeria, Senegal, Togo, Uganda, and Zimbabwe. Four of these countries also participated in the WFS. The sparsity of data on our specific topic poses one problem. Much of the material we use is gleaned from studies that did not focus on adolescents. The current round of DHS focuses on women of 15 to 49 years. Obviously, surveys must draw a line somewhere. But these lines omit a crucial slice of the group in whom we are interested: girls under age 15, many of whom, in some countries, are already married and have begun bearing children. Conversely, by viewing individuals 15 and older as mature "women," these surveys often fail to elicit information that is basic to adolescents' concerns—for example, whether they are still in school. Besides lacking focused information on adolescents, we also face some gaps in knowledge about differences within specific populations in Africa. Such gaps develop because reliable data on almost any topic that has been critical to the interpretation of the demographic transition in Europe are historically shallow for Africa; they develop also because not all current national studies cover the key variables in the same way. Large-scale population data sets are not usually set up to support the kinds of points that small-scale ethnographic studies make best. In the DHS, ethnicity, religion, occupation, social class, and property endowment are covered in some studies and not others. The same is true for data from the International Labour Office (ILO), upon which we draw. Because we use economic data on large populations, mainly up to 1980, and because the data on economic well-being are thin, we do not address poverty and destitution systematically. It is extremely difficult to find reliable economic or demographic information about refugee camps, homeland/reserve situations, guerilla camps, and urban streets, all of which obviously contain many youth. The DHS, as valuable as it is, has other limitations. Concerning marriage, for example, some countries omitted questions about cohabitation or consensual union, whereas others included a separate category for them (van de Walle, 1993). Because the DHS is forced to categorize many of the phases of the marriage process into discrete units ("never married," "married," and so on), it loses much of the richness and diversity of the African marriage process. Another problem area in the DHS is education. In some countries, it is apparently assumed that women 15 and over are no longer in school; hence, there is no question in the DHS on whether young women are still enrolled, although clearly many are nowadays. Moreover, because the DHS was based on household samples, it likely missed women who
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa were in boarding schools at the time of the survey. Obviously no survey can anticipate all important issues or include questions on every demographically relevant aspect of a woman's life. With all its shortcomings, the DHS is far more comprehensive than other major sources, as well as more recent, and so we rely heavily on it. INTEGRATION OF DATA SOURCES Because the sources on which we depend vary widely in coverage of topics about adolescence and fertility, we face the task of integrating a number of different types of data: national-level data for female labor force participation from International Labour Office materials, regional and national data on fertility from the Demographic and Health Surveys, data on religion or social class, and information on cultural and social practices based largely on individual ethnographic studies of ethnic or subethnic groups. Especially because of problems with the availability and coverage of data, interpreting demographic dynamics for Africa demands the use of methods besides quantitative inference; for example, logical inference from known social patterns and from history. The need to integrate such vastly different sources of data creates several problems. One derives from the enormous variation in the sizes of national populations. Although national and regional data have great strengths, they can also be misleading when we make comparisons as if, say, Botswana, with a population of 1.2 million people dominated by a single ethnic group, were equivalent to Nigeria, with a population of 89 million people and 250 different languages. Each of the three major ethnic groups in Nigeria is at least six times as large as Botswana's entire population. Although the experience of small countries highlights important processes, one should not necessarily accord them equal analytical weight. A second problem in trying to integrate data sets is that they almost invariably use different units of study. How, then, do we categorize indigenous variables or even social groups in order to draw reliable quantitative inferences about demographic regimes in Africa? There are two major contesting visions. One of the founders of African studies in America, Melville Herskovits, attempted to divide the continent into several vast "culture areas" that were linked by culture, history, and livelihood (Herskovits, 1930). Beginning in the 1950s, classifications were based on more narrowly defined criteria, often without reference to historical and geographical connections among peoples. The practice of designating ethnic groups within defined national boundaries—"the Yoruba of Nigeria," ''the Akan of Ghana"—grew largely out of the administrative needs of colonial governments and continues to reflect political pressures. The Human Relations Area Files and the subset of societies chosen to form the Ethnographic
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa Atlas (Murdock, 1967), which together make up the largest coded body of ethnographic information, are based on such classifications. Many of the demographic data have been collected not by the cultural area or the ethnicity framework, but with reference to political boundaries of nations and their constituent administrative units that colonial states defined. Only for some DHS countries, for example, can one isolate specifically ethnic categories of the population, because this variable was coded or because there is a close match between the ethnic and the administrative boundaries. As a result, hypotheses about the influence of indigenous property-control variables cannot be reliably addressed in quantitative terms at all. One of the most difficult problems we encounter in compiling sources is ambiguous or varying definitions. Our reservations about the term adolescence have been spelled out already. Other puzzles focus on terms that seem clear at first, yet become obscure upon reflection. What, for example, is meant by "sexual activity"? Besides intercourse that may lead to pregnancy, this phrase can refer to nonpenetrative and nonreproductive means of sexual expression, including those within marriage to a young bride with whom full sexual relations may not be consummated immediately. Another term that poses problems for this report is "work," a subject for which a principal source of data is the International Labour Office. The ILO definition of employment clearly presents problems for Africa. Over the past decade, the basis for defining the labor force has shifted from those considered to be "economically active'' (that is, usually employed) to those "currently active'' (employed during a specified period). As the ILO itself acknowledges, "The results may be significantly different, depending on the approach taken, for male youths and elderly males and women of all ages" (International Labour Office, 1990:5). Finally, any attempt to use multiple sources of data in interpreting the dynamics of persistence and change over long periods quickly runs up against problems of historical specificity. At this time, the greatest richness and precision in the ethnographic sources tend to be found in studies conducted up to the mid-1980s, before the DHS studies, because of the time it takes to collect, digest, compose, and publish contextual material. For the demographic data, the greatest confidence can be placed in the most recent data, but published ethnographic material that would illuminate these most recent changes is still sparse. DATING OF CHANGE Related to the problem of accuracy of the historical record is the problem of assigning dates to trends. Any statements that we can make with confidence about trends in the statistics per se cannot date back more than
OCR for page 178
Social Dynamics of Adolescent Fertility in Sub-Saharan Africa 30 to 40 years. Almost no quantitative material on adolescent fertility patterns predates 1960; for labor force participation, there is almost nothing before 1950. Many of the current trends, however, were set in motion long before. In trying to use ethnographic information to trace trends, on the other hand, we confront the old analytical problem of the "ethnographic present." The ethnographic literature on Africa often implies, through an enduring present tense, that certain practices, often referred to as "traditional," have existed unchanged for centuries, an implication that may not be true in every case. The term "modern" presents the opposite problem: It is often applied freely to patterns that are assumed to reflect the effects of wage labor, export crop production, conversion to world religions, and the spread of formal education, yet may in fact have their origin in the unfolding of trends that predated the introduction of these factors. Although these problems are to some extent irresolvable, we define our terms as carefully as possible. When we refer to traditional or "indigenous," we refer to practices that appear repeatedly in the ethnographic literature that appeared before the twentieth century. In most cases we specify the time period as precisely as the sources allow. Chapter 2 refers to the past 30 years or so. Chapter 4, on the social context, refers to values and processes we call indigenous to Africa—those that developed most coherently under precolonial conditions but continued to be relevant well into the twentieth century. The chapters on marriage (Chapter 3), education (Chapter 5), and training (Chapter 6) use quantitative data from 1950 to 1990. However, they are interpreted through recourse to studies extending further back in time in the search for appropriate qualitative material to complement the shorter-run quantitative data.