African Households: Censuses and Surveys (review)

Peter Laslett first showed us the sociological information available in historical census data. African census and survey data are rarely employed to such ends, so the African Census Analysis Project, a collaboration between the University of Pennsylvania and African institutions, arranged an Internet conference on African households in November 2001. These are the published results.

Here the "household" is merely a unit of enumeration. It generally contains members of at least one family, plus, often, unrelated others. Censuses take one of two definitions of the household. Francophone researchers tend to adopt a de jure, or social, definition of the household, thus including those who are absent on the enumeration day, while Anglophone researchers prefer a de facto, or residential, definition, and count only those physically present. The former runs the risk of double counts, while the latter tends to manufacture extra households and female heads. All of this is explained in van de Walle's masterful introduction. He notes that African censuses are weak when it comes to coding for intrahousehold relationships. In consequence, grandchildren, children's spouses, and domestic employees are often lumped together under the impenetrable category of "other relative." The most useful censuses are those that link husbands and wives, and mothers to their children.

Three of the chapters here examine the merits of various censuses and surveys. Van de Walle and Gaye's comparison of the Senegalese and Gambian censuses finds that Senegal's de jure census, just by including questions on marriage type and the rank of polygynous wives, is far more informative than the Gambia's more elaborate de facto census. Next, Labov shows the limitations of Tanzania's minimalist de facto census, which, however, does reveal that Tanzania's solitary females are usually separated, divorced, or widowed, and that the urban ones are more likely to be household heads than their rural counterparts. Then Hosegood and Timaeus's work on KwaZulu, Natal, shows how their time-consuming de jure survey accurately captures the emic realities of household composition, the high levels of individual and household mobility, the number of nonresidential members, and memberships in multiple households.

Vimard and Fassassi's chapter on the 1975–98 structural changes in Côte d'Ivoire's households is the only one with a central focus upon economics. They show that the recent increase in family breakups has less to do with a drift toward nuclear families than with households' relative prosperity. Single-parent households and coresident relatives are now permanent social fixtures.

Four of the chapters focus upon children. Townsend, Madhavan, Collinson, and Garenne's work in South Africa's Northern Province successfully traces personal relationships within and across household boundaries by training enumerators to use the six basic anthropological kinship terms. Households have become smaller, but more numerous and compositionally complex; and children with fathers away at work and those living with their parents get more schooling. Noël-Miller's study of Gambian grandparents and grandchildren finds no support for the hypothesis that children best survive weaning when sent to live with their maternal grandparents. They are, in fact, twice as likely to live with their paternal grandparents, and one-half of those times in their own parents' households. Mokomane, Baker, and van de Walle's chapter on extramarital childbearing and children's residence in Botswana attributes the high rate of single motherhood to the pressures put on young women to bear children regardless of their (or their partners') marital status. While economics determines urban mothers' ability to live with their children, rural mothers' ability to do so rests upon their living with parents, grandparents, uncles, or brothers. Then Kabore and Pilon's study of households and schooling in Burkina Faso finds that only one-half of a household's schoolage children are sent to school, and that the prospect for schooling is inversely proportional to the number of schoolage children in it. Schooling is 2.5 times more...