The Demography of South Africa (review)

The Demography of South Africa is a valuable reference book because it provides insights regarding quality of demographic data from censuses and vital registration at various points in South Africa's history. The stated purpose for the volume is "to outline the demographic contours of South Africa," but as part of providing this general demography of South Africa, many of the authors delve into issues of data quality, representativeness and comparability. Given how the political history of the country shaped the collection of demographic data, addressing such questions is of course necessary. For example, the first chapter by Khalfani et al. gives a history of racial classification across South African censuses. It has a detailed section on the last census to take place under apartheid (1991), and describes the failure of that census to successfully enumerate the population. I can easily imagine anyone seeking to analyze change over time needing to refer back to the detailed information on sources of undercount.

This level of technical detail is maintained throughout the book. Perhaps this is not surprising since the book resulted from collaboration between the African Census Analysis Project and Statistics South Africa: both groups have a vested interest in making the available data as useful to researchers as possible. As such, strengths and limitations of the data are consistently documented, even in chapters that have more substantive aims. To again give an example, Udjo's fertility chapter seeks not only to document fertility levels, differentials and trends, but also to assess whether apartheid era projections of the future size of the African population were inflated by overestimation of fertility.

My description of this volume as a valuable reference work is not entirely kind because I found some of the analyses fairly boring to read. Bah's technical appraisal of official South African life tables and how their methodologies changed over time in keeping with advances in demography again gives information vital for further time trend analysis, but holds no sociological interest. As a further example, Udjo documents "peculiar" patterns in mortality that are left unexplained. These data anomalies are good to know about when using the data in future work, but alone they just seem peculiar. Bah also has a chapter that estimates the size of the effect of HIV on mortality three different ways, only to conclude that although current data do not permit accurate assessment of cause of death, HIV/AIDS is a major health problem.

The final two chapters of the book showed a very high quality of writing; I also found them straightforward and simply more interesting than most of the rest. Sibanda and Lehloenya provided a descriptive analysis of gender, race and regional gaps in education that seemed to me exactly what I should expect from a book purporting to provide a general demography. They showed little gender inequality in education, but much according to race and region of residence. Zuberi and Sibanda provided an analysis of migrant status and employment probabilities with special attention to whether migration had occurred after the collapse of apartheid. They compared internal migrants, those from within the South African Development Community, and other international migrants. In so doing, they documented labor force returns to migration within the context of political history.

Two other interesting chapters on household structure – Noumbissi et al. and Zulu and Sibanda – seemed a little detailed for a general demography. In both of these chapters as well as elsewhere in the book, it seemed like part of the point was to do as much as possible with census data. On the one hand, demonstrating that census data can address questions of sociological import has the potential to stimulate further research. On the other hand, the rigor the authors' brought to the analysis also highlighted much that cannot be addressed with census data, and therefore might have unintentionally cast doubt on this as a valuable data source. For example, Noumbissi et al. show differences between surviving children reported by...