Problems of Interpretation and Specification in Analysing Fertility Differentials: Illustrated with Kenyan Survey Data

Abstract

High fertility rates and high population growth rates in the developing world have stimulated many studies of the factors determining fertility rates. Due to data limitations, much of the early multivariate research on fertility differentials was carried out at a fairly high level of aggregation and abstraction; in numerous regression analyses of fertility differentials, countries and regions of countries were used as the units of observation (see, for example, Adelman, 1963; Anker, 1978). Results from these analyses indicated that fertility rates tend to decline along with socioeconomic development and accompanying changes in the economic costs and benefits associated with children, thus supporting, for example, the socio-economic theory of fertility and the demographic transition theory (Becker, 1960; Easterlin, 1975; Notestein, 1945).