in mortality among all provinces since the 1950s, with a reduction of about 70 percent. Coast Province did rather poorly during the same period, with a decrease of only 19 percent. The four remaining provinces, other than Nairobi, experienced declines closer to the national average of 54 percent.

Attempts to link these trends to socioeconomic factors revealed a strong relationship to female education and adult literacy. Although individual education is probably key in sustaining differentials in mortality over time, evidence from this study indicates the importance of district-wide educational levels in the mortality declines, as demonstrated by the reductions in mortality of children with poorly educated mothers living in districts with relatively highly educated residents. Little association was found between decreases in child mortality and health indicators, although cross-sectional analyses found associations between levels of mortality and the incidence of malaria and malnutrition, as well as the ecological zone of residence.

Fertility in the 1960s and 1970s was high, about eight births per woman and seemed to be rising slightly. In the decade from the late 1970s to the late 1980s, fertility fell approximately 20 percent to a little more than 6.5 births per woman. This decrease was unexpected by many and was due principally to an increase in contraceptive use. What is striking about the reduction in fertility was its occurrence across almost all subgroups. Declines in fertility occurred within all age groups, with the middle and later reproductive ages contributing more to the reduction than the earlier ages. Other than Western Province (where the evidence for the magnitude of the decline is inconclusive), there were moderate to substantial decreases across all provinces, regardless of level of socioeconomic development. Central Province had the largest decrease of 31 percent.

Fertility fell 17 percent in rural areas, not much less than the 23 percent decline that occurred in urban areas. The reductions by level of education were almost equal in percentage terms, even with marked initial differentials in fertility levels. Decline occurred at all birth orders, a pattern that is clearly distinct from the fertility decline in Latin America and Asia, where fertility reduction began in the middle parities and spread to the higher and then to the lower birth orders. Analysis of other sub-Saharan African countries experiencing changes in fertility indicates that the pattern observed in Kenya is also occurring in Botswana, Nigeria, and Zimbabwe.

The near universality of the decreases in fertility in Kenya indicate that the determinants of the decline have a strong central component affecting geographical and social categories of the population in a similar way although not to the same extent. This conclusion is supported by the weak associations found between the fertility reductions and district-level socio