Content area
Abstract
The degree to which women in developing countries replace children who have died is an issue as yet unresolved in the demographic literature. This dissertation addresses the question of whether reproductive behavior is consciously altered by the death of a child using World Fertility Survey data from Colombia, Costa Rica, and Korea. These countries are chosen primarily because at the time of the survey each was at a somewhat different stage in its transition from high to low fertility.
Alternative strategies are proposed by which women replace dead children. They may choose to contracept for a shorter period of time following a death or they may refrain from contracepting altogether. For the first strategy, the dependent variable is the hazard of having a birth in a given interval, and contraceptive use is one of several covariates. A woman is excluded from the analysis if the child whose birth initiated the interval under consideration does not survive. This procedure is followed to avoid the possibility of a biological effect which is known to exist if a breastfed infant dies.
For the second strategy, the dependent variable is use of contraception in the interval. Both models are analyzed separately for selected birth intervals, both control for mother's education, region of residence, age and time period, and both are estimated using log-linear techniques.
Because information on contraception is available only for the last and penultimate birth intervals in Colombia and Costa Rica, comparative analysis of child replacement must be restricted to the last closed and open intervals. Results from an examination of Korean data where information on all birth intervals is available, indicate that limiting the analysis to the period five years before the survey, as is commonly done to avoid sample selection bias, is unnecessary. Thus analysis here is restricted to intervals begun since 1965, eleven years before the Costa Rican and Colombian surveys and nine years before the Korean survey.
The major substantive finding of this dissertation is that the timing and nature of the response to child mortality appear to depend on a country's stage in its fertility transition.
