Effect of birth spacing on infant survival in Thailand: two-stage logit analysis

We formulated a two-stage causal model for infant survival and applied it to data drawn from the 1987 Thai Demographic and Health Survey covering the fate of 5,074 index children. The following six variables were considered as the explanatory variables: maternal age, maternal education, birth order, preceding birth interval, survival of the preceding child, and place of residence. The analysis suggests that the birth interval not only directly affected the chance of infant survival but it played the role of the filtering factor through which other variables indirectly operate on infant mortality. The effect of preceding child's death was very strong, the odds ratios for the following infant's death and short birth interval both exceeding three.