Infant mortality by cause of death: main and interaction effects.

We examine infant mortality among the 1980-1982 live birth cohorts in the state of Florida, specific to five categories of underlying cause of death: infections, perinatal conditions, delivery complications, congenital malformations, and sudden infant death syndrome. The gross and net effects of eight categorical and continuous independent variables, along with 11 first-order interactions, are examined with microlevel data through the use of multinomial logit regression. Findings suggest the complexity of variable effects by cause of death and indicate the simultaneous importance of biological and social factors. It is important that the pattern of interactions suggests an overall dependence of infant life chances on social circumstances. It also suggests that these effects are attenuated for some variables and causes of death at lower birth weights, probably due to advances in health care organization, access, and technology.