Investigating Child Mortality in Malawi Using Family and Community Random Effects

The Malawi Demographic and Health Survey conducted in 1992 collected the retrospective birth histories for a national sample of 4,878 women aged between 15 and 49 years. The sample was randomly selected by a two-stage sampling design. The data consist of biological, demographic, and social variables collected for each birth. This article models the infant and early childhood survival using family and community random effect multipliers on the fixed effect proportional hazards model, which allows the dependence between observations in the same family and community into the model. A Markov chain Monte Carlo sample from the posterior distribution of the parameters given the data is found. The standard errors of the fixed effect estimates are more correct than those found from the standard model, which are underestimated because of the ignored correlation structure.