Child Mortality Estimation by Time Since First Birth

Abstract

The Brass method for estimating child mortality from proportions dead of children ever born classified by age or duration of marriage of mother has revolutionized our knowledge of child mortality levels and trends in developing countries. However, the age-based version of the method suffers from selection biases for young women whose data provides the most recent estimates, and the duration of marriage-based method is only applicable where most births occur within stable marriages. This paper develops and illustrates a methodology based on time since first birth using a model of fertility by time since first birth and model life tables. This methodology avoids the selection bias of the age-based method, and is applicable to all populations regardless of marriage customs, but does require an extra question, on date of or age at first birth, on the census or survey instrument. An application to data from Haiti shows the method to give estimates similar to those from a subsequent birth history. The method is likely to be of most value in large survey operations such as censuses in populations with moderate to high levels of pre- or extra-marital fertility.