Imputing Birth Histories to Construct Direct Estimates of Child Mortality
Livia Montana, Harvard School of Public Health
We present a new methodology of imputing full birth histories from one sample population to another with only summary birth histories (SBH), by matching on unique combinations of exposure, total number of children ever born, and number of dead children. We control for exposure by matching on the age of mother, aggregated into 3-year groups. We randomly match each woman with a given SBH to multiple comparable women with the same SBH and a full birth history, and we repeat this m times. Once the matches are made, direct mortality estimates are derived from the imputed full birth histories. We make one assumption: controlling for exposure or duration of childbearing, the birth history of a woman given her CEB and CD does not change over time. We present results for five countries and show that direct estimates from imputed data match or are better than indirect estimates.
Presented in Poster Session 5