Evaluating the Consistency of under-Five Mortality Rate Estimates Using Full Birth Histories and Summary Birth Histories

Romesh Silva, University of California, Berkeley

This paper reviews direct and indirect methods used to estimate child mortality rates when comprehensive vital registration information is unavailable. I present an analysis using data from 115 demographic and health surveys (DHS) in thirty-five countries spanning West Africa, East Africa, Latin America and South and Southeast Asia. This empirical evaluation entails: (i) an examination of the paired difference of direct and indirect estimates using DHS data, (ii) a characterization of these differences based on diagnostic analysis of and adjustment for data errors in the full birth histories and violations of the model assumptions in the indirect method and (iii) a comparison of the Rajaratnam et al. (2010b) indirect method (applied to census data) with corresponding full birth histories from DHS surveys. I conclude with a detailed description of the performance of the methods under different epidemiologic and demographic conditions.