Evaluating the Performance of Death Distribution Methods for Estimating Death Registration Completeness: Applications to Data from High-Income Countries

Kevin J.A. Thomas, Harvard University

In countries where vital registration systems are underdeveloped, mortality estimates derived from conventional methods can be badly biased. Analytical methods that estimate levels of death registration completeness are among the alternative approaches used for mortality measurement in these contexts. Although the use of such methods is increasing, few empirical studies have evaluated how they perform. One evaluation strategy is to apply them to settings where the recording of deaths and population is thought to be essentially complete. In this paper, we evaluate how three death-distribution methods perform with accurate data from the Human Mortality Database, and explore whether additional information can be used to improve performance. We also assess the sensitivity of death registration methods to the age groups used to estimate adjustment factors, and examine the association between registration completeness estimates and net international migration flows.