Can Estimating Completeness of Death Registration be used as Evidence of Inaccuracy of Population Size Estimates from a Census? The Case of the 2011 South African Population Census

Abstract

Background: Knowledge about the size of the population is important for planning in any population. The census is the traditional source of information about population size.  Accuracy of census figures can be in doubt even when a post-enumeration survey is used to adjust census figures.  Results from the three post-apartheid censuses in South Africa are controversial.  South Africa has a long history of death registration. Given the controversies surrounding South Africa’s censuses, this study examines whether estimating completeness of death registration can be a tool for assessing the accuracy of the 2011 official census population estimate.  Implication of inaccuracy of census figures on life expectancy at birth estimates is also examined.  Data Sources and Methods: The data for the study are South Africa’s 2001 and 2011 Censuses, 2001 and 2011 Deaths registration. Estimation of completeness of death registration was based on the Growth Balance method.  Findings and Conclusion: The results indicate that using the official figures, the trend in completeness of death registration is illogical largely due to over-estimation of the population size in 2011.  This in turn produces over-estimation of life expectancy at birth in South Africa.  It is important to get population figures as accurate as possible to avoid inefficient allocation of scarce resources