Testing the orphanhood method against benchmark mortality rates in South Africa since 1996

• Main Author: Magadzire, Andrew
• Other Authors: Dorrington, Rob
• Format: Dissertation
• Language: English
• Published: University of Cape Town 2014
• Subjects: Demography
• Online Access: http://hdl.handle.net/11427/5894

Includes abstract. === Includes bibliographical references (leaves 71-73). === This study utilises the 1996 and 2001 Censuses, and the 2007 Community Survey to test the orphanhood method mortality estimates against consensus estimates of mortality produced using other methods. In all the two censuses and the survey there were questions on the survival status of the respondent’s biological parents. The orphanhood method is widely used in developing countries where the vital registration is incomplete. In countries where the data on survival of parents has been compared with other mortality estimates, it has been observed that the orphanhood method estimates are biased by adoption effect, selection effect and age exaggeration. In the late 1980s and early 1990s, the advent of HIV/AIDS has also biased estimates obtained from the orphanhood method. Non-independence of the mortality of children and their mothers, relationships between HIV infection and fertility, and changes in age-specific mortality result in biases which affect the accuracy of the method. These biases have been observed to have a net effect of underestimating mortality especially female mortality. An adjusted method has been proposed which reduces error, when working with data taken from populations with a significant HIV prevalence. This adjustment can be applied, but further research to identify revised adjustments would further improve the accuracy of the method. The current research applies four variants of the orphanhood method, the regression variant, UN Manual’s two-survey variant, Timaeus’s synthetic cohort variant and the one allowing for HIV/AIDS proposed by Timaeus and Nunn.