Evaluating the plausibility of the method of using both the civil registration and census data in estimating adult mortality at district level in South Africa, circa 2011

The challenge in estimating mortality, both at national and sub-national levels, in developing countries such as South Africa is that neither of the death data sources (vital registration and census) are one hundred percent complete, that is, vital registration data is prone to incompleteness and deaths reported by household are subject to over- or under-reporting which may vary by age. Also, apart from issues with data sources, there is no method that estimates mortality accurately at subnational level and the methods for estimating the level of completeness of reporting of deaths cannot be applied at subnational level (due to issues with migration). Thus, measuring mortality rates at subnational level is a challenge. This research seeks to employ a method used by Dorrington, Moultrie and Timæus (2004) that makes use of both data sources in combination so as to overcome the weakness and makes use of the strength of each data source. To estimate the level of completeness in the year prior to the 2011 Census (to correct the number of deaths registered), first, the Death Distribution Methods (Synthetic Extinct Generations +delta and General Growth Balance method) are used to estimate the level of completeness of the vital registration deaths for the intercensal period 2001-2011 by population group. Thereafter, the level of completeness for each of the years in the intercensal period is estimated by fitting a logistic curve to the level of completeness for the intercensal period of 1996-2001 and 2001-2011 (derived by both Chinogurei (2017) and Richman (2017)). Thus, the number of deaths registered in the year prior to the 2011 census are then corrected for either under- or over-reporting using the estimates of completeness to obtain the true number of deaths by population group and age group for each sex. The corrected true numbers of registered deaths are then used to determine the age-specific correction factors by population group for correcting the household reported deaths at district level and thereafter estimates of mortality at district level are determined. Comparison of estimates derived in this study to estimates by other studies indicated that the method produces plausible estimates at district level, thus, findings in this research strengthens the reasonability of the method.

Grantor dc:publisher.institution

Centre for Actuarial Research (CARE)

Year dc:date.issued

2018

Author dc:creator

• Marozva, Nicola

Advisor dc:contributor.advisor

• Dorrington, Robert

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/29820

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/29820