Is the 2016 South African Community Survey data suitable to estimate fertility indicators?

Previous studies from South African censuses and community surveys observed some errors in reported fertility data. Erroneous data on fertility often produces implausible indicators which affects decision making processes in policy planning and the review of programmes. This paper appraises the quality of fertility data from the Community Survey 2016 and employs the Censuses 1996, 2001, 2011 and Community Survey 2007 to determine consistencies overtime. Demographic techniques of evaluating data were applied to lifetime and current fertility to measure errors and to indicate ages where errors were more concentrated. The techniques included were average parities, the Brass P/F ratio method, parity progression ratios, age specific fertility rates, sex ratio at birth and the relational Gompertz model. Distributions of women in single ages were inspected and levels of unspecified parities and childlessness were examined. There were indications of age heaping and preference at ages ending with 0 and 5. Data on current fertility conformed to the traditional pattern of fertility distribution amongst African populations. The pattern of average parities indicated consistencies in reporting lifetime fertility. Levels of average parities and age specific fertility rates suggested a fertility decline. The pattern of current fertility and mean children ever born from diagnostic plots of the relational Gompertz model attested to the declining fertility, however data showed indications of an overstatement of children ever born by elderly women. In each age group less than 2% of women did not specify parity. [ABSTRACT FROM AUTHOR]