Statistics South Africa (Stats SA) has firmly defended the accuracy of Census 2022 data, rejecting demands from the academic and scientific communities for a recount.
The agency faced scrutiny after deciding not to release critical data, such as statistics on mortality, fertility, employment, and household income, based on expert recommendations.
This withheld data is essential for enriching the Census 2022 findings, which help researchers, policymakers, and government bodies better understand societal dynamics and allocate taxpayer resources effectively.
The decision to withhold the data has raised concerns among demographers about the reliability of the census, conducted in February 2022 during the Covid-19 pandemic and publicly released in October 2023, with a total expenditure of R2.3 billion.
In a recent statement, an acting senior official from Stats SA explained that it is standard practice to withhold data when there are concerns about reporting and coverage accuracy.
The official noted that while most census data has been published, the remaining information is still undergoing quality checks. This approach is consistent with practices for other surveys, where some data is also withheld until it meets quality standards.
The choice to delay the release was influenced by lower response rates in door-to-door surveys, with concerns that incomplete data could mislead the public. “We ensure data quality before making it available,” the official said.
Alternative Proposals for Data Reliability
In response to the withheld key data, Stats SA has directed researchers, policymakers, and government departments to its mid-year population estimates, which include fertility and mortality statistics derived from administrative records provided by the Department of Home Affairs.
For insights into employment and household income trends, the agency has recommended its annual Income and Expenditure Survey and Quarterly Labour Force Survey.
Criticism has emerged from UCT demographers Rob Dorrington and Tom Moultrie, who argue that Census 2022 exhibited significant anomalies, including a 31% undercount, compared to a 14% undercount in the 2011 census. They emphasize the need to address these issues before planning the 2031 census.
Solly Molayi, acting deputy director-general at Stats SA, acknowledged the 31% undercount discovered through post-census surveys and attributed it to challenges faced during Covid-19 lockdowns. He noted that data collection methods inherently have limitations and criticized the reliance on models that do not involve direct household surveys.
Molayi reaffirmed the agency’s commitment to refining the withheld data and emphasized that the census results, including sections on water access, education, migration trends, and disability, remain valuable. He rejected calls to discard the entire census due to the issues with specific variables.
