A census is one of the most expensive and logistically challenging exercises the government undertakes and is one of its most important planning tools. It is used to determine budget allocations, plan where to build schools and hospitals, and measure the reach of interventions such as childhood vaccination programmes.
So we should all be concerned when the country’s chief statistics agency dismisses questions about the integrity of Census 2022, fails to release data on key variables and refuses to publish the technical reports on which it bases its insistence that all is well.
Stats SA certainly faced unique challenges in conducting the census in October 2022. SA was still reeling from the disruption caused by the Covid-19 pandemic, and the project was carried out in the wake of deadly riots and the distraction of local government elections. Stats SA also decided to switch to a digital questionnaire, and it may have had to rush the job because the Treasury refused to let it roll funds over from one fiscal year to the next, and delay the project by a few months.
The results were astounding. Stats SA reported a 31% undercount, a figure that was more than double the 14.6% undercount in Census 2011 and the highest recorded yet by the UN Population Council. After adjustments it concluded the total population was 61-million — a million more than its 2021 midyear population estimate — and released provincial and municipal population figures that immediately raised eyebrows.
The Gauteng City Region Observatory said the results were bewildering, as they indicated an apparent halt to the growth of Johannesburg while the population of the historically low-growth Ekhurhuleni surged by an average of 2.5% a year since 2011.
Reporting biases
In July, the Medical Research Council published an assessment by UCT demographers Tom Moultrie and Rob Dorrington that concluded the census was so deeply flawed it was not fit for purpose. They identified problems with the post-enumeration survey and said it led to a national count that was 1-million too high and resulted in grave errors in district and municipal population figures that were clearly at odds with other information sources such as local voter registration data and satellite imagery of the geographical areas covered by dwellings.
To cap it all, Stats SA announced last week that its technical experts had advised it against releasing figures on fertility, mortality, employment and household income due to reporting and coverage biases in the data.
There is, unfortunately, no immediate alternative to the census. The administrative data held by government departments, such as records on births, deaths, taxes and school attendance, could in theory be harmonised to build a picture of the population but there is no guarantee it will be feasible or affordable soon.
Stats SA has an unfortunate history of digging in its heels when challenged to defend its work. In 2013, faced with questions about the integrity of Census 2011, which contained implausible surges in fertility and in the number of young white women, it insisted the figures were accurate and that the results had been signed off by the Statistics Council.
It’s running exactly the same playbook now. That is a great pity, for time is fast running out to figure out what went wrong with the most recent census and make sure it isn’t repeated in 2031.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.