Census ’22 data is ‘fundamentally flawed’ as DA calls for new exercise

The failure of Census 2022 to produce accurate data has direct implications for policies and interventions to benefit the lives of all South Africans – and the DA says the situation is so catastrophic that the whole exercise should be rerun.

Statistics South Africa (Stats SA) this week admitted that valuable data collected from its Census 2022 is fundamentally flawed and not of good enough quality to be released for official or public use.

Amid questions about the data’s credibility, Stats SA released an updated version of its publication Census 2022 in Brief, which now excludes income and earnings, labour and employment, and mortality and fertility.

Implications of flawed Census is huge

The data was scheduled to be released later this month. The exclusion of the data amplifies the concerns previously raised by University of Cape Town demographers, Tom Moultrie and Rob Dorrington, about its integrity.

Speaking to Saturday Citizen, Moultrie said the implications of the flawed data is huge.

“The problems came from the very start when they revealed in October last year, that there’s been a 31% undercount, which is a world record for undercount censuses where they’ve tried to estimate what that undercount is.

ALSO READ: ‘Census 2022 failure a national scandal’ − DA calls for a re-run

“It is more than double the amount of the undercount in 2011 at the time of the last census.

“The fact that they’re not even releasing the data points to a greater scale of the problems in the census and which kind of amplifies our concerns about the census results,” Moultrie said.

The flawed census data does raise concerns over the allocation of resources across South Africa, he said.

Questions raised over allocation of resources

“For example, if the census shows there are a whole lot more schoolchildren in a municipality, that might lead them to build another school or another clinic, but those people might not actually be there.

“You could see a misallocation of resources, schools or clinics or roads or infrastructure being built in places where there actually might not be the need for them, whereas there might be other parts of the country where they are required.”

Moultrie said municipal population sizes was one of the factors that determined the allocations from provincial governments.

ALSO READ: ‘A travesty’: Census 2022 fundamentally flawed, Stats SA confirms poor data quality and bias

“If those municipal population numbers are incorrect, some municipalities might be benefitting, while other municipalities will be suffering because of those misallocated of financial resources.”

He said census data is used by researchers, policy makers and planners to do a “deep dive” into the spatial patterns and spatial legacies of apartheid and unemployment in the country. This can now not happen.

“That is a loss for being able to design programmes or to make sure that childhood vaccines are in the right place at clinics,” Moultrie added.

Stats SA needs to investigate

“All of those are concerns. They might not affect an individual directly, but they will affect the ability of the government to deliver services where they are most needed to benefit the lives of South Africans.”

Moultrie said Stats SA needs to investigate what went wrong.

“Given that the next census is due in 2031, in seven years’ time, the effort now should be spent in the next, two to three years on really trying to understand what went wrong with this census.

ALSO READ: South Africa’s population surpasses 63 million, females live longer than males

“It’s too easy, to blame Covid or blame government funding. There are indications other things went wrong as well.

He said running a census in the middle of the Covid pandemic was a mistake.

“Had we been asked, we would have certainly advised Stats SA not to run the census then and to delay it by a year or two. But they went ahead and we can’t really correct the results of the census now because the mechanism by which they corrected the census is itself out of date.”

Not official census numbers

Moultrie said Stats SA might be able to produce alternative estimates, but those aren’t going to be the official census numbers.

“Stats SA needs to formulate a very clear plan for understanding what went wrong and a clear plan for the next census and to express the willingness to engage with users on the data to produce that alternative set of estimates.”

Moultrie and Dorrington previously produced a report for the Medical Research Council that pointed out the flaws in the census process. He said Stats SA never engaged with them about this.

ALSO READ: Danger of sharing customers’ data without permission

“Stats SA’s response at the time was to say, ‘no, it’s all fine. We followed the recipe.’ They have never engaged with the substance of the problems identified in that 60-page report.

“They’ve never tried to actually engage with the substance of the critiques which we issued in July. Nor indeed now.

“Their answers are still along the lines of, ‘we think the data is fit for purpose’. We have demonstrated using their own data that there are significant inconsistencies and incoherence in the data,” Moultrie said.

Concerns ‘unfounded and misleading’ – Stats SA

“We think part of that process of acknowledging that the census was not a good census is a willingness to engage with users – not only us, but economists and other researchers and policy makers and planners and government and National Treasury, the department of basic education – all of those users of the data should basically be giving input and feedback to Stats SA, saying this is what they found wrong with the data.”

Stats SA in July was adamant that the Census 2022 met data collection standards, despite warnings by Moultrie and Dorrington. At the time, it said the concerns were “unfounded and misleading”.