Statistician-general Risenga Maluleke has come out guns blazing against critics of South Africa's 2022 census, dismissing them as representing a minority view in a scientific world that widely endorsed the project.
The census relied on the agency's scientists to work rigorously on various levels of statistical methodology and the outcome had the stamp of approval of internationally acclaimed peer reviewers. The South African Statistics Council reviewed the census and approved it before it was released, he said.
“We should not behave as if the monopoly of science, or the word scientist, befits only people when they raise issues about the census, and that other people, and those who worked on the census, are not scientists,” he said in an interview this week.
At the centre of the row about the veracity of the census results released last year is a population undercount of 31%, and logistical difficulties during the Covid pandemic. Business and governments use the census to make key decisions.
Due to the undercount, Tom Moultrie, professor of demography at the University of Cape Town, and Rob Dorrington, professor emeritus at the University of Cape Town, have described census 2022 as just an estimate instead of an account of a credible enumeration.
The Statistics Council, an advisory board made up of industry experts, brought in local and international statistics gurus to assess both the methodology and outcome of the census, Maluleke said.
There are times when experts and councils have advised countries, 'this data doesn't make sense, you can't release it'. In our case, it has never happened
“The council came back with a statement that 'you can release the figures'. And so when council came back and said the data makes sense, as in the data is fit for purpose, why would I sit on that data?
“There are times when experts and councils have advised countries, 'this data doesn't make sense, you can't release it'. In our case, it has never happened.”
Maluleke said the undercount was identified through the agency's own initiative — a post-enumeration survey — which empowered it to effectively estimate a credible count of 61-million people, from 51.8-million in 2011.
“We haven't deviated from the same methods we have used since 1996.”
The undercount in census 2001 was 11% and in 2011 it was 14%.
“But there were three things that affected us [in conducting the 2022 survey]. There was Covid. You and I have forgotten that we used to wear masks when we moved about. When you send field workers into villages or into settlements, or suburbs, [there were] people who would not allow strangers into their houses,”
“So we count everyone within the borders of South Africa. And that is in line with the UN's fundamental principles of official statistics, as well as in line with UN guidelines on conducting a census,” said Maluleke.
“In our regular quarterly labour force survey, we take 30,000 households, and we reach between 67,000 and about 100,000 people. And out of those, we're able to estimate what unemployment is for the whole country. So with 42-million people, there's no way you would fail to estimate for the whole country,” he said.
The 2022 census would not be redone. “We stand by the census numbers, we are solid, we have put them through a battery of tests.
“There are some countries that do not even conduct the post-enumeration survey. And because they don't conduct the post-enumeration survey, they don't know what their undercount is, but they're able to release their numbers. So in our case, according to the guidelines of the UN, if your undercount is 5% or less, it's what you would desire. You don't even need to adjust. But once it's higher, you have to adjust.”
Maluleke said budget constraints did not affect the outcome of the enumeration. “Budget constraints have always affected statistics South Africa, it's not new in the census.”
While coping with competing finance needs, Stats SA had decided to protect its research on indicators of the health of the nation.
"[We] will protect national indicators, your GDP will come out, your quarterly labour force will come out, your quarterly employment survey will come out, your consumer price index will come out, all of them will come out.”
Where finances were tough, the agency would consider reducing the frequency of some of the products. For example, the quarterly labour force survey could come out twice or once a year. Or the monthly CPI could be released every three months.
“So those are considerations we can make. But we haven't reached that point.”
