Is Census 2011 accurate? Still depends on who you ask
The census is solid, Statistics South Africa says. But its critics are still not agreeing.
It is strange that South Africans suddenly started having babies again, Statistics SA said on Thursday, and the distribution by province of the population comes as a surprise. But that should not be reason to doubt the accuracy of the 2011 census.
"We still stand by our decision that the census results are accurate," said Howard Gabriels, chair of the Statistics Council that certified the results. "It is in fact very timely, and it is fit for use in policy-making and decision-making."
Gabriels called an urgent press conference on Thursday to address issues with the census data raised by Cape Town University demographers Tom Moultrie and Rob Dorrington. As part of a group of consultants contracted to consider the accuracy of the data, Dorrington and Moultrie called for a delay in the release of the census data, then went public with their concerns when that was not heeded.
The council, which provides oversight of the work of Stats SA, took a report by Moultrie and Dorrington "very, very, very seriously", Gabriels said, but on the balance of evidence and with the support of other experts, decided that the results were accurate and ready for relase.
Among the findings that have been released so far, and have raised eyebrows, are numbers that show a sudden surge in fertility across the population, including among the more affluent white group. That puts South Africa's fertility rate at 2.7, well above the rate of 2.1 needed to keep the population steady. Numerous other surveys and models, as well as indicative trends – including earlier numbers from Stats SA – directly contradict that.
But independent data, drawn from sources including records of births and deaths confirm that trend, said Griffith Feeney, a retired American demographer and part of the evaluation team. "Indeed that pattern in the age distribution is real," he said on Thursday. "It's surprising, certainly not what we expected."
Feeney was among the experts who joined Gabriels and statistician general Pali Lehohla to defend the results. The census outcome is a better reflection of reality than models now proven incorrect, said Eric Udjo of the University of South Africa. And there is "nothing that indicates here that there is a problem" with the age data, said Jacky Galpin, a retired statistician from Wits University.
Those arguments have not mollified Moultrie and Dorrington, however. Both are highly respected in their field – Feeney on Thursday described Dorrington as someone he has "known for many years and respect very much" – and both remain disturbed by what they see in the data, and the process that led to its publication.
"This is a logical process you have to work through," said Moultrie. "You can't just have five people saying it is right, and that makes it right." Statistics SA had received reports from different eminent demographers, he said, and chosen those reports it agreed with. In the meanwhile, both historic and internal inconsistencies remain.
"If you look at the numbers, they imply that white fertility rates have increased massively over the last five years," said Moultrie. "The white population is trivial in the broader demography, but it is a bellwether. Everything suggests that didn't happen. To take a solipsistic view and say anything inconsistent with the census data is wrong, that's very dangerous."
The report submitted by Dorrington and Moultrie remains confidential, and the statistics council could not on Thursday say whether it would be made public. The two plan to compile an analysis of the publicly-available data not covered by confidentiality in due course, but say that could take several months.
See the Mail & Guardian this Friday as we unpack how the census data could influence hundreds of millions of rands in government.