CENSUS 2022 ANALYSIS
How much can we rely on Census 2022?
The latest South African Census, released this week, is a hugely important set of numbers that will be used to shape policy and budgets. But is it accurate enough?
Three experts expressed concern to Daily Maverick this week about the extent of the undercount of the just-released Census 2022. The undercount refers to the proportion of the population who were missed during counting.
For Census 2022, that figure is 31%.
The experts, who did not wish to be named, stressed that a fuller picture of what happened would be available only once the relevant Census data has been worked through. But the undercount figure of 31% is considered a significant cause for alarm.
The previous closest contender was in 2001 when the undercount was at 17% – a figure former Statistician-General Pali Lehohla would later characterise as Stats SA having “lost the plot”.
For Census 2012, the undercount of 14.6% was considered a concern – particularly because the previous year, Lehohla had promised that the double-digit undercount would be a “thing of the past”.
In this latest census, the undercount has more than doubled.
In developed countries, the census undercount usually hovers around one to 2%. Different countries employ different methodologies to calculate the undercount, so direct comparisons are difficult, but research suggests that the average in African countries is less than 5%.
Stats SA bullish, regardless
Stats SA has downplayed the problem of the undercount on this occasion.
Asked by Daily Maverick whether the fact that the undercount had doubled from 2011 was considered a concern, a joint response from the post-enumeration survey and Census 2022 project teams said:
“Census 2022 was affected by unprecedented challenges, including riots, ongoing Covid-19 lockdowns and climate change issues such as flooding in some parts of the country.
“Stats SA endeavours in all their collections to have improved coverage rate. We will continue to innovate with the aim of improving coverage generally in our surveys and census.”
Elsewhere, Statistician-General Risenga Maluleke has been quoted as saying that the validity of the Census numbers should not be doubted because the necessary adjustments had been made to reflect the undercount.
“The provinces with the higher undercounts mean they will have higher adjustments when we are dealing with adjustments out of what we got and what we are adjusting, and this is done in line with the United Nations methods and that makes sure that we have the quality of our data guaranteed,” Maluleka was quoted as saying by EWN.
Two of the experts Daily Maverick spoke to this week drew attention to the particularly high undercount of white and Indian South Africans.
The Census 2022 findings showed the white population of South Africa in marked decline, dropping to 7.3% of the population. In certain age groups, such as 0 to 4-year-olds, whites are recorded as low as 4% of the population. The Indian and Asian population accounts for 2.7% of the total population.
The post-enumeration survey records the undercount for white South Africans as 61.64% and for Indian South Africans as 72.3% – as compared with an undercount for black South Africans of 36.74%.
[UPDATE: On 1 November 2022, Stats SA contacted us to tell us that the undercount figures contained in the post-enumeration survey were incorrect because they were preliminary estimates, not final estimates. Stats SA says that the final undercount figure for white South Africans is 24,86% and the final undercount figure for Indian South Africans is 42,10%].
Stats SA hinted to Daily Maverick that white and Indian South Africans may have been more reluctant to open the door to Census counters, or less accessible, saying: “Stats SA continues to have challenges with conducting fieldwork in high-walled and gated communities. Census was no exception.”
Part of the complexity is that the post-enumeration survey, which takes place after the Census and is used to estimate the undercount, had a very high undercount itself.
“According to the results from the matching process, a member of the in-scope population had an approximately 58.54% chance of being enumerated in the Census, 65.96% chance of being enumerated in the post-enumeration survey, and 38.61% chance of being enumerated in both,” it states.
Figures around homelessness and migration
Two aspects of the Census findings have already raised eyebrows: the figures around homelessness and international immigrants.
The Census recorded 55,719 homeless people nationally.
Daily Maverick asked Richard Bolland, the founder of New Hope SA – an organisation that works with unhoused men in Cape Town – what he made of that figure.
“I don’t have any concrete figures for South Africa but there have been a number of research findings in Cape Town to suggest that the number of unhoused people is around 14,000,” Bollard said. (Census 2022 records 9,743 homeless people in Cape Town.)
If one were to extrapolate from that, Bollard suggested, 55,719 “seems a very low estimate” for total numbers nationally.
The Census figures for international immigrants, meanwhile, have surprised many. It records 2,418,197 international migrants – which would suggest that in the decade between 2001 and 2011, South Africa gained more than a million migrants, but in the decade between 2011 and 2022, less than 300,000.
Making these results more confusing is the fact that the 2020 mid-year population estimates by Stats SA estimated that there were 3.9 million migrants in South Africa.
This was a figure Stats SA reiterated in August 2021, with Statistician-General Maluleke quoted as saying: “If one uses the output of foreign-born persons enumerated in Census 2011 and adds to it the net international migrants for the period 2011-2016, as well as the period 2016-2021 from the 2021 mid-year population estimates, one would get an estimation of 3.95 million persons.”
What explains the almost 1.5 million discrepancy?
Diego Iturralde, from Stats SA’s mid-year population estimates team, said there were several reasons why foreign-born people might want to avoid being counted as such – including the periodic upticks in xenophobic violence.
“In essence, we share your concern about migrant numbers, but we do not feel this is due to them being mainly missed out, but rather that foreign nationals may have reported themselves as South African-born,” Iturralde said.
“We are quite confident around the population count and do not feel that we would need to add to that count the foreign-born people who are not captured as such. We will continue to monitor any migration data that we collect so as to update this number…” DM