The South African Statistics Council, an independent body that oversees Statistics South Africa, plans to subject the 2006 census to a thorough review following concerns about certain aspects of the 2001 census.
Since the release of the 2001 census experts in the field have been debating whether the benefits of conducting a five-yearly head count outweigh the costs.
It costs R1-billion over five years to conduct the census. Those arguing for a five-yearly census say it is necessary for development. Those arguing against it say it needs to be justified. Other interlocutors say that if it continues its quality must improve.
Hillary Southhall, chair of the council, says the review of the 2006 census is designed to determine whether the benefits outweigh the costs.
”Council suggests that the conducting of a census in 2006 — the level of detail of the information to be collected, and the number of questions to be included — be subjected to a review designed to determine whether the anticipated benefits outweigh the costs.”
She suggested that Statistics SA should consider conducting the census every 10 years.
”If they decide to go ahead with it in 2006 though, we need to have the costs justified,” she said.
But various statisticians and academics have lambasted her ideas. They say the infancy of our democracy demands continual reassessment.
”Stats SA is spending between R22 and R25 [divide the R1-billion spent on the census by the 44,8-million population count] on each South African to gauge their needs. This makes the census money well spent,” said Professor Tim Dunne, head of the department of statistical science at the University of Cape Town.
”The complex problems of poverty, hunger, Aids and the provision of water and power in the far-flung regions of the country can be best informed by regular and well conducted censuses. We owe it to the poor and dispossessed to quantify their basic needs for a simple and comfortable human life.”
Minister of Finance Trevor Manuel said R1-billion is a realistic amount, but that the nature of the census needs to be carefully considered.
”It [the census] is not only a cost thing. We need the information to help plan and quantify the improvement of life. But it is something that one wants to do quite sparingly. We need to decide what we need, at what interval,” he said.
Debbie Bradshaw, an epidemiologist at the Medical Research Council, said, ”We should seriously consider if we can [conduct the survey] every 10 years instead of every five years and in between do surveys to provide additional information.”
Jacky Galpin, head of the school of statistics and actuarial science at Wits University, said that there is merit in conducting the censuses less frequently and complementing them with surveys, such as All Media Product Surveys and household surveys.
”The censuses can paint a demographic picture of the country and then regular surveys which draw on representations of the population can be grossed up as an equivalent of the census statistics. The big question is whether there is a point in asking all those questions [the 2001 census asked about 75 questions]. People get fed up with filling them in and start making mistakes.”
Galpin said shorter surveys are easier to fill in and quicker to capture. ”But other people argue that if you are going to the place anyway, once you’ve got yourself in the door, you may as well ask the whole raft of questions,” she said.
Despite the advantage of surveys, ”I feel that we need to conduct the 1996 census because the population is still extremely mobile particularly because of the high unemployment rate and the rise in the land claims and restitution. Once the population is more stable we can begin conducting the censuses every 10 years.”
Meanwhile, there is concern in some circles about the quality of the 2001 census. Does it reach all remote areas of South Africa?
”Regardless of how often it is done, we need to look at the quality and why so many people were missed. One in six [not counted] is quite a poor response,” said Bradshaw.
The statistics council said that 17,6% of South Africans are unaccounted for in the latest census. This is compared to 10,5% in 1996.
Ros Hirschowitz, deputy director of Statistics SA, said the figures were within a ”95% confidence limit”.
”The nature of a census this large is that there will be people missed. The difficulty of an ‘almost industrialised’ country is that there are pockets of poverty and wealth and these disparities make it difficult to balance the statistics. The figures therefore have to be adjusted according to the post-enumeration survey,” she said.
The post-enumeration survey determines the degree of under- or overcount of the population.
Dr Gugu Gule, Statistics SA deputy director general, said plans for the 2006 census had already started despite the debate on its future.
”We do not want to wait for a decision and only after it has been made we find ourselves running around at the last minute,” she said.