Why Census ’96 results are the best ever Census ’96, whose results were launched by President Mandela in Pretoria on Tuesday, "was a decisive improvement over recent censuses in several crucial respects", according to Stats SA Head Dr Mark Orkin. "All the residents in our new democracy were treated equally by the census for the first time" Dr Orkin noted. The nine million households were visited in the same way in all parts of the country, including rural areas and informal settlements. In the previous census, many of these areas were estimated from aerial photographs, or omitted altogether and estimated afterwards. In 1991 the Census questionnaires were printed in English or Afrikaans. This time, the respondents could select the language of their choice, when supplying the information in each household by interview or by filling in the form. "Questionnaires were available in all eleven official languages" noted Census chief director Mr Pali Lehohla. "The print run weighed 740 tons - just getting the questionnaires into the hands of the 100,000 enumerators was a massive undertaking." Census ’96 was also the first time that a a nationwide post-enumeration survey was undertaken immediately afterwards. "This is done to estimate the persons or households who are unavoidably missed in any census", according to Dr Ros Hirschowitz, chief director of Research and Development at Stats SA. The most skilled of the Stats SA fieldworkers re-visited one percent 1% of the census "enumeration areas", listed the households afresh, and visited them again to check the count. The results from this exercise were compared with those from the wider census by a laborious process of "matching" by household and individual. This yielded an improved adjustment for "undercount", to take account of the people who were missed. The undercount was found to have been 10,6%. In the ensuing period, the data on the questionnaires was coded and punched into computers, by a total of 5 000 contract staff working in shifts in nine provincial processing centres. During the 1991 census, roughly half the population were incompletely counted, or not counted at all. Instead, a demographic model reaching back twenty years was used to estimate the size of the population. |