In October and November 2008, the Afrobarometer surveyed a nationally representative, random, stratified probability sample of 2400 South Africans. Face-to-face interviews were conducted in the language of the respondent’s choice by the Afrobarometer team based at Citizens Surveys. The Institute for Democracy in South Africa (Idasa), a Core Partner in the Afrobarometer Network, planned and provided technical backstopping for the survey.
The sample was drawn by taking the smallest geographic units, Census Enumeration Areas (EAs), and stratifying all EAs across the country into separate lists according to province, geographical area (urban/rural) and race. 600 EAs were then randomly selected from these lists with the probability proportionate to its size in the overall population as represented in the 2001 population census and 2008 projections from Stats SA. This ensured that every eligible adult had an equal and known chance of being selected. Four households were then randomly selected within each EA, and a respondent 18 years of age or older was randomly selected from each household. A gender quota ensured that every other interview must be with a female. A sample of this size is sufficient to yield an overall margin of error of +/- 2 percent at a confidence level of 95 percent.