An omnibus survey is done quarterly and its purpose is to give clients an opportunity to participate in a national survey at low cost. A number of clients’ questions are combined into one questionnaire. This questionnaire is usually administered to probability sample of about 2200 respondents in the whole country (South Africa). The 1996 February omnibus consisted of two separate samples of 2200 each – one sample having a yellow questionnaire and the other sample having a pink questionnaire. The February/March 1996 omnibus survey was undertaken over the period 27 February to 24 March 1996. The data from this survey was available in May 1996. The fieldwork was done on a countrywide basis including all nine provinces.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
Units of analysis in the survey included individuals
v1: Edited, anonymised data for licensed distribution
The 1996 Omnibus survey collected demographic data (sex, age, marital status, education, language, income, occupation, population group), as well as data on the following topics: The Truth Commission, politics, the economy, the police services, non-South Africans living in the country, elections, perceptions of the government, local government, and cancer
The lowest level of geographic aggregation for the data is Magisterial district.
The lowest level of geographic aggregation covered by the data is magisterial district
The universe included all household residents 18 years old or older
Producers and sponsors
Human Sciences Research Council
The South African population of persons 18 years and older was stratified according to:
Province (Western Cape, Eastern Cape, Northern Cape, Orange Free State, Natal/KwaZulu, Eastern Transvaal, PWV, North Western Province, Northern Transvaal)
Rural areas in former self-governing and TBVC states
Squatter areas in former non-white urban (metro and non- metro areas)
Hostels and hotels
Former urban areas for coloureds
Former urban areas for a Asians
Former urban areas for blacks
Former urban (non- metro) areas for whites
Former urban (metro) areas for whites
Rural areas, excluding the former self-governing and TBVC states
The sample allocation to these strata was done roughly proportional to the adjusted 1991 populatio n census figures with a few exceptions, among which was to ensure a minimal provincial total of 120. Multistage stratified cluster (probability) sampling was used to draw the respondents with the adjusted 1991 population census figures as measure of size. Census enumerator areas and similar areas were used as the clusters in the pen-ultimate sampling stage, from which an equal number, viz. one or two by four households were drawn. All clusters were drawn with probability proportional to size, whilst households were drawn from the final clusters with equal probability (systematically). Respondents were drawn at random from qualifying household members. In addition, population of live-in domestic workers was sampled in relation to their residence in already drawn households.
The sample design allocation per population group was not proportional to the population figures as given in the adjusted 1991 census figures as well as updated figures for certain areas such as the former TBVC states. The population total for each region and socioeconomic category was divided by the actual realisation in the particular cell. The figure derived is the pre-weight that was used. The data was weighted for the age group 18 years and older, in other words, the population that qualified as respondents.
The aim of the factor weighting was to correct the disproportions which were incorporated in the sample design (minimum number of 120 respondents per province, over-sampled population of Asians) and the only weighting targets were stratification variables: ‘province’ and ‘socio-economic category’.
Dates of Data Collection
Data Collection Mode
Data Collection Notes
MarkData of the Human Sciences Research Council conducted the fieldwork for the survey