The victims of crime survey 1998 was commissioned by the South African Department of Safety and Security (DSS), and undertaken by Statistics South Africa (Stats SA). The first national survey of its kind in South Africa, this countrywide, household-based survey examines crime from the point of view of the victim. While surveys of crime victims cannot replace police statistics, they can provide a rich source of information which will assist in the planning of crime prevention. A victim survey can also examine the extent of reporting of crime, explore the perceptions that different people have about the police and police services, and act as a benchmark against which future surveys of the same nature can be compared.
Kind of Data
Sample survey data
Unit of Analysis
Households and individuals
v1: Edited, anonymised dataset for licensed distribution
The Victims of crime survey covered patterns and extent of crime, types of crime, and reporting of crime, as well as citizens’ perceptions of police and police services.
The survey has national coverage
The lowest level of geographic aggregation for the data is magisterial district
The survey covered all households in South Africa
Producers and sponsors
Statistics South Africa
Department of Safety and Security
United Nations Interregional Crime and Justice Research Institute
United Nations Development Programme
Department of Safety and Security
The sample consisted of 4 000 people aged 16 years or more. It was drawn in three stages. Firstly, a probability sample of 800 enumerator areas (EAs) was drawn from the sampling frame of 86 000 EAs, as demarcated for the 1996 population census. This sample was stratified explicitly by province, and implicitly by the 42 police districts of the country. Secondly, within each of the 800 EAs, five households were selected for interviewing, using systematic sampling. Thirdly, one respondent aged 16 years or more was selected to be interviewed in each of the five households in each sampled EA. This person was chosen using a table of random numbers. Once a respondent had been selected, fieldworkers were instructed to make sure that they interviewed only that specific person and nobody else. In case of non-contacts with that person, repeated callbacks (at least three) had to be made. There were no substitutions for refusals or non-contacts.
The 1996 population census formed the basis for weighting the data to the population of the country. Two different sets of weights were used for this study, household and individual weights. The questions posed on crimes committed against households were weighted to the population of households in the country, while those concerning crimes committed against individuals were weighted to the population of individuals aged 16 years or more. Factors taken into account in weighting households were province, police area and EA type (urban formal, urban informal, non-urban traditional, commercial farms and other types of nonurban areas, for example small villages or mission stations). Additional factors taken into account for individual weights were population group, age, gender and population growth.
Dates of Data Collection
Data Collection Mode
Data Collection Notes
Fieldwork was conducted over a two-week period, between 16 and 27 March 1998. An extra week was allowed for carrying out check-back procedures and finalising interviews. Fieldwork was organised through teams of five people, each consisting of one supervisor and four fieldworkers. There were 49 such fieldwork teams, each covering an average of 16 EAs. An approach letter bearing the fieldworker's name and ID number was sent out with each fieldworker, introducing him or her to the household. To reduce the incidence of a selected female respondent being confronted by a male interviewer, the ratio of male to female interviewers in a team was kept at 1:3. Teams were provided with contact telephones for organisations with specialist expertise in case their advice was needed during the course of fieldwork.
The survey questionnaire was based on a standard international questionnaire, but with certain modifications for use in South Africa. The international questionnaire covered eleven main crimes, including theft of a car or other motor vehicle, theft from a car or other vehicle, car vandalism, theft of a motor cycle or scooter, theft of a bicycle, burglary or housebreaking, attempted burglary, robbery with force, personal theft, sexual incidents and assault and two supplementary crimes (consumer fraud and corruption). In the South African questionnaire, the following crimes were added on the recommendation of the advisory committee to meet specific South African needs: theft of livestock, poultry and other animals, hijacking or attempted hijacking of vehicles, deliberate damage, burning or destruction of dwellings and deliberate killing or murder.
A control questionnaire was administered by the fieldwork supervisor in one of the five households selected for participation in each enumerator area. This served as a check on the accuracy of the random selection process of the individual in the household, and of the quality of information collected. The survey was favourably received, and 97% of the sample was realised.
The processes of computer programming, data capture and data analysis involved several steps:
A tabulation plan was drawn up beforehand to assist with writing the computer programme for data capture. The data-input programme, containing both range and consistency checks, was written by a programmer working in Stats SA's Directorate of Household Surveys. Coding of the questionnaires and data capture were handled by temporary staff. Once the capturing was completed, additional editing programmes were written, and then the data-cleaning process was completed. Tables from the dataset, based on the tabulation plan, and the data set itself were then made available for analysis and report-writing.
Statistics South Africa. Victims of Crime Survey 1998 [dataset]. Version 1. Pretoria: Statistics South Africa [producer],1998. Cape Town: DataFirst [distributor], 2011. DOI: https://doi.org/10.25828/r9w2-9n50