Project for Statistics on Living Standards and Development 1993
South Africa, 1993 - 1994
Get Microdata
Reference ID
zaf-saldru-pslsd-1993-v2
Producer(s)
Southern Africa Labour and Development Research Unit
Metadata
Created on
Nov 07, 2008
Last modified
Mar 13, 2023
Page views
321721
Downloads
25155
Identification
Survey ID Number
zaf-saldru-pslsd-1993-v2
Title
Project for Statistics on Living Standards and Development 1993
Country
| Name | Country code |
|---|---|
| South Africa | zaf |
Study type
Living Standards Measurement Study [hh/lsms]
Abstract
The Project for Statistics on Living standards and Development was a countrywide World Bank-sponsored Living Standards Measurement Survey. It covered approximately 9000 households, drawn from a representative sample of South African households. The fieldwork was undertaken during the nine months leading up to the country's first democratic elections at the end of April 1994. The purpose of the survey was to collect data on the conditions under which South Africans live in order to provide policymakers with the data necessary for development planning. This data would aid the implementation of goals such as those outlined in the Government of National Unity's Reconstruction and Development Programme.
Kind of Data
Sample survey data
Unit of Analysis
Households and individuals
Version
Version Description
v2: Edited anonymised dataset for public distribution
Version Date
1995
Version Notes
Version 1 of the data was the original version released by the World Bank in 71 separate files and 4 folders (folders A-D, representing the 4 stiffy disks holding the data). In verison 2 of the dataset these files were updated for two reasons: Additional data cleaning; and the removal of bogus clusters identified during field research for the KwaZulu-Natal Income Dynamics Study in May 2001. Data from all 39 household interviews in these clusters - 217 and 218 - was shown to be fabricated and was removed from the data files. Weights calculated by the World Bank and provided with the original data were not updated to reflect these changes.
Scope
Notes
The PSLSD covered demographic, economic, education, and health data of enumerated households.
Keywords
| Keyword |
|---|
| agriculture |
| consumption |
| demography |
| education |
| expenditure |
| fertility |
| health |
| housing |
| income |
| land use |
| literacy |
| numeracy |
| poverty |
Coverage
Geographic Coverage
The survey had national coverage
Geographic Unit
The lowest level of geographic aggregation of the data is province.
Universe
The survey covered all household members. Individuals in hospitals, old age homes, hotels and hostels of educational institutions were not included in the sample. Migrant labour hostels were included. In addition to those that turned up in the selected ESDs, a sample of three hostels was chosen from a national list provided by the Human Sciences Research Council and within each of these hostels a representative sample was drawn for the households in ESDs.
Producers and sponsors
Primary investigators
| Name | Affiliation |
|---|---|
| Southern Africa Labour and Development Research Unit | University of Cape Town |
Producers
| Name | Role |
|---|---|
| World Bank | Technical assistance |
Funding Agency/Sponsor
| Name | Role |
|---|---|
| Government of Denmark | Funder |
| Government of the Netherlands | Funder |
| Government of Norway | Funder |
Sampling
Sampling Procedure
The sample size for the survey is 9,000 households. The sample design adopted for the study was a two-stage self-weightingdesign in which the first stage units were Census Enumerator Subdistricts (ESDs, or their equivalent) and the second stage were households.
The advantage of using such a design is that it provides a representative sample that need not be based on accurate census population distribution.in the case of South Africa, the sample will automatically include many poor people, without the need to go beyond this and oversample the poor. Proportionate sampling as in such a self-weighting sample design offers the simplest possible data files for further analysis, as weights do not have to be added. However, in the end this advantage could not be retained and weights had to be added. The sampling frame was drawn up on the basis of small, clearly demarcated area units, each with a population estimate. The nature of the self-weighting procedure adopted ensured that this population estimate was not important for determining the final sample, however. For most of the country, census ESDs were used. Where some ESDs comprised relatively large populations as for instance in some black townships such as Soweto, aerial photographs were used to divide the areas into blocks of approximately equal population size. In other instances, particularly in some of the former homelands, the area units were not ESDs but villages or village groups.
In the sample design chosen, the area stage units (generally ESDs) were selected with probability proportional to size, based on the census population. Systematic sampling was used throughout that is, sampling at fixed interval in a list of ESDs, starting at a randomly selected starting point. Given that sampling was self-weighting, the impact of stratification was expected to be modest. The main objective was to ensure that the racial and geographic breakdown approximated the national population distribution. This was done by listing the area stage units (ESDs) by statistical region and then within the statistical region by urban or rural. Within these sub-statistical regions, the ESDs were then listed in order of percentage African. The sampling interval for the selection of the ESDs was obtained by dividing the 1991 census population of 38,120,853 by the 300 clusters to be selected. This yielded 105,800. Starting at a randomly selected point, every 105,800th person down the cluster list was selected. This ensured both geographic and racial diversity (ESDs were ordered by statistical sub-region and proportion of the population African). In three or four instances, the ESD chosen was judged inaccessible and replaced with a similar one.
In the second sampling stage the unit of analysis was the household. In each selected ESD a listing or enumeration of households was carried out by means of a field operation. From the households listed in an ESD a sample of households was selected by systematic sampling. Even though the ultimate enumeration unit was the household, in most cases "stands" were used as enumeration units. However, when a stand was chosen as the enumeration unit all households on that stand had to be interviewed.
Census population data, however, was available only for 1991. An assumption on population growth was thus made to obtain an approximation of the population size for 1993, the year of the survey. The sampling interval at the level of the household was determined in the following way: Based on the decision to have a take of 125 individuals on average per cluster (i.e. assuming 5 members per household to give an average cluster size of 25 households), the interval of households to be selected was determined as the census population divided by 118.1, i.e. allowing for population growth since the census. It was subsequently discovered that population growth was slightly over-estimated but this had little effect on the findings of the survey.
Individuals in hospitals, old age homes, hotels and hostels of educational institutions were not included in the sample. Migrant labour hostels were included. In addition to those that turned up in the selected ESDs, a sample of three hostels was chosen from a national list provided by the Human Sciences Research Council and within each of these hostels a representative sample was drawn on a similar basis as described abovefor the households in ESDs.
The advantage of using such a design is that it provides a representative sample that need not be based on accurate census population distribution.in the case of South Africa, the sample will automatically include many poor people, without the need to go beyond this and oversample the poor. Proportionate sampling as in such a self-weighting sample design offers the simplest possible data files for further analysis, as weights do not have to be added. However, in the end this advantage could not be retained and weights had to be added. The sampling frame was drawn up on the basis of small, clearly demarcated area units, each with a population estimate. The nature of the self-weighting procedure adopted ensured that this population estimate was not important for determining the final sample, however. For most of the country, census ESDs were used. Where some ESDs comprised relatively large populations as for instance in some black townships such as Soweto, aerial photographs were used to divide the areas into blocks of approximately equal population size. In other instances, particularly in some of the former homelands, the area units were not ESDs but villages or village groups.
In the sample design chosen, the area stage units (generally ESDs) were selected with probability proportional to size, based on the census population. Systematic sampling was used throughout that is, sampling at fixed interval in a list of ESDs, starting at a randomly selected starting point. Given that sampling was self-weighting, the impact of stratification was expected to be modest. The main objective was to ensure that the racial and geographic breakdown approximated the national population distribution. This was done by listing the area stage units (ESDs) by statistical region and then within the statistical region by urban or rural. Within these sub-statistical regions, the ESDs were then listed in order of percentage African. The sampling interval for the selection of the ESDs was obtained by dividing the 1991 census population of 38,120,853 by the 300 clusters to be selected. This yielded 105,800. Starting at a randomly selected point, every 105,800th person down the cluster list was selected. This ensured both geographic and racial diversity (ESDs were ordered by statistical sub-region and proportion of the population African). In three or four instances, the ESD chosen was judged inaccessible and replaced with a similar one.
In the second sampling stage the unit of analysis was the household. In each selected ESD a listing or enumeration of households was carried out by means of a field operation. From the households listed in an ESD a sample of households was selected by systematic sampling. Even though the ultimate enumeration unit was the household, in most cases "stands" were used as enumeration units. However, when a stand was chosen as the enumeration unit all households on that stand had to be interviewed.
Census population data, however, was available only for 1991. An assumption on population growth was thus made to obtain an approximation of the population size for 1993, the year of the survey. The sampling interval at the level of the household was determined in the following way: Based on the decision to have a take of 125 individuals on average per cluster (i.e. assuming 5 members per household to give an average cluster size of 25 households), the interval of households to be selected was determined as the census population divided by 118.1, i.e. allowing for population growth since the census. It was subsequently discovered that population growth was slightly over-estimated but this had little effect on the findings of the survey.
Individuals in hospitals, old age homes, hotels and hostels of educational institutions were not included in the sample. Migrant labour hostels were included. In addition to those that turned up in the selected ESDs, a sample of three hostels was chosen from a national list provided by the Human Sciences Research Council and within each of these hostels a representative sample was drawn on a similar basis as described abovefor the households in ESDs.
Weighting
A self-weighting sample design should in principle eliminate the need for weighting. A number of factors intervened, however, which made it essential to use weights after all. Amongst these was violence, which prevented survey teams from conducting interviews in two clusters on the East Rand; failure to continue interviewing in a cluster until the required take had been interviewed; and systematic under-representation of whites in the sample. This last problem resulted both from systematic non-response (whites were found to be more likely to refuse to be interviewed, or to be absent than other groups) and from sampling problems themselves.
The importance of race in determining living standards in South Africa is such that the racial distribution of the population has a major bearing on measures of living standards and inequality. It was thus regarded as essential that the problems mentioned above should be overcome by applying appropriate weights to the data. The most appropriate weights to apply would usually be the average values obtained in a cluster for the missing questionnaires from that cluster in order to capture the homogeneity usually inherent in residential contiguity. However, that presented some difficulty for the two clusters in which violence prevented surveying and for those clusters in which there were only a small number of questionnaires completed. It was felt that this method would therefore not be appropriate.
Accordingly it was decided to use weights as far as possible at the level of the old provincial/homeland boundaries and race. The listing of households in each cluster combined with the sampling interval was used to determine how many households should have been interviewed. Where this deviated from the number actually interviewed, this was taken into account. The assumption was that the households left out were racially distributed in the same proportion as the actual households interviewed. When these numbers were then calculated at the provincial level, a weight could be calculated for each race group to rectify errors made in the field work. These errors typically resulted from the fact that most field work organizations involved had little experience of using anything but a weighted sample and were used to replacements that could easily be added ex post, not necessarily in the same area. When these mistakes were discovered, it was too late to go back to the field.
The sample of 360 clusters of 25 households each based on an expected household size of 5 should have yielded a resident population of 45,000. In fact, a different household size should not affect the results. In any particular cluster, the expected take of individuals would remain the same if the census population were accurate, irrespective of household size, for a smaller household size (as in the case of whites) would only have yielded more households, of whom a given proportion would have been interviewed. If in a particular cluster the census population was 472, every fourth household should have been interviewed (based on a sampling interval calculated to produce 125 persons per cluster in 1993, the expected take based on the census data of 118.1 per duster divided into the same population size). Irrespective of household size, then, one quarter of the cluster population would have been included in the survey. An average household size of 5 would have given 94 households of whom 23 would have been interviewed, i.e. 115 resident household members would have been found. If the household size were only three, on the other hand, one-quarter of the 157 households would have been 39, representing 117 household members. Only small differences from the expected take of 118 should thus arise, due to rounding. Only if the estimate of population based on the census is wrong, however, would the actual number of households deviate substantially from the expected take. In such a case, one quarter of the actual (i.e. listed or enumerated) rather than of the census population would have been included in the survey, i.e. there would have been an automatic adjustment. This gives the sample design its self-weighting character.
The census population for the survey data was estimated by applying Sadie's population growth rates to the adjusted 1991 census figures. The resultant racial and geographic distribution of the population of 40.1 million was presuming, of course, that no migration across provincial and homeland boundaries had occurred since the census. This implies that a raising factor of 891.4154 (40.1 million divided by an expected take of 45,000) should be applied to the results weighted by enumeration to obtain the population it represents. Applying the weights according to enumeration, 38.1 million people were covered by the survey, i.e. there was a 2 million under-enumeration amounting to about 5 per cent. Broken down by race, the under-enumeration was particularly large amongst whites, for whom the best census data exists, indicating that the problem did not lie so much with the census as with the survey. However, this is to be expected - a survey of this nature is better at capturing inequality and living standards than population size. Nevertheless, the margin of error in aggregate population estimates is relatively small, considering the presence of some homeless people, uncertainties about ESD boundaries in some areas and the likelihood of incomplete listings of households for various reasons. These results are therefore encouraging regarding the accuracy of the survey and also confirm that the adjusted census does not deviate substantially from population estimates obtained in a different manner.
However, the raised enumeration results deviate more from the census results where the provincial breakdown is concerned. The reason for this is not hard to find. The sample design introduced stratification only by geographic area (statistical regions) and proportion of the ESD population that was black. South African population clusters are still predominantly racially homogeneous, inter alia, because of past controls on residential patterns. It is therefore not surprising that in particular regions too few or too many clusters of a particular group were selected. In Natal, for instance, Coloureds and Indians are over represented in the data, even when weighted by enumeration, while Whites are under-represented. At the aggregate level, this should have little effect on the validity of the conclusions drawn, but it emphasizes the fact that care should be taken when drawing implications from the survey for Small populations. In small provinces (for instance, the new Northern Cape), only a small number of clusters has been included, with the result that little can be concluded about living standards there, even though these clusters are important in determining overall distribution.
As a final comment on weights, the data provided for the user contains weights to correct for the enumeration difficulties discussed above as well as census based weights. If the user of the data wishes to use these weights they are found in the data file named "weight02". The variable name for the enumeration-based weight is "rsweight" and the name for the census-based weight is "rcweight". (Do not use the "sweight" and "cweight" variables.)
The importance of race in determining living standards in South Africa is such that the racial distribution of the population has a major bearing on measures of living standards and inequality. It was thus regarded as essential that the problems mentioned above should be overcome by applying appropriate weights to the data. The most appropriate weights to apply would usually be the average values obtained in a cluster for the missing questionnaires from that cluster in order to capture the homogeneity usually inherent in residential contiguity. However, that presented some difficulty for the two clusters in which violence prevented surveying and for those clusters in which there were only a small number of questionnaires completed. It was felt that this method would therefore not be appropriate.
Accordingly it was decided to use weights as far as possible at the level of the old provincial/homeland boundaries and race. The listing of households in each cluster combined with the sampling interval was used to determine how many households should have been interviewed. Where this deviated from the number actually interviewed, this was taken into account. The assumption was that the households left out were racially distributed in the same proportion as the actual households interviewed. When these numbers were then calculated at the provincial level, a weight could be calculated for each race group to rectify errors made in the field work. These errors typically resulted from the fact that most field work organizations involved had little experience of using anything but a weighted sample and were used to replacements that could easily be added ex post, not necessarily in the same area. When these mistakes were discovered, it was too late to go back to the field.
The sample of 360 clusters of 25 households each based on an expected household size of 5 should have yielded a resident population of 45,000. In fact, a different household size should not affect the results. In any particular cluster, the expected take of individuals would remain the same if the census population were accurate, irrespective of household size, for a smaller household size (as in the case of whites) would only have yielded more households, of whom a given proportion would have been interviewed. If in a particular cluster the census population was 472, every fourth household should have been interviewed (based on a sampling interval calculated to produce 125 persons per cluster in 1993, the expected take based on the census data of 118.1 per duster divided into the same population size). Irrespective of household size, then, one quarter of the cluster population would have been included in the survey. An average household size of 5 would have given 94 households of whom 23 would have been interviewed, i.e. 115 resident household members would have been found. If the household size were only three, on the other hand, one-quarter of the 157 households would have been 39, representing 117 household members. Only small differences from the expected take of 118 should thus arise, due to rounding. Only if the estimate of population based on the census is wrong, however, would the actual number of households deviate substantially from the expected take. In such a case, one quarter of the actual (i.e. listed or enumerated) rather than of the census population would have been included in the survey, i.e. there would have been an automatic adjustment. This gives the sample design its self-weighting character.
The census population for the survey data was estimated by applying Sadie's population growth rates to the adjusted 1991 census figures. The resultant racial and geographic distribution of the population of 40.1 million was presuming, of course, that no migration across provincial and homeland boundaries had occurred since the census. This implies that a raising factor of 891.4154 (40.1 million divided by an expected take of 45,000) should be applied to the results weighted by enumeration to obtain the population it represents. Applying the weights according to enumeration, 38.1 million people were covered by the survey, i.e. there was a 2 million under-enumeration amounting to about 5 per cent. Broken down by race, the under-enumeration was particularly large amongst whites, for whom the best census data exists, indicating that the problem did not lie so much with the census as with the survey. However, this is to be expected - a survey of this nature is better at capturing inequality and living standards than population size. Nevertheless, the margin of error in aggregate population estimates is relatively small, considering the presence of some homeless people, uncertainties about ESD boundaries in some areas and the likelihood of incomplete listings of households for various reasons. These results are therefore encouraging regarding the accuracy of the survey and also confirm that the adjusted census does not deviate substantially from population estimates obtained in a different manner.
However, the raised enumeration results deviate more from the census results where the provincial breakdown is concerned. The reason for this is not hard to find. The sample design introduced stratification only by geographic area (statistical regions) and proportion of the ESD population that was black. South African population clusters are still predominantly racially homogeneous, inter alia, because of past controls on residential patterns. It is therefore not surprising that in particular regions too few or too many clusters of a particular group were selected. In Natal, for instance, Coloureds and Indians are over represented in the data, even when weighted by enumeration, while Whites are under-represented. At the aggregate level, this should have little effect on the validity of the conclusions drawn, but it emphasizes the fact that care should be taken when drawing implications from the survey for Small populations. In small provinces (for instance, the new Northern Cape), only a small number of clusters has been included, with the result that little can be concluded about living standards there, even though these clusters are important in determining overall distribution.
As a final comment on weights, the data provided for the user contains weights to correct for the enumeration difficulties discussed above as well as census based weights. If the user of the data wishes to use these weights they are found in the data file named "weight02". The variable name for the enumeration-based weight is "rsweight" and the name for the census-based weight is "rcweight". (Do not use the "sweight" and "cweight" variables.)
Data Collection
Dates of Data Collection
| Start | End |
|---|---|
| 1993-01-01 | 1994-04-01 |
Data Collection Mode
Face-to-face [f2f]
Data Collection Notes
The Bureau of Market Research was responsible for data collection in the rural and the predominantly non-African urban areas of the Transvaal excluding the homelands. MarkData conducted surveys in the Orange Free State, Qwa-Qwa, Bophuthatswana and Lebowa. Social Surveys covered the African townships in the PWW as well as Venda, Gazankulu and Kwandebele. Data Research Africa from Natal was responsible for the fieldwork in Kwazulu and Kangwane. The rest of Natal and the Ciskei was covered by the HSRC in Durban. The HSRC in Cape Town covered the Northern, Western and Eastern Cape. Finally, a team from the University of the Transkei covered theTranskei.
Data Collectors
| Name |
|---|
| Bureau of Market Research |
| Data Research Africa |
| Human Sciences Research Council |
| MarkData |
Questionnaires
Questionnaires
The main instrument used in the survey was a comprehensive household questionnaire. This questionnaire covered a wide range of topics but was not intended to provide exhaustive coverage of any single subject. In other words, it was an integrated questionnaire aimed at capturing different aspects of living standards. The topics covered included demographics, household services, household expenditure, educational status and expenditure, remittances and marital maintenance, land access and use, employment and income, health status and expenditure and anthropometry (children under the age of six were weighed and their heights measured). This questionnaire was available to households in two languages, namely English and Afrikaans. In addition, interviewers had in their possession a translation in the dominant African language/s of the region.
In addition to the detailed household questionnaire, a community questionnaire was administered in each cluster of the sample. The purpose of this questionnaire was to elicit information on the facilities available to the community in each cluster. Questions related primarily to the provision of education, health and recreational facilities. Furthermore there was a detailed section for the prices of a range of commodities from two retail sources in or near the cluster: a formal source such as a supermarket and a less formal one such as the "corner cafe" or a "spaza". The purpose of this latter section was to obtain a measure of regional price variation both by region and by retail source. These prices were obtained by the interviewer. For the questions relating to the provision of facilities, respondents were "prominent" members of the community such as school principals, priests and chiefs.
A literacy assessment module (LAM) was administered to two respondents in each household, (a household member 13-18 years old and a one between 18 and 50) to assess literacy levels.
In addition to the detailed household questionnaire, a community questionnaire was administered in each cluster of the sample. The purpose of this questionnaire was to elicit information on the facilities available to the community in each cluster. Questions related primarily to the provision of education, health and recreational facilities. Furthermore there was a detailed section for the prices of a range of commodities from two retail sources in or near the cluster: a formal source such as a supermarket and a less formal one such as the "corner cafe" or a "spaza". The purpose of this latter section was to obtain a measure of regional price variation both by region and by retail source. These prices were obtained by the interviewer. For the questions relating to the provision of facilities, respondents were "prominent" members of the community such as school principals, priests and chiefs.
A literacy assessment module (LAM) was administered to two respondents in each household, (a household member 13-18 years old and a one between 18 and 50) to assess literacy levels.
Data Processing
Data Editing
Completed questionnaires were sent to SALDRU where data entry management and cleaning were centralized. All the questionnaires were checked when received. Where information was incomplete or appeared contradictory, the questionnaire was sent back to the relevant survey organization. As soon as the data was available, it was captured using local development platform ADE. This was completed in February 1994. Following this, a series of exploratory programs were written to highlight inconsistencies and outlier. For example, all person level files were linked together to ensure that the same person code reported in different sections of the questionnaire corresponded to the same person. The error reports from these programs were compared to the questionnaires and the necessary alterations made. This was a lengthy process, as several files were checked more than once, and completed at the beginning of August 1994. In some cases questionnaires would contain missing values, or comments that the respondent did not know, or refused to answer a question.
These responses are coded in the data files with the following values:
-1 : The data was not available on the questionnaire or form
-2 : The field is not applicable
-3 : Respondent refused to answer
-4 : Respondent did not know answer to question
These responses are coded in the data files with the following values:
-1 : The data was not available on the questionnaire or form
-2 : The field is not applicable
-3 : Respondent refused to answer
-4 : Respondent did not know answer to question
Data Appraisal
Data Appraisal
Removal of fabricated data:
During fieldwork for the KwaZulu-Natal Income Dynamics Study in 2001 the data from all 39 household interviews in clusters 217 and 218 was shown to be fabricated and the data for these clusters was removed from the PSLSD data files.
Other quality issues:
Cluster 206 is now coded as urban as this was incorrectly coded as rural in the first release of the data.
Weights calculated by the World Bank and provided with the original data are not updated to reflect changes to release one of the data files
During fieldwork for the KwaZulu-Natal Income Dynamics Study in 2001 the data from all 39 household interviews in clusters 217 and 218 was shown to be fabricated and the data for these clusters was removed from the PSLSD data files.
Other quality issues:
Cluster 206 is now coded as urban as this was incorrectly coded as rural in the first release of the data.
Weights calculated by the World Bank and provided with the original data are not updated to reflect changes to release one of the data files
Access policy
Contacts
| Name | URL | |
|---|---|---|
| DataFirst Helpdesk | support@data1st.org | Link |
Access conditions
Public use files, available to all
Citation requirements
Southern Africa Labour and Development Research Unit. Project for Statistics on Living Standards and Development 1993 [dataset]. Version 2. Cape Town: SALDRU [producer], 1995. Cape Town: DataFirst [distributor], 2013. DOI: https://doi.org/10.25828/7247-6w78
Access authority
| Name | Affiliation | URL | |
|---|---|---|---|
| DataFirst | University of Cape Town | support@data1st.org | Link |
Metadata production
Producers
| Name | Affiliation | Role |
|---|---|---|
| DataFirst | University of Cape Town | Metadata Producer |
Date of Metadata Production
2020-07-19
DDI Document version
Version 9