Documentation for the South Africa Integrated Household Survey

In order to use the data from the South Africa Integrated Household Survey, it is necessary to download supporting documentation, which includes the household questionnaire, community questionnaire, a list of variable labels, and an overview of the Survey. Also available for downloading are a file of summary statistics for the data files, a table which matches data file names to either a section in the questionnaire or to a brief description of the file, and a set of program files for the constructed data.

The materials which have little formatting are available only as ASCII files (or compressed ASCII). The more heavily formatted documents are available in both Postscript and PDF format. If you do not know whether you prefer PDF to PostScript, return to the page on Document Formats for a description of these formats.

I. Household and Community Questionnaires

• The most important documents for use of the data are the questionnaires. The questionnaires contain, with limited exception, all response codes. The questionnaires also contains all the skip codes, which are essential for properly interpreting blank values.

II. Overview, Variable Labels and Summary Statistics

The Overview of the Survey contains brief descriptions of the origins, questionnaires, sample design, field work, data management, and weighting of the survey. This document also contains some important definitions and codes used in the Survey, such as the definition of a household member and the special codes for missing answers.

The Codebook of Variable Labels contains four items of information for each variable in the data files.

• Variable Name--Every variable has a name which can be up to eight characters in length.
• Variable Label--Variable labels are up to 30 characters in length. They are more descriptive, but do not exist for all variables.
• Label Name--Some variables are assigned labels to help interpret response codes. For instance, the variable named gender is assigned a label which matches the numeric responses to male or female.
• Label Codes or Values--These are the code values for the label names.

The file of Summary Statistics contains summary data on each of the variables in the data files. If a variable takes on nine or fewer unique values, then a tabulation of the values is presented. If more than nine, then information on the distribution of values is presented (e.g. minimum, median, maximum, etc.).

III. File Name Descriptions and Programs for Constructed Data

• Datamap.txt--An ASCII file which matches the data file names to sections in the questionnaires or to a brief description of its contents. (This information is also listed in the section on Details of the Group files from the data page.)
• Readme.txt--An ASCII file which describes the contents of Za94pgm.zip, and matches program file names to constructed data files.
• Za94pgm.zip-- This compressed file archives 42 program files (written for the Stata software) which were used to create the constructed data files. Our assumption is that most users will prefer to construct their own estimates of household income and expenditure from the raw data and not rely on the constructed data available at this site. These users will not want to download this file, the readme.txt file, nor the Group D data files. For those users who do choose to use the constructed data (Group D) available at this site, these program files (along with the readme.pgm file) will allow the user to determine what assumptions were made in the construction of the data files.

Revised 06/27/07

Back to the South Africa Integrated Household Survey Page