Sampling Procedure
The National Department of Transport (NDOT) identified 342 national Transport Analysis (TA) zones in consultation with provincial transport departments. From these TA zones, a sample of 50 000 households had to be interviewed in order to collect information about travel patterns and habits. Stats SA overlaid the analysis zones with the Census 2001 Enumerator Areas (EAs) to facilitate the sampling.
The sample design is outlined in the steps which follow:
• The 342 TA zones were overlaid with the Census 2001 Enumeration Areas. Each of the analysis zones will be treated as a reporting domain. Where possible, the zones were further stratified by geography type such as urban formal, urban informal, rural formal and tribal. Each zone was allocated a number of EA's proportional to its size using the enumerated "dwelling unit count" on the Stats SA books as a measure of size.
• The number of EAs allocated to each TA zone was determined using the number of dwelling units in the EAs as a measure of size. The power allocation method was used to determine the number of primary sampling units (PSUs) to be allocated per zone.
• Each PSU number consist of 12 digits, the first 8 representing the EA number and the last 4 the number of the PSU. Ten dwellings had to be enumerated in each of the 5 000 PSUs to arrive at the required sample of 50 000 households. This worked out to an average of 13.3 EAs per zone, based on 10 dwelling units per EA, assuming an equal allocation of EAs to the zone.
• Additional EAs were pooled where the selection criterion of a minimum number of 80 dwelling units per EA was not met, such as in farming areas in the Northern Cape. In this case, the extra EA was taken from the same township, suburb and area as the original EA.
• Once the dwelling units in each PSU were verified and the dwelling unit total was established, a sample of 10 dwelling units was selected from each PSU. Systematic sampling of EAs was used in all cases and equal probability systematic sampling was used to select the dwelling units that were enumerated in each of the EAs that was drawn. This allows for approximate selfweighting of the selected planning units within each of the zones.
• In each household, one person, aged 15 years or older was selected for an interview about their attitudes to transport services. Kish's grid for selection of individuals aged 15 years and older, was used to determine who would be required to complete the attitudinal section of the questionnaire.
