Survey ID Number
Knowledge, Attitudes and Practices Survey 1994
Sample Design and Implementation
The sample for the 1994 TKAPS was national in scope, with the exclusion of only Zanzibar. In order to maximize efficiency and enhance the measurement of trends, sample points for the TKAPS were selected from those which had been chosen for the 1991/92 TDHS. The TDHS sample was a three-stage design, consisting of wards/branches at the first stage, census enumeration areas (EAs) at the second stage, and households at the third stage. Of the total of 357 EAs used in the TDHS, 203-57 urban and 146 rural were selected for the TKAPS.
The ratio of the sub-selection of TKAPS sample points from TDHS sample points was not uniform across the country. Although the TKAPS sample size was too small to obtain separate estimates for each of Tanzania's 20 mainland regions, estimates of most variables were obtained for groups of regions.
Regions were grouped into three geographically contiguous zones, as follows:
Coastal Zone: Tanga, Coast, Dares Salaam, Lindi, Mtwara, Ruvuma, and Morogoro;
Central Zone: Arusha, Kilimanjaro, Singida, Dodoma, Iringa, and Mbeya; and
Western Zone: Kagera, Mwanza, Mara, Shinyanga, Tabora, Rukwa, and Kigoma.
Based on TDHS results, the three zones correspond to medium, high and low contraceptive prevalence rates, respectively. Moreover, to meet the secondary objective of the TKAPS, namely to provide data on AIDS knowledge and sexual behaviour, it was decided to provide separate estimates of certain variables for Mwanza, Dodoma, lringa and Dares Salaam regions. Thus, the TKAPS sample over-selected EAs from the TDHS for these four regions relative to the other regions, such that there would be an expected minimum of 350 women interviewed in each. Whereas in the other regions, one-half of the selected households were designated for the men's survey, in these four regions, all households were selected for both the women's and men's surveys. Due to the oversampling of households in certain regions, the TKAPS sample is not self-weighting at the national level; consequently, the data presented in this report have been weighted to compensate.
After the selection of the TKAPS sample points, field staff from the Bureau of Statistics conducted a household listing operation in May and June 1994. A systematic sample of households was then selected from these lists, with an average "take" of 22 households in both urban and rural clusters for an expected total of about 4,466 households selected. As already mentioned, every second household was identified as selected for the male survey, meaning that, in addition to interviewing all women age 15-49, interviewers were also to interview all men age 15-59; in Dodoma, Iringa and Dares Salaam regions, all selected households were eligible for the male survey. 4 It was expected that the sample would yield interviews with approximately 4,500 women age 15-49 and over 2,500 men age 15-59.