The sample of the 2006 UDHS was designed to allow separate estimates at the national level and for urban and rural areas of the country. The sample design also allowed for specific indicators, such as contraceptive use, to be calculated for each of nine sub-national regions. Portions of the northern region were oversampled in order to provide estimates for two special areas of interest: Karamoja and internally displaced persons (IDP) camps. At the time of the survey there were 56 districts. This number later increased to 80. The following shows the 80 districts divided into the regional sampling strata:
- Central 1: Kalangala, Masaka, Mpigi, Rakai, Lyantonde, Sembabule, and Wakiso
- Central 2: Kayunga, Kiboga, Luwero, Nakaseke, Mubende, Mityana, Mukono, and Nakasongola
- Kampala: Kampala
- East Central: Bugiri, Busia, Iganga, Namutumba, Jinja, Kamuli, Kaliro, and Mayuge
- Eastern: Kaberamaido, Kapchorwa, Bukwa, Katakwi, Amuria, Kumi, Bukedea, Mbale, Bududa, Manafwa, Pallisa, Budaka, Sironko, Soroti, Tororo, and Butaleja
- North: Apac, Oyam, Gulu, Amuru, Kitgum, Lira, Amolatar, Dokolo, Pader, Kotido, Abim, Kaabong, Moroto, and Nakapiripirit (Estimates for this region include both settled and IDP populations.) Karamoja area: Kotido, Abim, Kaabong, Moroto, and Nakapiripirit IDP: IDP camps in Apac, Oyam, Gulu, Amuru, Kitgum, Lira, Amolatar, Dokolo and Pader districts
- West Nile: Adjumani, Arua, Koboko, Nyadri, Nebbi, and Yumbe
- Western: Bundibugyo, Hoima, Kabarole, Kamwenge, Kasese, Kibaale, Kyenjojo, Masindi, and Buliisa
- Southwest: Bushenyi, Kabale, Kanungu, Kisoro, Mbarara, Ibanda, Isingiro, Kiruhura, Ntungamo, and Rukungiri
A representative probability sample of 9,864 households was selected for the 2006 UDHS survey. The sample was selected in two stages. In the first stage, 321 clusters were selected from among a list of clusters sampled in the 2005-2006 Uganda National Household Survey (UBOS, 2006c). This matching of samples was conducted in order to allow for linking of 2006 UDHS health indicators to poverty data from the 2005-2006 UNHS. The clusters from the Uganda National Household Survey were in turn selected from the 2002 Census sample frame. For the UDHS 2006, an additional 17 clusters were selected from the 2002 Census frame in Karamoja in order to increase the sample size to allow for reporting of Karamojaspecific estimates in the UDHS. Finally, 30 IDP camps were selected from a list of camps compiled by the United Nations Office for the Coordination of Human Affairs (UN OCHA) as of July 2005, completing a total of 368 primary sampling units. Figure 1.1 shows the geographical distribution of the 368 clusters visited in the 2006 UDHS.
In the second stage, households in each cluster were selected based on a complete listing of households. In the 321 clusters that were included in the UNHS sample, the lists of households used were those generated during the UNHS listing operations April-August 2005. The UNHS sampled ten households per cluster. All ten were purposively included in the UDHS sample. An additional 15 to 20 households were randomly selected in each cluster. The 17 additional clusters in Karamoja were listed, and 27 households were selected in each cluster. The selected IDP camps were divided into segments because of their large size, and one segment selected in each camp. Then a listing operation was carried out in the selected segment, and 30 households were selected in each camp from the segment of the map that was listed. All women age 15-49 who were either permanent residents of the households in the 2006 UDHS sample or visitors present in the household on the night before the survey were eligible to be interviewed. In addition, in a sub-sample of one-third of all the households selected for the survey, all men age 15-54 were eligible to be interviewed if they were either permanent residents or visitors present in the household on the night before the survey. Indicators such as total fertility rate, childhood mortality rates, and the maternal mortality ratio require a larger sample size than other indicators. These indicators are all calculated from the data provided by female respondents only. For this reason, the number of male respondents required in the sample to obtain acceptable precision in estimates of desired indicators is lower than the number of female respondents.
Biomarkers collected in the UDHS included height and weight measurements for children under 6 years, women age 15-49, and men age 15-54; anaemia testing in children age 6 to 59 months old, women age 15-49, and men age 15-54; and dried blood spot collection for vitamin A testing in children age 6 to 59 months old and women age 15 to 49 years. All of these biomarkers were measured only in those households selected for the male interview-that is, one in three households.
COMPARABILITY OF THE 2006 UDHS SAMPLE WITH SAMPLES FROM PREVIOUS UDHS SURVEYS
The 2006 UDHS is the first UDHS to include the entire country in the sample. In previous surveys, it was necessary to exclude groups of districts because of security problems. In the 2000-2001 UDHS, areas making up the current districts of Amuru, Bundibugyo, Gulu, Kasese, Kitgum, and Pader were excluded from the sample. According to the 2002 Census, these areas comprise around 7 percent of the population of Uganda (UBOS 2006a). The 1995 UDHS excluded Kitgum and Pader, while the 19881989 UDHS excluded most of the Northern region.
To show trends using comparable data, the 2006 UDHS data were run without the districts that were excluded in previous surveys. For some key indicators, the report presents two estimates from the 2006 data: one covering the entire country, and a second covering the geographic area surveyed in the 2000-2001 UDHS. Differences between these two estimates are small, seldom exceeding one or two percentage points.
Because it was not possible to run every indicator twice, the report includes many comparisons between the 2000-2001 and 2006 surveys in which the 2006 data have not been adjusted. The report states explicitly when the 2006 data presented are adjusted; otherwise, the data are unadjusted. Comparisons that include unadjusted 2006 data should be interpreted with caution.