RWANDA - Demographic Health Survey 2005

The sample for the RDHS-III used a stratified, two-stage cluster selection. The primary sampling unit is the EA as defined in the 2002 census. Each province is separated into urban and rural areas to create the sampling strata and the sample was drawn independently in each stratum. There were therefore 23 strata in total, because the City of Kigali comprised one strata, as it had no rural component. In the first stage, 432 EAs were selected with probability proportional to size, the size being the number of households in the EA.

An updating operation listed all the households in each selected EA and this list was used to select the households for the second stage. Before this updating of the households, the larger EAs were divided into segments, of which only one was selected for the survey. In the second stage, in each EA selected in the first stage, a fixed number of households (20 households in each urban cluster, 24 households in each rural
cluster) were selected using a systematic selection based on the new list of households created during the household listing. In all, 10,644 households were selected for the women’s interview.
All members of each selected household were listed in the Household Questionnaire. Every woman age 15-49 in the household was interviewed using the Women’s Questionnaire. Half of the households
selected for the women’s interview were also selected for the men’s interview. In this subsample of households all men age 15-59 were interviewed. All men age 15-59 and all women age 15-49 in this subsample of households were also asked to consent to an HIV test.

In all, 462 EAs were selected (111 in urban areas and 351 in rural areas) and 10,644 households were selected (2,220 in urban areas and 8,424 in rural areas.)


The sample for the RDHS-III survey covered the population residing in ordinary households across the country. A national sample of 10,644 households was selected. The sample was first stratified to provide adequate representation of urban and rural areas as well as all 12 provinces including the “City of Kigali,” the nation’s capital. Decentralization reforms were introduced after this sample was drawn, resulting in new geographically-based divisions that regroup the former districts into five new provinces. However, the sample used posed no obstacle to adequate representation of the new provinces and the data in this report present key indicators corresponding to the five recently created provinces (South, West, North, East, and the City of Kigali).

The survey used a two-stage sample design. The first stage involved selecting primary sampling units (PSUs) based on the list of enumeration areas covered in 2002 General Population and Housing Census (RGPH) prepared by the National Census Bureau. These enumeration areas provided the master frame for the drawing of 462 clusters (351 rural and 111 urban), selected with a representative probability proportional to their size. A strictly proportional sample allocation would have resulted in a very low number of urban households in certain provinces such as Umutara. It was therefore necessary to slightly over-sample urban areas in order to survey a sufficient number of households to produce reliable estimates for urban areas.

The second stage involved selecting a sample of households in these enumeration areas. In order to adequately guarantee the accuracy of the indicators, it was necessary to control the total size of the households drawn by setting the number of households to be surveyed at 20 in urban clusters and 24 in rural clusters. Because of the nonproportional distribution of the sample among the different strata and the fact that the number of households was set for each cluster, weighting was used to ensure the validity of the sample at both national and regional levels.
All women age 15-49 years who were either usual residents of the selected household or visitors present in the household on the night before the survey were eligible to be interviewed (approximately 11,500 women). In addition, in a subsample of every second household selected for the survey, a sample of 5,000 men age 15-59 years was selected to be interviewed. In this subsample, all men and women eligible for the individual survey were also eligible for the HIV test. In addition, in this subsample of households, all women eligible for the survey and all children under the age of five were eligible for the anemia test. Finally, in this same subsample of households, all women eligible for the survey and all children under the age of five were eligible for height and weight measurements to determine their
nutritional status.


All of the 462 clusters selected for the sample were able to be surveyed for the RDHS-III. A total of 10,644 households were selected, of which 10,307 households were identified and occupied at the time of the survey. Among these households, 10,272 completed the Household Questionnaire, yielding a response rate of nearly 100 percent (Table 1.1).
In the 10,272 households surveyed, 11,539 women age 15-49 years were identified as being eligible for the individual interview; interviews were completed with 11,321 of these women, yielding a response rate of 98 percent. Male interviews were conducted in every second household. A total of 4,959 men age 15-59 years were identified in the subsample of households. Of these 4,959 men, 4,820 completed the individual interviews, yielding a response rate of 97 percent.
The response rates were slightly higher in rural areas for both men and women.