South Sudan - Household Health Survey 2006
Reference ID | SSD-SSCCSE-SHHS-2006-v01 |
Year | 2006 |
Country | South Sudan |
Producer(s) | National Bureau of Statistics - Government of South Sudan |
Sponsor(s) | United Nations Children's Fund - UNICEF - Financial and technical support The World Food Programme - WFP - Financial support and logistics United States Agency for International Development - USAID - Financial and technical support World H |
Created on
May 23, 2016
Last modified
Jul 20, 2016
Page views
20412
Sampling
Sampling Procedure
The sample selection methodology for the 2006 SHHS was based on a stratified multi-stage sample design. The steps involved in the sample selection included the following:
Selection of sample Primary Sampling Units (PSU). For the first stage of selection of the sample for the SHHS was a frame of PSUs which covered as much of the population as possible. The PSU was defined as the smallest area or administrative unit which could be identified in the field with commonly recognized boundaries.
Any areas that could not be included in the survey because of problems of security or accessibility were excluded from the frame before the first stage selection of sample PSUs. The villages or quarters constituted the PSU for the SHHS. Therefore, the list of villages was used as the most effective sampling frame of PSUs for the first stage of sampling.
For some states the list of villages appeared to be fairly complete, and population estimates Probability Proportional to Size were available for all villages, this frame was used for the first stage selection of villages with PPS. In the case of these states, at the first sampling stage, the sample PSUs within each state were selected with (PPS) for each stratum, where the measure of size was based on the estimated total population. An Excel file was used for selecting the sample of 40 sample villages in each state for the SHHS, based on the allocation of 40 sample villages per state. The Excel file included a separate spreadsheet for each state, showing the ordered frame of villages with the corresponding information on population estimates.
When most of the villages in the state had population estimates but figures were missing for some villages, an average measure of size was imputed for these villages; in this way such villages had an equal probability of selection in the frame. In other words, the sampling frame of villages was compiled separately for each state based on the best available sources. When the estimated population was not available, an average measure of size was imputed; in this way such villages had an equal probability of selection in the frame. In the case of a few states, where the sampling frame did not include population estimates, it was decided to select the sample villages with equal probability. There were four states in Southern Sudan (Upper Nile, Jonglei, Unity and Lakes) which did not have population measures in the frame. In these four states the sample villages were selected systematically with equal probability. The same type of sample selection spreadsheet was used for these states, but each village was assigned a measure of size of 1. In cases where a selected village could not be found in the field or could not be reached because of security or access problems, it was replaced by a neighboring village in the sampling frame. All 40 villages within the sampled segments in each state were fully covered with the exception of only 12 segments in two states in Southern Sudan (7 segments in Upper Nile and 5 in Western Bahr Al Gazal states) that had to be substituted due to insecurity, influencing accessibility during the fieldwork period.
Segmenting of large sample villages: Some of the villages in the frame had 500 or more households. In the case of a sample village with a large number of households (for example, greater than 200), the village was subdivided into smaller segments of similar size (with about 80 to 120 households each) with clear defined boundaries in order to facilitate the listing process and avoid coverage problems. Following this, one sample segment was selected at random with equal probability for the listing of households at the second sampling stage.
Listing of households in sample villages or segments: A listing of the households was undertaken in each sample segment prior to the SHHS data collection in order to enumerate all housing units and households within the boundaries of each sample village or segment. At the last sampling stage the households were selected systematically with a random start from this household listing for each sample segment. The supervisor was responsible for verifying the boundaries of the sample village or segment in order to ensure good coverage of the sample households.
Selection of sample households within sample village or segment: At the last sampling stage, a sample of 25 households was selected systematically for enumeration with a random start from the household listing for each sample village or segment. If a village had less than 25 households, all of them were selected. Once the listing was completed, the supervisor referred to the sample selection table to find the row.
Response Rate
The 24,527 households selected for the sample, 24,507 households were found to be occupied. Of these, 24,046 households were successfully interviewed for a household response rate of 98.1 percent. In the interviewed households, 32,599 women (age 15-49 years) were identified. Of these, 26,923 were successfully interviewed, yielding a response rate of 82.6 percent. In addition, 22,512 children under age five were listed in the household questionnaire. Questionnaires were completed for 19,870 of these children, which corresponds to a response rate of 88.3 percent. Overall response rates of 81.0 percent and 86.6 percent are calculated for the women's and under-5 children's interviews respectively.
Weighting
Appropriate weights were applied to the sample data based on the probabilities of selection. Measures of sampling variability for key survey estimates were also calculated. The Sudan Household Health 2006's sample was not self-weighted. Essentially, by allocating equal numbers of household to each of the regions, different sampling fractions were used in each region since the size of the regions varied. For this reason, sample weights were calculated and these were used in subsequent analyses of the data.
HHweight = Household
WMweight = Women
CHweight = Children under five