The primary objective of the sample design for the Sierra Leone MICS3 was to produce statistically reliable estimates of most indicators at the national level, for urban and rural areas, and at the province level. The design of the sample allows the estimation of indicators at district level - however, such estimates are likely to be very imprecise, since the sample size was not determined to enable district-level estimates.
A multi-stage, stratified cluster sampling approach was used to select the survey sample. The 2004 census frame was used for the selection of clusters. Census enumeration areas (EAs) were defined as primary sampling units (PSUs), and were selected in each district using pps sampling procedures. The stages of the sampling approach are described below.
Description of sampling approach for Sierra Leone MICS3
Stage 1: Selection of EAs
The list of all EAs in Sierra Leone was ordered using implicit stratification according to the following variables: province; district; chiefdom; and, population size. 320 EAs were then selected using stratified systematic sampling, thus yielding a self-weighting sample. Selected EAs were then classified as rural (population of the settlement were the EA is located is < 2,000) or urban (population of the settlement where the EA is located is = 2,000).
Stage 2: Selection of households
A list of all households in each of the 320 selected EAs as enumerated during the 2004 census was prepared using data contained in the 2004 Population and Housing Census registers.
A team of listers/verifiers visited each of the 320 EAs to update the household lists in the EA by verifying each of the households on the list and adding any new households that have been formed in order to control for out-movers, non-existent households, and/or new households. This task produced an updated listing of households in all selected EAs.
The newly updated listing of households in each EA was then sequentially numbered from 1 to n (the total number of households in the enumeration area of interest) at the Statistics Sierra Leone Office. Sampling experts then selected 25 households in each EA using systematic selection procedures.
(Information extracted from final report: Statistics Sierra Leone and UNICEF-Sierra Leone 2007. Sierra Leone Multiple Indicator Cluster Survey 2005, Final Report. Freetown, Sierra Leone: Statistics Sierra Leone and UNICEF-Sierra Leone.)
Of the 8,000 households selected for the sample, only 7,125 were found to be occupied. Of the 7,125 occupied households, 7,078 were successfully interviewed for a household response rate of 99.3 per cent. In the interviewed households, 9,257 eligible women (aged 15-49) were identified. Of these, 7,654 were successfully interviewed, yielding a response rate of 82.7 per cent. The response rate for the Questionnaire for Children Under Five was 88.9 per cent; mothers/caretakers of 5,246 children under five were successfully interviewed, from among 5,904 children under five who were identified in the interviewed households. Overall response rates of 82.1 percent and 88.3 percent are calculated for the women's and under-5's interviews, respectively
Sample weights were calculated for each of the datafiles.
Sample weights for the household data were computed as the inverse of the probability of selection of the household, computed at the sampling domain level (urban/rural within each region). The household weights were adjusted for non-response at the domain level, and were then normalized by a constant factor so that the total weighted number of households equals the total unweighted number of households. The household weight variable is called HHWEIGHT and is used with the HH data and the HL data.
Sample weights for the women's data used the un-normalized household weights, adjusted for non-response for the women's questionnaire, and were then normalized by a constant factor so that the total weighted number of women's cases equals the total unweighted number of women's cases.
Sample weights for the children's data followed the same approach as the women's and used the un-normalized household weights, adjusted for non-response for the children's questionnaire, and were then normalized by a constant factor so that the total weighted number of children's cases equals the total unweighted number of children's cases.