Sampling

The principal objective of the sample design was to provide current and reliable estimates on a set of indicators covering the four major areas of the World Fit for Children declaration, including promoting healthy lives; providing quality education; protecting against abuse, exploitation and violence; and combating HIV/AIDS.  The population covered by the 2005-2006 MICS is defined as the universe of all women aged 15-49 and all children aged under 5.  A sample of households was selected and all women aged 15-49 identified as usual residents of these households were interviewed.  In addition, the mother or the caretaker of all children aged under 5 who were usual residents of the household were also interviewed about the child.The 2005 MICS collected data from a nationally representative sample of households, women and children.  The primary focus of the 2005 MICS was to prodvide estimates of key population and health, education, child protection and HIV related indicators for the country as a whole, and for urban and rural areas separately.  In additon, the sample was designed to provide estimates for each of the 5 regions for key indicators.  Tajikistan is devided into 5 regions: Dushanbe (the capital), Direct Rule Districts (DRD), Sogd, Khatlon and Gorno Badakhshan (GBAO).  Each of such division, for last census purposes, was further subdivided in convenient areas called “census section”, “instructor's sector” and “enumeration sector”(ESs). Each ES is either totally urban or rural.  The list of ESs has census information on population information for each ES, and the ESs list are grouped by administrative unit. The size of the rural ES was 342 persons and the size of the urban ES was 378 persons. There were established 17923 ES in Tajikistan. The available demarcated cartographic material for each ES in localities with more than 5000 population made last census material as the most appropriate sample frame for the MICS2005. 

The primary sampling unit (PSU) - or cluster - for the MICS2005 is defined on basis of ES from the 2000 census frame, as having one (or more) ES per PSU.  The number of clusters selected in each region was not allocated proportional to the size of the region, but approximately equal number of clusters was allocated to each region due to the need to present estimates for each of the 5 regions.  A total of 290 clasters and 6,960 households were selected.  The primary sampling unit variable is the cluster (HH1).  The sampling stratas can be identified using a combination of the region (HH7) and area (HH6) variables, e.g. stratum = HH7*10 + HH6.

A two-stage, stratified cluster sampling approach was used for the selection of the survey sample. The sampling domains are defined as urban and rural areas of each of the regions of Tajikistan.  In the first stage, 290 PSU were selected in each stratum with equal probabilities. Within each cluster a fixed number of 24 households were selected.  Updating of the list of households in each cluster was done prior to selection of the households.  Since the distribution of clusters between sampling domains was not proportional to the census distribution of population and, consequently neither was the final household distribution, the sample is not self-weighting. For reporting national level results, sample weights are used. 

Following standard MICS data collection rules, if a household was actually more than one household when visited, then a) if the selected household contained two households, both were interviewed, or b) if the selected household contained 3 or more households, then only the household of the person named as the head was interviewd.

No replacement of households was permitted in case of non-response or non-contactable households.  Adjustments were made to the sampling weights to correct for non-response, according to MICS standard procedures.

The sampling procedures are more fully described in the sampling design document and the sampling appendix of the final report.

No major deviations from the original sample design were made.  All sample enumeration areas were accessed and successfully interviewed with good response rates.

Of the 6.968 households selected for the sample, 6.961 were found to be occupied. Of these, 6.684 were successfully interviewed for a household response rate of 96 percent. In the interviewed households, 10.626 women (age 15-49) were identified. Of these, 10.243 were successfully interviewed, yielding a response rate of 96 percent. In addition, 4.370 children under the age of five were listed in the household questionnaire. Questionnaires were completed for 4.273 of these children, which corresponds to a response rate of 98 percent. Overall response rates of 93 and 94 percent are calculated for the women's and under-5's interviews respectively.

The household response rates are slightly higher in rural than in urban areas, 97 compared to 94 percent. The response rates in Dushanbe are a little lower than in others regions, due to the busy lifestyle of the respondents living in the capital city. The lowest women's response rate is noted in GBAO region, 88 percent. Differences between number of sampled and occupied households almost does not exist, thanks to the previously updated household listings.

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.