Multiple Indicator Cluster Survey 2000

The survey used multistage sampling method. A sampling frame was provided by the State Department for Statistics listing all regions, districts, strata, census enumeration units and census areas. The size of the smallest unit, census area, is 20-60 households (HHs), the following unit by size is the census enumeration unit incorporating 4-5 census areas with the size from 67 to 900 households; strata is the combination of 3-5 census enumeration areas and so on for the remaining units.

At the first stage census enumeration units (which were the primary sampling unit or PSU) were selected from the sampling frame, from each of the survey region using probability proportional to household size (PPS) method. At the second stage 35 housholds (40 in Tbilisi) were selected in each PSU using systematic sampling method, selecting every n-th HH depending on the size of the PSU. The sample size calculations were performed based on immunizations for children 15-26 months as it required the largest sample size. The details of the sample size calculations can be found in Appendix C. The decision was to sample 35 households in each PSU (40 in Tbilisi) and that a total of 474 clusters would be included in the survey.

The collection of data at the cluster level was divided into two stages. During the first stage 150 SDS interviewers obtained lists of households residing in 474 PSUs and using systematic sampling method (every n-th HH), selected 35 HHs (40 for Tbilisi) in each. For example, if the PSU size contained 350 HHs, every 10th HH was selected for the mini-interview. At this stage the SDS interviewers collected information only on HH composition for the selected HHs.

At the second stage, which took place during 7-30 June, 40 interviewers and 40 assistants from the National Center for Disease Control (NCDC) visited all the HHs having underfives using the HH composition information generated during the first stage. Some of the HHs with eligible children for the education/child labour module (5-16 years of age) were sampled. It was assumed that each HH had only one child 5-16 years of age and thus 150 HHs would need to be visited per region, with the exception of Tbilisi where 160 HHs were to be sampled because of a higher non-response rate. For the entire country, this resulted in the target of visiting 150 HH with children 5-16 years of age in six of the regions and 160 HH in Tbilisi, for a total of 1060 HH (1506 + 160 = 1060). Because there were sometimes more than one child 5-16 years of age in the HH, the final sample size was approximated to 1400 for the education module and 1300 for the child labour module. A sample of HHs with no children was also obtained in each PSU to allow for more valid regional and national estimates for the water- and sanitation-related goals. In each survey region, 125 HHs (135 for Tbilisi) with no children were selected for a total of 885 HHs (1256 + 135 = 885). These HHs without children were randomly selected from the SDS list.

Therefore, prior to initiating the second stage of sampling, the NCDC supervisors, using the SDS generated lists, determined the number of HHs to be selected in each PSU. All HHs with children less than five years of age were selected; HHs with children 5-16 years of age were systematically sampled to assure 150 were selected for each survey region (160 in Tbilisi); and HHs with nochildren were systematically sampled to assure 125 HH for each survey region (135 in Tbilisi). Thus, the NCDC interviewers had the identification numbers for each HHs they were to visit. Questionnaire modules for the household, underfives, and women's questionnaire were to be completed in all HHs visited. The education/child labour modules were to be completed only forthose HHs which were identified for this purpose.

The interviewer for his/her cluster had the ID numbers of the HHs with information on whichmodule to be completed in each of them.