Sampling

Sampling Procedure

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 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 provide 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 addition, the sample was designed to provide estimates for each of the 8 regions for key indicators: Skopski, Pelagoniski, Vardarski, North East, South West, South East, Poloski and East.  Each region was divided into municipalities, and addition, in the 2002 census, each municipality was subdivided into enumeration areas.  In total Republic of Macedonia includes 123 municipalities, and 7105 enumeration areas.  The sample frame for this survey was based on the list of enumeration areas developed from the 2002 population census.

The primary sampling unit (PSU), the cluster for the 2005 MICS, is defined on the basis of the enumeration areas from the census frame.  Census enumeration areas were defined as primary sampling units (PSUs).  The sampling strata can be identified using a combination of the region (HH7) and area (HH6) variables, e.g. stratum = HH7*10 + HH6.

A multi-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 Republic of Macedonia. The sample was further stratified to include specific strata for the Roma population, dividing all strata into Roma and non-Roma strata. The selection of clusters was performed by generating a list of all clusters in each stratum, ordered by the total number of women aged 12-46 at the time of the census in 2002 (who would be 15-49 at the time of the survey in 2005), with the cluster with the largest number of women listed first.  From this list, the first K clusters were selected, where K is the number of clusters to be selected in the stratum, according to the selection table above. In total there were 350 clusters selected with a fixed take of 15 households per cluster selected for interview, expected to yield 5250 households.  The selection of households was performed by sorting the list of households in each cluster into two groups: households with children under 5, and those without children under 5.  From these two groups, 12 households were to be selected from the first group and 3 households from the second group.  

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 interviewed.

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.

Deviation from Sample Design

The household completion rate was less than expected (household surveys such as MICS target a completion rate of 95%, or better). Looking at the distribution of incomplete households by stratum and by interviewer, there is neither apparent bias in the incompletion rates nor any obvious systematic incompletion of household interviews, with similar response rates in each of the two household selection groups (households with children under 5 and households without children under 5).  There are however, a substantial number of “Refusals”, a similar number recorded as “Other”, and slightly smaller numbers categorized as “Not at home”, or “Households or addresses not found/Dwelling destroyed”.

Response Rates

5250 households were selected for the sample. During the fieldwork the number of households identified increased to 5379, due to the fact that in some cases, more than one household were found in one dwelling. In these cases, the MICS3 guidelines were followed, and in the cases where two households are found in one dwelling, both households were included in the survey. Of the total of 5379 households, 5287 were found to be occupied. Of these, 4701 were successfully interviewed for a household response rate of 89 percent. In the interviewed households, 7539 women (age 15-49) were identified. Of these, 7397 were successfully interviewed, yielding a response rate of 98 percent. In addition, 4578 children under age five were listed in the household questionnaire. Of these, questionnaires were completed for 4548 which corresponds to a response rate of 99 percent. Overall response rates of 87 and 88 percent are calculated for the women's and under-5's interviews respectively.

Weighting

The Republic of Macedonia Multiple Indicator Cluster Survey sample is not self-weighted.  Sample weights were calculated for each of the datafiles.  

Initially, for weighting the data, stratum level weights were calculated, based on urban-rural within region, and separately for non-Roma and for Roma, and for households with children under-5, and for households without children under-5 based on the sampling information collected. Stratum level weights would be applicable if sampling with probability proportional to size (PPS) method of selecting clusters had been used.  As the method of selecting clusters was based on selecting the largest K clusters in the stratum, where K is the number of clusters to be selected, it was not possible to use this approach, and so cluster level weights have been calculated.

Weights computed at the cluster level applying the inverse of the probability of selection were calculated.  These were then adjusted for non-response at the stratum level, and then normalized to the total sample of households to produce the final household weights.  The household weight variable is called HHWEIGHT and is used with the HH data and the HL data.

For the women's and for the children's data, the household weights were adjusted for non-response to the women's and the children's questionnaires respectively, and normalized to the total number of women and children interviewed, as is standard in MICS (and other household surveys).

However, due to the biased age/sex distribution found in the survey data (resulting from the selection of clusters being based on those clusters with the largest number of eligible women), the data for the household members and for women were further weighted to adjust the age/sex distribution to the age/sex distribution found in the census.  This adjustment was performed based on five-year groups of age.  These adjusted weights were then normalized to the total number of household members and women, respectively.  The same adjustment was not done for the children's data as there is only one five-year age group in the children's data.  Although there are some differences in the sex distribution from the survey data compared to the census data, the differences were not considered sufficient to adjust the weighting for this group.
Generated: APR-28-2008 using the IHSN Microdata Management Toolkit