Children’s Activities Survey 2009

Sample Design
The CAS was conducted as a module of a larger, regularly conducted Labour Force Survey (LFS), which is the main survey instrument used to collect labour market data in Moldova. The sample size was chosen so as to allow for representative estimates of key child-labour indicators for the country at large as well as for urban and rural areas and for the four statistical regions of the country, namely, North, Centre, South and Chisinau Municipality, which includes the capital city.

Sample Selection
The LFS employs a complex, two-stage sample design based on probabilistic sampling. In Stage I, all primary sampling units (PSU) from the Multi-dimensional Sample for Social Research (EMDOS) master sample were included, with the PSUs in the EMDOS selected using probability-proportional-to-size (PPS) sampling. Population Census lists (2004) combined with the lists of electricity consumers (2007) were also used as the sampling frame. In Stage II, households were initially selected using simple random sampling (SRS); from these, households identified as having children aged 5-17 – the target population – were selected for interviewing in the CAS. The sampling frame for Stage II was comprised of the LFS sample for the first, second and fourth quarter.

The CAS survey was conducted as an ad-hoc survey using the LFS sample for the first, second and the fourth quarter of 2009. The LFS sample comprises 12,000 households in 150 Primary Sampling Units (PSUs). Due to the rotational structure of the LFS, some households included in the LFS are interviewed in more than one consecutive month/quarter. Consequently, the total sample size for the CAS would be 36,000 households (LFS: Q1+Q2+Q4), if it were to include the multiple appearances of each household in the rotation scheme. However, in calculating the CAS sample size, individual households were counted only once. Thus, the CAS sample size consisted of 18,400 households, comprised of 23 rotational groups (sub-samples) of 800 households each, and was estimated to include 5,054 households with at least one child aged 5-17 years. This sample was obtained by combining a sub-sample of 3,534 households that had been sampled in the LFS for the first and second quarters of 2009 (Rotational Groups a-g and j-s) with a sub-sample of 1,520 households that were not included in the LFS until after the second quarter of 2009 (Rotational Groups t-y). Estimated numbers of children aged 5-17 expected to be interviewed included 2,194 in the first sub-sample and 5,246 in the second sub-sample, for a total estimate of 7,440 children.