5. Annex: The survey and the sampling methodology
Annex: The survey and the sampling methodology
The third round of the Life in Transition Survey (LiTS III) was conducted between the end of 2015 and the beginning of 2016 in 34 countries, comprising 29 transition countries, the Czech Republic and two western European comparator countries (Germany and Italy). Cyprus and Greece were also covered for the first time. The survey was implemented by the EBRD and has benefited from the joint collaboration with Transparency International and the World Bank.
The Life in Transition Survey (LiTS) is a combined household and attitudinal survey which collects information on the socio-economic status of respondents and includes asking perception-based questions on economic, political and social topics. The first round of the LiTS (LiTS I) was conducted in 28 transition countries and the Czech Republic in 2006, while the second round (LiTS II) was conducted in 29 transition countries, the Czech Republic and 5 western European comparator countries (France, Germany, Italy, Sweden and the United Kingdom) in 2010. Both rounds were implemented by the European Bank for Reconstruction and Development (EBRD) in collaboration with the World Bank.
The third round of the Life in Transition Survey (LiTS III) was conducted between the end of 2015 and the beginning of 2016 in 34 countries, comprising 29 transition countries, the Czech Republic and two western European comparator countries (Germany and Italy). Cyprus and Greece were also covered for the first time. The survey was implemented by the EBRD and has benefited from the joint collaboration with Transparency International and the World Bank. Chart 5.1 illustrates the countries included in the latest round and in the first two.
For the third round, LiTS III, the target number of interviews was set at 1,500 households per country, which was a 50 per cent increase over the past two surveys. The achieved number of interviewed households and respondents differs by country and is reported in Tables 5.1 and 5.2.
The LITS III survey instrument was largely based on the LiTS II questionnaire and was designed so that the interview would be no longer than 60 minutes. The questions were translated into and made available in all the national languages of the surveyed countries. In addition, they were piloted prior to the beginning of field work in 10 to 15 households in each country.
The final questionnaire consists of nine modules (10 in Greece). The first and the second modules collect data at the household level and include a household roster with demographic characteristics as well as information on the household’s dwelling, the assets they own, the utilities and the services they have access to, and their consumption patterns. These modules were completed by the head of the household or any other household member who was knowledgeable of the household characteristics and finances at the time of the interview.
The remaining modules were completed by the primary respondent, with the exception of the third and the fifth ones, which were asked of both the primary and the secondary respondents so as to allow for gender comparisons on asset ownership and labour participation. The third module elicits information on any dwellings or land plots that the respondent may own in addition to their main property. The fourth module includes questions on attitudes and values. The fifth, sixth and seventh modules leverage information on the employment situation of the respondent. The fifth module is specifically concerned with the respondent’s working history, while the other two gather information on the unemployment history of the respondent, if applicable, and on any entrepreneurial activities that the respondent may have been involved with. The eighth module focuses on governance, utilisation of public services, incidence of bribery practices and perceptions on corruption. A final module includes a host of miscellaneous questions. In the Greek version, an additional module is dedicated to the impact of the economic crisis and the voting behaviour of the respondent in the recent elections.
The survey was designed by means of a multi-stage random probability stratified clustered sampling. The sample was stratified by geographical region and level of urbanity (urban or rural areas).
In the first stage of the sampling exercise, 75 Primary Sampling Units (PSUs) were selected in each country. A panel element was built into the survey design by requesting interviewers to revisit the localities that were sampled during the second round of the survey in 2010.1 ,2 A mapping exercise preceding the sampling was carried out to match the borders of the 2010 PSUs to the current ones and to identify those PSUs that had changed, merged into bigger conglomerates or split into smaller localities. Full sample frames with updated population information stratified by region and urbanity were also obtained.
In the countries where LiTS II was conducted, 50 PSUs were selected from the previous round of the survey.3 The mapping exercise was instrumental in making sure that the exact locations visited in 2010 could be pinpointed. The remaining 25 PSUs were drawn from the new sample frames in an attempt to rebalance the old sample based on the updated population information. In Cyprus and Greece, where LiTS took place for the first time, the sample was drawn afresh.
Given the target of 1,500 interviews per country, 20 households had to be visited, on average, in each of the selected PSUs. This means that approximately two-thirds of the interviews were conducted in the same areas where fieldwork took place in 2010. The remaining 500 interviews were carried out in the newly selected PSUs. In the second stage of the sampling, therefore, 20 households were selected with equal probability within each PSU. Two different procedures were adopted. In a few countries (Armenia, Estonia, Georgia, Hungary, Latvia, Lithuania, Poland and the Slovak Republic), households were directly selected from a register of individuals or a list of addresses. In the remaining countries, the addresses were enumerated prior to the start of field work by means of a random walk procedure. In such instances, PSUs were split into four segments by drawing two lines on a map, crossing from north to south and from east to west. A starting point was then identified in each of the four segments. Enumerators were instructed to conduct random walks beginning from the points indicated in the maps and write the address details of the households encountered. When more than one household was resident at a particular address (for instance, in multi-flat blocks), interviewers were instructed to not enumerate more than 13 of them. The enumerated addresses were finally used to draw a random sample of households to visit.
Selected addresses were handed over to interviewers, who were instructed to attempt to make contact with the households if, upon inspection, the dwellings were inhabited and were the main residence of the individuals occupying them. During the first visit, interviewers explained the purpose of their visit and the structure of the survey and collected information on the household composition. For the purpose of the survey, a household was defined as “a group of individuals who live together, put resources in common and share meals”.
The survey was conducted face-to-face in each selected household with one, two or three respondents. As the interviews were all administered by means of Computer-assisted Personal Interviewing (CAPI), the respondents were automatically selected by the software. Any adult household member aged 18 or over who had been living in the household in the six months prior to the survey and could speak at least one of the national languages was eligible to take part. The primary respondent was selected randomly from all eligible respondents in the household and was asked to complete modules 3 to 9 of the questionnaire (3 to 10 in Greece). A secondary respondent was selected randomly from all eligible respondents in all households that were composed of at least two adults of opposite gender. The secondary respondent was requested to complete modules 3 and 5. Lastly, the head of the household or any other member who was knowledgeable of the household characteristics and finances (potentially also the primary or the secondary respondent) were asked to complete modules 1 and 2.
An interview was considered complete only when all sections of the questionnaire were answered. In the case of non-response or temporarily absent respondents, interviewers were requested to return to each address a minimum of four times and attempt to complete the questionnaire before they could replace the household.
The number of interviews conducted with primary and secondary respondents is reported in the last column of Table 5.1 and Table 5.2. A breakdown by gender and age of the respondent is also shown in the remaining columns. Gender, age and location of residence (geographical region and urbanity) of the respondents were also used to benchmark the sample against the population and compute survey weights that allow the sample proportions to be reflective of the population ones.