Eu Survey of Income and Living Conditions (Eu-Silc) 2011

NATIONAL INSTITUTE OF STATISTICS - ROMÂNIA
Quality reporton EU-SILC Survey in 2011

EU SURVEY OF INCOME AND LIVING CONDITIONS (EU-SILC) 2011

QUALITY REPORT

GENERAL DESCRIPTION

At European level, the main statistical tool for data collection to provide information to measure poverty and social exclusion is the survey on income and living conditions (EU-SILC). The Romanian survey on income and living conditions, named Quality of Life Survey, represents the implementation of EU-SILC Survey in Romanian statistical system.

1. Coverage

The survey covers all the members of the selected households including the persons absent from home for a longer period (over 6 months), if they are preserving family relations with the household to which they belong, such as:

-pupils and students away for study;

-persons left for work;

-prisoners;

-persons temporary in hospitals or sanatoria for medical cure, and they are surveyed via the relatives living in the selected households.

Persons permanently living in collective units (elderly, handicapped, workers hostels, sanatoria etc.) are not included in the survey.

2. Reference periods and registration

According to the specificity of questions there were several reference periods: week before the interview to some questions about economic activity, 12 months for questions related to the need to consult a doctor, the previous calendar year for income etc.

For 2013, data collection period was in 16 May – 3 June.

3. Periodicity of the results

Starting with 2007, the quality of life survey is annually carried out as a continuous research allowing getting detailed statistical data used to characterize the living conditions of the population. This is done by collecting data on various social issues: housing, health, education, labour information, income etc.

4. Sampling Plan

The survey implemented the methodology described in the EU-SILC Regulation (EC) no 1177/2003 of the European Parliament and of the Council concerning Community Statistics on Income and Living Conditions.

An integrated design with a rotational sample was applied, in which the sample is divided in sub-samples, each of them similar in size and design and representative for the whole population. The survey uses the integrated four years rotational panel design. From one year to another three sub-samples are ratained, one is dropped and one new sub-sample is included in the survey. In this way, the cross-sectional and longitudinal statistics are produced from the same set of sample observations.

Because of the lack of appropriate registers (dwelling register, population register etc), the household surveys carried out by NIS-Romania are based on the repeated use of a master sample, which involves further the use of multi-stage sampling designs.EU-SILC sampling design has been founded on a two-stage sampling technique.

The sampling plan is a two-stage probability sampling of clusters of housing units.

In the first stage, a stratified random sample of 780 areas, Primary Sampling Units (PSUs), wasdesigned after the 2002 Census, using as stratification criteria the residence area and county.This is the Multifunctional Sample of Territorial Areas, so called the master sample EMZOT.

In the second stage sampling, the Secondary Sampling Units represented by dwelling units are systematically selected from the initial sample of PSUs. Each Secondary Sampling Unit requires the inclusion of all households in the sample and hence all persons in the household. The sample is divided into four sub-samples of the same size and structure, using a rotational scheme of panel (recommended by Eurostat) with duration of 4 years.

Figure1. Illustration of the rotational pattern

YEARS
2008 / 2009 / 2010 / 2011 / 2012
Sub-samples: /
2 / /
3 / / /
4 / / / / /
5 / / / / /
6 / / / /
7 / / / /
8 / /
9 / / / /

5. Sample frame

During 2002-2003 years, the master sample EMZOT was designed on the basis of the information provided by the Census of the Population and Dwellings from 2002, aiming to obtain a sampling reserve for the household surveys to be conducted in the inter-censual period.

EMZOT was created by including the 780 groups of census sections.

The 780 census sections composing EMZOT at the time being are the Primary Sampling Units (PSU), being distributed on the two areas as follows:

- 427 PSU in urban area;

- 353 PSU in rural area.

EMZOT is a stratified sample. Stratification criteria were county and urban/rural area, resulting 88 strata. In order to select UP, within every stratum, the balanced method was used, by macro SAS CUBE.

6. SamplingUnit

The Primary Sampling Unit, corresponding to the selection of the master sample, has been a group of census sections.

The Secondary (ultimate) Sampling Unit, corresponding to the selection of the survey sample, has been the dwelling.

7. The sample size

The survey was conducted on a sample of about 10,000 dwellings in all counties.

According to the methodology adopted by Romanian NIS for its household surveys, a dwelling may contain one or many households (still, not very frequent cases, usually a dwelling comprising only one household). Thus, in case of EU-SILC, all the households belonging to the selected dwellings, as well as all the persons of 15 years old and over, living in those households, are surveyed.

Concerning the SILC instrument, three different sample size definitions can be applied:

- the actual sample size which is the number of sampling units selected in the sample

- the achieved sample size which is the number of observed sampling units (household or individual) with an accepted interview

- the effective sample size which is defined as the achieved sample size divided by the design effect with regards to the at-risk-of poverty rate indicator

Given that the effective sample size has been treated in the section dealing with sampling errors, in this section the attention focuses mainly on the achieved sample size.

Table 1. Actual and achieved sample size

Obs. / Actual S_Size / Achieved_S
1 / 8151 / 7675

Table 2. Achieved sample size

Obs. / Number of household
2012 / Number of household
2013 / Percent 1 / Persons 16 years and over / Last_rot_
group / Num_of_
rot__hh_
2013 / Percent2
1 / 7718 / 7675 / 0.99 / 15974 / 4 / 1956 / 25.49

8. Stratification

Stratification concerns only the first stage sampling. There are 88 strata; the criteria used being the area (urban or rural) and the county (NUTS-3 level) where a certain PSU is located.

9. Method of weighting procedure

The final weights, used to obtain estimates at household level and person level, as well, are obtained following, basically, a 3-step procedure. The first step assigns the inverse of the selection probabilities to each sampled dwelling unit. The second step adjusts for non-response. The third and final step consists of calibrating the secondary weights, for each wave, by region, to the best latest available population totals.

• Design factor

Wave 1(subsample selected in 2011)

The design factor of the household is the inverse ofinclusion probability. The design factor for households and for individuals are the same, because in each selected dwelling, all persons are selected for the survey.

In case of the householdsat the second, third and fourwave, an indirect sampling of householdsis donethrough the panel (of persons aged 14+ at the time of the panel selection). In this case, the inclusion probabilities cannot be calculated. Then, the solution consists of applying the Weight Share Method.

Wave 2(subsample selected in 2010)

The design factors of households are calculated through the individual base weights. The individual base weights are obtained from cross-sectional weights calculated in previous year 2011 inflated with attrition.Co-residents are given zero base weight.

Wave 3(subsample selected in 2009)

There are two situations:

a. The sample person was a respondent in 2010. The base weight is calculated taking into account the base weight of previous year and then corrected both: attrition between 2010 and 2011 and compensation of the re-entrees.

b. The sample person was a non-respondent in 2010 (re-entrees) In this case the base weight is obtain taking into account the cross-sectional weight RB050 calculated in 2009 corrected for the attrition between 2009-2011. For co-residents the weight is equal with zero.

Wave 4(subsample selected in 2008)

The approach is similar with the previous wave and two cases are distinguished, too:

a. The sample person was a respondent in 2010.The base weight is calculated
taking into account the base weight of previous year and then corrected both: attrition
between 2010 and 2011 and compensation of the re-entrees.

b. The sample person was a non-respondent in 2010. In this case the base weight is obtaining taking into account the base weight calculated in2009 corrected for the attrition between 2009-2011. For co-residents the weight is equal with zero.

• Non-response adjustments

In order to contra balance the non-respondent households, it is proceed at a re-weighting, by adjusting the weights of the respondent households with the inverse of the response rate.

The non-response are not globally adjusted, at the entire sample level, but separately, on groups of households, groups generated by the variable considered as explicative variable of the non response: region (NUTS 2 level) and residential areas (urban/rural). This correspond to the so-called 'response-homogenous groups" method, which assumes that in a certain group all the units have the same probability. In order to minimize the effects induced by the presence of non-response another adjustment is done: re-weighting by calibration of the weights.For wave 1 we used as explicative variables for non-response region (NUTS II level) and area of residence (urban/ rural) and for the second, third and fourth wave - the region. In order to minimize the effects induced by the presence of non-response another adjustment is done: re-weighting by calibration of the weights.

• Adjustments to external data (level, variables used and sources)

We applied an integrative calibration that means that we used both households and personal variables in the procedure. The calibration is performed at the household level using the household variables and individual variables in their aggregate form as calibration variables. This technique ensures that all members in the same household receive the same weight. Adjustments were made using the SAS macro CALMAR. Calibration variables were: “distribution of the population by age group (0-15; 16-24; 25-34; 35-49; 50-64; 65-74; 75 and over), area of residence (urban\ rural) and gender” using Romanian Population Estimates at the end of the income reference period and ’’households totals by region’’.

Three cross-sectional weights were calculated:

1) Household cross-sectional weight (DB090)

2) Personal cross-sectional weight for all household members (RB050)

3) Personal cross-sectional weight for all household members aged 16 and over (PB040)

10. Data collection

The method of data collection was face-to-face personal interviews, using paper questionnaires. The interviewers visited the addresses selected in the sample and fulfilled the questionnaires, based on the interviews. The household questionnaire was fulfilled by interview with the household head and individual questionnaire by interview with each household member 15 years old and more.

Data collection is achieved using statistical questionnaires designed according to European Parliament and Council Regulation no. 1177/2003. They do not support major changes from one year to another.

Statistical questionnaires used:

MG – Household file;

CG – Householdquestionnaire;

CI – Individual questionnaire;

LG – List of the dwelling and household included in the sample.

HOUSEHOLD FILE(MG) is a very important questionnaire of EU-SILC, particularly for its longitudinal component. Longitudinal data concerns individual-level changes over time, observed periodically over a four-year period.

HOUSEHOLD QUESTIONNAIRE(CG)was fulfilled by interviewing the household respondent and includes the following sections regarding the household:

Household Identification;
General Data;
Data regarding child care for all the children less than 13 years;
Household economical situation;
Loansandcredits;
Household endowment with durable goods;
Living conditions;
Minimum household income to make ends meet;
Taxes and household income.

On the first page of the questionnaire CG (cover) adding the following identification:

Address – from LG;
Centre Code – from MG;
Dwelling Code- 5 digits code from MG;
HQ ranking number inside dwelling– is a number assigned sequentially (starting with 1) completed questionnaires to every household in a dwelling;
Household ID – from MG;
Household status;
Contact at address;
Average interview duration.

INDIVIDUALQUESTIONNAIRE(CI) was fulfilled for each person 15 years or more, in order to record better the incomes of the people less than 16 years.

The questions relate to:

Interviewee person identification;
Information about interview;
Demographical data;
Health status;
Education;
The situation during the teenage years;
Economic activity;
The intention to move somewhere else;
Cash transfers;
Individual incomes.

Explanations for a big number of questions from all the questionnaires were included into an handbook. It is divided into the following chapters:

General characteristics;
General recommendations for operators;
Specifications for the survey questionnaires.

Tabel 3. Distribution of households members 16 years old and over by data status

Number / %
Total / 16010 / 100.0
Information of interview completed / 15974 / 99.8
- information completed only from interview (RB250=11) / 15974 / 99.8
- information completed only from registers (RB250=12) / na / na
- information completed both from interview and registers (RB250=13) / na / na
Interview not completed, though contact made / 3 / 0.02
-individual unable to answer and no proxy possible (RB250=21) / 1
- failed to return the self-administrated questionnaire(RB250=22) / na / na
-refusal to cooperate(RB250=23) / 2
Individual not contacted because: / 33 / 0.2
-person temporarily away and no proxy possible(RB250=31) / 16 / 0.1
-no contact for other reasons(RB250=32) / 12 / 0.08
Interview not completed, though contact made(RB250=33) / 5 / 0.02

Tabel 4. Distribution of household members by the respondent status

Number / %
Total / 18347 / 100.0
- Current household member aged 16 years and over (RB245=1) / 16010 / 89.2
- Selected respondent (RB245=2) / na / na
- non-selected respondent (RB245=3) / na / na
- not eligible respondent (RB245=4) / 1940 / 10.8

Tabel 5. Distribution of households members aged 16 years old and over by the type of interview

Number / %
Total / 15974 / 100.0
Questionnaire completed –face-to-face interview PAPI (RB260=1) / 13498 / 84.5
Questionnaire completed –face-to-face interview CAPI (RB260=2) / na / na
Questionnaire completed –CATI (RB260=3) / na / na
Self-administrated by respondent (RB260=4) / na / na
Proxy interview (RB260=5) / 2476 / 15.5

The mean interview duration per household is calculated as the sum of the duration of all household interviews plus the sum of the duration of all personal interviews, divided by the number of household questionnaires completed. Only households accepted for the database have to be considered.

Average interview duration = 30.8 minutes.

11. Survey staff for field operations

Staff responsibilities

The National Institute for Statistics of Romania (NIS) is responsible for the overall co-ordination of the survey, with the organisation of data collection, supervision of field operations and survey co-ordination at the county and BucharestMunicipality levels falling under the responsibility of regional statistical offices. Staff involved in carrying out the survey consisted of:

-42 local co-ordinators (one each for 42 counties according to NUTS2 level);

-47 supervisers;

-780 interviewers (one for each enumeration area).

Local co-ordinators: Local co-ordinators are responsible for organising and co-ordinating of the survey field operations over all research centres of the county. Tasks included:

-Organise and plan the field work;

-Allocation of enumeration areas to supervisers;

-Receipt of survey materials (methodological guides, questionnaires, list of selected addresses, identification cards etc.) and distribution to supervisers;

-Supervision of supervisers and enumerators;

-Carry out the activities for identification of the sample persons who have moved into the county in the last year;

-Inspection of fieldwork (data collection);

-Co-ordination of quality control of survey data from questionnaires;

-Supervision of data codification and transmission of the fulfiled questionnaires to the National Institute for Statistics.

Supervisors: Supervisors are directly responsible for conducting the survey in the allocated enumeration areas. Tasks included:

-Allocation of research areas to interviewers;

-Dissemination of materials (methodological guides, questionnaires, list of selected addresses etc.) to each enumerator;

-Supervision and periodic inspection of data collection in the field;

-Receipt and control of questionnaires filled in by enumerators;

-Codification of the data and correction of the errors related to data registered and flows within the individual questionnaire;

-Carry out the activities for identification of the sample persons who have moved into the county in the last year.

Interviewers: Interviewers are responsible for carrying out tasks related to the following activities:

-Identification of dwelling addresses based on the list of selected addresses received from the supervisors;

-Conduct preliminary visits to identify households within the research centres;

-Visiting (up to 3 times) the households from the list to identify the eligible persons and to conduct the interviews;

-Conducting the interviews and recording the data collected in the questionnaires;

-Providing the supervisors with the completed questionnaires and the situation of the results of households visited;

-Complete the list of the dwellings and households surveyed.

Table 6. Relevance of the main EU-SILC indicators at national level

For policy

makers

/ For social

actors

/ For the
media / For
researchers
and students
High / Low / High / Low / High / Low / High / Low