Official name of the survey/data source:
ESRI Survey of Income Distribution, Poverty and Usage of State Services

LIS name:
ESRI Income Survey

Administrative unit responsible for the survey:
Economic and Social Research Institute (ESRI)

Address:

Funding for the ESRI Income Survey is provided for by the Directorate General for Employment, Industrial Relations and Social Affairs of the Commission of the European Union, the Irish Combat Poverty Agency and the Economic and Social Research Institute.

Copies of the original codebook and other documentation can be obtained from the Economic and Social research Institute at the above address. The principle users of the data collected in this survey are the research staff of the Economic and Social Research Institute.

The main purpose of the ESRI Income Survey is to provide information on the income distribution, poverty, non-cash transfers and the labor market.

The ESRI Income Survey sample was extracted from the Irish electoral lists (adults aged 18+). The process of data collection for the 1987 ESRI Income Survey was begun in december 1986 and completed in september 1987; the information being collected retrospectively for the 12 months prior to the interview.

Use of this data is restricted. Users of the data must not disclose information which would allow individuals or families to be identified. No other special restrictions apply to LIS users.

A bibliography of the most important publications based on the ESRI Income Survey is provided in the bibliography in Section M. Also included in the bibliography in Section M is a list of available user documentation for the ESRI Income Survey.

B. POPULATION AND SAMPLE SIZE, SAMPLING METHODS back

The survey was designed to provide a national sample from the population resident in private households. The sampling frame, from which a sample of names and addresses was drawn, was the Register of Electors. This is a annually revised listing of all those eligible to vote in local, national or European elections.

Therefore, marginal populations like the homeless and itinerants were excluded from the sampling frame. The sampling was performed using the RANSAM program developed at the ESRI. This implements a multi-stage random sample incorporating both stratification and clustering, and giving each individual on the Register an equal probability of being selected.

The target sample selected for the survey comprised 5,850 households, selected as 225 clusters of 26 each. Within clusters, respondents were selected on a systematic basis, giving an implicit geographical stratification. Since the initial sample of names and addresses was on the basis of persons on the Register, households had a probability of selection proportional to the number of electors they contained. Thus households with larger number of voters were oversampled. The weighing applied to the results, discussed in point E, was designed inter alia to adjust for this bias.

Although the population living in institutions or other "group" arrangements such as nursing homes for the aged or ill, prisons. military installations, etc., were not excluded from the sampling frame, interviewers were instructed to exclude them.  

C. MEASURES OF DATA QUALITY back

The total sample selected was 5,850 households. Of these, a total of 615 were not successfully contacted. For the majority of these - 421 cases - this was because the household had moved and their new address could not be found. A further 114 of the addresses no longer existed, and in 80 cases the person selected was deceased, In addition, 70 of the addresses selected were found to be in institutions, and therefore did not form part of the private household population.

Excluding these cases left an effective sample of 5,165 households. Of these, 3,321 households (64.3%) responded to the survey. The refusal rate was 24.1 per cent (1,246 households). A further 486 (9%) were never available when the interviewer called despite repeat visits. For 112 cases (2.2%) they were too ill or senile to take part. Of the responding households, 27 were excluded from the sample for analysis due to completely or substantial missing information in key areas; notably income. This left a sample for analysis of 3,294 households. This constitutes 63.8 per cent of the effective sample and 56.3 per cent of the overall sample originally selected.

The refusal rate was somewhat higher than that found in most other ESRI surveys, presumably due to the sensitivity of the subjects covered in this survey and the complexity of the questionnaires involved. Comparison with the response rate achieved in the Household Budget Surveys, the principal data source hitherto on household incomes in Ireland, is relevant. The two national Household Budget Surveys, carried out in 1973 and 1980, had response rates of 57% and 56% of the effective sample.

D. DATA COLLECTION AND ACQUISITION   back

The survey was carried out by the Institute's own Survey Unit and panel of trained interviewers. All interviews were carried out by personal visits, often entailing repeat visits to households, and interviewers were instructed to keep calling until a response or definite non-response was obtained.Each adult member's cooperation was sought for direct info. The head of household was asked to provide information for absent household members. Respondents were asked to consult or use pay records, tax returns, or other documents when providing income information. Participation in the survey was voluntary.

It was necessary to take special account of the problems in obtaining reliable measures of farm income because of the relatively high proportion of the Irish labor force engaged in farming activity. Most respondents were not likely to have detailed farm accounts. various methods of overcoming these problems were explored. After consulting 'An Foras Taluntais' (AFT), now 'Teagasc', and the Central Statistics office, it was decided to develop an additional farm questionnaire to collect data on the major elements of output and costs.  

E. WEIGHTING PROCEDURES back

Survey weights have been assigned to each sample case. These weights were assigned in order to adjust for survey units that were selected but not interviewed and for the deliberate oversampling of larger households. The reweighting was implemented on the basis of four key variables: 1) household location (urban vers. rural); 2) the number of adults in the household; 3) occupation of the household head; and 4) age of the household head. The sum of the survey weights is equal to the total number of sampled units.

The representativeness of the sample was assessed by reference to data from a number of external sources. As one independent check, data from the 1986 Labor Force Survey were obtained from the CSO, showing the breakdown of the households by the number of members in paid work. The sample when reweighted reflects very closely the pattern shown by the Labor Force Survey in terms of this variable.

The same is true of the distribution of households by number of persons unemployed. The demographic composition of the sample was also compared with the population figures from the 1986 Census. The sample reflects the population pattern well. Where there are differences, the sample has a higher proportion of children (up to 14 years of age) and a lower proportion of those aged between 15- 25, 35-44 and 75 and over than the Census. These differences are not substantial, though, and in the case of the elderly would be largely attributable to the fact that the sample refers only to those in private households while the Census of course refers to the entire population.

Administrative statistics on the numbers in receipt of social welfare payments of different types provided a further external source against which the representativeness of the sample was assessed. The numbers in the sample in receipt of payment from the various schemes, grossed-up to implied population totals ware compared with the number of individual recipients shown by the administrative records at end-1986 and end-1987. This shows a quite close correspondence between the sample and administrative figures for the major schemes.

F. DETERMINATION OF SURVEY UNIT MEMBERSHIP back

The survey gathers information at individual and household level. The household unit itself is defined as a person or a group of persons who all live regularly together (at the address selected) and for whom food is provided (at least one meal a day) by the same person of group of persons.

G. CHILDREN AND SPOUSES back

Alternative definitions are available. One key distinction is between adults aged 15+ and not in full time education and children aged <15 or in full time education. Only the former are individually interviewed. Children at college/university are regarded as members of the household, even when they live away from home during term. Children who are not the biological children of the head and/or spouse are included. Information identifying parent of each child was coded and keyed.

H. AVAILABILITY OF BASIC SOCIAL AND DEMOGRAPHIC INFORMATION back

In Table 1 are summarized the basic social and demographic information which is available in the ESRI Income Survey.

Table 1:
Basic Social and Demographic Information

I. AVAILABILITY OF LABOUR MARKET INFORMATION back

Labor force status information is available for this survey. No groups were excluded at the time of interview through filter questions relating to the employment status of the respondent, although only income recipients were requested to provide answers to the income schedule. The labor market information which is available in the survey is summarized in Table 2.

Table 2:
Labor Market Information

Indiv: All adults 15+, not in full time education
Findiv: All adults, but must have completed full

J. AVAILABILITY OF GEOGRAPHIC INFORMATION back

It is possible to identify the geographic region and city size.  

K. SOURCES AND AMOUNTS OF CASH INCOME back

Sources and amounts of income from employment are available for the weekly periods only. For adults aged 15+ and not in full time education, the amounts received are usual amounts. For Findiv actual and usual amounts are recorded.

L. TAXES back

Income tax and payroll tax (Findiv) is identified in the survey and recorded individually for each person.

M. BIBLIOGRAPHY OF MAIN PUBLICATIONS back

Callan, T.; Nolan, B. & B. Whelan 1988, Poverty and the Social Welfare System in Ireland.

Callan, T.; Nolan, B. & B. Whelan 1989, Measuring Poverty and the impact of the Social Welfare System, ESRI General Research Series Paper.