Conferences

The demographic development is one of the greatest challenges for Europe in the 21st century. As older people constitute larger and larger proportions in the population, not only does their political clout grow, but also the interest for this group from various other sources, like the media, advertising or medicine. Social research can play a key role in understanding the challenges of demographic change in its various facets, including health, economics and social developments. However, it is not quite clear how the growing proportion of the elderly influences particularly the collection and quality of survey data and thus empirical research. While increasing age is generally associated with declining cognitive ability, it is not quite clear in the literature whether survey quality is indeed affected by this decline or to what extent (see for example Andrews and Herzog, 1986, and Calahan, 1969).

The goal of this paper is to investigate how quality of survey data is influenced by age and whether any differences found are large enough to warrant differences in data collection depending on age. We use the Survey of Health, Aging and Retirement in Europe, SHARE. SHARE concentrates on respondents 50 years old and older in 15 European countries, and is constructed similar to other aging surveys, like the Health and Retirement Study (HRS) in the US or the English Longitudinal Study on Aging (ELSA) in the UK.

The task of measuring quality is approached mainly in considering fractions of item non-response, i.e. the amount of questions a respondent answers with “Don’t know” and / or “Refusal” in relation to the overall number of questions (see for example Guadagloni and Cleary, 1992). This will be split by different topics in the questionnaire. In a preliminary analysis, we found that between the ages of 70 and 90, the item non-response almost doubles in a linear fashion from 6.3% to 11.9% in SHARE’s employment and pension module. Other measures of quality we consider include response scale effects and variance analyses. The analyses will be elaborated upon by controlling possibly confounding variables, like gender, education, health and other demographic variables but also paradata, like the duration of the interview when the question was asked to cover effects of fatigue. Finally, SHARE allows assessing country specific differences in these effects, as the ex-ante harmonized survey ensures that all interviews are conducted in the same manner.