Individual and neighborhood determinants of survey nonresponse : an analysis based on a new subsample of the German Socio-Economic Panel (SOEP), microgeographic characteristics and survey-based interviewer characteristics

This study examines the phenomenon of nonresponse in the first wave of a refresher sample (subsample H) of the German Socio-Economic Panel Study (SOEP). Our first step is to link additional (commercial) microgeographic data on the immediate neighborhoods of the households visited by interviewers. These additional data (paradata) provide valuable information on respondents and nonrespondents, including milieu or lifestyle, dominant household structure, desire for anonymity, frequency of moves, and other important microgeographic information. This linked information is then used to analyze nonresponse. In a second step, we also use demographic variables for the interviewer from an administrative data set about the interviewers, and, in a third step, we use the results of a special interviewer survey. We use multilevel statistical modeling to examine the influence of neighborhoods and interviewers on non-contacts, inability to participate, and refusals. In our analysis, we find our additional variables useful for understanding and explaining non-contacts and refusals and the inability of some respondents to participate in surveys. These data provide an important basis for filling the information gap on response and nonresponse in panel surveys (and in cross-sectional surveys). However, the effect sizes of these effects are negligible. Ignoring these effects does not cause significant biases in statistical inferences drawn from the survey under consideration.