Assessing the effect of social integration on unit nonresponse in household surveys

Abstract

The social integration hypothesis concerns one cause of survey nonresponse. Individuals who are more integrated in society will be more likely to respond to a survey, while individuals who are socially isolated will be less likely to participate. While much research has demonstrated support for the social integration hypothesis, the strength of the findings has been mixed. We expect this is because previous research has treated social integration and nonresponse as singular concepts and has not analyzed their various components (e.g., civic vs family engagement or noncontact vs refusal). Using the rich frame information available on the American Time Use Survey (ATUS) and the Survey of Health, Ageing and Retirement in Europe (SHARE) Wave II, we applied confirmatory factor analysis to identify and build measures of various routes to integration. We regressed survey response onto these integration measures to detect which routes to integration (e.g., civic engagement) are most predictive of response. We also decomposed nonresponse into refusals and noncontacts to test whether the different integration routes made individuals more likely to be a given type of nonrespondent. Overall, the models suggest that not all routes to integration improve the probability of response. In fact, some routes may lower the probability. These findings hold moderately well regardless of the type of nonresponse.