Let’s Talk About Money! How Do Interviewer Expectations Affect Item Nonresponse to Income and Asset Questions?

Authors

Sabine Friedel, Technical University of Munich (TUM), Chair for the Economics of Aging Munich; Center for the Economics of Aging (MEA), Max Planck Institute for Social Law and Social PolicyFollow

Date of this Version

2-26-2019

Document Type

Article

Citation

Presented at “Interviewers and Their Effects from a Total Survey Error Perspective Workshop,” University of Nebraska-Lincoln, February 26-28, 2019.

Comments

Copyright 2019 by the authors.

Abstract

Personal income and assets are sensitive topics to discuss and the discussion of money tends to be a taboo. This phenomenon is reflected by high nonresponse rates to items that address income and assets questions in interviewer-mediated surveys. However, such information is important to obtain, as household income and different types of assets are used as core variables in socio-economic models. Such item nonresponse is influenced by interviewers. Although interviewers are trained to conduct standardized interviews, some interviewers obtain a higher number of item nonresponses than others. This study examines interviewer effects on nonresponse to several income and asset questions in a face-to-face survey- the Survey of Health, Ageing and Retirement in Europe. First, we investigate the extent of interviewer effects on different types of nonresponse (“don’t know” and “refuse to answer”) to items that address household income, bank balance, and interest and dividend income. Second, we investigate whether interviewer expectations about providing answers to financial questions matter. For this purpose multilevel logistic regressions are used to separate respondent and interviewer characteristics. The results of the multilevel models show that interviewers have a significant influence on both types of item nonresponse. The interviewer variances are significant at the five percent level and the intraclass correlation coefficients in the random intercept model without any explanatory variables range between 0.42 and 0.54. In addition, interviewer expectations are significantly correlated with “don’t know” responses and ”refuse to answer” of items addressing household income, bank balance, and interest or dividend income. Respondents are more likely to state their income and assets when interviewed by an interviewer who expects a higher share of his or her respondents to provide answers to financial questions than by an interviewer who expects a low share of his or her respondents (50 percent or fewer) to provide answers to financial questions. Our results aim to inform about interviewer effects on income and asset questions and support survey practitioners in designing interviewer training sessions.