Inconsistent Respondents and Sensitive Questions

Ruben Castro, University of Pennsylvania

Little attention has been put on the respondents as the main cause of measurement error, regardless of the question being asked, the interviewer effect, or other factors that influence survey quality. Some individuals simply provide worse answers than others, a behavior called satisficing. There is much to be discovered about the consequences of satisficing on the quality of answers. An especially troublesome case is that of sensitive questions, where misreporting is quite common. In this study I conduct an empirical analysis of the effects of satisficing on answers to sensitive questions. I find a significant role for individual-level time-consistency as a predictor for answers to sensitive questions related to children, sex, condoms and HIV; Couple’s reports, where couple-level fix effects can be ruled out, confirm this result. Therefore, it is possible that satisficing has an impact on the way individuals “lie” on sensitive questions.