Question Threat and Response Bias

Abstract

ONE CRITICISM of using self-reports to estimate the frequency of different activities is that there may be systematic bias in such reports. The bias is believed to be toward overreporting for socially desirable behaviors (e.g., voting), and toward underreporting for socially undesirable behaviors (e.g., intoxication) and for those of a personal nature about which respondents may feel uneasy talking with others, particularly strangers (e.g., sexual behavior). Because investigators often have no practical alternative to reliance on self-reports, it is important to understand as fully as possible the sources.of these reporting biases, and, where possible, to use question formats and control variables that will enable the investigator to minimize or adjust for the reporting biases. This article presents the results of a study designed to investigate factors related to behavioral reports typified by underreporting. We can distinguish two kinds of questions which might lead respondents to distort their responses: (1) anxiety-arousing questions about, for example, behaviors that are illegal or contra-normative or about