ABSTRACT
In their 1950 study that included a comparison between survey responses and records, Parry and Crossley (1950) showed that respondents overreported voting, having a library card, and several other socially desirable characteristics as compared to information about the respondents in external records. In a 1976 record-check study, Locander, Sudman, and Bradburn reported similar findings and also showed that respondents underreport bankruptcy and being charged with drunken driving. These classic studies addressed a key concern of those who rely on survey data-the accuracy of responses to questions that respondents might perceive as threatening.
