Social researchers tend to take it for granted that respondents are, or can be, motivated to give them the information they require. In this chapter, however, we will look at the possibility that questions per se sometimes threaten respondents in a way that either influences their willingness to give answers at all or determines the nature of the answers that they do give.

While it is generally recognised that respondents sometimes give biased answers and sometimes refuse to answer at all, it is nevertheless generally assumed that these problems can be overcome by one means or another.

In fact, a number of discussions of how to overcome the effects of question threat have been published over the years (e.g. Barton, 1958; Williams, 1964; Blair et al., 1977; Bradburn et al., 1978; Sudman and Bradburn, 1974, 1982; DeLamater, 1982). In Barton's words (1958:67), ‘the pollster's greatest ingenuity has been directed to asking embarrassing questions in nonembarrassing ways’.