A large body of literature exists on nonresponse to questionnaires, particularly in relation to the most frequently used surveys conducted by mail (Dillman, 1991). Research assessing response rate is ‘dominated by hundreds of tests of specific procedures and techniques for improving response’ (Dillman, 1991: 229). Nonresponse to questionnaires is generally seen to occur in two ways. First, the error can be created by questionnaires which are not returned by the chosen sample, sometimes referred to as noninterviews. Second, it can occur when members of the sample population do not complete particular questions, also called response omissions (Warwick and Lininger, 1975). Many different methods, both subjective and quantitative, have been suggested to compensate for these forms of nonresponse (e.g. Brook, 1977; Dillman, 1991; Jolliffe, 1986; Viswesvaran et al., 1993). Some researchers (e.g. Brook, 1977) have tried to create profiles of nonrespondents suggesting that they may be elderly or disengaged; live in urban areas; may feel a judgement is being made about their responses; or may feel inadequate about contributing. Bradburn and Sudman (1979) suggest that nonresponse may be more of a problem when the questions being asked are of a sensitive or threatening nature or involve ‘social definitions of “desirable” behavior’ (Bradburn and Sudman, 1979: 65). Through a variety of experiments and tests they determine that the ‘perceived normative threat of a topic influences responses to questions’ (Bradburn and Sudman, 1979: 83) including both noninterview and response omissions. None of this research addresses the situation in which nonrespondents do not answer questionnaires, but do in fact send letters explaining why they are not responding. Some of the explanations received may easily fall into the nonresponse categories outlined above, but others may challenge both these categories and the research questions. We 1 are conducting research on feminism, sex and the media, which involves sending questionnaires by mail to women identified as spokespersons for women’s issues in selected mainstream media. 2 More than 180 questionnaires were sent out. About 45 percent of these were returned completed. Of the remaining 55 percent, about half (27 percent of the total) were not completed, but the women to whom we had sent them responded
Read full abstract