Interviewers' Perceptions of Reasons for Participation Refusal in a National Longitudinal Survey, 1979-1980

Surveys based on scientific sampling methods began in the mid-1930s with market research studies and public opinion polls. The public's response to such efforts was one of enthusiastic participation, perhaps reaching a peak during World War II for government-sponsored surveys as part of the patriotic zeal attending the war [3: I0-ii]. However, for the years since World War II there has been a serious concern that nonresponse rates have increased for surveys conducted under various auspices--government, academia, market research firms, and poll-taking organizations [4]. i_/ Several reasons for the trend in survey nonresponse have been posited. For example, Goldfield, et al. [4:219] attribute it to such factors as, fear of crime, over-surveying, disillusionment about the validity and usefulness of survey results, salesmen and con artists misrepresenting themselves as survey interviewers, and issues of privacy and confidentiality. Some concern has been expressed that non-response, due mainly to refusal [3: ii], jeopardizes the future of the entire survey enterprise [4: 219]. Much of this concern centers on the public's fear of violation of privacy and compromise of confidentiality. This culminated in the formation of the Panel on Privacy and Confidentiality in Survey Response, which was convened under the sponsorship of the Committee on National Statistics of the National Academy of Sciences. One emphasis was on the general paucity of empirical data on reasons for nonresponse and quantitative studies on strategies for improving and maintaining "acceptable" response rates [4]. Of late, several empirical studies have been undertaken to explore the issues surrounding survey refusal. In addition to two studies resulting from the Panel's recommendations (see [4] ), both the National Center for Health Statistics [8] and the Center for Human Factors Research of the Bureau of the Census (CHFR) [8] have conducted research on refusals. The former concentrated on an assessment of privacy concerns and hostility toward the government as reasons for refusing to participate in the National Medical Care Expenditure Survey of 1977-78. Their data indicate that concerns with privacy and hostility towards the government were relatively minor in terms of explaining overall refusal. Only about 6% and 9% of refusers gave these two reasons respectively, as motives for refusal. 2_/ Privacy and hostility ranked well below such other reasons as, "lack of interest in participating" and "no time to spend answering the questions" [8:511-512]. DeMaio's [2] research on refusal in the Current Population Survey expands on earlier work done by the Census Bureau with CPS data in that it investigates characteristics of refusers in households in which no interviews have been conducted. Previous work was limited to respondents who refused after having cooperated in earlier waves [9]. One major concern addressed in this recent research is the possible increase in refusal rates for governmentsponsored surveys due to the Privacy Act, which requires explicit notification of the voluntary nature of the survey to potential respondents. The data indicate, however, that knowledge of the voluntary nature of the survey was cited much less frequently than other factors as the reason for refusal, suggesting that the Privacy Act may not have the detrimental effect on response rates that had been feared [2]. Research on CPS refusals is of particular importance since the CPS is a major longitudinal survey for which problems of non-response are compounded by the attrition of respondents due to refusal over the life of the panel. This paper explores several dimensions of refusal for another Census-conducted longitudinal survey, the 1979 Panel of the Income Survey Development Program (ISDP). With information obtained from interviewers on non-response due to refusal, we can bring new data to bear on several issues concerning refusal and non-response. Among the questions to be explored are: i. What are interviewers' perceptions of the reasons for refusal to participate in the survey, i.e. which reasons are most commonly cited; which reasons are least conmDnly cited? 2. To what extent are demographic and perceived socioeconomic characteristics of refusers related to reasons for refusal? 3. How do reasons for refusal in the initial wave of the panel differ from reasons given in subsequent waves? How do refusers who have never cooperated with the survey differ from those refusers who have cooperated in at least one wave of the panel? 4. At what point during the interview are refusals most likely to occur? Although answers to some of these questions may have implications for the analysis of statistical bias in survey responses, of central interest to us are the possible implications of our findings for operational aspects of survey research. Specifically, do our findings suggest any strategies for maximizing respondent cooperation?