The threatening questions, which concerned such subjects as gambling, smoking marijuana, and sexual behavior, were placed within the framework of a leisure-activity study. Interviewers were all female, age 35-50, and generally had substantial previous experience in the field. Interviewer expectations were measured by a self-administered questionnaire which each interviewer completed after the training for the study but before any interviewing had been conducted. It was hypothesized that response effects on threatening questions would be influenced by question form, respondents' perceptions of social norms against discussing the topic, and the effects of interviewers' expectations. A major problem in interpreting the results was that interviewers were not randomly assigned to cases but were assigned within their home areas. However, results showed that the effects of interviewers perceptions generally accounted for about 7 percent of the total response variance. This finding was controlled for the respondent's education. The proportion of variance explained by interviewers' effects was generally stable across items and locations. Interviewers' expectations about the general difficulty of the interview appeared to be weakly related to levels of reporting. Interviewers who expected difficulties with the questionnaire obtained lower percentages of respondents who reported ever engaging in the activities surveyed than interviewers who thought that the survey would be easy. In addition, interviewers who expected their respondents to underreport behavior obtained lower levels of reporting, thus confirming the notion of the self-fulfilling prophecy. Other interviewer expectations had no additional impact in the data. Results indicated that in most practical situations, interviewer effects are trivial and can be ignored. However, results also suggested that interviewer effects could be minimized by not hiring interviewers who expect a study to be difficult and by not training interviewers to expect underreporting. Tables and nine references are provided.