Revising Test Items to Increase Validity

Eight female interviewers varied their interview styles in interviewing 207 adults who had been involved in automobile accidents during a four-year period. A “professional” and a rapport or “interpersonal” style was used. Dependent variables consisted of reporting of events related to an automobile accident in which the respondent was the driver, and for which independent criterion data were available. In addition, the amount of health information reported was also analyzed. In spite of the respondents' more favorable attitudes towards the interpersonal interviews, accuracy was not significantly greater in these rapport interviews. In fact, significant interaction effects between interviewer style and recency of the accident were obtained. Respondents, however, reported more completely under rapport conditions. The results are interpreted within a cognitive, role-learning framework.