Privacy and Differences among Methods of Self-Administration

Mick Couper, Eleanor Singer & Roger Tourangeau

University of Michigan & University of Maryland

Abstract

Self-administered surveys generally reduce reporting errors on sensitive topics relative to interviewer administration, but there is mounting evidence that the gains from self-administration depend on the exact method used and the data collection setting. Several survey studies find differences among methods of self-administration-for example, between audio computer-assisted self-interviews (ACASI) and paper questionnaires. In addition, a growing body of evidence suggests that features of the interface of automated systems can evoke reactions similar to those triggered by live interviewers, including social desirability effects. We are currently conducting a study that compares three versions of a computerized questionnaire: an ACASI version that presents the questions both visually and aurally, a text-only version, and an audio-only version. The questions concern a variety of sensitive topics, ranging from exercise and diet to sexual behavior and illicit drug use. Respondents in the three mode groups completed the questions in a laboratory setting. Half the respondents in each mode group completed the computerized interviews under conditions of high privacy, with the experimenter waiting in a different room while the respondent answered the questions. The other half completed the questions under conditions of low privacy, with the experimenter present as the respondent completed the questions and several interruptions by a confederate posing as a computer technician. Our hypothesis was that the three modes would show reporting differences but only under conditions of low privacy. We present some preliminary findings from the study, focusing on respondents' subjective reactions to the mode and privacy variables.