Gender-of-voice effects in an ACASI study of same-sex behavior

Abstract

Audio computer-assisted self-interviewing (ACASI) is a popular method for administering surveys of sex and other sensitive behaviors. One advantage of ACASI is that it removes the requirement that respondents divulge sensitive behaviors directly to another person. Thus, ACASI may reduce the extent to which interviewer effects compromise response accuracy. However, the literature on computers as social actors suggests that even subtle humanizing cues, such as the gender of the computer-generated voice, may cause users to react to the computer as they would to another person. The present randomized experiment examined whether the gender of the ACASI voice affects the reporting of socially undesirable HIV-risk behaviors in a sample of 405 men who have sex with men across 12 US cities. We randomly assigned participants to hear either a male or a female recorded voice. We tested for gender-of-voice effects on reports of same sex behaviors and number of sexual partners. The male ACASI voice elicited fewer instances of unprotected receptive anal sex with an HIV-status-unknown partner (p = .002), whereas the male voice elicited marginally higher numbers of HIV-negative partners in the past 30 days (p = .052). Overall, our results suggest that interviewer effects stemming from the gender of the ACASI voice were minimal in our study. The significant (or marginally significant) effects that we did detect are consistent with findings in prior research and seem to suggest that a female voice may elicit more accurate reports.