Influences of Survey Instrument Format and Social Desirability on the Reliability of Self-Reported High Risk Sexual Behavior

Abstract

This study examined instrument-related and respondent-related self-presentation effects on the reliability of HTV sexual risk self-reports. Sixty-five men who have sex with men (MSM) were randomly assigned to assessment conditions differing only on the format for questions about their sexual behavior. For Group 1, the instrument tested responses to items directly addressing the issue of unsafe sex; men reported absolute frequencies of sexual activities engaged in “with” and “without a condom.” For Group 2, questions were framed more positively with respect to condom use; participants were instructed to report total number of intercourse occasions, and then indicate on a scale the relative frequency (percentage) of times “condoms were used.” It was anticipated that Group 1 participants would feel more inhibited about reporting episodes of unprotected sex, particularly unprotected anal intercourse (UAI). The Marlowe–Crowne scale was used to control for general tendencies toward socially desirable responding within the sample. Results confirmed that men in Group 1 tended to report lower rates (percents) of UAI, and that, among participants with higher frequencies of anal intercourse, this difference was statistically significant. Weighted least squares regression analysis showed that hypothesized indicators of measurement error—assessment group and social desirability—together accounted for more than half of the explained variance in percent UAI. The findings suggest that HIV prevention researchers need to give higher priority to methodological issues involved in the data collection process. Also, more research is needed on the implications of socially desirable responding for the reliability of self-reports of HIV sexual risk behavior.