| Validity of self reports is an enduring problem in HIV research, especially with injection drug users. Recently, audio computer assisted self-interviews (ACASI), have been shown to elicit higher reports of HIV behaviors. Higher frequencies are assumed more valid and are attributed to ACASI's ability to address socially desirable responding. We sought to prove the validity of ACASI using construct validation, and to evaluate the utility of ACASI as a gold standard in determining the extent of mis-reporting on traditional IAQ methods.;A randomized study design was used, drawing participants from a longitudinal study that has been investigating HIV in injection drug users since 1988. Participants at their follow-up visit were randomized to either IAQ or ACASI, stratified by gender, HIV status, and entry into the cohort. Based on the hypotheses that socially desirable responding is both addressed by ACASI, and HIV positive individuals are more prone to it compared to HIV negative individuals, we compared IAQ and ACASI reports, using univariate, stratified, and logistic regression analyses, including an interaction term (ACASI*HIV). In addition, to quantify mis-reporting on IAQ by gender, we calculated percent reporting errors for drug use and sexual risk behaviors (% IAQ--% ACASI).;1142 participants were enrolled, 586 on IAQ, 556 on ACASI. Randomization was successful in the similarity between the two groups regarding mean age (43 yrs), race (95.2% African American), gender (71% male), and HIV (34.4% HIV positive). HIV positive respondents were significantly less likely to report preventive behaviors on ACASI compared to IAQ than HIV negative respondents. Both gender were significantly more likely to report commercial sex work on ACASI vs. IAQ; males regarding selling it, and females buying it. Males tend to underreport more drug use behaviors than females, particularly high risk behaviors. Males tended to overreport both sex and condom use, while women were just as likely to over as to underreport them.;ACASI identified both differential subgroup reporting (including HIV positive and male individuals) and mis-reporting differences by gender; thus evidence of its validity. Future studies are necessary to evaluate ACASI's relative validity compared to IAQ's when administered longitudinally. |
