UCLA

This paper examines the concordance of reported shared sexual behaviors, including condom use, among 535 heterosexual, African American, serodiscordant couples and identifies factors that might predict discordant reports. Percentages of agreement, Kappa and McNemar's statistics and conditional probability indices are used to measure concordance. Logistic regression models identify predictors of couples' discordant sexual reports. Analyses revealed Kappa statistics for reporting anal sex, fellatio and cunnilingus indicated moderate to substantial agreement. The effects of demographics and the couples' relationship contexts on concordance of reported sexual behaviors were found to vary somewhat by gender and type of sexual behavior. Findings showed that concordance of reporting between the couples was consistent for the past 90 and 30 days. Findings from this paper provide new scientific insights into the knowledge base of self-reported couples' data and suggest that these data can be used to evaluate their accuracy and serve as a proxy for validity.