The use of couple data to determine the reliability of self-reported sexual behavior

Unreliable self-reporting of sexual behavior has been suggested asa major c ause of variance in survey research. In addition, differing gender-related biases are thought to account for discrepancies in reports of oral and anal sex. Insufficient attention, however has been given to assessing the relia bility of the sexual partner's reports. Data obtained from 70 couples conce rning 68 sexual behaviors was reviewed Four different reliability indices y ielded moderate to substantial agreement. These included a simple percentag e criterion of 87%, a Kappa of .67, and conditional probabilities of .77 an d .91. Previously reported discrepancies in the reporting of oral and anal sex were not found. Kappa was found to be superior to simple percentage agr eement but produced overly conservative reliability estimates when derived from behaviors that occur at either high or low base-rates. The two conditi onal probability indices are proposed for use in conjunction with Kappa to evaluate the reliability of partner data.