The reliability and validity of drug users' self reports of amphetamine use among primarily heroin and cocaine users.

Relatively few studies have addressed the psychometric properties of self-report measures of amphetamine use. This study examines the reliability and validity of the Risk Behavior Assessment's (RBA) lifetime and recent amphetamine-use questions. To evaluate validity, 4027 out-of-treatment primarily cocaine and heroin users provided urine samples that were compared to self-report data; to evaluate reliability, 218 completed the RBA at two time points, 48h apart. In the overall sample, self-reports demonstrated moderately high validity, with a 95% accuracy rate (kappa=.54). When analysis was restricted to recent amphetamine users validity was slightly lower (71.5% accuracy; kappa=.41). Test-retest data indicated good reliability for self-reports of ever having used amphetamine (kappa=.79), and amphetamine use in the past 30 days (.75<r<.91). Out-of-treatment drug users provided accurate self-reports of amphetamine use. Reliable and valid measures are essential for describing and predicting trends in amphetamine use, evaluating the effectiveness of interventions, and developing policies and programs.