Improving the measurement of criminal sexual behavior: the application of randomized responding technique

Randomized responding technique (RRT), a method for maintaining anonymity, was used with a sample of sexual offenders recruited into a treatment outcome study. Prior to release from incarceration, three groups--those randomly assigned to treatment, those randomly assigned to no-treatment control, and those who refused participation in the treatment study but consented to a prerelease interview--participated in an alternate-questions RRT procedure. This procedure pairs a nonsensitive question with the sensitive question of interest, in this case, the number of prior sex offenses. Respondents answered either the sensitive or nonsensitive question, depending on the results of a randomizing device (roll of dice). The distributions for the nonsensitive questions were highly skewed; therefore, the outliers were removed and RRT estimates calculated. RRT estimates of prior offending (2.20 prior offenses) were significantly higher than officially recorded prior offenses (0.51 prior offenses). The pattern of differences between treatment and control groups were similar in RRT estimates and officially recorded priors, as were the correlations between RRT estimates and other self-report scales. These results provide preliminary evidence that RRT is a useful method for generating reoffending data that are more sensitive than officially recorded offenses and that contain less bias than other self-reports.