Scoring Algorithms to Optimize Viewing Time Measures of Sexual Interest in Children.

Viewing times (VTs) are established indirect measures of sexual interests. They are based on the effect that stimuli of sexually attractive target groups are scrutinized longer than stimuli of non-attractive groups in sexual attractiveness rating tasks. When assessing sexual interest in children and adults, decision latencies for target sexual maturity categories are typically compared by subtracting average latencies for children from average latencies of adults. In this study, several ways of improving the scoring algorithm were explored (e.g., using standardized differences, excluding the longest VTs of each category from the scoring, or conducting a profile analysis based on peaks in certain target categories). Scoring algorithms were compared in terms of reliability, convergent validity, and the influence of confounding variables in a secondary analysis of a German male community sample (N = 8554). Additionally, the influence of the scoring on the number of participants who would be classified as exhibiting (increased levels of) sexual interest in children was investigated. The VT task's validity proved to be quite robust to changes in scoring, but the reliability and classification proportions differed substantially between scores. The best performing score was the mean difference of adult and child target groups with the exclusion of the longest VT in each target category. The results imply that the scoring approach should be chosen based on the research or diagnostic goal of using the VT task (e.g., whether one prioritizes sensitivity or specificity). The study underlines the influence of measurement and scoring approach on prevalence estimates of (increased) sexual interest in children.