The Validity of Implicit Association Test (IAT) Measures of Sexual Attraction to Children: A Meta-Analysis

Abstract

The current study presents a quantitative review of the discriminative and convergent validity of Implicit Association Test (IAT) measures adapted to assess sexual interest in children. IAT measures were able to distinguish sex offenders against children (SOC) from non-SOC (M weighted d from random-effects = 0.63, 95 % CI [0.42-0.83], N = 707, k = 12). The largest group differences were found between SOC and non-offenders, followed by non-sex offenders and rapists. IAT measures using sex versus not sex (and similar attribute categories, such as sex vs. neutral) provided superior discrimination compared to IAT measures using sexy versus not sexy (and similar attribute categories, such as erotic vs. non-erotic). The IAT measures had a moderate relationship to self-report (r = .27, 95 % CI [.13-.40], N = 182), sexual offense history variables (r = .27, 95 % CI [.08-.43], N = 145), and viewing time (r = .30, 95 % CI [.16-.43], N = 180) measures of sexual interest in children. Although these IAT measures can discriminate between groups and show convergence with other measures of sexual interest, a better understanding of the construct validity of these tools is required before their use in the assessment, treatment, and supervision of sex offenders.