How well do indirect measures assess sexual interest in children? A meta-analysis.

Abstract

Objective: We quantitatively reviewed the construct validity evidence for all cognitively based indirect measures of sexual interest in prepubescent children (pedophilic interest) and pubescent children (hebephilic interest) using meta-analysis. Method: Studies were included if they presented scores on a cognitively based indirect measure of pedohebephilic interest for a sample of adolescent or adult males who had committed a sexual offense against a child 16 years of age or younger, or who reported sexual interest in children, and for a comparison group. Studies were also included if they reported on the strength of association between scores on an indirect measure and an independent indicator of pedohebephilic interest in a sample of males. We used meta-analysis with robust variance estimation to summarize effect sizes and metaregression to test potential moderators. Results: Cognitively based indirect measures of pedohebephilic interest showed a moderate difference between pedohebephilic (n = 2,552) and nonpedohebephilic males (n = 2,434), d = 0.61, 95% CI [0.46, 0.76], k = 39. A small-to-moderate correlation was also observed between indirect measures and independent indicators of pedohebephilic interest, r = .23, 95% CI [0.17, 0.28], k = 23, n = 3,623. These effects were qualified by substantial heterogeneity; however, most moderators we tested did not account for a significant amount of heterogeneity. Conclusions: Findings suggest that publication bias did not substantially distort the results. However, the lack of significant moderators suggests more research is needed to understand the conditions under which indirect measures best reflect pedohebephilic interest. Public Health Significance StatementThe body of evidence on cognitively based indirect measures of sexual interest in children (e.g., viewing or reaction time measures) suggests that they are promising and worthy of continued research, but more evidence is required before we can rely on these measures alone to make decisions that impact public safety or civil liberty.

Neurodevelopmental Differences, Pedohebephilia, and Sexual Offending: Findings from Two Online Surveys

“…Furthermore, with exception of the VT measure and the IQ test, the online environment forced us to rely completely on self-report, which puts restraints on the validity of the data assessed. The tendency of men to over-report their height is well-documented (Gorber et al, 2007 ), and pedophilic and teleiophilic men who have committed sexual offenses are no exception to this rule: Fazio et al ( 2014 ) report differences between actual and self-reported height varying between 2 and 4 cm, similar to differences observed in non-clinical samples (e.g., Palta et al, 1982 ). This means that the data we obtained on height in both our pedophilic and teleiophilic samples are likely to be biased.…”

“…For further details on how we dealt with statistical outliers see the section on VT classifications. Viewing time measures of pedohebephilia have repeatedly been shown to produce reliable (Welsch et al, 2021 ) and valid results (Pedneault et al, 2021 ; Schmidt et al, 2017 ).…”

Viewing Time Measures of Sexual Interest and Sexual Offending Propensity: An Online Survey of Fathers

“…Penile plethysmography is still subject to conscious manipulation and is also more expensive and intrusive (Wilson & Miner, 2016). As a result, researchers have developed indirect methods to assess sexual interests, including relative viewing time (VT) measures (Pedneault et al, 2021;Schmidt et al, 2017).…”

“…This practice is based on the assumption that a history of sexual offending against children can be a proxy—albeit imperfect—for sexual interest in children, which is what VTs ultimately aim to detect. VTs have been shown to moderately discriminate men with compared to without a history of sexually offending against children, and to have small-to-moderate convergent validity with self-reports, penile plethysmography, and other indirect measures of sexual interest in children (Pedneault et al, 2021 ; Schmidt et al, 2017 ). Further supporting the construct validity of VT measures, one study found that VTs for child stimuli significantly predicted sexual recidivism among men who had sexually offended (Gray et al, 2015 ).…”

Testing Causal Explanations of Viewing Time Measures of Sexual Interests

“…Over the last 25 years, numerous studies have shown that individuals' unobtrusively measured viewing times are prolonged for sexually preferred (versus non-preferred) stimuli when they are instructed to rate pictures of persons in terms of their subjectively perceived sexual attractiveness (e.g., Banse et al, 2010;Dawson et al, 2012;Ebsworth & Lalumière, 2012;Harris et al, 1996;Israel & Strassberg, 2009;Jahnke et al, 2021;Larue et al, 2014;Lippa, 2017;Oberlader et al, 2017;Petterson et al, 2018;Rönspies et al, 2015;Quinsey et al, 1996;Xu et al, 2017). This basic finding-in the literature commonly referred to as viewing time (VT) effect-is so robust, that it is frequently used as an easy-to-apply latency-based indirect measure of individual differences in sexual interests, particularly, in forensic contexts where self-report measures are not the most viable option due to positive self-presentation concerns (Schmidt & Banse, in press; for systematic overviews see Pedneault et al, 2021;Schmidt et al, 2017).…”