Interrater reliability and concurrent validity of the sexual violence risk–20 (SVR–20) V2 in South Korean sexual offenders

The Sexual Violence Risk-20 Version 2 (SVR-20 V2) is a structured professional judgment tool for formulating intervention plans. This study investigated its psychometric properties in a South Korean context using 27 incarcerated male sexual offenders. Four forensic clinical psychologists independently rated each case. Interrater reliability ranged from ‘good' to ‘excellent’. A generalizability study revealed minimal variance attributable to raters or occasions, confirming high consistency and temporal stability. The largest variance source was person-item interaction, suggesting score variability was driven by idiosyncratic item responses. A decision study showed high reliability for relative decisions, but moderate for absolute ones. Concurrent validity was supported by strong correlations with the Psychopathy Checklist-Revised and the dynamic-focused Hallym Assessment Guide for Sex Offender Risk (HAGSOR), outperforming the static-based Static-99R. These findings support the cross-cultural utility of the SVR-20 V2 in Korea and emphasize the necessity of team-based consensus models to maximize risk judgment dependability.