Psychometric Issues in SVP Risk Assessment

Abstract

Predicting sex offenders’ risk of recidivism is critical in the United States, where courts can commit high-risk sex offenders as sexually violent person (SVP). We address psychometric issues in sex offender risk assessment that have emerged in the published literature, professional forums, and in our own forensic practice. We consider absolute risk thresholds, prediction, single-case probability, reference groups, and margin of error. We present a Bayesian view of risk assessment and contend that it is the most coherent way to both define and quantify individual’s risk of sexual recidivism.