Development and Validation of an Actuarial Risk Assessment Tool for Juvenile Sex Offenders

This grant is funded under NIJ's 2013 Basic Scientific Research to Assess Youth with Sexual Offending Behavior solicitation. This study proposes to advance the field of juvenile sex offender research and treatment by developing and validating a state-of-the-art actuarial risk assessment instrument that can effectively predict the risk of sexual recidivism for juveniles with a history of sexual offending and can be widely adopted and implemented by juvenile justice agencies. In partnership with four jurisdictions around the country, Pennsylvania, Virginia, New York, and Oregon, a pilot study will assess a minimum of 1,200 juvenile sex offenders whose sexual recidivism will be tracked over 12-18 months. For this 36-month project, led by a multidisciplinary team of researchers and in consultation with an external review panel, the project team will develop a prototype instrument that considers static and dynamic risk factors and protective factors of juvenile sexual offending. The project will develop prediction models in seven different forecasting frameworks, including classic multivariate regression and machine learning algorithms. Based on cross-validation and bootstrapping strategies, the instrument will be validated locally and globally. Furthermore, the project will assess a sense of the practical utility and external validity of the instrument by applying the risk models to additional jurisdictions that are not part of primary data collection. Results from the project will develop empirical knowledge and practical guidance regarding the risk classification of juvenile sex offenders. Research findings will be disseminated in a variety of forms for diverse audiences including but not limited to conference presentations, research briefs, journal articles, and interim and final reports to NIJ. Additionally, the fully validated instrument and detailed technical documentation will be made available to the public as well as a blueprint for the widespread, accountable distribution of the instrument through web training and self-test modules.ca/ncf