We examined statewide data of persons evaluated for sexually violent predator (SVP) commitment and compared risk-relevant data of three groups: those committed as SVPs (n = 374), those not recommended for commitment (n = 2,707), and those nearly committed (recommended for commitment but ultimately not committed; n = 117).
Consistent with legal language for SVP commitment, binary and multinomial regression analyses revealed that risk scores predicted SVP commitment recommendations in addition to some historical factors (e.g., psychiatric history, never being married, prior sex offenses). For those recommended for commitment, prior sexual offenses predicted ultimate commitment. Those nearly committed had significantly higher sexual recidivism rates than did others who were not committed; however, these recidivism rates were still low (11.5%). Findings suggest that evaluators’ SVP decisions incorporate risk data and follow empirically supported trends; however, the observed recidivism rates of a subset of people that SVP commitment appears to target suggests that the potential for reducing sexual recidivism effectively and efficiently through commitment appears to be quite limited.
- The great methods bake-off: Comparing performance of machine learning algorithms
- Developing Knowledge about What Works for Make Schools Safe: Implementation and Evaluation of Tools for Life in Jackson Public School District in Mississippi
- Examining Criminogenic Risk Levels Among People with Mental Illness Incarcerated in US Jails and Prisons