RESUMO
The Post Conviction Risk Assessment (PCRA) was developed for use in the U.S. Probation System in 2009. Although previous publications have demonstrated the predictive validity of the PCRA using development and validation samples, this study uses assessments completed by U.S. probation officers on 113,281 offenders during the course of supervision. The current research assesses the PCRA's validity in predicting arrest for any new criminal conduct and arrest for violent offenses at 6, 12, 18, and 24 months. Bivariate and multivariate models were estimated based on race, gender, and ethnicity. The area under the receiver operating characteristic curve (AUC-ROC) values ranged between .70 and .77 depending on the subsample, outcome being predicted, and follow-up time. Overall, this research indicates that the PCRA predicts equally well across race, gender, and ethnicity and for differing follow-up time periods. In addition, the PCRA is a valid predictor of arrest for any criminal behavior and arrest for violent criminal behavior.
Assuntos
Crime/estatística & dados numéricos , Criminosos/estatística & dados numéricos , Psicometria/instrumentação , Violência/estatística & dados numéricos , Adolescente , Adulto , Idoso , Idoso de 80 Anos ou mais , Seguimentos , Humanos , Pessoa de Meia-Idade , Reprodutibilidade dos Testes , Medição de Risco , Adulto JovemRESUMO
Offender assessment has been and remains the cornerstone of effective community supervision. This article presents the development of and tests the predictive validity of a 4th-generation risk assessment instrument designed for U.S. probation. A large administrative data set was used to create the assessment instrument and conduct an initial validation. Subsequent data generated from officer-completed assessments were used to conduct a prospective validation. Finally, data from case vignettes scored by trained officers were used to test the interrater agreement of the assessment instrument. Overall, analysis revealed that the assessment instrument predicted rearrest reliably when using the assessment results based on administrative data or officer-completed assessments. Analysis also revealed high rates of interrater agreement. Recommendations for future research and policy implications are presented.