Purpose: This study describes the development and predictive validity of the North Carolina Assessment of Risk – Revised (NCAR-R), a new risk assessment instrument for juvenile justice settings. The NCAR-R follows a trend in risk assessment of blending two assessment functions, risk assessment and needs assessment, into a single procedure. It was developed to improve the predictive validity of the North Carolina Assessment of Risk (NCAR).

Method: Administrative data on risk factors for recidivism was collected for a sample of 590 juveniles (32% female, 47% non-Latino White, 45% African American) adjudicated delinquent by the North Carolina juvenile courts. Recidivism status and length of time in non-detention out of home placements were collected during a follow-up period of up to one year. Receiver Operator Characteristic Curve analysis (ROC) and event history analysis were used to estimate the relationship between the NCAR-R risk score and recidivism.

Results: Predictive validity of the NCAR-R was greater than the NCAR (AUC=.692 vs. .660) and is comparable to other risk assessment instruments in juvenile justice, criminal justice, and violence prediction. Non-Latino White youths (AUC=.737) and males (AUC=.709) had higher AUC scores than African American youths (AUC=.676) and females (AUC=.650). Multivariate Cox regression models, estimated separately by gender, explain these differences. Results show that certain types of risk factors are less predictive of recidivism for African American youths and that length of time in out of home placement suppresses recidivism of higher risk females but not of higher risk males.

Implications for research and practice: The NCAR-R appears useful for classifying risk of recidivism of delinquent juveniles. Racial/ethnic differences in predictive validity may be explained by the disproportionate involvement of African American youth in the juvenile justice system. Gender differences appear to be associated with different patterns of intake offenses which make out of home placement more effective for females in reducing recidivism over short periods of time. These differences notwithstanding, the NCAR-R identifies juveniles at risk of repeat offending for all groups and identifies malleable risk factors that are potentially targeted in case planning.