Assessing the predictive validity of the Network Inventory of risk, victories, and needs assessment (NIRVANA) for youth receiving prevention services

The current study assesses the predictive validity of a newly-developed holistic risk/needs assessment for youth receiving prevention services. Employing a multiyear statewide sample of over 17,000 prevention youth, multivariate and Area Under Curve techniques are used to examine predictive validity in predicting future formal juvenile justice system involvement (new law violations) across sex, race/ethnicity, and multiple youth age ranges. Furthermore, predictive validity across the severity of the subsequent offense type (misdemeanor, non-violent felony, violent felony) was examined. Findings demonstrate large effect sizes for static risk only, dynamic risk only, and total risk plus protective scores on future offending. Sex and race/ethnicity differences in predictive validity were not evident, through Area Under Curve or multivariate interaction models, while some evidence was found that the NIRVANA performed better for youth under the age of twelve at the time of assessment. Predicted probability analyses reveal stronger prediction of subsequent misdemeanor and violent felony offenses than non-violent felonies, net of demographic controls. Policy implications surrounding the feasibility of holistic risk/needs assessment for youth receiving prevention services are discussed.