Juvenile diversion is offered in most counties throughout Nebraska to eligible youth; and although state guidelines require the use of a screening or assessment tool, the tools utilized are not standardized or uniform statewide. This report quantifies whether the various tools are being reliably administered and are effectively predicting diversion completion and future system involvement. The Juvenile Justice Institute gathered item-level risk/needs screener and assessment data from all juvenile diversion programs receiving Community-based Aid (CBA) funds. Overall, 3,916 youth were assessed for a juvenile diversion program between July 1st, 2015 and June 30th, 2017. The Youth Level of Service Inventory/Case Management Inventory (YLS) comprised the largest number of completed assessments (n = 2,193), followed by the Nebraska Youth Screen (NYS; n = 1,512), and the Arizona Risk-Needs Assessment (ARNA; n = 211). First, we tested the reliability of each tool, which is how well it is consistently performing at predicting risk (i.e., less error in measurement). Reliability analyses revealed the YLS/CMI had the strongest internal consistency of the three measures, which means the items are grouped well together to measure the construct (i.e., risk level). The items within the NYS, however, demonstrated the strongest item-total correlations, which means these items were most related to the overall construct (i.e., risk level). Both the NYS and ARNA had poor internal consistency. Second, we performed Receiver Operating Curve (ROC) analyses to determine the predictive validity of each tool, utilizing both unsuccessful discharge from diversion and future system involvement as outcomes. Results revealed all three tools had predictive validity for unsuccessful diversion completion with large effect sizes (i.e., measure of strength of the relationship). Furthermore, while all three tools demonstrated predictive validity for future system involvement with small to moderate effect sizes, when we tested predictive validity by both gender and race/ethnicity, only the YLS accurately predicted future system involvement for girls, none of the tools accurately predicted future system involvement for Black/African American youth, and only the ARNA accurately predicted future system involvement for Hispanic youth. While it is always recommended to screen and/or assess youth, the tools currently being utilized in Nebraska juvenile diversion programs are not reliably and validly measuring risk for all youth assessed. The most problematic items within each tool were those relating to prior convictions or prior contacts with the legal system. Because this is a diversion population, presumably with little to no prior juvenile justice system involvement, these items poorly capture risk in this population, which in turn contributes to lower reliability and predictive validity. While these findings do not provide definitive results for us to whole-heartedly recommend a tool for juvenile diversion programs at this time, our recommendation is to explore creating/utilizing a risk assessment tool that removes items that measure previous legal system involvement or norming current tools without those items. Future research and practice should continue to explore gender and racial/ethnic differences within youth assessment.
Rufino, Katrina; Wylie, Lindsey; Harris, Chelsea W.; Carpenter, Jordan; and Juvenile Justice Institute, University of Nebraska at Omaha, "Reliability and Predictive Validity of Screener/Assessment Tools in Nebraska Juvenile Diversion" (2018). Reports. 11.
Files over 3MB may be slow to open. For best results, right-click and select "save as..."