Examining the Item-Level Performance of a State-Mandated Juvenile Risk Needs Assessment

• In Event: Can we Enhance Youth Justice and Policing through Assessment and Measuring Perceptions?

Abstract

Juvenile risk-needs assessment (JRNA) instruments guide legal decision-making such as whether to place youth on probation, the degree of probation surveillance intensity, and the level of service provision. As such, it is critically important JRNAs are well-developed and are upheld to the highest psychometric standards. Until recently, JRNAs have been evaluated primarily at the test level (e.g., reliability, predictive validity). Although test-level performance is certainly important, it is equally important that each item performs adequately. Using a sample of youth (n=2,361) on juvenile probation in a large Southwestern County, this study conducts an item-response theory assessment of a state-mandated JRNA tool. Examining the item-level characteristics of the JRNA tool, two-parameter logistic regression models assess item discrimination and difficulty. Potential differences in item-level performance across racial/ethnic groups were also assessed using tests of differential item functioning. Results have direct implications and recommendations for the calibration and use of the JRNA tool.

Authors

• Kelsey Tom, School of Criminology and Criminal Justice, Arizona State University

• Savanna Allen, School of Criminology and Criminal Justice, Arizona State University

• Adam Fine, Arizona State University