An Item-Level Examination of a Youth Assessment: Practical Implications of Alternate Measurement and Analytic Procedures

Abstract

ABSTRACT To date, research has focused on the scale-level properties of juvenile risk assessments, such as their predictive validity across populations. Although these studies are valuable, they provide little item-level information to inform future developments in assessment strategies. Item-level information, including the difficulty and discrimination parameters of the items that comprise the scale, provides insight into patterns of endorsement across risk levels. The current study examines the item-level characteristics of the Ohio Youth Assessment System-Disposition Tool (OYAS-DIS) in a sample of 459 justice-involved youths across a Midwestern state. Using the two-parameter logistic model of Item Response Theory, discrimination and difficulty parameters were estimated for each item that comprises the OYAS-DIS. A test information curve was also examined to determine the levels at which the OYAS-DIS provides the most information. Results indicated that prosocial decision-making items had low difficulty, meaning that even low-risk youths endorsed them. Overall, OYAS-DIS provides the most information for low- to high-risk youths, but additional items should be created to examine prosocial decision-making, as these items may be characteristic of justice-involved youths as a group and not an indicator of greater recidivism risk.