Reliability and Validity of the RNR Program Tool

Research over the last 25 years has identified a number of features of programs for justice-involved individuals that consistently predict better outcomes. The challenge for programs comes in translating this research into useful practices for their specific context. This study describes the development and pilot testing of a web-based self-report program classification and assessment tool. The RNR Program Tool was designed to identify programs’ ideal target behavior, and to assess the degree to which program operations are consistent with the best available evidence on practices in the areas of risk, need, responsivity, dosage, implementation, and restrictiveness. The current study used data from 53 programs in 7 jurisdictions to test the internal consistency of the RNR Program Tool. Pilot testing of the RNR Program Tool for Adults found that programs had an average score of 58.7%. This finding is consistent with the research conducted by Lowenkamp, Latessa, and Smith (2006) using the CPAI in which they found that halfway house programs scored 45.5%. Findings also suggest that the RNR Program Tool has respectable internal reliability (α = 0.703 for the Program Tool Total Score). Additionally, exploratory factor analysis suggests that the RNR Program Tool includes two factors reflecting a treatment-matching or individual component, and a program operations or system’s level component. Future research based on this study should test