Best Practices for Neglected Assumptions: Multi-Site Confirmation of the MPACT-6

Abstract

In the assessment world, risk determines “who” to treat and needs determine “what” (Bonta & Andrews, 2016). Yet, for youth, greater emphasis is placed on needs that target recidivism reduction interventions. Unlike risk, needs represent dynamic domains, or latent constructs, requiring testing to assure adequate measurement. We conducted a multi-level, multi-group analysis of the Modified Positive Achievement Change Tool (MPACT) with a 10-state sample of youth ( N = 258,464). Findings confirm the validity and reliability of needs domains, and the development of a novel “Global Needs Factor,” a composite summary of needs. Created without criminal history/static measures, needs scales demonstrate predictive accuracy. Further, measurement invariance and aspects of gender and race/ethnicity prediction parity are observed.