Caution or Warning? a Validity Study of the Maysi-2 with Juvenile Offenders

Background and Purpose: Most detained youth, regardless of gender and race, have at least one diagnosable mental health disorder. Given these high prevalence rates, the juvenile justice system has an emphasis on identifying these youth and in 47 states the Massachusetts Youth Screening Instrument version 2 (MAYSI-2), has been adopted to accomplish this task. Given such wide use, ongoing empirical scrutiny is necessary to assess its continued usefulness. Therefore, the aim of this study is to further validate the MAYSI-2 by using LPA to explore two questions. 1) Will identified latent classes reflect existing MAYSI-2 cut-off categories? 2) Can the categorical approach of the MAYSI-2 be further validated by comparing identified subgroups across a range of external covariates?

Methods: This study uses data originally collected to explore the usefulness of the California Youth Authority’s (CYA) Treatment Needs Assessment battery designed to assess the mental health problems of their institutionalized delinquent population. Respondents were wards who completed screening, were subsequently placed in mental health programs, prescribed psychotropic medications, and/or identified by staff as requiring those services. Assessments were machine scored and supplemented official mental health records. The data included: demographics, prior juvenile history and juvenile justice system involvements, psychopathology, and problem behaviors, type of commitment, and official misconduct records. Respondents (N = 836) were committed to the CYA between 1997 and 1999. The sample is 81% male; mean age of 16.88 years. The race and ethnicity breakdown was as follows: Hispanic (N = 375, 46%), African American (N = 226, 28%), and White (N = 140, 17%), and Asian, Pacific Islander, American Indian, and Other (N = 72, 9%). Finite mixture modeling was employed to identify distinct latent classes based on MAYSI-2 scale scores. Identified classes were then compared across a range of covariates.

Results: We identified clusters of youth in a three-class gradient structure: Low Distress (n = 380), Moderate Distress, (n = 327), and High Distress (n = 129) and the latent classes mapped onto the range of scores representing the three MAYSI-2 scales cut offs: Below Cut-off, Caution, and Warning. External validation of the three classes was demonstrated in analyses comparing the three classes across covariates: demographic factors, juvenile justice system involvement, conceptually similar measures, and mental health treatment and psychotropic medication use.

Conclusions and Implications: Our analysis essentially found the MAYSI-2 based latent classes follow a severity gradient (i.e., ordinal-like pattern) suggesting that the Caution or Warning categorical distinctions may not provide additional information over and above a relatively high or low score. However, the current MAYSI-2 scoring format is user friendly for practitioners; as such our results support the continued use of the MAYSI-2 as a mental health screener. The MAYSI-2 is not designed to and does differentiate youth related to offending patterns, but the ability to understand such an intersection could be useful for practitioners in their decision-making process. Therefore, future research should explore whether the MAYSI-2 can be strengthened to address this area of need.