Using Latent Transition Analysis to Explore Changes in Risk Profiles of Youth on Probation

Background and Purpose: While scholars have focused on classifying youth based on risk for recidivism through standardized risk assessments, much less attention has been paid to how scores on risk assessments change. Some studies have described changes in risk using traditional variable-centered approaches. We extend current knowledge of changes in risk during probation by using a person-centered approach. This study sought to identify distinct subgroups of youth on probation at intake and exit and the probability that youth will transition between groups over time.

Methods: Data were collected at intake and exit using the Youth Assessment and Screening Instrument (YASI) administered to youth on probation (N=7,024) in a mid-Atlantic state. The YASI covers 10 domains (e.g., legal history, family, school, aggression, attitudes). We used 8 static risk, 2 static protective, 8 dynamic risk, and 7 dynamic protective subscales from the YASI. We dichotomized the subscales to indicate whether a youth was medium or high risk (=1) on each subscale. We used latent class analysis to identify subgroups at intake and exit, then estimated probabilities of moving between groups using latent transition analysis (LTA).

Results: Based on the LCA models at intake and exit, we estimated 5, 6, and 7 classes at both time points in our LTA. We selected the model with 5 groups at both intake and exit based on model fit indices (AIC and BIC) and considerations for distribution and meaningful groups. We characterized the five groups, in order of declining levels of risk: Highest Risk (HR), Social Drug Risk (SDR), Individual Risk (IR), Drug Risk (DR), and Low Risk (LR). While both the DR and SDR groups reported high risk in the alcohol and drug subscale, the SDR group also reported high risk in the family, school, and community and peers subscales. The IR group reported high risk in mental health, aggression, attitudes, and skills. Overall, the majority of youth remained in the same group at both intake and exit (78%). LR youth were the most likely to remain (96.8%) and HR youth were the least likely to remain (67.1%). The most likely transitions were from SDR to DR (19.0%), HR to DR (15.3%) and HR to SDR (12.2%).

Conclusion and Implications: Most youths remained in the same risk category at the end of probation, highlighting the difficulty in affecting risk profiles. It was encouraging to find that youth in the lowest risk categories were most likely to remain in that category while the highest risk group were most likely to transition between categories. It is also important to note that few youth moved directly into the LR category except youth in the IR category. This suggests that interventions should strategically seek to reduce risk for recidivism, realizing that it is unlikely youth will leave probation LR (unless they began LR). Future research should identify the services provided to the youth in these transition groups to understand the role probation plays in facilitating these transitions.