The Examination of Changes in Dynamic Risk Factors and Recidivism Over Time in Youth Offenders

Risk assessments that include dynamic risk factors are increasingly being utilized within the youth justice system to predict a young person's likelihood to reoffend, to assist with case management, and to better inform intervention services. Based on the principles of the Risk-Need-Responsivity (RNR) framework, the Youth Level of Service/Case Management Inventory (YLS/CMI) is a widely used risk assessment tool that demonstrates strong psychometric properties among youth offenders. However, most studies to date have relied solely on single-wave, cross-sectional research designs that essentially treat dynamic risk factors as static. The purpose of the present study was to determine whether the RNR dynamic risk factors assessed by the YLS/CMI do in fact change over time and whether identifying these changes improved the prediction of recidivism. To answer these research questions, 200 12-18 year-old youth were tracked from the time of their first risk assessment conducted at probation to their transition out of the youth justice system. Information pertaining to post-assessment re-offences was collected from official conviction records. A series of generalized linear mixed modelling and latent class growth modeling analyses were used to assess change in dynamic factors over time and to examine the effect of youth-specific predictors on these changes. A hierarchical Cox regression with time-dependent covariates and Receiver Operator Characteristic (ROC) analyses were used to assess the predictive accuracy of the YLS/CMI on recidivism rates. Results indicated that most dynamic risk domain scores increased over time but that there was significant individual variation among youth at initial status and in the rate of change. Even when controlling for youth-specific factors, results showed that lower risk youth at the time of initial assessment increased in risk at a greater rate than higher risk youth; this finding was unlikely related to ceiling effects since most youth fell in the moderate risk range. The inclusion of dynamic risk scores improved the predictive accuracy of recidivism over and above static risk factors and risk scores obtained from the initial assessment. Findings from the current study have implications for enhancing the RNR framework, improving the accuracy of risk assessments, and informing treatment implementation.Ph.D