Realist Evaluation of Manchester Youth Justice Service in England: Continuous Repeated Analysis of Big Data to Investigate What Works and for Whom in Reducing Recidivism

Purpose: Although factors such as sex, age, race, substance abuse, family structure and past criminal behavior play a significant role in juvenile delinquency  (Ford, 2005), these don’t tell the full story and researchers continue struggling to find effective means of intervening.  Studies have shown that some form of early intervention across programs seemed to be the key to lowering recidivism (Sedlak & Bruce, 2010).  This paper reports on 100% natural samples from 2 years, (followed up in a further two years) to investigate what works and for whom in reducing recidivism.

Methods: A 100% naturally occurring sample was selected of all service recipients in the two years 2011 and 2012, and then followed through into the next two years, to investigate recidivism and what factors were predictors of returning to MYJS following the first episode of the intervention.   A quasi-experimental design was used, comparing recidivism amongst those who completed programs based on desistance theory and those that did not. Data analysis methods included the development of binary logistic regression models to investigate the factors that were predictors for reducing recidivism.

Results: The 2011 and 2012 MYJS service recipients (n = 1052) were followed into the next two years, 2013 & 2014.  It was found that 59% of this cohort had not re-offended. It was also found that the predictors for recidivism included age at first offence, sex and the baseline risk assessment score. The predictors for not returning to crime included the early intervention services, the completion of the MYJS desistance theory program, and whether the youth were living with parents/relatives. Males were generally more likely to return, but the youth living with parents and/or relatives were more likely not to return. It was found that completing the MYJS Program of Intervention (based on the desistance theory) was an important predictor for not returning to recidivism in the two years following the intervention.

Conclusions and Implications

There is strong evidence from the study that early intervention helps to divert the youth from crime and to reduce recidivism, but more needs to be done to develop effective programs. The repeated evaluation analysis of big data including entire populations directly helps the agencies to better target their interventions, and to develop new strategies in the circumstances where the interventions are less successful. This paper demonstrates how partnerships can be built between evaluators and service providers to develop program designs, evaluation designs and information designs to enhance the utility of evaluation in practice in a longitudinal study that includes regular analysis of data undertaken with the service providers. The fact that race and ethnicity were no longer predictors emphasized the MJYS commitment to equal opportunity, equity and justice, and demonstrated how continuous analysis of big data can help agencies to better strive to achieve their goals.