Realist Evaluation Of Youth Justice Services in England, Scotland and USA: Utilizing 100% Agency Data to Investigate What Works and for Whom in Reducing Recidivism

Purpose: Some of the court interventions that have shown positive results include teen courts (Harrison et al., 2001), community probation (Bechtel, Lowenkamp, & Latessa, 2007) as well as the use of drug courts (Rodriguez & Webb, 2004). These studies shave shown that some form of probation and early intervention across programs seemed to be the key to lowering recidivism.  This paper reports on 100% natural samples of the last 10 years’ data in a County Probation Service, NY; Youth Offending Team, England; and a Youth Justice Team, Scotland. Data analysis tools were from both the efficacy and epidemiology traditions to investigate patterns between outcomes, interventions and the contexts of practice.

Methods: A quasi-experimental design was used, comparing the achievement of  goals, in naturally occurring 100% samples from the same 10-year period in each of the three agencies. Both the retrospective data of the last 9 years and prospective analysis every three months in the 10^th year were included. Youth justice agencies in both USA and United Kingdom have invested in management information systems that record extensive data on the youth circumstances, the offences, the court dispositions, and the interventions.  Realist evaluation involves the systematic collection of data on 1) the offender circumstances (e.g. demographic characteristics such as age, race and gender; as well as contextual data such as mental health diagnoses, type of crimes, age at first, employment and education status); 2) the dosage, duration and frequency of the intervention received, e.g. the content and number of supervision sessions; and 3) the changes in the outcomes as observed through the repeated use of reliable outcome measures such as risk assessment scores, the number of arrests, and the number of alleged incidents of unacceptable behavior. Data analysis methods include the development of binary logistic regression models and regression discontinuity designs.

Results: 10-year data were analyzed in three youth justice agencies: a Youth Offending Team (England; n = 11,668) indicated that 46% had not offended again, as compared with a County Probation Service (n = 2416, New York State) where 76% had not offended again; and a rural Youth Justice Team in Scotland (n = 252) where 43% did not offend during the intervention. In all three agencies, predictors for not offending included offence type (e.g. burglary, motoring offences), program of intervention (e.g. diversion), baseline risk assessment, and demographic factors (e.g. age at first offence, alcohol and substance misuse).

 Implications: Conclusions and Implications

There is strong evidence from the three studies that the collaboration between the Courts and Probation  in Rensselaer County (NY state) has helped to divert the youth from crime and to reduce recidivism when compared with similar agencies in the United Kingdom. The evaluation 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.