Recidivism, sanctioning, structural equation modeling, restorative justice


With the rapid growth of juvenile offender diversion programs, which use many nontraditional sanctions, the effectiveness of sanction combinations in juvenile diversion programs and in each individual program needs to be evaluated. Those making sanctioning decisions currently do so based on intuition rather than using an evidence- or theory-based approach. Considerable research has examined the relationship between offender risk factors and recidivism (who is more likely to reoffend?) and between offender risk factors and sanctions (who is more likely to receive what sanctions?), but little is known about the relationship between sanctions and recidivism (which sanctions best reduce recidivism and for whom?). Furthermore, recidivism studies vary drastically in how they measure or quantify recidivism. This variability of approach makes comparing studies difficult and provides a less-than-complete picture of recidivism in general. The present study used data from one specific youth diversion program to test certain hypotheses of sanctioning by developing and testing a model for assigning sanction combinations to certain offenders on the basis of their individual characteristics. The study first developed measurement models for Offender Risk Propensity, Multiplicity of Sanctions, and Recidivism using structural equation modeling (SEM). Then predictive models were developed to test specific relationships. Understanding the effectiveness of certain sanction packages on certain offenders can form the basis for effective sanctioning in youth diversion programs. This study sought to answer three research questions: What is the best way to measure recidivism? Does completion of a restorative justice program reduce recidivism? Which sanctions, if any, reduce recidivism for specific offender types? To answer the first question: a iv multi-indicator latent construct of recidivism did a very good job of measuring variation in recidivism. Multiple indicators analyzed simultaneously produced a robust tool that can be used in other recidivism studies and help to reduce comparability issues between studies. The recidivism construct, when tested as a function of completion of the restorative justice program, was seen to produce a significant model having an overall good fit with the data. Thus to answer the second research question: offenders’ completion status for the restorative justice program was shown to be a significant predictor of the latent construct of recidivism at the 0.05 level (two-tailed), with those who failed to complete (or chose not to participate) having higher recidivism than did those who completed the program. To answer the third research question: the assignment of specific sanctions (both those suggested by research and theory and those traditionally assigned by this and similar programs) on the entire data set (and on various subsets) of this study have no statistically significant impact on recidivism at the 0.05 level (two-tailed). The findings suggest many policy implications. Consistency is all but nonexistent in recidivism measurements in the academic literature and in program review studies. A multiindicator latent construct of recidivism, such as the one proposed and proven effective in this study, provides a more complete picture than simply conceptualizing recidivism by one dummy variable. This recidivism model can be used as the endogenous variable to evaluate programs and their practices and could reduce the problem of study comparability. This could lead to a better understanding of program characteristics and their impact on offender success. This study also found that completion of the Neighborhood Restorative Justice Program was a significant predictor of recidivism, yet none of the eleven most commonly assigned v sanctions were seen to have a significant impact on recidivism for any subgroup. Proponents of restorative justice argue that it is the programs’ characteristics and not their specific activities that make the programs successful. Reintegrative Shaming Theory and Labeling Theory support this claim and suggest the best approach to address youth criminal behavior is to admonish the act and not the actor, have the offender and community agree on a plan to make the community whole after that criminal act, and prevent repeated interaction with the formal criminal justice system which encourages the youth to see themself as a deviant and engage in further deviant behavior. These characteristics should be further examined and widely employed if confirmed.


If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at

Graduation Date





Wan, Thomas T. H.


Doctor of Philosophy (Ph.D.)


College of Health and Public Affairs

Degree Program

Public Affairs








Release Date

August 2013

Length of Campus-only Access

1 year

Access Status

Doctoral Dissertation (Open Access)


Dissertations, Academic -- Health and Public Affairs, Health and Public Affairs -- Dissertations, Academic