The Effects of Sanction Intensity on Criminal Conduct: A Randomized Low-Intensity Probation Experiment

Probation is a well-established part of our criminal justice toolkit, but we know surprisingly little about the circumstances under which it is effective. Attempts to increase supervision intensity for crimeand cost-saving purposes have yielded mixed results at best. This dissertation examines the theory and scientific evidence on the effectiveness of probation, and the impact of changing the intensity of probation sanctions on recidivism. First, we conduct a rigorous search and synthesis of the existing literature on intensive probation programs. We utilize meta-analysis to identify the circumstances under which such programs might be effective. We find no evidence that probationers in these programs fare better than their counterparts under traditional supervision. We call for further research into supervision approaches that emphasize behavioral management over contact frequency and caseload size. Second, we employ a range of statistical procedures to examine the viability of saving resources by reducing supervision for low-risk offenders. In a randomized controlled trial comparing low-intensity probation to traditional practice, we find no evidence that reducing supervision increases recidivism. We find that low-risk probationers are heterogeneous in their characteristics but homogeneous in their propensity to reoffend. They appear to respond well regardless of the intensity of the sanction. Finally, we use epidemiological methods to evaluate the low-risk prediction model used in the experiment. We find that the model successfully identifies offenders who are unlikely to commit serious offenses, and is therefore a useful tool for diverting probationers to low-intensity supervision. In turn, lowintensity supervision is not associated with changes in offending severity. Chapters 2 and 3 both conclude that low-intensity supervision is a safe strategy that works very well for a probation agency’s lowest-level offenders. This dissertation contributes to knowledge by changing perceptions of the characteristics of offenders and resource allocation in criminal justice supervision. We find that ‘more’ does not always mean ‘better,’ and there is no need to distribute expensive services equally. In a given probation population, the majority of offenders will respond well no matter how little supervision they receive, so it makes sense to focus our attention on the minority that will not. Degree Type Dissertation Degree Name Doctor of Philosophy (PhD) Graduate Group Criminology First Advisor John M. MacDonald Second Advisor Lawrence W. Sherman This dissertation is available at ScholarlyCommons: http://repository.upenn.edu/edissertations/121 Third Advisor Paul D. Allison