Recidivism reduction is the primary goal of many correctional programs, and "recidivism" is the most prevalent outcome measure in related program evaluation research. Many different operationalizations of recidivism are used without a clear delineation of how these variations may impact conclusions. This study explores how the definitions of recidivism may impact research findings and resultant policy recommendations regarding the efficacy of parole.

Data from prisoners released in 2008 (n = 12,132) to parole or unconditional release are analyzed according to 10 different operationalizations of recidivism. We compare recidivism rates, time to failure, and hazard rates between groups through the presentation of descriptive statistics and the use of multivariate Cox proportional hazards survival models.

Our findings indicate that parole supervision could be deemed either effective or ineffective depending on which definition of recidivism is employed. These findings are largely driven by whether technical parole violations are included into more traditional criminal outcome measures, such as rearrests, reconvictions, or reincarcerations for new crimes, and if court processing times are factored into measures of time to failure.

Our results raise questions about the consistency of findings within the corrections literature. These conclusions, given the role that technical violations and court processing times can play, suggest a need for increased specificity when using recidivism as an outcome measure.