Is Something Better Than Nothing? The Effect of Short Terms of Mandatory Parole Supervision

Abstract

This article reports the findings of a quasi-experimental evaluation of a mandatory early parole program. New Jersey’s Statutory Early Release (SER) law required that all inmates, with few exceptions, be released onto parole six months before the expiration of their sentence. SER participants (n = 405) were matched using propensity scores to two groups of similar offenders who left prison in the year prior to the implementation of the SER statute. Comparison groups include former inmates that were released either unconditionally (n = 4,507) or onto standard parole (n = 5,657). At the conclusion of the matching routines, 395 SER-parolee pairs and 394 SER-unconditional release pairs were identified. Outcomes measures, including comparisons of rearrest and reconviction rates, are reported pre and post matching, and at 6-, 12-, and 18-month intervals. Individuals enrolled in the SER program demonstrated significantly reduced arrest rates during the SER supervision period when compared to unconditionally released offenders, though differences do not persist past six months. Offending rates for both SER participants and matched parolees were generally similar, with some significant variation in arrest rates emerging after 12 months.

Keywords:

• parole
• early release
• unconditional supervision
• propensity score matching
• community corrections

Notes

¹ Post hoc analyses revealed that SER parolees were approximately 4 times more likely to be transitioned through residential community programs when compared to traditional parolees after matching on pertinent predictors of SER participation. This suggests that increased program participation may have been a de facto, albeit unofficial, component of the SER “treatment.” While the Parole Board could have deployed its community programming resources to mitigate the risks posed to the community by members of the SER program, we are hesitant (and unable) to include a detailed discussion of the Board’s casework decisions during the course of supervising the different parole groups. We cannot use program referral data to inform our propensity scoring-and-matching routines because, in our research setting, the decision to transition an inmate through a residential program is made temporally after the attachment of both the SER and traditional parole statuses.

² We refer to this group as “parolees,” rather than “discretionary” or “mandatory” parolees. The SPB’s data that was used for this study did not include an indicator of whether a given parolee was released via a discretionary decision or statutory mandate. Hence, the parole comparison group likely represents a mix of discretionary and mandatory releases, and we are not able to precisely specify the level of mixture. However, the inclusion of the manifold background factors within the propensity scoring-and-matching routine controlled for many underlying differences between the groups of interest that were likely inclusive of factors associated with the different types of parole release in the state (e.g. instant offense types, criminal history, and a proxy for severity of prior offending). Further, after controlling for similar covariates, Ostermann (2011a) found no statistically significant differences in the likelihood of experiencing new arrests or convictions within three years of release between mandatory and discretionary parolees. Thus, the potential mixture of different parole release mechanisms (i.e. discretionary and mandatory) within the larger parole group would likely not have marked impacts upon our findings.

³ Our recidivism measures do not include parole revocations for technical violations because the data abstracting system from which we gathered our recidivism data does not include information about technical violations of parole. Both our SER program participants and parolee comparison group cases could have experienced technical parole violations during the course of their supervision. While we were not able to obtain individual-level technical parole violation information for our cases, the Parole Board was able to produce a report that presented rates for our matched SER and parolee comparisons across the follow-up times that we explored. The rates of technical parole violations did not exceed 10% for either group and the rates between the matched groups did not differ at a p ≤ .05 level at the 6-, 12-, or 18-month follow-up periods (particular results are available upon request). As a result, it is unlikely that the inclusion of individual-level technical parole violation information would lead to substantively different findings.

⁴ Different specifications of caliper distances, matching routines, and one-to-many matching led to substantively similar findings.

Additional information

Notes on contributors

Michael Ostermann

Michael Ostermann is an assistant professor at the Rutgers University School of Criminal Justice. His research interests primarily lie within the fields of corrections and re-entry and how they intersect with public policy. His work has recently been published in Criminology and Public Policy, Criminal Justice and Behavior, Crime and Delinquency, Law and Human Behavior, and Justice Quarterly.

Jordan M. Hyatt

Jordan M. Hyatt is a postdoctoral research fellow at the Evidence-Based Institute for Justice Policy Research at Rutgers University and a senior research associate for the Jerry Lee Center of Criminology, University of Pennsylvania.