Examining the Effects of Community-Based Sanctions on Offender Recidivism

Abstract

Offenders who violate their conditions of release (parole violators) pose a threat to public safety and a unique challenge for parole officials. Historically, parole officers have simply revoked the parole of these offenders and returned them to prison; however, increases in state correctional populations have forced many jurisdictions to experiment with sanctions administered by parole officers in the community. Community‐based sanctions are sanctions administered by parole officers as consequences for noncompliance with release conditions that are ultimately intended to reduce parole violators’ odds of recidivism. Yet, the effects of these types of sanctions are still relatively unknown. Drawing from perspectives on formal and informal social control, this study involved an examination of the effect of community‐based sanctions on parole violators’ odds of recidivism. Findings revealed that formal controls such as community‐based sanctions that were applied with certainty, severity, and in a swift manner along with informal controls such as measures of offenders’ stake in conformity had effects on parole violators’ odds of recidivism and time to recidivism.

Keywords:

• sanctions
• social control
• recidivism
• parole
• violation

Acknowledgments

This project was indirectly supported by award number 2005‐IJ‐CX‐0038 from the National Institute of Justice, Principal Investigators: Ohio Department of Rehabilitation and Correction.

Notes

1. Scholars have used the term “community sanctions” interchangeably with intermediate sanctions (e.g., Gendreau et al., 2000), sanctions that are typically imposed by judges and fall somewhere in between probation and prison. There has been considerable research on the efficacy of intermediate sanctions (e.g., drug courts, boot camps, ISP) (see Gendreau et al., 2000; MacKenzie, 2006 for reviews of this literature). In contrast, community‐based sanctions (as defined for this study) have received very little empirical attention.

2. The relationship between sanctions and recidivism could be considered within perspectives on specific deterrence. In our view, however, perspectives on specific deterrence can be subsumed under broader perspectives on social control (see also Gibbs, 1977; Sherman et al., 1992).

3. Technically, all violations of offenders’ release conditions could be considered “recidivism.” Since this study is focused on the effects of sanctions for violations of release conditions; however, recidivism is conceptualized here as behavior that reflects reoffending.

4. An important caveat to the findings regarding sentence severity, however, is that most of the researchers who have observed a negative relationship between sentence severity and recidivism have used a measure of sentence severity inclusive of sentences other than imprisonment (e.g., fines, jail) (see, e.g., Dejong, 1997; Thistlethwaite et al., 1998; Wooldredge & Thistlethwaite, 2002; Yu, 1994). The majority of the studies examining the effects of imprisonment on recidivism suggest that imprisonment does not reduce offenders’ likelihood of recidivism (see Nagin et al., 2009 for a review of this literature).

5. The data used for this study were collected as a part of a larger project designed to evaluate the effects of a change to the ODRC parole violation sanction policy. The project involved the use of samples that were representative of all offenders released on discretionary parole or post‐release control (mandatory parole) in Ohio during a three‐month period before (October–December 2003) and after (August–October 2005) the violation sanction policy was implemented statewide. The larger study revealed that the change to the sanctioning policy had no effect on various measures of offender recidivism (see Martin & Van Dine, 2008), and so the two samples were combined for the analyses described here. Using the subsample of the data used for this study, we also examined whether the measure of sanctions differed between the time periods (pre‐ and post‐policy). No significant differences were observed. Additionally, sanctions did not become more certain, swift, or severe (as measured for this study) after the policy was implemented. Finally, we also added a measure reflecting pre‐ or post‐policy sample to the final models, but it was ultimately dropped because it was not related to any of the measures of recidivism examined in this study.

6. As opposed to examining all offenders released from prison, the focus on the effect of community‐based sanctions required that the sample be restricted to only those offenders who violated the conditions of their release, thus providing the opportunity for parole officials to sanction them. In other words, the effect of the intervention (i.e., sanction) could be observed only on the population for whom the intervention could have been applied (i.e., parole violators). In many respects, the logic is similar to examining the effects of other criminal justice sanctions (e.g., arrest, prosecution) on only those individuals who come to the attention of the justice system actor (e.g., individuals police encounter, individuals who were arrested). It is important to note, however, that the findings reported in this study are only generalizeable to parole violators. In other words, the results of the analyses are not applicable to all offenders released from prison, but only those offenders who violate their conditions of release.

7. According to Ohio Revised Code (5120:1‐1‐41), the parole board must order a period of post‐release control of five years for offenders sentenced for sexual offenses or felony one offenses and three years for offenders sentenced for felony two offenses and felony three offenses where the offender caused or threatened to cause physical harm to a person. The parole board may order a period of post‐release control of up to three years for all nonviolent felony three offenses, felony four offenses, and felony five offenses. (In practice, a one‐year period of post‐release control is typically imposed.) The parole board may also terminate or modify the period of post‐release control before the period supervision is concluded. For example, during the years 2004–2008 the average length of post‐release control served by offenders in Ohio was about 1.4 years.

8. Offenders whose first (or only) violation involved an arrest for a new felony did not provide parole officials with an opportunity to sanction them for a violation prior to the study end date (the date of arrest for these offenders). As such, the effects of the intervention (sanctions) could not be examined for this group. Offenders whose first violation resulted in their return to prison were removed for similar reasons; the one‐year follow‐up period for this study prohibited us from examining behaviors that occurred after the offenders who were reincarcerated for violation of release conditions were released. Therefore, inclusion of the few offenders who were sanctioned to prison terms in the analyses would have biased the results (as a result of an incapacitation effect) in favor of finding sanctioning effects.

9. Removing these offenders whose first violation involved an arrest for a new felony or the offenders whose first violation resulted in their return to prison for a technical violation raised concerns about whether the sample was generalizeable to the target population, parole violators. This possibility was investigated by comparing the offenders who were removed from the sample to those offenders who were retained in the final study sample. The groups did not differ significantly (p ≤ .01) on measures of age, gender, race, committing offense, risk level, or gang membership.

10. Mastrofski, Worden, and Snipes (1995) described a similar problem when attempting to examine the effects of encounter‐ and officer‐level factors on police decisions.

11. Celerity, the number of weeks between an incident and sanction, was restricted to vary between one and seven in order to reduce the skew in the distribution and capture more meaningful variation. Over 88% of the sanctions included in this study were imposed within seven weeks.

12. The fact that parole officers did not sanction some offenders for violations of their release conditions raised the possibility that those parole violators differed from the parole violators who were sanctioned. However, the offenders who were sanctioned were compared to those who were not sanctioned for violations, and the two groups did not differ significantly on measures of age, gender, race, offense incarcerated for, risk level, or gang membership.

13. A potentially important step in the application of the adjustment for sample selection bias is correcting for heteroskedasticity of error variance. This is typically done by using a consistent standard error estimator (see Bushway et al., 2007). Although this is easily done for probit models, the related adjustment cannot be performed for survival analyses. As such, the significance tests generated from the models of time to felony arrest are potentially biased. In order to ease concerns, we estimated the survival models without including the inverse Mills ratio. The results from these models were very similar to those reported in Tables 2 and 3. Still, the problem could not be completely overcome, and so readers should interpret the findings with the potential limitations of these analyses in mind.

14. Prior to estimating the final models, all of the measures were subjected to tests for multicollinearity. Multicollinearity was not determined to be a problem for any of the measures examined in this study. For the analyses that included the product terms, multicollinearity between the product terms and the main effects was avoided by centering the values for sanctions and residential mobility around their means prior to creating the product terms.