Assessing the influence of risk/need domains on probation completion among a sample of offenders with mental illness

ABSTRACT

A significant body of research has illustrated that the Risk/Need/Responsivity (RNR) paradigm is predictive of recidivism for a variety of offenders, including those with mental illness. This paper builds upon the extant literature in this area by examining how RNR domains specifically predict successful completion of a mental health specialty caseload. To do so, we use data collected from offenders on the Seriously Mentally Ill probation caseload in Maricopa County, Arizona to examine the completion of probation. Using the Offender Screening Tool (OST) to examine how risk/need predicts probation completion (n = 1,430), our findings demonstrate that the OST score overall, as well as participants’ age, whether the probationer received a Petition for Revocation for a technical violation or new crime during their time on probation, and multiple specific OST risk domains (e.g., educational risk, drug risk, mental health risk, and criminal behavior risk) are meaningfully associated with probation completion. We close by contextualizing these findings and discussing their policy implications.

KEYWORDS:

• Mental illness
• community corrections
• specialty caseloads

Acknowledgments

The authors would like to thank Maricopa County Adult Probation Department for providing the data analyzed in the current project. Points of view in this document are those of the author and do not necessarily represent the official position or policies of the Maricopa County Adult Probation Department.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. This time frame accounts for when the more recent version of the OST was implemented to the time data collection for the current research project began in the fall of 2012.

2. On occasion the sentencing judge will note specific mental health needs or ‘terms’ in the case file of an offender and request that a screening for the SMI caseload be completed.

3. Functional impairment can entail a wide-variety of determining factors but some examples might include: inability to work, significant problems with relationships, inability to reliably self-administer psychiatric medication, history of developmental delays or psychiatric hospitalization, etc. An offender managing their mental illness day to day without incident would not be screened onto the SMI caseload.

4. Although the vast majority of offenders on the SMI caseload are individuals with significant Axis I disorders, on more rare occasions an individual with an autism spectrum disorder or traumatic brain injury may also be placed on the caseload.

5. Individuals with mental illness can also be referred to the mental health court to gauge the offender’s progress and even to celebrate the offender’s accomplishments when doing well.

6. The mental health court judge is allowed to give a set number of jail days for any individual on the SMI caseload across the OMI’s probation term as a graduated sanction for not adhering to the terms of their probation sentence without formally revoking their probation sentence.

7. At the time of the current publication, the OST was undergoing revalidation and was slated to conclude at the end of 2018. A full copy of the OST questions can be found as part of this document: www.nradc.com/home/showdocument?id=12005.

8. Information about obtaining a copy of the OST and its scoring can be found here: https://www.azcourts.gov/apsd/Evidence-BasedPractice/RiskNeedsAssessment/OffenderScreeningTool(OST).aspx.

9. Despite physical health only being a responsivity factor for the current sample, the authors left this domain in Table 3 models because it may be that it remains an important area of consideration that influences the delivery of services, and thus plays an important role for success for individuals with mental illness in the current sample.

10. We also performed a series of logistic regression analyses that predicts probation revocation overall, as well as PTRs for a technical violation and PTRs for a new crime. We completed these supplemental analyses for the convenience of the reader as these outcomes have traditionally been used in the empirical research to examine the RNR framework. We do not discuss findings from these analyses in the body of this paper but added three supplemental tables (each with three models) in the appendix of the paper. The models in the appendix exclude the physical health variable of the OST (a responsivity only domain) and also exclude the PTR for technical and new crime violation as predictors to see how RNR domains influence failure for this sample overall. In general, the results of these models show that there are differences across predictors of probation revocation, PTRs for technical violations, and PTRs for new crimes. First, we find that age, gender, race, drug abuse risk, mental health risk, and criminal behavior risk are associated with revocation. Second, we find that only age, residential/neighborhood risk, and mental health risk are associated with PTR for technical violations. Finally, we find that age, gender, race, probation offense, attitude risk, and criminal behavior risk are associated with PTR for new crimes. Considered together, these models suggest that there is reason for research in this area to effectively examine pathways to completion or failure.

11. Anti-social/procriminal attitude risk, widely considered one of the strongest predictors of failure in RNR research (e.g., see Latessa and Lovins 2010) was not significantly predictive in the modeling. Although there is no clear explanation of this, one hypothesis is it may be a result of the way some of the attitude questions are measured in the OST. For instance, the screener rates the offender’s motivation and attitude at the time of intake in some of the questions. It may also be a consequence of the current study using only the OST risk/need domain scores at the time of intake (pre-sentence) as criminal attitudes have the potential to fluctuate as a dynamic OST domain.

Additional information

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Notes on contributors

Philip Mulvey

Philip Mulvey is an assistant professor in the School of Criminal Justice Sciences at Illinois State University. He earned his PhD from Arizona State University in criminology and criminal justice, and also has degrees in psychology from University of Illinois, Boston College, and Northwestern University. Prior to entering academia, he was the project manager in the Health Disparities & Public Policy Program at Northwestern University. His focal research interests revolve around the interface between mental illness and the criminal justice system, issues of social control for disenfranchised populations, and criminal justice policy.

Matthew Larson

Matthew Larson is an assistant professor at Wayne State University whose recent work has focused on life-course criminology, romantic relationships, mental illness, and violence. His research has appeared in Criminology, Journal of Youth and Adolescence, and Criminal Justice & Behavior.