Does Change in Risk Matter?

• Author: Scott W. VanBenschoten,Christopher T. Lowenkamp,Thomas H. Cohen
• Published date: 01 May 2016
• DOI: http://doi.org/10.1111/1745-9133.12190
• Date: 01 May 2016

RESEARCH ARTICLE

OFFENDER RISK CHARACTERISTICS

Does Change in Risk Matter?

Examining Whether Changes in Offender Risk

Characteristics Influence Recidivism Outcomes

Thomas H. Cohen

Christopher T.Lowenkamp

Scott W.VanBenschoten

Administrative Office of the U.S. Courts

Research Summary

The Post Conviction Risk Assessment (PCRA) is a correctional assessment tool used by

federal probation officers that identifies offenders most likely to commit new crimes

and the criminogenic characteristics that, if changed, could reduce the likelihood

of recidivism. We explored how changes in offender risk influence the likelihood of

recidivism by tracking a population of 64,716 offenders placed on federal supervision

with multiple PCRA assessments. In general, offenders scoring in the high-, moderate-,

and low/moderate-risk categories at their initial assessment and experiencing decreases

in their risk classifications were less likely to recidivate comparedwith their counterparts

whose risk levels remained unchanged or increased. Conversely, increases in offender

risk were associated with higher rates of reoffending behavior. Notably, we saw no

recidivism reduction for offenders in the lowest risk category if they received decreases

in their overall PCRA scores.

Policy Implications

This analysis provides officers with crucial information about how changes in offender

risk can influence the likelihood of arrest. Probation officers should consider adjusting

The authors would like to thank Laura Baber and Matthew Rowland at the Administrative Office of the United

States Courts for their helpful suggestions and comments. Jeremy Luallen from Abt Associates also provided

helpful guidance. Finally, this publication benefited from the careful editing of Ellen Fielding. Direct

correspondence to Thomas H. Cohen, Administrative Office of the U.S. Courts, One Columbus Circle NE,

Washington, DC 20544 (e-mail: thomas_cohen@ao.uscourts.gov).

DOI:10.1111/1745-9133.12190 C2016 American Society of Criminology 263

Criminology & Public Policy rVolume 15 rIssue 2

Research Article Offender Risk Characteristics

downward the amount of time and resources devoted to offenders with decreasing risk

levels once those decreases have stabilized. Alternatively, probation officers should pay

particular attention and allocate more resources to those offenders reclassifiedinto higher

risk categories. Finally, probation officers should be cautious about providing resources

to low-risk offenders who do not seem to benefit from efforts aimed at reducing their

criminal risk factors.

Since the early 2000s, the federal probation system has adopted an outcome-based

approach that emphasizes crime reduction (Alexander and VanBenschoten, 2008).

The grounding of the federal probation system in a results-based framework occurred

in part as a result of a comprehensive strategic assessment conducted in 2004. After the

recommendation of this assessment, the federal probation system developed policies that

clearly articulated the goals of post-conviction supervision and developed a framework for

measuring the system’sprogress toward its goals. One primary goal of federal supervision was

defined as the protection of the community through the reduction of risk and recurrence of

crime (that is, recidivism) both during and after an offender’ssuper vision period (Alexander

and VanBenschoten, 2008; Cohen and VanBenschoten, 2014; Hughes, 2008).

Meeting the key goal of recidivism reduction required the federal probation sys-

tem to adopt the risk, needs, and responsivity (RNR) model of correctional treatment

(Alexander and VanBenschoten, 2008; Andrews and Bonta, 2010; Andrews, Bonta, and

Hoge, 1990). Developed by a group of community corrections scholars in Canada, this

treatment paradigm has shown that one of the most effective methods for reducing criminal

behavior is to employ actuarial risk assessment instruments that allow officers to target of-

fenders most likely to reoffend and to identify specific criminogenic (i.e., crime-producing)

factors that when changed have the potential for reducing the likelihood of recidivism

(Andrews and Bonta, 2010; Andrews et al., 1990; Cullen and Gendreau, 2000; Gendreau,

Smith, and French, 2006).

In specific terms, the risk principle mandates that officers work most intensively with

offenders who are at the highest risk of recidivating. The needs principle directs officers to

focus on offender characteristics that are criminogenic (e.g., antisocial attitudes, poor social

networks, substance abuse problems, and employment issues). Since these factors are asso-

ciated with criminal behavior, ameliorating them has the potential to reduce crime. Finally,

the specific responsivity principle guides officers to match interventions with the ability

and learning styles of offenders (Administrative Office of the U.S. Courts [AOUSC], 2011;

Andrews and Bonta, 2010; Andrews et al., 1990; Lowenkamp, Johnson, VanBenschoten,

Robinson, and Holsinger, 2013; Van Voorhis and Brown, 1996).

The federal probation system embraced the RNR model of supervising offenders

with the aim of reducing recidivism and protecting the community by developing and

implementing the Post Conviction Risk Assessment (PCRA) instrument. The PCRA is

264 Criminology & Public Policy

Cohen, Lowenkamp, and VanBenschoten

a dynamic actuarial risk assessment tool developed for use by federal probation officers.

This instrument identifies offenders who are most likely to recidivate, ascertains dynamic

criminogenic characteristics that need to be addressed, and providesinformation on obstacles

that would prevent successful offender reintegration and/or treatment (AOUSC, 2011).

Requiring probation officers to be trained to use this actuarial tool meant that the PCRA

was gradually implemented and brought to scale, with the first assessments occurring in

2010. More than 95% of offenders released into federal supervision over the last 12 months

have had at least one PCRA assessment.

Data from the PCRA make it possible to investigate changes in an offender’s risk of

recidivism over time. As the PCRA is a dynamic risk tool, we can explore whether the risk

levels and overall risk scores of offenders under federal supervision are changing and the

relationship between these changes and an offender’sreoffending behavior. When measuring

changes in the risk of recidivism, an important component is assessing whether certain

dynamic PCRA risk factors are more likely to change than others. Specifically,are risk factors

such as employment status more amenable to change than other risk factors such as social

networks? Also important is whether changes in risk are related to an offender’s reoffending

behavior. Stated differently, are offenders whose risk levels or scores decrease being arrested

less frequently than their counterparts whose risk levels or scores remain stable or increase?

A related issue is which reduced PCRA risk domains are significantly associated with the

likelihood of arrest. For example, does getting a job significantly reduce the probability

of arrest to the same extent as obtaining support from a network of prosocial friends or

mentors? These issues are explored more fully in this study examining federally supervised

offenders with multiple PCRA assessments. Before discussing this study’smethodology and

findings, we provide a brief overview of the literature examining the relationship between

changes in offender risk characteristics and arrest outcomes.

What Do We Know About Changes in Dynamic Risk and Its Relationship With

Recidivism?

There is a growing but still limited body of literature examining changes in risk over

time and the relationship between these changes and offender recidivism. Of crucial im-

portance for the U.S. probation system and the field of correctional risk assessment in

general are the empirical efforts that examine changes in the dynamic PCRA risk do-

mains and the association between changes in PCRA risk levels and supervision outcomes

(Cohen and VanBenschoten, 2014; Johnson, Lowenkamp, VanBenschoten, and Robin-

son, 2011; Lowenkamp et al., 2013). Initially, we review these studies and provide back-

ground information about the development and validation of the PCRA. We also briefly

summarize other research—specifically in the area of the Level of Service Inventory—

Revised (LSI-R)—that has been done in the area of dynamic risk assessment and crime

prediction.

Volume 15 rIssue 2 265