One State’s Use of Program Evaluation to Improve Correctional Practices

• Author: Edward J. Latessa,Lori Brusman Lovins
• Date: 01 February 2018
• Published date: 01 February 2018
• DOI: 10.1177/1043986217750639
• Subject Matter: Articles

https://doi.org/10.1177/1043986217750639

Journal of Contemporary Criminal Justice

2018, Vol. 34(1) 81 –96

© The Author(s) 2018

Reprints and permissions:

sagepub.com/journalsPermissions.nav

DOI: 10.1177/1043986217750639

journals.sagepub.com/home/ccj

Article

One State’s Use of Program

Evaluation to Improve

Correctional Practices

Lori Brusman Lovins1 and Edward J. Latessa2

Abstract

Trends in criminal justice programming can sometimes lead one to question the

direction the field is taking related to rehabilitation. This article presents one state’s

attempt to evaluate the quality of residential treatment services in the state across

a primary and then follow-up outcome evaluation. This article outlines study results

specific to treatment effects by risk. The study finds consistent evidence in support

of the risk principle across both studies. The article also outlines what the state did

with the recommendations from these studies, and how that affected the delivery of

correctional programming in that state.

Keywords

correctional program evaluation, evidence-based practices, risk principle, RNR

Introduction

Research on the principles of effective intervention, otherwise known as the risk, need,

responsivity (RNR) principles, has existed for nearly three decades (Andrews, Bonta, &

Hoge, 1990; Andrews et al., 1990; Gendreau, 1996; Gendreau & Andrews, 1990). The

quantitative reviews of correctional interventions that make up this body of knowledge

contributed to the shift from the Martinson’s (1974) “nothing works” era to the “what

works” period we are currently in (Andrews & Bonta, 2010). These research findings

have been widely used to inform evidence-based correctional policy and practices.

There has been a myriad of research to support the principles of effective interven-

tion (see Andrews & Bonta, 2010). In general, this model subscribes to a

human service philosophy. Accordingly, there are few studies that empirically support

1University of Houston–Downtown, TX, USA

2University of Cincinnati, OH, USA

Corresponding Author:

Lori Brusman Lovins, Department of Criminal Justice and Social Work, University of Houston–Downtown,

One Main St., Houston, TX 77002-1001, USA.

Email: lovinsl@uhd.edu

750639CCJXXX10.1177/1043986217750639Journal of Contemporary Criminal JusticeBrusman Lovins and Latessa

research-article2018

82 Journal of Contemporary Criminal Justice 34(1)

a punitive or “control-based” philosophy. Correctional interventions based on punish-

ment and deterrence in absence of rehabilitation efforts tend to produce negative treat-

ment effects (Gendreau, Goggin, Cullen, & Andrews, 2000; Lowenkamp, Flores,

Holsinger, Makarios, & Latessa, 2010; MacKenzie, 2000; Petersilia, 1998; Smith,

Goggin, & Gendreau, 2002). The same can be found with juveniles. For example,

Lipsey (2009) concluded that interventions using control-based strategies increased

recidivism by up to 8%, whereas those using a therapeutic or treatment-oriented

approach had an average recidivism reduction of 12%.

Although there is evidence to support a human service approach in corrections, not

all treatment approaches pack the same punch. There are numerous studies that support

the importance of incorporating the RNR principles into correctional interventions. The

risk principle asserts that we must (a) accurately classify offenders by likelihood of

recidivism, via use of actuarial risk assessment, and (b) use that information of match

offenders to the appropriate supervision and treatment program, commensurate with

risk level (Andrews et al., 1990). High to moderate risk offenders respond best to inten-

sive interventions, whereas low risk individuals generally respond poorly to these same

programs (Andrews & Dowden, 2006; Latessa, Brusman Lovins, & Smith, 2010;

Lowenkamp & Latessa, 2004; Lowenkamp, Latessa, & Lemke, 2006). This finding has

also been supported with specialized populations, such as juveniles (Lipsey, 2009),

females (Lovins, Lowenkamp, Latessa, & Smith, 2007), and sex offenders (Hanson,

Bourgon, Helmus, & Hodgson, 2009; Lovins, Lowenkamp, & Latessa, 2009).

The second principle is the need principle. This principle asserts that programs

should target the dynamic risk factors correlated with recidivism, namely, criminogenic

needs. The primary criminogenic need factors include antisocial attitudes, personality

and peers, lack of employment/education, family issues, substance abuse, and unstruc-

tured leisure time (Andrews & Bonta, 2010). In addition to ascertaining risk level,

when a dynamic risk assessment is used, criminogenic needs can be identified and used

to develop treatment or case plans (Andrews, 1999; Bonta, 2002). When programs fail

to target a range of criminogenic needs, treatment outcomes are affected. Gendreau,

French, and Taylor (2002) found that programs that target four to six more crimino-

genic than noncriminogenic needs show a significantly stronger effect size (ES = .31).

The final component of the RNR model is the responsivity principle. This principle

has two components. Specific responsivity suggests that one must address personal

learning styles and potential barriers to successful completion of a correctional inter-

vention. For example, although needs such as mental health, trauma history, and intel-

lectual functioning may not be predictive of future offending, these factors can serve

as barriers to successful completion of treatment or supervision. As such, they should

be addressed, with the goal of increasing the likelihood of retention and completion of

a program (Andrews & Bonta, 2010). The general responsivity principle carries much

more empirical backing than specific responsivity. General responsivity asserts that

correctional programs use a therapeutic model deemed effective at changing offender

behavior; the cognitive-behavioral and social learning models have shown consistent

effectiveness across a range of populations (Izzo & Ross, 1990; Landenberger &

Lipsey, 2005; Lipsey & Wilson, 1998; Pearson & Lipton, 1999; Pearson, Lipton,