Psychometric Examination of Treatment Change Among Mentally Disordered Offenders: A Risk–Needs Analysis

• DOI: 10.1177/0093854817743539
• Author: Mark E. Olver,Drew A. Kingston
• Date: 01 February 2018
• Published date: 01 February 2018
• Subject Matter: Articles

743539CJBXXX10.1177/0093854817743539Kingston, Olver / PSYCHOMETRIC EXAMINATION OF TREATMENT CHANgE
research-article2017
Psychometric examination of
treatment change among mentally
DisorDereD offenDers
a risk–needs analysis
DREw A. KINgSTON
Royal Ottawa Health Care Group
MARK E. OLVER
University of Saskatchewan
The present study examined the association of psychiatric symptomatology, criminal attitudes, and treatment changes within
these domains to violent and general recidivism in a sample of 614 mentally disordered offenders. Significant pre–post
changes were found on multiple measures of criminal attitudes, symptomatology, and readiness for change. Antisocial
Intentions and Attitudes Toward Associates (from the Measure of Criminal Attitudes and Associates [MCAA]) predicted
general recidivism and covaried with the Big Four criminogenic need domains on the Level of Service Inventory–Ontario
Revision; none of the remaining psychometric measures significantly predicted violent or general recidivism. Although pre–
post changes were seldom linked to changes in recidivism, positive changes in Antisocial Intentions (MCAA) significantly
predicted reductions in general recidivism via Cox regression survival analysis, controlling for baseline risk and pretreatment
attitudes score. Risk and need implications of psychometric assessments of treatment change in mentally disordered offender
populations are discussed.
Keywords: psychometric; mentally disordered offender; risk; need; treatment change; recidivism
It is widely accepted that mental illness is highly prevalent and significantly overrepre-
sented in the criminal justice system. Prevalence estimates vary considerably, however,
depending on myriad factors such as sample composition, type of mental disorder, and dif-
ferences in assessment and diagnostic methodology. Several studies have reported an
approximate 4% prevalence rate of psychosis among male offenders (Beaudette, Power, &
Stewart, 2015; Fazel & Danesh, 2002; Fazel & Seewald, 2012), which increases to about
15% when the definition of serious mental illness incorporates major mood and bipolar
disorders (Beaudette et al., 2015; Steadman, Osher, Robbins, Case, & Samuels, 2009).
There is also a high level of psychiatric comorbidity among male offenders, particularly
authors’ note: The authors thank R. Karl Hanson and Kelly M. Babchishin for their helpful consultation
on change analyses. Correspondence concerning this article should be addressed to Drew A. Kingston,
Integrated Forensic Program, Brockville Mental Health Center, 1804 Highway 2, Brockville, Ontario, Canada
K6V 5W7; e-mail: drew.kingston@theroyal.ca.
CRIMINAL JUSTICE AND BEHAVIOR, 2018, Vol. 45, No. 2, February 2018, 153 –172.
https://doi.org/10.1177/0093854817743539
DOI: 10.1177/0093854817743539
© 2017 International Association for Correctional and Forensic Psychology
153

---

154 CRIMINAL JUSTICE AND BEHAVIOR
with substance misuse and personality disorders (Diamond, wang, Holzer, Thomas, & des
Anges Cruser, 2001; Fazel & Seewald, 2012). Such rates have undoubtedly contributed to
a growing interest in establishing best practices in rehabilitation with mentally disordered
offenders (MDOs).
Interventions with MDOs have traditionally been guided by the clinical or psychopatho-
logical model of criminal behavior (see Bonta, Law, & Hanson, 1998). This perspective
underscores the importance of untreated serious mental illness, particularly schizophrenia
and other psychotic disorders, as a salient risk factor for violence (Douglas, guy, & Hart,
2009; Fazel, gulati, Linsell, geddes, & grann, 2009; Hodgins, 2008). Interventions guided
by the clinical model incorporate a number of techniques that are ultimately directed at
improving mental health functioning as the primary means with which to reduce the likeli-
hood of recidivism (Skeem, Manchak, & Peterson, 2011).
Despite the widespread implementation of the clinical model with MDOs (Morgan et al.,
2012; Skeem et al., 2011), some have questioned the utility of this approach for recidivism
reduction. Although serious mental illness, namely schizophrenia and other psychotic dis-
orders, are associated with recidivism in the general population (Brennan, Mednick, &
Hodgins, 2000; Douglas et al., 2009; Hodgins, 2008), the effect typically fails to generalize
to offender samples. Indeed, the vast majority of studies with offender samples have shown
psychiatric diagnoses and related clinical constructs, such as previous psychiatric hospital-
izations, to be weakly (and in some cases inversely) related to recidivism (Kingston et al.,
2016; Kingston, Olver, Harris, wong, & Bradford, 2015; Rezansoff, Moniruzzaman, gress,
& Somers, 2013; Skeem, winter, Kennealy, Louden, & Tatar, 2014).
Andrews and Bonta (1994, 2010) presented the general personality and cognitive social
learning (gPCSL) model of criminal behavior, which underscored eight robust predictors
of criminal behavior that were found to reside within the individual or their immediate
social learning environment. The gPCSL model is differentiated from the clinical model
due to the type of variables identified as salient predictors (Bonta, Blais, & wilson, 2013).
The specific predictors identified in the gPCSL model include criminal history, procriminal
companions, procriminal attitudes, antisocial personality pattern, education/employment,
family/marital, substance abuse, and leisure/recreation. The variables identified in the clini-
cal model, such as psychiatric diagnoses, are considered minor or negligible risk factors.
In their most recent meta-analysis with MDOs, Bonta et al. (2013) compared the predic-
tive utility of the central eight risk/need factors of the gPCSL model with variables sub-
sumed within the clinical model, such as psychosis, mood disorder, and past psychiatric
treatment. Results showed that the central eight risk/need factors were better predictors of
general and violent recidivism when compared with clinical variables. A notable exception
was that a diagnosis of personality disorder, namely, antisocial personality disorder, was a
reliable predictor of recidivism, although the authors correctly noted that this variable is
consistent with a central theme of the gPCSL model. However, this underscores an impor-
tant point that there is considerable overlap between certain mental health and gPCSL
constructs, such as substance misuse and various antisocial traits.
Based on the research noted earlier, some researchers and policy makers have suggested
embedding principles of effective correctional intervention into treatment protocols used
with MDOs. The risk–need responsivity (RNR) model (Andrews & Bonta, 1994, 2010) is
the predominant offender rehabilitation model, and posits that treatment is most effective
when it is matched to the risk posed by the offender (risk principle); when it specifically

---

Kingston, Olver / PSYCHOMETRIC EXAMINATION OF TREATMENT CHANgE 155
targets criminogenic needs, that is, the seven dynamic factors of the gPCSL (need princi-
ple); and when it is based on empirically supported modalities, but maintains sufficient
flexibility so that it is applicable to diverse learning styles and abilities in offending popula-
tions (responsivity principle).
There has been substantial support for the RNR model in non-MDOs, such that pro-
grams adhering to these three principles are more effective than programs that do not (see
Andrews & Bonta, 2010, for a review). The RNR approach has also demonstrated some
support with MDOs (Skeem et al., 2014; also see Skeem, Steadman, & Manchak, 2015, for
a review). Although studies have repeatedly shown that evidence-based mental health
treatment has little effect on criminal recidivism, targeting criminogenic needs, such as
antisocial attitudes, demonstrates the strongest effect in reducing recidivism (Morgan
et al., 2012; Skeem et al., 2011). In a randomized controlled trial, Cullen et al. (2012a)
found that MDOs assigned to a cognitive skills program targeting thinking styles and
criminal attitudes showed significant improvements in social problem solving relative to
the treatment-as-usual comparison group and that such changes were maintained up to 12
months, postrelease. In a follow-up study, Cullen et al. (2012b) reported that the treatment
group engaged in fewer incidents of verbal aggression and leave violations during treat-
ment compared with the treatment-as-usual comparison group. Despite the accumulation
of evidence supporting the RNR approach in both MDOs and non-MDOs, caution is war-
ranted in fully adapting this model with MDOs, given the limited evidence base with this
group (Skeem et al., 2015).
Although considerable progress has been made in identifying criminogenic needs in
MDOs that inform risk assessment, intervention, and management, the risk-change litera-
ture is far less developed. As noted earlier, the need principle of effective correctional inter-
vention posits that intervention should target causal risk factors that when changed are
associated with reductions in recidivism.
Relatively few investigations have examined the association between within-treatment
change and recidivism (e.g., Beggs & grace, 2011; Olver, Stockdale, & wormith, 2014;
Olver, Kingston, Nicholaichuk, & wong, 2014). Ashford, wong, and Sternbach (2008)
examined MDOs diagnosed with psychotic or bipolar disorders who participated in a cogni-
tive skills program targeting antisocial attitudes. Results...