A Field Study of a Comprehensive Violence Risk Assessment Battery

Date01 September 2015
DOI10.1177/0093854815572252
Published date01 September 2015
Subject MatterArticles
/tmp/tmp-17AsCf8IgMHqAm/input 572252CJBXXX10.1177/0093854815572252Criminal Justice and Behaviorneal et al. / Comprehensive violence risk Assessment Battery
research-article2015
A Field Study oF A ComprehenSive
violenCe riSk ASSeSSment BAttery

TESS M. S. NEAL
Arizona State University
SARAH L. MILLER
R. CLAyTON SHEALy
The University of Alabama
We used archival data to examine the predictive validity of a prerelease violence risk assessment battery over 6 years at a
forensic hospital (N = 230, 100% male, 63.0% African American, 34.3% Caucasian). Examining “real-world” forensic deci-
sion making is important for illuminating potential areas for improvement. The battery included the Historical-Clinical-Risk
Management–20, Psychopathy Checklist–Revised, Schedule of Imagined Violence, and Novaco Anger Scale and Provocation
Inventory. Three outcome “recidivism” variables included contact violence, contact and threatened violence, and any reason
for hospital return. Results indicated measures of general violence risk and psychopathy were highly correlated but weakly
associated with reports of imagined violence and a measure of anger. Measures of imagined violence and anger were corre-
lated with one another. Unexpectedly, Receiver Operating Characteristic curve analyses revealed that none of the scales or
subscales predicted recidivism better than chance. Multiple regression indicated the battery failed to account for recidivism
outcomes. We conclude by discussing three possible explanations, including timing of assessments, controlled versus field
studies, and recidivism base rates.
Keywords: anger; violence; risk assessment; recidivism; incremental validity; field study
Mental health providers in correctional, hospital, and community settings often rely on
violence risk assessments to make decisions about client placements. “Dangerousness
to others” became a primary criterion in the 1960s in federal law for involuntary hospital-
ization of individuals with mental disorders in the United States and continued to be used in
subsequent case law (e.g., Barefoot v. Estelle, 1983; Baxstrom v. Herold, 1966; O’Connor v.
Donaldson
, 1975). Tort liability required mental health treatment providers to consider
patients’ potential for violence beginning in the 1970s, and the dangerousness standard was
expanded to authorize involuntary outpatient treatment in the 1980s, typically through state
statutory and case law (Appelbaum, 1994).
Although the field of risk assessment initially began with unstructured clinical judg-
ments that yielded poor predictions (Monahan, 1981; Steadman & Cocozza, 1974), signifi-
cant progress has been made since that time. The extensive and impressive body of research
AuthorS’ note: Correspondence concerning this article should be addressed to Tess M. S. Neal, New
College of Interdisciplinary Arts & Sciences - SBS, Arizona State University, 4701 West Thunderbird Rd,
Glendale AZ 85306. E-mail: tneal2@nebraska.edu
CRIMINAL JUSTICE AND BEHAVIOR, 2015, Vol. 42, No. 9, September 2015, 952 –968.
DOI: 10.1177/0093854815572252
© 2015 International Association for Correctional and Forensic Psychology
952

Neal et al. / COMPREHENSIVE VIOLENCE RISK ASSESSMENT BATTERy 953
regarding modern violence risk assessment has provided significant practical advances for
the clinicians tasked with performing such assessments in real-world settings. For instance,
several structured assessment instruments have been developed to assist clinicians in for-
mally assessing violence risk (e.g., Douglas, Hart, Webster, & Belfrage, 2013; Hare, 2003;
Harris, Rice, & Quinsey, 1993).
Despite these advances, the utility of the research on violence risk assessment tools is
limited by the extent to which the tools can translate to clinical sites (especially those
unaffiliated with academic organizations or researchers) given that almost all of the
research demonstrating the reliability and validity of these measures has been conducted
in controlled settings. Fewer studies have examined how these instruments perform in
“the real world,” though it is to be expected that poorer reliability and validity values will
be seen when these tools are administered by everyday practitioners. Given that most risk
assessments and subsequent treatment- and placement-related decisions are done in the
field rather than in more controlled research settings, understanding the strengths and
weaknesses of these tools in the field is critical for targeting potential areas for
improvement.
In the present study, we analyzed existing assessment data from a forensic psychiatric
hospital in which clinicians are frequently tasked with conducting prerelease violence risk
assessments to inform discharge decisions. The hospital policy specified a standard battery
for violence risk assessments, which included not only well-known and frequently used
violence risk assessment tools but also an index of anger, an approach not often discussed
in the violence risk assessment literature. We sought to add to the violence risk assessment
literature by (a) testing the field reliability and predictive validity of well-researched mea-
sures and (b) examining the degree to which measuring anger adds incremental validity to
the risk assessment battery. We thought the anger measure might improve the assessment
because it measured a different trait (anger) via a different method (self-report as opposed
to clinician-rated) than the measures most often studied in the violence risk assessment
literature.
AdvAntAgeS oF multitrAit, multimethod meASurement
The potential advantage of our approach is supported by the methodological and sta-
tistical benefits of using a multitrait, multimethod approach to measurement (Campbell &
Fiske, 1959; Walters, 2006). For example, in a meta-analysis examining multimethods by
comparing risk tools with self-report measures in the prediction of criminal justice out-
comes, Walters (2006) found that both kinds of measures accounted for incremental vari-
ance. That is, clinician-completed tools and self-report tools each contributed valuable
and unique predictive validity information. Using a multitrait, multimethod approach is
particularly beneficial when the variables or tools used are independent of one another
and provide unique and incrementally valid information (see Faust & Ahern, 2012;
Kahneman, 2011). Using more tools is not better, especially when the tools or variables
provide overlapping or redundant information. Thus, an investigation like this of the
strongest components of an assessment battery will yield data not only for the hospital
itself (e.g., to inform policy and practice at the particular institution where these data
were gathered) but also for other facilities regarding the reliability and validity of these
tools in real-world settings.

954 CRIMINAL JUSTICE AND BEHAVIOR
Anger And violenCe riSk
Anger has long been associated with aggression and violence from both a theoretical
perspective and as demonstrated through empirical evidence (Craig, 1982; Novaco, 2003;
Novaco & Jarvis, 2002). Recent studies have shown anger to be related to physical aggres-
sion in inpatient settings (Doyle & Dolan, 2006a; McDermott, Quanbeck, Busse, yastro, &
Scott, 2008) as well as in the community (Doyle, Carter, Shaw, & Dolan, 2012; Doyle &
Dolan, 2006b; Monahan et al., 2001; Skeem et al., 2006). As a result, the assessment of
anger has increased in forensic psychiatric and correctional settings, and treatment pro-
grams targeting anger have been developed in hopes of reducing violence (see, for example,
Haddock et al., 2009).
One self-report measure developed to assess the internal experience of anger is the
Novaco Anger Scale and Provocation Inventory (NAS-PI; Novaco, 2003), which was the
measure used in this study (see “Method” section). The NAS-PI has been shown to substan-
tially correlate in the expected direction with other measures of anger and hostility, observ-
ers’ ratings of angry behaviors, and occurrence of violent behaviors (Grisso, Davis,
Vesselinov, Appelbaum, & Monahan, 2000; Novaco, 2003; Steadman et al., 2000). Doyle et
al. (2012) found the NAS to be predictive of community violence following discharge from
acute psychiatric hospitals in England up to 20 weeks postdischarge and noted that anger
was more predictive of violence than impulsiveness and psychopathy.
hypotheSeS
We developed several a priori hypotheses. First, we expected the established violence
risk assessment measures would demonstrate acceptable unique and conjoint predictive
validity on the outcomes of interest given the extant research demonstrating good predictive
validity in more controlled research settings. Second, we expected the anger measure would
correlate with the existing measures but would add unique incremental validity to the risk
assessment battery when included with the existing risk assessment tools. Third, we
expected the field reliability of the well-researched violence risk assessment measures
would be lower than samples documenting the reliability of the tools’ more controlled
research settings, given the typical decrease in cross-validation samples and in applied ver-
sus research samples (see, for example, DeMatteo et al., 2014; Monahan et al., 2005).1
method2
pArtiCipAntS
Participants were male inpatients in a maximum security forensic psychiatric hospital in the
southern United States. Patients who were considered for discharge between 2003 and 2009
were included in the current study....

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT