Full text of DOI:10.1375/bech.18.1.18

Offender Risk and Needs Assessment:
Some Current Issues and Suggestions
Mitchell K. Byrne
University of Wollongong
Stuart Byrne, Katherine Hillman, and Emma Stanley
University of South Australia
Crime impacts upon the community at multiple levels, causing distress and loss for the
victims, and feelings of insecurity for the public, as well as adding to the drain on financial
resources for governments. This makes the accurate identification of risk of reoffending and
the determination of efficacious rehabilitation strategies imperative. Key principles in cognitive
and behavioural psychology can contribute to this task. This paper will review the issue
of risk assessment and describe the applicability of functional analysis to forensic psychology.
Two studies by the authors will be used to illustrate concepts raised in this review. The paper
will conclude with a model that may help guide the realistic implementation of detailed
individual functional analyses of offenders’ behaviour.
Forensic psychology has undergone some- the development of rehabilitation programs,
what of a renaissance over the last decade.
This follows a lengthy period in the wilderness
during the “nothing works” period of the 70s
and 80s. The nothing works belief was fuelled
by Martinson (1974), who reviewed 231 studies
of the outcomes of treatment programs
for offenders between the years of 1945
to 1967. His conclusion that correctional reha-
bilitation is ineffective became the dominant
mindset amongst administrators, professionals,
and forensic psychology lagged behind other
psychologies, especially clinical psychology, in
the development of effectivetreatment strategies
.
While clinical psychologists developed
a formulation-based approach to problem anal-
ysis and treatment design, forensic psycholo-
gists were limited by theories of crime that
attributed responsibility for offending behav-
iour to sociological, biological, and/or nonde-
and the public. This perception persisted from terminist causes. This effectively hamstrung the
the mid-70s through to the early 90s.
application of scientific principles to the proces
s
The appeal of Martinson’s conclusions
was ubiquitous. Left-wing crime theorists
attributed the cause of crime to inequities
in society while right-wing protagonists
saw support for their retribution-orientated
approach to offenders. The systemic outcome
of the broad acceptance of Martinson’s conclu-
sions was that little or no effort was put into
of changing individual offending behaviour.
The change in this nothing works mentality
was initiated by Thornton (1987), who re-exam-
ined the papers reviewed by Martinson using
a more methodologically sound set of criteria. For
each of the 231 studies he asked three questions:
•
Was recidivism used as an outcomemeasure?
An earlier version of this paper was presented at the 22nd Annual Conference of the Australian Association
for Cognitive and Behaviour Therapy in Perth, Western Australia, July 1999.
Address for correspondence: Mitchell K. Byrne, Department of Psychology, University of Wollongong,
Northfields Avenue, Wollongong NSW 2522, Australia. Email: mbyrne@uow.edu.au
BEHAVIOUR CHANGE
Vol. 18, No. 1 2001 pp. 18–27
18

RISK AND NEEDS ASSESSMENT
•
•
Were subjects randomly allocated to treat- 4. Treatment programs should be designed to
engage high levels of offender responsivity.
5. Treatment programs conducted in the
community have a stronger effect than
residential programs. While residential
programs can be effective, they should
be linked structurally with community-
based interventions.
ment/control conditions, or were treatment
and control subjects matched?
Was the methodology of the study sound?
Only 34 of the 231 studies met these criteria.
Of these, 1 study demonstrated a worse
outcome for the treatment group, 17 studies
showed no difference between groups, and 16
(47%) demonstrated significant advantage
following treatment.
Since Thornton’s re-examination of the data,
other research has demonstrated treatment effi-
cacy of forensic rehabilitation. In a recent
paper, McGuire (1998) reported that meta-ana-
lytic studies of offender rehabilitation provide
evidence of an effect on recidivism, ranging
from 10% to 36%. McGuire noted that effect
sizes such as these were only marginally less
than those obtained for psychological interven-
tions into mental health problems, that they
were comparable to that of AZT in the treat-
6. The most effective programs have high
treatment integrity in that they are carried
out by trained staff, and the treatment initia-
tors are involved in all the operational
phases of the treatment programs.
Risk and Needs
This paper focuses on the first of Hollin’s state-
ments: the targeting and focus of programs.
Specifically, it reflects the risk (of reoffending)
presented by an offender and the identified
(criminogenic) needs of that offender that
should be targeted in order to ameliorate such
ment of AIDS, and that they were greater than a risk. Risk is conventionally construed
that of Aspirin or bypass surgery for the preven- as “stable” or nonchangeable aspects of an indi-
vidual that have been statistically associated
with a high likelihood of recidivism. Examples
of such variables include age at first offence and
current age of the offender. Needs are
in fact risk variables that are “dynamic”
or changeable. Often referred to as crimino-
genic needs (Howells, Watt, Hall, & Baldwin,
1997), these dynamic risk variables include
skills deficits, current behaviours, personality
dispositions, attitudes, cognitions, beliefs,
and general life circumstances.
Risk has been the subject of significant
research and the development of actuarial
measures of risk has occurred both in Australia
(Cumberland & Boyle, 1997; Ward & Dock-
erill, 1999) and abroad (Hart, 1998; Monahan
& Steadman, 1994; Prins, 1996; Steadman et al.,
1994; Webster, Eaves, Douglas, & Wintrup,
1995; Webster, Harris, Rice, Cormier, & Quinsey,
1994). Research such as this has equipped
forensic psychologists to identify high-risk
offenders based on their past behaviours
and other static (nonchangeable) variables.
tion of myocardial infarction. The implication
is that research and effort into forensic rehabili-
tation, given the broad impact of crime on the
community, may lay claim to resources compa-
rable to those expended elsewhere.
Enhancing Treatment Efficacy
To achieve the effect sizes quoted by McGuire
(1998), a number of key features of successful
interventions have been identified. Hollin has
summarised the core components of successful
programs in six key statements, as follows
(Hollin, 1999, cited in Howells et al., 2000):
1. Indiscriminate targeting of treatment
programs is counterproductive in reducing
recidivism; medium- to high-risk offenders
should be selected and programs should
focus on criminogenic targets.
2. The type of treatment program is important,
with stronger evidence for structured
behavioural and multimodal approaches
than for less focused approaches.
3. The most successful studies, while
behavioural in nature, include a cognitive
component to focus on attitudes and beliefs.
Dynamic risk factors have also been the sub-
ject of considerable research, with a number of
measures being developed to assess an off-ender’s
19

MITCHELL K. BYRNE, STUART BYRNE, KATHERINE HILLMAN, AND EMMA STANLEY
level of criminogenic need. For example, Hemphill,
Hare, and Wong (1998) assessed the predictive
utility of a personality dimension, “psychopa-
thy”, using the Hare Psychopathy Checklist
(Hare, 1991). They found that psychopaths
were three times more likely to recidivate and
four times more likely to violently recidivate
than nonpsychopaths. Further-more, Hemphill
Functional Analysis of Offending Behaviour
With the range of risk variables thus far identi-
fied, a comprehensive process of risk assess-
ment appears unclear. However, one strategy
incorporating both actuarial and clinical mea-
sures that has the potential to identify static,
stable dynamic, and acute dynamic risk
variables is functional analysis. We undertake
and colleagues (1998) found that the PCL a functional analysis to determine why a be-
haviour occurs. A clear and simple description
of functional analysis has been provided
by Tustin (1993) who suggests that “(a) func-
tional analysis involves a process of identifying
possible functions of a response by gathering
objective information about the environmental
events immediately surrounding it, especially
the antecedents and consequences of the
response” (p. 39).
Howells (1998) has detailed the process
of applying functional analyses to offending
behaviour by proposing 12 variables that should
be included in offender needs assessment,
an example of which is applied to aggressive
behaviour in Table 1. This approach is multi-
modal, involving both environmental and intra-
personal factors, and examines both operant and
antecedent factors in the establishment
and maintenance of offending behaviour.
Howells (1998) also emphasises the need
to assess contextual factors in the development
of the offending behaviour, and in doing
so extends the import of triggering events
from immediate triggers to the behaviour
to the general environmental context in which
the proximal trigger occurred.
However, it is not possible to undertake
a detailed individual functional analysis with
each offender. In practice, correctional service
agencies tend to apply general knowledge of the
relationship between certain offender variables
and offending behaviour to identify those
who pose the greatest risk to the community,
then target scarce resources at those individuals.
This is achieved essentially by using conven-
tional, primarily static, measures of risk assess-
ment. The exploitation of new insights
into offender risk assessment is a developing
process, and scarce services may potentially
-R scores were more strongly correlated with
recidivism than many standard risk scales con-
structed specifically for that purpose.
A range of dynamic or needs-based risk
scales are used widely by criminal justice agen-
cies. Some, such as the Violent Risk Appraisal
Guide (VRAG; Rice & Harris, 1995) and Rapid
Risk Assessment for Sex Offence Recidivism
(RRASOR; Hanson, 1997) focus on particular
categories of offending. Others provide a more
general overview of propensity to reoffend,
being somewhat ubiquitous across offence
categories. Of these more general dynamic
risk measures, the Level of Service Invent-
ory (Andrews, 1982) has been presented
as amongst the more reliable of measures
(Andrews & Bonta, 1995; Gendreau, Goggin,
& Paparozzi, 1996).
The sophistication of risk measurement
has recently been enhanced by the delineation
of two classes of dynamic risk variables:
stable dynamic and acute dynamic factors
(Hanson & Harris, 1998, 2000). Acute dynamic
risk factors refer to those aspects of the individ-
ual offender that are observed to be immedi-
ately antecedent to reoffending. Useful for
monitoring risk during community supervision,
acute dynamic risk factors might include nega-
tive mood or alcohol intoxication. Stable
dynamic factors, which Hanson and Harris
(1998) identify as most strongly identifying
recidivists, include personality disorders and
deviant attitudes. This is consistent with the
observations of Zamble and Quinsey (1997),
who have identified “dynamic antisociality”
as most predictive of violent recidivism in men-
tally disordered offenders.
20

RISK AND NEEDS ASSESSMENT
TABLE 1
Department of Correctional Services (DCS).
It was designed as a pilot study from which the
DCS might draw inferences, and as such em-
ployed a limited data set. The focus is on vio-
lent behaviour because of the greater impact
of such offences and the low base rate, which
effects statistical risk prediction.
Variables to Include in a Functional Analysis
of Aggression and Violence
Frequency, intensity, and form (of current and
previous violent behaviour)
Environmental triggers (including background
stressors)
Cognitive antecedents (including biases in appraisal
of events, dysfunctional schemata, and underlying
beliefs and values supporting aggression)
Study 1
In the first study (Stanley, 1998) it was our
intention to examine the current risk assessment
used by the DCS and explore the idea that
it may be enhanced by the inclusion of a simple
measure of the personality of the offender, thus
tapping into more (stable) dynamic aspects
of risk. Of 250 male community corrections
clients invited to participate, 36 agreed to be in-
volved. The sample included only clients
with “new” offences who had yet to receive
an assessment of risk.
Affective antecedents (emotions preceding aggressive
acts, e.g., anger or fear)
Physiological antecedents
Coping problem-solving skills
Personality dispositions (e.g., anger-proneness,
impulsivity, psychopathy, general criminality,
overcontrol, undercontrol)
Mental disorder variables (mood, brain impairment,
delusions, hallucinations, personality disorders)
Consequence/functions of aggressive acts
Community corrections staff in several
states in Australia currently rely on an adapted
version of the Wisconsin Risk–Need
Assessment Scale (WRAS; Baird, 1981) to de-
termine offender risk and needs. The WRAS
is a clinical interview-based assessment which
is used primarily to assess risk and determine
the appropriate level of supervision for each
client. The focus of the WRAS is on the assess-
ment of static risk factors. While it also identi-
fies some dynamic risk/need factors, these
are used primarily to inform level of supervi-
sion, and not as targets for change or indices
of program efficacy.
(for the perpetrator and others, short term and long
term; including emotional consequences such as
remorse and peer group or institutional reinforcement)
Buffer factors (good relationships, family support,
achievement in some area)
Opportunity factors (weapons, victim availability,
restrictions)
Disinhibitors (alcohol, drugs)
Note. From “Cognitive behavioural therapy for anger, aggres-
sion and violence”, by K. Howells, 1998, in N. Tarrier
(Ed.), Cognitive behavioural therapy for complex cases
(pp. 295–318), Chichester: Wiley. Copyright 1998
by Tarrier. Adapted with author’s permission.
Various studies have supported the use
of the WRAS in a community correctional set-
ting (National Institute of Corrections, 1981;
Payne, McCabe, & Walker, 1974). Gendreau,
Goggin, and Little’s (1996) meta-analysis
reported that the WRAS was not significantly
better or worse than other composite scales.
The main difficulty with the WRAS lies in its
reliance on static variables in predicting risk.
Static variables, by definition, do not change
over time, and as such, the level of risk
be wasted through the absence of dynamic risk
assessment and screening.
In the rest of this paper, we will describe
two studies that explore strategies for the inclu-
sion of dynamic independent variables into the
conventional assessment processes used
by a majority of Australian correctional service
providers. This inclusion, which aimed to min-
imise disruption to day-to-day staff activity,
is evaluated for its potential contribution to the as measured by the WRAS rarely changes.
determination of reoffence risk. This research Thus, whilst it may be an acceptable instrument
was undertaken by the University of South Aus- for measuring risk, an assessment based solely
tralia in cooperation with the South Australian on the WRAS is unlikely to offer guidance
21

MITCHELL K. BYRNE, STUART BYRNE, KATHERINE HILLMAN, AND EMMA STANLEY
as to program determination or treatment need to program delivery. The CPS has the
outcome. Whilst the WRAS does measure some added advantage of being useful in measuring
dynamic factors, there have been very few stud-
ies that report on their predictive validity
(Bonta, 1996). Furthermore, in many jurisdic-
tions (including South Australia), the WRAS
has been adapted and modified to suit local
needs, without adequate evaluation of the
efficacy of either the original form or the
adapted version.
treatment effects (via pre/post treatment mea-
surements), and could thus be used to link risk
and need to intervention outcomes.
In this study, we administered both the
WRAS and the CPS to the sample. The results
were limited by the sample size but, as a pilot
study, it did provide an indication of possible
data trends. The most important statistic that
emerged was from a comparison between CPS
subscale Antisocial Tendencies and the WRAS
item Assaultive Offence. As a rule, past be-
haviour predicts future behaviour. Assaultive
Offence should relate to offenders’ assessed
personality characteristics derived from Anti-
social Tendencies. Furthermore, both past
violence and characteristics of antisocial
personality disorder have considerable support
in the literature as predictors of future recidi-
vism (Howells et al., 2000). However, there
was no significant relationship between the CPS
subscale Antisocial Tendencies and the WRAS
item Assaultive Offence in this study (r = .068,
p = 0.69). We interpreted the lack of a relation-
ship as indicating the different aspects
of risk/need measured by the two instruments:
static in the case of the WRAS Assaultive
Offence item and (stable) dynamic in the case
of the CPS Antisocial Tendencies item.
The stable dynamic variable of offender per-
sonality has been previously identified
as most informative when assessing recidivism
(Hanson & Harris, 1998; Zamble & Quinsey,
1997). One such measure is the Carlson
Psychological Survey (CPS, Carlson, 1981,
1982). The CPS is a psychometric instrument
that was developed to investigate dimensions
of psychopathology of persons who come to the
attention of criminal justice systems. It is a self-
report scale that contains four clinical scales:
chemical abuse, antisocial tendencies,
self-depreciation, and thought disturbance.
It was designed specifically for use with foren-
sic populations, and the wording for items
has been drawn directly from prisoner dialogue.
There is some evidence to suggest that
the CPS is an accurate measure of risk, incorpo-
rating both stable and acute dynamic variables.
Carlson (1982) has reported a significant
relationship between the antisocial tendencies
(stable) scale and recidivism. Black (1993)
found that repeat incarcerates had a statistically
significant higher score on antisocial tendencies
compared with first-time incarcerates. He also
found that these scores correlated closely
to static risk factors such as frequency of previ-
ous offending. The CPS has been shown
to assist in the prediction of behaviour disor-
der/violence (antisocial tendencies – stable),
mental health vulnerability (thought disturbance
– acute), suicidality (self-depreciation –
acute/stable), and chemical dependency (chemi-
cal abuse – acute) (Black, 1993; Byrne, 1993;
Carlson, 1982). It has been used by mental
health professionals to screen for psychopathol-
ogy on dimensions relevant to offenders.
The CPS is primarily a measure of dynamic
We hypothesised that some of the offenders
had changed in terms of their propensity to off-
end, and that the CPS reflected this change.
Study 2
The second study (Hillman, 1998), which was
also exploratory in nature, examined this
change potential further, as it relates specifi-
cally to violent offending. In this study,
we examined two instruments used to assess
for the potential to be violent: the Violent Off-
ender Treatment Program Risk Assessment
Scale (VOTP RAS; Howells et al., 1997) and
the Violence Potential Index (VPI) of the Per-
sonality Assessment Inventory (PAI; Morey,
1991; Morey & Henry, 1994). The VOTP RAS
targets primarily static or contextual variables,
variables, so it offers the potential for a better while the VPI is a measure of (primarily stable)
assessment of offender need and for linking dynamic risk variables.
22

RISK AND NEEDS ASSESSMENT
This study was a case analysis involving of future harm. The probability measure
some retrospective factors (as scored by the is derived from information about number
VOTP RAS) and some current factors (as mea- of past offences, age at onset of offending
sured by the PAI). Participants were 57 violent behaviour, and alcohol and other drug misuse.
male offenders (as defined by the Justice
The PAI is a self-administered inventory
Administration National Classification of Off- of some 344 questions. It has non-overlapping
ences; JANCO), of whom 23 were parolees scales measuring clinical, interpersonal,
and 34 current prisoners.
and treatment constructs as well as four validity
The VOTP RAS was designed to screen scales. Items from within the various scales
offenders for inclusion in violence prevention were drawn to develop the VPI, which Morey
programs (Hall, 1998). It was modified by Ward and Henry (1994) suggested will assess
(1997), and this version was used in the current the likelihood of an individuals future violent
study. Ward reported that the VOTP RAS behaviour. The VPI consists of 20 items drawn
has achieved between 72% and 76% accuracy from scales relating to aggression, antisocial
in identifying violent recidivism, comparing personality, borderline personality, mania, para-
favourably with the Statistical Information noia, schizophrenia, and substance abuse.
on Recidivism Scale (SIR), which achieved For example, from the Schizophrenia scale,
65% accuracy in predicting violent recidivism items are drawn from the subscales relating
(Bonta, Harman, Hann, & Cormier, 1996). to psychotic thoughts and social detachment,
The false positive rate for the VOTP RAS but not from the items relating to thought disor-
has been reported as 13.9%, compared with der. The VPI has correlated highly with the
rates reported in the literature of 26% to 67.9% Minnesota Multiphasic Personality Inventory
(Ward, 1997). In the interests of reducing false scales of psychoticism (r = .65) and hostility
negatives, the current study adopted a cut- (r = .63), as well as the self-report version
off score lower than that used by Ward. of the Hare Psychopathy Checklist (r = .67)
Only a marginal increase in false positives (Morey, 1991).
resulted (16.3%).
By obtaining a score for each participant
The VOTP RAS is scored using case- on both measures, this study explored the con-
note (archival) information, and as such tention that the VOTP RAS (static) and the VPI
requires no direct interaction with the offender. (dynamic), measure different aspects of risk and
The scale is divided into two sections: level of therefore may inform different approaches
harm caused by violent offences and probability to treatment. Spearman correlation coefficients
TABLE 2
VOTP RAS and VPI Coefficients
AGG-P
.945**
.597**
.530**
.703**
.021
AGG
ANT-A
ANT-S
VPI
PVO
MSO
AGG
ANT-A
ANT-S
VPI
.620**
.534**
.643**
.096
.598**
.655**
.189
.671**
.188
PVO
0.40
–.112
.119
MSO
.082
.157
–.095
.186
–.109
.116
.513**
.573**
VOTP RAS
.198
.259
.524**
Note. AGG-P = physical aggression. AGG = aggression. ANT-A = antisocial behaviours. ANT-S = stimulus seeking.
PVO = past violent offence. MSO = most serious offence.
* p < .05, two-tailed. ** p < .001, two-tailed.
23

MITCHELL K. BYRNE, STUART BYRNE, KATHERINE HILLMAN, AND EMMA STANLEY
TABLE 3
Rotated Factor Analysis — Varimax.
Factor
1
2
3
Variable
.855
.836
.774
.588
.585
.537
Poor emotional control, feelings of betrayal, impulsivity (BOR)
Suspiciousness, belief that treated unfairly (PAR)
Delusional beliefs, social isolation, and discomfort (SCZ)
Lack of empathy, egocentrism, and instability (ANT)
Overinvolvement in a variety of activities, inflated self-esteem (MAN)
Characteristics and attitudes related to anger, hostility, and aggression (AGG)
Past non-violent Offences
.858
.758
.627
.529
Age at 1st offence
Alcohol problems
Drug problems
.866
.845
Most serious offence
Past violent offences
were computed between the total VOTP RAS lence did not load on this factor. Factor 1 was
labelled Future Risk.
and the two subscales most highly related
to past violence: most serious offence and past
violent offences. The total VPI score and
the associated violence scores of physical
aggression, aggression, antisocial behaviours,
and stimulus seeking were included, and the
results are contained in Table 2.
The data presented in Table 2 suggest
that the VPI factors that aim to predict future
violence are not related to those of the VOTP
RAS. That is, the VPI and the VOTP RAS
did not measure violence potential or aggres-
sion in the same manner.
To explore the data further, a factor analysis
using principle components analysis with
a varimax rotation was applied to a “composite”
test formed by combining the two assessments.
Table 3 describes the factors obtained
after varimax rotation. From the table, it can be
seen that three factors emerge with Eigenvalues
greater than one.
Factor 2 primarily included past non-
violent offences,age at onset of offending,
and substance abuse. Factor 2 was labelled
as Past Delinquency.
The third factor was also static in nature
and encompassed the most serious offence, past
violence and aggression. We refer to this factor
as Past Violence.
The results suggest that the VOTP RAS
and the VPI measure different aspects of poten-
tial to reoffend, which might best be described
as a potential toward violence (static variables)
and triggers for violence (dynamic variables).
Static variable assessments, such as the VOTP
RAS, ignore the potential for change in the
offender’s risk profile. A focus on offenders’
criminal past and exclusion of factors such
as personality characteristics and support
networks will result in certain offenders being
consistently labelled violent even if their most
recent crimes, and indeed their future crimes,
are not violent. The results of this study added
weight to a distinction between static
and dynamic variables and are consistent with
Factor 1 includes variables relating to poor
mental health and to antisocial and aggressive
tendencies. This factor represents the dynamic
and changeable predictor variables. Past vio- the views of Andrews and his colleagues,
24

RISK AND NEEDS ASSESSMENT
who suggest that the potential to reoffend
is assessed differently by the two types of vari-
ables (Andrews, 1996; Andrews & Hoge, 1995;
Andrews et al., 1990). This means that an ade-
quate assessment of risk would need to include
measures of background factors as well as crim-
inogenic needs in some form of composite test.
criminogenic needs assessment, is not without
precedent. Melton, Petrila, Poythress,
and Slobogin (1997) referred to the “anamnes-
tic” approach, which “depends on the identifica-
tion of factors that have distinguished
a particular subject’s prior displays of aggres-
sive behavior” (p. 284). The problem for the
practitioner is the time required to undertake
such a process with each offender, making such
an approach unrealistic for most practitioners.
A possible solution would be the develop-
ment of a composite measure that may enable
a screening process to be established. The best
screening tool would involve a two-step
process, beginning with the static variables
(to establish a background for an offence) and
progressing to the dynamic variables (to estab-
lish a current likelihood the offence). The infor-
mation generated by such an approach would
enable a classification of offenders into four
groups, as outlined in Figure 1. In this matrix
model of offender classification, resources
would be directed to those offenders occupying
Cells 1 and 2, with a thorough functional analy-
sis determining the intervention required.
Resources would be deflected from Cell 4,
while successes would be acknowledged, and
celebrated, in Cell 3.
Conclusions and Recommendations
The risk literature has focused on two key
forms of risk assessment: clinical and actuarial.
Actuarial assessments have received significant
support in recent times, and research has delin-
eated three types of variables incorporated
in such studies: static, stable dynamic,
and acute dynamic. This study, in accordance
with others cited, advocates the incorporation
of dynamic variables in the prediction of future
risk, and suggests that the inclusion of such
variables may assist in targeting rehabilitation
resources. While the studies reported in this
paper were limited by small sample sizes, they
do suggest that standard correctional practice
could be enhanced by the inclusion relevant
(dynamic variable) assessments. However, it is
our view that the best way of obtaining useful
information about an offender’s potential
to reoffend is to undertake a functional analysis
of his offending behaviour, incorporating
the risk variables identified in past research
as part of this process.
Given an accurate identification of those
offenders likely to reoffend, and a thorough
functional analysis of why they offend, over
time more and more offenders would find their
Such an application cognitive-behavioural
principles, which we have referred to as way into Cell 3.
S TAT I C
High Risk
Low Risk
High Risk
Low Risk
‚ High risk of reoffending
ƒ Emerging risk of reoffending
ꢀ Little risk of reoffending
„ Changed level of risk
of reoffending (reduced)
FIGURE 1
Matrix Model of Offender Classification
25

MITCHELL K. BYRNE, STUART BYRNE, KATHERINE HILLMAN, AND EMMA STANLEY
Gendreau, P., Goggin, C., & Little, T. (1996). Predicting
References
Andrews, D.A. (1982). The Level of Supervision
Inventory (LSI). Toronto: Ontario Ministry
of Correctional Services.
Andrews, D.A. (1996). Criminal recidivism
is predictable and can be influenced: An update.
Ottawa, Canada: Carleton University, Department
of Psychology.
Andrews, D.A., & Bonta, J. (1995). The Level of Service
Inventory – Revised. Toronto: Multi-Health Systems.
Andrews, D.A., & Hoge, R.D. (1995). The psychology
of criminal conduct and principles of effective
prevention and rehabilitation. Forum on Corrections
Research, 7, www.esc-scc.gc.ca.
Andrews, D.A., Zinger, I., Hoge, R.D., Bonta, J.,
Gendreau, P., & Cullen, F.T. (1990). Does correc-
tional treatment work? A psychologically informed
meta-analysis. Criminology, 28, 369–404.
Baird, C.S. (1981). Probation and parole classification:
The Wisconsin model. Corrections Today, 43,
36–41.
Black, Q.C. (1993). A review of behaviour disorder and
the use of the Carlson Psychological Survey (CPS)
as a diagnostic screening instrument in the forensic
setting. Thesis submitted for the degree of Master
of Applied Psychology, University of Adelaide,
Australia.
adult offender recidivism: What works! (Use Report
No. 1996-07). Ottawa: Department of the Solicitor
General of Canada.
Gendreau, P., Goggin, C., & Paparozzi, M. (1996).
Principles of effective assessment for community
corrections. Federal Probation, 60, 64-70.
Hall, G. (1998). The violent offender treatment program:
Practice and issues. Paper presented at the 18th
Annual Congress of the Australian and New Zealand
Association of Psychiatry, Psychology and Law,
Melbourne, Australia.
Hanson, R.K. (1997). The development of a brief actuar-
ial risk scale for sexual offence recidivism (User
Report No. 1997-04). Ottawa: Department of the
Solicitor General of Canada.
Hanson, R.K., & Harris, A.J.R. (1998). Dynamic predic-
tors of sexual recidivism (User Report No. 1998-01).
Ottawa: Department of the Solicitor General
of Canada.
Hanson, R.K., & Harris, A.J.R. (2000). The Sex Offender
Need Assessment Rating (SONAR): A method for
measuring change in risk levels (User report).
Ottawa: Department of the Solicitor General
of Canada.
Hare, R.D. (1991). The Revised Psychopathy Checklist.
Toronto: Multi-Health Systems.
Hart, S.D. (1998). The role of psychopathy in assessing
risk for violence: Conceptual and methodological
issues. Legal and Criminological Psychology, 3,
121–137.
Hemphill, J.F., Hare, R.D., & Wong, S. (1998).
Psychopathy and recidivism: A review. Legal and
Criminological Psychology, 3, 139–170.
Hillman, K.J. (1998). Predicting violent recidivism:
Examining static and dynamic predictors using the
Violent Offender Treatment Program Risk
Assessment Scale (VOTP RAS) and the Personality
Assessment Inventory (PAI). Unpublished honours
thesis, University of South Australia.
Bonta, J. (1996). Risk-needs assessment and treatment.
In A. Harland (Ed.), Choosing correctional options
that work: Defining the demand and evaluating
the supply (pp. 18–32). Thousand Oaks, CA: Sage.
Bonta, J. (1997). Predicting adult offender recidivism.
Ottawa: Department of the Solicitor General
of Canada.
Bonta, J., & Cormier, R.B. (1999). Corrections research
in Canada: Impressive programs and promising
prospects. Canadian Journal of Criminology, 41,
235 – 247
Bonta, J., Harman, W.G., Hann, R.G., & Cormier, R.B.
(1996). The prediction of recidivism among feder-
ally sentenced offenders: A re-validation of the SIR
scale. Canadian Journal of Criminology,
38, 61–79.
Byrne, M. (1993). Predicting behaviour disorder and
reducing recidivism: A follow up to the E Division
Planning Group (Report). South Australia:
Department of Correctional Services.
Carlson, K.A. (1981). A modern personality test for
offenders: The Carlson Psychological Survey.
Criminal Justice and Behaviour, 8, 185–200.
Carlson, K.A. (1982). Carlson Psychological Survey
manual. Michigan: Sigma Assessment Systems.
Howells, K. (1998). Cognitive behavioural therapy for
anger, aggression and violence. In N. Tarrier (Ed.),
Cognitive behavioural therapy for complex cases
(pp. 295–318). Chichester: Wiley.
Howells, K., Day, A., Byrne, M.K., White, L.J., Hart, K.,
& Nakos, J. (2000). Best practice literature reviews:
Rehabilitative strategies for preventing re-offending
(Vol. 1). Adelaide: University of South Australia.
Howells, K., Watt, B., Hall, G., & Baldwin, S. (1997).
Developing programmes for violent offenders. Legal
and Criminological Psychology, 2, 117–128.
Martinson, R. (1974). What works? Questions
and answers about prison reform. The Public
Interest, 35, 22–54.
Cumberland, A.K., & Boyle, G.J. (1997). Psychometric
prediction of recidivism: Utility of the Risk Needs
Inventory. Australian and New Zealand Journal
of Criminology, 30, 72–86.
McGuire, J. (1998). Alternatives to custodial sentences:
Effectiveness and potential for development
(Memorandum of Evidence to the Home Affairs
Committee). United Kingdom: House of Commons.
26

RISK AND NEEDS ASSESSMENT
Melton, G.B., Petrila, J., Poythress, N.G., & Slobogin,
C. (1997). Psychological evaluations for the courts
(2nd ed.). New York: Guilford.
Monahan, J., & Steadman, H.J. (Eds). (1994). Violence
and mental disorder: Developments in risk assess-
ment. Chicago: Chicago University Press.
Morey, L.C. (1991). The Personality Assessment
Inventory: Professional manual. Florida:
Psychological Assessment Resources.
Morey, L.C., & Henry, W. (1994). Personality
Assessment Inventory. In M.E. Maruish
(Ed, The use of psychological testing for treatment
planning and outcome assessment (pp. 185–216).
New Jersey: Lawrence Erlbaum.
of risk assessment research. In J. Monahan & H.J.
Steadman (Eds.), Violence and mental disorder:
Developments in risk assessment. Chicago: Chicago
University Press.
Thornton, D. (1987). Treatment effects on recidivism:
A reappraisal of the “nothing works” doctrine.
In B.J. McGurk, D.M. Thornton, & M. Williams
(Eds.), Applying psychology to imprison-
ment: Theory and practice (pp. 191–217).
London: HMSO.
Tustin, R.D. (1993). Functional analysis of challenging
behaviours of two people with intellectual disability.
Behaviour Change, 10(1), 39–46.
Ward, A. (1997). The predictive accuracy of the Violent
Offender Treatment Program Risk Assessment Scale.
Unpublished honours thesis, Edith Cowan
University, Western Australia.
National Institute of Corrections. (1981). The Model
Probation and Parole Management Program.
Washington, DC: Author.
Payne, C., McCabe, S., & Walker, N. (1974). Predicting
offender-patients’ reconvictions. British Journal
of Psychiatry, 125, 60–64.
Prins, J. (1996). Risk assessment and management
in criminal justice and psychiatry. Journal
of Forensic Psychiatry, 7, 42–62.
Rice, M.E., & Harris, G.T. (1995). Violent recidivism:
Assessing predictive validity. Journal of Consulting
and Clinical Psychology, 63, 737–748.
Ward, A., & Dockerill, J. (1999). The predictive accu-
racy of the Violent Offender Treatment Program
Risk Assessment Scale. Criminal Justice and
Behaviour, 26(1), 125–140.
Webster, C.D., Harris, G.T., Rice, M.E., Cormier, C.,
& Quinsey, V.L. (1994). The Violence Prediction
Scheme: Assessing dangerousness in high risk men.
Toronto: University of Toronto, Centre
for Criminology.
Webster, C.D., Eaves, D., Douglas, K., & Wintrup,
A. (1995). The HCR-20 scheme: The assessment
of dangerousness and risk. Vancouver Canada:
Simon Fraser University.
Stanley, E. (1998). The comparison and utility of two
risk assessment forms for offenders at a correctional
centre. Unpublished honours thesis, University
of South Australia.
Zamble, E., & Quinsey, V.L. (1997). The criminal
recidivism process. Cambridge: Cambridge
University Press.
Steadman, H.J., Monahan, J., Appelbaum, P.S., Grisso,
T., Mulvey, E.P., Roth, L.H., Robbins, P.C.,
& Klassen, D. (1994). Designing a new generation
ꢀ
27