Actuarial Sentencing: An “Unsettled” Proposition
(For Sept 2010 - University at Albany Symposium on Sentencing)
Kelly Hannah-Moffat, University of Toronto 1
[email protected]
The use of actuarial risk-need assessment tools is increasing in many Western penal,
social welfare, health, and human service contexts. Although considerable variation exists in
how these tools are applied in criminal justice settings, many international jurisdictions are
beginning to use risk instruments to structure, inform, and/or determine an increasing range of
corrections-management practices including arrest, diversion, bail, pre-sentence reports,
sentencing, prison classification, and parole decisions. The most common use of these tools is to
predict recidivism in cases involving violent, youth, or sexual offenders (Cunningham and
Reidy, 2002; Dutton and Kropp, 2000; Grann and Langstrom, 2007; Harris and Rice, 2007;
Lodewihka, Doreleijers, and De Ruiter, 2008; Looman and Abracen, 2010; Lucken and Bales,
2007; Manchak, Skeem, Douglas, and Siranosian, 2009; Maurutto and Hannah-Moffat, 2007;
Monahan, 2006; Murrie and Balusek, 2008; Rettenberger, Matthes, Boer, and Eher, 2010; Roehl
and Guertin, 2000; Schwalbe, 2007; Scott, 2009; Vess, 2009; Viljoen et al., 2008). In addition to
these applications, some jurisdictions have also incorporated risk tools into new sentencing
guidelines for non-violent offenders (Kleiman, Ostrom, and Cheesman, 2007; Ostrom, Kleiman,
Cheesman, Hansen, and Kauder, 2002). This new application reflects a broader movement
toward evidence-based penality.
Internationally, the current penal context is characterized by rapidly growing prison
populations, declining resources, and rising demands for public accountability and security.
Within this context, actuarial tools are being promoted as offering a rational, objective, and
empirically sound method of predicting recidivism. The emphasis on actuarial risk prediction
tends to accompany demands for evidence-based “smart” sentencing (Etienne, 2009),
alternatively known as “crime prevention jurisprudence” (Andrews and Dowden, 2007), which
relies on the use of social scientific evidence to determine a sentence that is more likely to
improve public safety” (Chanenson, 2009; Heilbrun, 2009; MacKenzie, 2001; Marcus, 2009a;
2009b; Warren 2007;Wolfe, 2008). Particularly relevant to sentencing is evidence about
1|Page
recidivism and correctional interventions that reduce criminal propensities. The evidence and
techniques informing this sentencing framework are being imported from a particular type of
correctional research and practice. Proponents of evidence-based sentencing and corrections
argue that research can identify ‘what works’ to reduce the risk of recidivism, and that lower risk
offenders can be efficiently, economically, and effectively managed in community settings
(Andrews and Bonta, 2003; Andrews et al., 1990; Clements, 1996; Cullen and Gilbert, 2001;
Etienne, 2009; Mcguire, 1995; 2002). They also maintain that penal resources should be
redirected into correctional programs “proven” to work and that all interventions should be
accredited and evaluated to ensure that they facilitate the reduction of recidivism. These
perspectives are highly influential in correctional policy sectors.
Although the law, policy and principles of sentencing vary considerably by jurisdiction
and country, the introduction of risk into sentencing is part of an established effort to curtail
judicial discretion and regulate sentencing. The indeterminate sentencing policies popular
throughout the 1970s gave judges broad discretionary power and parole boards considerable
flexibility in determining release. Concerns about the global effects of these practices produced
significant changes to sentencing practices, which were introduced by the late 1970s: (1)
“determinate sentencing;” (2) the abolition of parole in many jurisdictions; (3) mandatory
sentencing laws; (4) “truth” in sentencing; and (5) sentencing guidelines. The general
disillusionment with rehabilitation and the popularised view that “nothing works” that
characterized this era contributed to additional shifts in sentencing theory and policy. By late
1999, several jurisdictions had adopted sentencing guidelines that sought to: (1) reduce judicial
disparity; (2) promote consistent sentencing; (3) prioritize and allocate correctional resources; (4)
adjust punishments for certain categories of offenders; (5) reduce prison overcrowding; and (6)
encourage the use of non-incarceration sanctions. The most recent sentencing trend in Canada
and most US states follows the recommendations of risk assessment advocates (Bonta, 2008b;
Marcus, 2009a; Wolfe, 2008) to increase and institutionalize the use of formal actuarial risk
assessment instruments by purchasing those already on the market, adopting those in the public
domain, or developing jurisdictional specific tools (Etienne, 2009).
The logic built into risk tools is different from the subjective professional or clinical
knowledge that previously informed the decision making of judges, correctional authorities,
2|Page
police, and parole boards (Ericson and Haggerty, 1997; Feely and Simon, 1992; 1994; Gray,
Lanig and Noakes, 2002; Hannah-Moffat, 2004 O’Malley, 2004; Simon, 1993). Actuarial risk
uses aggregate statistics to categorize defendants’ risk levels and then matches interventions to
those levels. The use of actuarial technologies promises improved safety/security, a more
efficient allocation of resources, and superior decision-making practices (Andrews and Bonta,
2003). Consequently, both the public and practitioners tend to “embrace risk” (Baker and Simon,
2002) as a technology for improving decision making and limiting discretionary powers
(Hannah-Moffat et al., 2009; Ostrom et al., 2009). Risk assessment tools are appealing because
of their perceived ability to engender reliable, valid, and objective determinations of future risks
and to enhance managerial accountability; however, there are numerous conceptual and
methodological concerns associated with their use in corrections, including concerns about racial
and gender discrimination and the individualization of social problems (Hannah-Moffat, 2009).
The use of risk tools in sentencing is especially problematic because when used in courts
they may offend moral and legal norms as well as country-specific constitutional values (cf.
Etienne 2009; CHRC, 2003; Hannah-Moffat and Shaw, 2001). Sentencing reports and statements
about an offender’s risk level remain with the offender for the entirety of his or her sentence,
affecting a range of correctional decisions from levels of surveillance and intervention to
eventual parole release. Because the tools classify and promote interventions based on categories
of offender risk (i.e. low, medium, high), risk technologies tend to de-individualize punishments
and can shift and reorient sentencing practices in unanticipated ways. Moreover, legal and
correctional professionals who use risk information in decision making are unlikely to have
considered the documented limitations about the science of risk, and frequently have only a
limited understanding of the actuarial technologies they are using 2 . The trend toward using risk
instruments in all sectors of the criminal justice system, therefore, merits further theoretical
deliberation and empirical study.
Most research on actuarial risk is highly compartmentalized; limited meaningful dialogue
takes place between risk theorists, legal scholars, sentencing and government researchers, policy
makers, and practitioners. This means that concerns about legal, ethical, and policy implications
arising from the use of risk-based instruments do not receive the benefit of cross-sectoral
dialogue. Consequently, my primary objectives in this paper are to highlight a number of these
3|Page
concerns in order to: (1) stimulate discussion about the potential impact of incorporating
actuarial risk logic into sentencing processes; and (2) identify questions requiring further
empirical examination.
I used multiple sources of data to inform the analysis I present in this paper. These
included: the international research on risk; criminal justice policy; a selection of Canadian legal
cases; a content analysis of common risk tools, their user manuals, and interview guides; and
interviews with 127 practitioners (crown attorneys; defense lawyers; policy makers; probation
officers; prison staff; and risk instrument developers, trainers, and researchers). My focus in this
paper is on how sentencing decisions may be affected by actuarial risk instruments that are
predictors of general recidivism. 3 While some risk tools can be administered only by
psychiatrists or psychologists, general risk tools can be used by any professional (i.e., probation
officer, social worker, police, and program personnel) who has completed required training. A
considerable number of examples used in this paper reference the Level of Service Inventory–
Revised (LSI–R). 4 The LSI–R is one of the most common tools used in Canada and the United
States (Maurutto and Hannah-Moffat, 2006; Peters and Warren, 2006) and is perhaps one of the
most researched instruments for predicting general recidivism.
In this paper, I first briefly outline the emergence and entrenchment of probabilistic
reasoning in criminal justice decision making and the more recent extension and application of
actuarial risk logic to sentencing. Then, acknowledging that risk has managerial and
organizational benefits, I argue that risk can be used to generate a culture of penality that relies
less heavily on incarceration and recognizes its criminogenic effects. 5 Prison minimization is a
laudable goal, but using actuarial risk to achieve it is not without consequence. Before risk is
adopted as an organizing principle for sentencing, I argue that further research about the
following is required: (1) the methodological structure and varied logics of risk; (2) the effect of
actuarial risk models on individuals and groups of criminal defendants; (3) the legal relevance
and epistemological basis of risk; and (4) the organizational and policy impact of risk-need
technologies.
4|Page
The Growth and Logic of Risk Assessment
Risk is a flexible construct that can support seemingly incompatible principles (Ewald,
1991; O’Malley, 2004; Hannah-Moffat and O’Malley, 2007). It is easily aligned with outwardly
incongruous managerial and jurisprudential concerns about public safety, rehabilitation, justice,
and efficiencies (Hannah-Moffat, 2004). Thus risk has considerable institutional appeal; in fact,
it has become a focal point in the criminal justice system. Risk instruments are used for a range
of different purposes. How risk is used in a given jurisdiction is determined by law, policy and
sentencing purposes. For example, in Canada sentencing laws for youth, Aboriginals and
dangerous offenders are different as is the assessment of ‘risk’ (Maurutto and Hannah-Moffat
2006; 2007). Accordingly, the use of ‘risk’ produces an assortment of problems. The nature and
severity of these complications will vary by, and within, the jurisdiction (or sentencing regime).
The assessment of risk and risk tools has significantly changed over the past 40 years.
Although current practices of risk assessment are influenced by local needs and law, it is possible
to identify four generations of risk assessment tools within this time frame. The first generation
of risk tools was based on clinical prediction—the unstructured clinical judgment of skilled
practitioners. This method was discredited as subjective and unempirical, and critiqued for its
poor predictive accuracy. Research supporting the use of actuarial instruments continues to
consistently report that actuarial risk scoring provides more accurate assessments of risk than
clinical judgments based on professional training and experience (Andrews and Bonta, 2006).
In the 1970s, new “evidence-based” risk technologies that relied on statistical prediction
emerged. These tools assign a quantitative risk score to an offender by assessing individual
factors (e.g., history of substance abuse, age at first offence) that have been statistically linked to
the risk of recidivism in correctional populations. These tools were developed using research on
large population samples. These second-generation risk assessments, which are still used in
many jurisdictions, use static historic factors, such as an offender’s age, gender, and number and
type of convictions in order to make predictions about the offender’s likelihood of recidivism.
The presence of a risk factor equates with a score of one; its absence with a score of zero. Factors
are tallied and the higher the score, the higher the probability of reoffending. Examples of these
tools currently being used are the Salient Factor Score (in the United States), Static 99 (United
5|Page
States, Canada) the Statistical Inventory on Recidivism (SIR) (in Canada), and the Risk of
Reconviction (in the United Kingdom). 6
While these tools are typically evaluated as “better predictors of recidivism than clinical
judgments” (Ægisdóttier, White, Spengler et al., 2006; Andrews, Bonta, and Wormith, 2004;
Grove, Zald, Lebow, Snitz, and Nelson, 2000), criminal justice and treatment professionals
critiqued them because of their rigidity and over-reliance on inert offence-based risk criteria.
They were thought to produce a “fixed” prediction of risk based on accumulated historic and
immutable factors. Static risk logic implies that an offender’s risk level cannot be reduced
because the variables used to predict it do not change. This conceptualization of risk and the
offending subject limited the scope of correctional management and provided little guidance
about interventions to correctional professionals (Hannah-Moffat, 2004). Nonetheless, many
static (second generation) risk instruments are being developed and used to inform decision
making.
Don Andrews (1989:5-6), a pioneer of third-generation risk assessment, claimed that
improving the predictive accuracy of risk assessments required that the tools move beyond their
reliance on static risk factors to include offender characteristics and circumstances that can be
changed during incarceration. This proposition, which is rooted in the presumption that some
correctional treatment actually works, stimulated a third generation of risk research. The logic of
‘what works’ rehabilitative rehabilitation is central to the design of third generation risk tools
that focus on areas of an individual’s life that can be altered through effectively delivered
correctional treatment. Although some static factors, such as criminal history, remain important
features of third-generation tools, additional factors in an offender’s life that can change over
time have been added to the tools. For example, the new tools attempt to capture and categorize
data on present employment (one can lose a job or find a job), criminal friends (one can make
new friends and lose old friends), and family relationships (which can change from supportive to
unsupportive or vice versa) (Bonta and Andrews, 2007:4). The changeable factors highlighted in
third generation tools are known as “dynamic risk factors” or “criminogenic need”. Some tools,
such as the Youth Level of Service Inventory, include “protective factors,” which are positive
influences that can improve the lives of individuals or enhance public safety. These factors can
decrease the likelihood of recidivism and counterbalance risk factors. Some examples of
6|Page
protective factors include stable employment or housing, access to social services, positive selfesteem, and positive attitudes, values, or beliefs. Additional factors for youth are parental
supervision and strong parenting skills, social support, and positive role models and peer groups.
Third-generation risk instruments are commonly referred to as “risk-need” instruments:
However, since offenders’ needs are predetermined as factors shown to be statistically co-related
with recidivism in aggregate offender populations, the needs assessment in these tools is not
“individualized” or self-reported. These tools are designed to align risk prediction with the
management of offender needs that are empirically designated as treatable; by default, this logic
categorizes some needs as illegitimate targets for intervention. In other words, a distinction is
made between criminogenic and non-criminogenic needs. While non-criminogenic needs (i.e.,
poor health) may be important, they have not been demonstrated in research to be related to
recidivism and are consequently considered low priority for intervention except for “humane”
reasons (see Hannah-Moffat 2004). The LSI–R, SARA (Spousal Abuse Risk Assessment),
SAVRY (Structured Assessment for Violence Risk among Youth), HCR-20, PCL–R
(Psychopathy Checklist revised), and SONAR (Sex Offender Need Assessment Rating) are
examples of these tools. These tools, which incorporate changeable risk and need factors, have
been gaining popularity since their introduction in the mid-1980s.
Before discussing the newest fourth generation risk-need technologies, it will be useful to
briefly examine the logic of dynamic risk (criminogenic need), which remains a central feature
of fourth-generation risk tools. The concept of dynamic risk is derived from the Risk-Needs
Responsivity Model (RNR) (Andrews and Bonta, 2006). This model is considered central to the
delivery of “effective” correctional treatment programs (Andrews, Bonta, and Hoge, 1990:19),
whereby “effective treatment” is understood to produce a measureable reduction in the
propensity for recidivism. Within the RNR model, the risk principle is an endorsement of the
premise that criminal behavior is predictable and that treatment services (often cognitive
behavioral interventions that claim to “teach” and not “treat,” as previous rehabilitative
connotations suggest) need to be matched to an offender’s level of risk; “high risk” offenders are
targeted for the greatest number of interventions. 7 The needs principle targets for treatment an
offender’s dynamic risk factors, or criminogenic needs. Correctional researchers established a set
7|Page
of criminogenic needs, some of which I identified earlier, by identifying variables that prior
research had empirically correlated with recidivism and that are amenable to intervention.
Proponents of this model argue that “evidence of dynamic validity, that is, changes in risk
scores signal changes in the likelihood of committing a new offence, is immensely important for
correctional programs and the staff charged with managing offender risk. The third-generation
risk-need instruments offer a way of monitoring the effectiveness, or ineffectiveness, of
programs and supervision strategies” (Bonta and Andrews, 2007). Knowledge of dynamic needs
allows correctional officials to target their interventions and prioritize scarce correctional
resources.
The targeting of interventions is critically linked to the responsivity principle, which
refers to the matching of styles and modes of intervention to the learning styles and abilities of
offenders (Andrews et al., 1990:20). It requires attentiveness to: (a) how diverse populations
respond to various treatment options; and (b) “specific” responsivity factors (i.e., self-esteem,
motivation, personality traits, life circumstances, and therapeutic relationships) that may
facilitate or impede an individual’s response to intervention (Ogloff and Davis, 2004:233). The
concept of responsivity is also critical to how non-white and female offenders are managed. As I
discuss in detail below, risk instruments rarely distinguish between racialized or ethno-cultural
groups of men and women—gender, culture, ethnicity, and race are only considered within the
context of responsivity (i.e., how an offender from a particular ethno-racial group and/or gender
may respond to an intervention based on population-level data) (Hannah-Moffat, 2009).
The responsivity principle features more prominently in the newest convention in risk
assessment and classification, which uses strategies and tools that “systematically bring together
information about an offender’s history and needs to develop a treatment plan and assign levels
of supervision” (Bonta, 2002:1). These fourth-generation tools (i.e., LSI–CMI) adopt the same
basic approach as third-generation tools, but refine assessments of risk and need so that they
align more directly with case management priorities (Andrews, Bonta, and Wormith, 2004;
Bonta and Wormith, 2008; Maurutto and Hannah-Moffat, 2006). The fourth-generation risk
assessments are yet not as widely used (Andrews, Bonta, and Wormith, 2006), but a considerable
amount of risk-need research is now devoted to determining, measuring, and categorizing
responsivity factors in keeping with their centrality to this newest risk logic.
8|Page
With the evolution of RN -based risk instruments, some authors claim that there is “little
justification for the continued use of professional judgment to make decisions related to risk”
(Bonta, 2008a:1) and that “any correctional agency that has the goal to reduce recidivism should
use, at a minimum, third generation risk-needs assessment instruments” (Bonta and Wormith,
2008). These authors also claim that risk-need tools can help corrections staff efficiently and
effectively allocate resources for case management. This logic has been accepted at a policy
level in a number of jurisdictions, resulting in the integration of the LSI–R (or variations) into
the preparation of pre-sentence reports and in sentencing guidelines that explicitly stress riskneed assessment as an evidence-based technology.
These types of actuarial instruments enable new ways of understanding risk and
“knowing the offender.” In particular, third- and fourth-generation risk instruments reassert the
premise that an offender’s risk of recidivism can be changed if knowledge of their needs is
integrated into assessment technologies and then “correctly” targeted in correctional
interventions. Importantly, the RNR model emerged when few supported the continuance of
rehabilitation. The RNR model has informed the development and refinement of a series of
administrative processes and risk tools currently used to target treatment, direct case
management plans, accredit treatment programs, and make decisions about institutional
resources. The RNR has played a pivotal organizational role in offender management by
advancing a necessarily narrow and targeted view of rehabilitation.
At the same time, there are significant problems with RNR derived risk logic 8 . Both
third- and fourth-generation risk tools are fundamentally different from second-generation static
risk assessments because they purport to guide the “treatment or rehabilitation” of the offender to
prevent reoffending, rather than simply predict recidivism. These tools embed the assessment of
risk and need in a utilitarian theory of punishment. The application of sanctions under this risk
model should support the reduction of recidivism. For example, probation conditions can be
aligned with the results of risk assessment and enable the probation office to stream an offender
into a program that targets his/her areas of risk-need.
Using this risk –need logic, sanctions reinforce the principle of effective correctional
interventions by using the results of the risk-need assessment to target treatment regimes that
have empirically demonstrated reduction in criminogenic need, and thus probabilities of
9|Page
recidivism. This approach to risk differs importantly from the correctional use of static risk
(second generation tools) for preventive or selective incapacitation or to deter recidivism through
the administration of harsh penalties, which is another possible outcome of risk-based
sentencing. Nonetheless, defendants who score high on third generation (rehabilitative oriented)
risk-need scales will still continue to endure incapacitation, especially if they are designated as
non-responsive to treatment and unwilling to participate in regimes “empirically proven”’ to
reduce their probable risk. Offenders are required to participate in programs designed to address
their risk-need area(s). Their success in these programs, which hinges on their ability to “gain
insight” into their criminogenic factors and to demonstrate a capacity for change and risk
management, plays a significant role in their correctional placement and access to parole
(Hannah-Moffat and Yule, forthcoming; Hood and Shute, 2000a; 2000b; Padfield, 2007;
Padfield and Liebling, 2000; Padfield, Liebling, and Arnold, 2003; Shute, 2007).
The Limitations of Risk in Sentencing
In this section, I argue, the fact that risk instruments do not use general population data
suggests that it is difficult, if not impossible, to determine the extent to which other underlying
contributors such as the social and economic disadvantage of certain groups, economic status, or
some other factor is erroneously influencing the risk prediction. Actuarial risk (both
rehabilitative of incapacitation oriented) de-individualizes the assessment of risk by categorizing
offenders on the basis of unalterable group characteristics. This means that decisions about
community or custodial punishments, the conditions of probation, and levels of supervision are
determined based not on what offenders did, but rather on how closely who they “are”
approximates subgroups of an offender population. Categorizing individuals as risky in
comparison with an aggregate group contradicts the jurisprudential value of individualism
(Simon, 1988:776).
Seduced by Science: A Misguided “Trust in Numbers” 9 ?
Risk scores impart a moral certainty and legitimacy into the classifications they produce,
“allowing people to accept them as normative obligations and therefore scripts for action”
(Ericson and Haggerty, 1997:7). Yet, most scholars agree that our present risk knowledge does
10 | P a g e
not allow us to provide an absolute statement about an offender’s likelihood of recidivism or the
timing of potential recidivism. Nor can risk scores tell us with certainty how an offender will
recidivate, whether violently, sexually, or simply as a violation of a condition. The possibility of
making a prediction error (false positive or false negative) using a risk tool is probable, but not
easily determined—a problem that is attracting considerable empirical testing (see Netter 2007).
Several scholars have additionally questioned the ethics of punishing someone for potential
future behavior (Harcourt, 2007; Hudson, 2003; Maurutto and Hannah-Moffat, 2006 Monahan,
2006; Netter, 2007; Simon, 1993). The danger is that we shift to what Silver and Miller (2002)
and Reichman (1986) label “statistical justice,” wherein dispositions are determined on the basis
of how closely an offender matches an actuarial profile, with less significance being given to
other relevant legal criteria.
The extent to which various tools can reliably predict recidivism within specific
offending categories is a key research focus (Kleiman et al., 2007; Schwalbe, 2007; Silver and
Chow-Martin, 2002), with the majority of studies focusing on the predictive validity of tools
assessing violent recidivism (Gendreau, Goggin, and Smith, 2002). Different risk assessment
instruments have been developed to address specific offending categories, but not all instruments
appear equally capable of discriminating between high- and low-risk offenders. A risk tools’
empirical foundations and psychometric properties determine the type and quality of predictions
possible with that tool.
The use of aggregate statistics to apply individualized punishments has been critiqued on
theoretical, methodological, and ethical grounds. The fact that actuarial risk assessments are
typically created from the case files of a subpopulation of incarcerated offenders raises concerns
about the ability of any instrument to make an unbiased prediction of risk. Prison populations are
not random (Netter, 2007); they are the products of past sentencing policies and patterns and they
disproportionately represent Blacks, Aboriginals, and other socially disadvantaged groups
(Blumstein, 1982; Bushway and Smith, 2004). Some researchers claim that the prison population
does not accurately reflect who is at risk of reoffending and question the predictive ability of risk
instruments that are based on these populations. These issues are important to the of ethics
decision making because the base rate estimates for recidivism may actually be lower in the
general offender population than what is predicted on risk assessment instruments. This may
11 | P a g e
result in the possibility that a more severe penalty is administered on the basis of a risk
assessment tool that inflates the actual risk posed by certain groups of offenders. Logically the
converse may also occur. Advocates of empirically based sentencing have positioned recidivism
as a central organizing principle of sentencing, 10 and risk-need assessments as necessary
sentencing aids. These tools are variously described as using hard data, social science and
empirical evidence, and actuarially sound data to make predictions relevant to sentencing.
However, there appears to be considerable misuse (and little understanding) of the “social
science” and “hard data,” which can lead practitioners using the tools to reproduce some of the
same difficulties the tools are purported to remedy.
Risk scholars have yet to examine how tools designed to predict recidivism function with
the general population. One of the intrinsic difficulties in these tools is that recidivism itself is a
notoriously slippery concept that is difficult to operationalize and reliably measure (Maltz,
1984). Within the pool of risk tools, recidivism is variably defined as rearrest, reconviction, or
reincarceration, and does not always refer to the original offence(s). The general definitions of
recidivism used in many actuarial risk instruments do not differentiate between types of
recidivism. For example, tools such as the LSI–R offer a generic prediction of recidivism that
cannot reliably indicate if the defendant will breach a probation condition or commit a serious or
violent offence.
A noteworthy slippage: predict to cause.
The issue of risk and prediction is further complicated by practitioners’ unfamiliarity
with probability statistics and the general tendency to conflate correlations with causality. My
interviews with criminal justice practitioners demonstrated that few understand and appropriately
interpret probability scores. Despite receiving training on these tools and their interpretation,
practitioners tended to struggle with the meaning of the risk score and the importance of the
items contained in the assessment tools (Hannah-Moffat and Maurutto, 2004; O’Malley 1998).
Rather than understanding that an individual who obtains a high risk score shares characteristics
of an aggregate group of high-risk offenders, the individual is likely to become known as a highrisk offender. Instead of being understood as correlations, risk scores are misconstrued in court
submissions, pre-sentence reports, and the range of institutional file narratives that ascribe the
12 | P a g e
characteristics of a risk category to the individual. Significantly, these reports follow the offender
through the system and can stick with them for the entirety of their institutional careers, and
risk–based characterizations can have significant managerial effects (e.g., classification levels,
institutional placement, treatment access, parole release, number and type of conditions, etc.).
In practical terms, correlation becomes causation and potential risk is translated into an
administrative certainty. When used at the pre-sentence stage, the courts may assume that a
“high-risk” offender poses a greater danger to society and sentence accordingly. Risk scores,
however, merely identify who is more likely to reoffend and, in the case of third-generation tools,
identify treatment targets for correctional programming that may reduce the likelihood of
recidivism. This slippage positions “some” areas of an individual’s life as legitimate targets for
corrective evidence-based interventions. Risk scores and accompanying quantifications of
“need” are no longer probability scores, but are instead a prescription for correctional treatment.
The “abstract” risk score is converted into a correctional artefact that enables the efficient
management of correctional populations and resources and defines a person’s criminal justice
experiences.
Risk scores have a range of purposes: targeting and streaming various types of offenders
into different kinds of programs; reducing penal costs; and regulating entry into alternative
measures, diversion, community sanctions, and custody. Risk schemes can be used by criminal
justice officials to systematically target groups of high to low offenders. For example, the
Virginia Criminal Sentencing Commission developed a method for diverting 25% of non-violent,
prison-bound offenders into alternative sanction programs using risk assessment to identify the
lowest-risk offender to help inform possible diversion options for non-violent offenders
(Kleiman, Ostrom, and Cheesman, 2007).
Concerns about prediction have not impeded the expansion, development, and use of risk
instruments. Actuarial instruments continue to appeal because of their purported ability to
classify offenders based on a set of statistically relevant factors, without a reliance on clinical
discretion and because they offer the prospect of “colonizing the future.” (Giddens 1991)
Proponents of these tools deflect some critiques by stressing that the practice of identifying an
individual’s level of risk based on their membership in a classification group results in superior
predictive validity of the tools over clinical judgement alone (Kim et al., 2008). However, they
13 | P a g e
have not conclusively demonstrated that these “classification groups” are free from biases or
other non-normalised contributors to recidivism (i.e., arrest bias, practitioners’ interpretations,
gender, race or other social factors). On the level of correctional practice, James Bonta, a staunch
advocate for risk assessment, cogently observes that, “It is one thing for scientists to demonstrate
that a risk instrument or a treatment program can work but it is a very different matter to make it
work in correctional agencies with a diverse work force in terms of education, values and
experience, conflicting criminal justice policies, and management practices that are not
conducive to selecting and training of staff in effective assessment techniques” (Bonta, 2008:np).
Further sentences predicted on risk will not guarantee the availability of correctional programs.
Regrettably, there is considerable evidence of governments’ not providing resources for
correctional programming (Quigley, 2008) and, on a very basic level; it is faulty to assume that a
cohesive and directive evidence-based rehabilitative framework on how to “treat” offenders,
guide practitioners, and organize correctional regimes exists.
Gender, Race, and “Evidence-Based Risk”
Since risk instruments cannot predict who will recidivate, what should be addressed and
minimized are the factors that contribute to the causes of crime. This is certainly easier said than
done, as evidenced by the considerable academic debate on what causes crime (and who is
detected, prosecuted etc.). Nevertheless, social context—gender, race, and economic and sociostructural factors—are generally agreed to play a role. Risk assessment does not and, by virtue of
its internal logic and structure, cannot account for these variables. This fact has contributed to
the extensive debate about whether conventional risk assessment instruments can be usefully, or
ethically, applied.
I do not think that risk instruments such as the LSI–R are universally applicable and
suggest that even if their “validity” were demonstrated, these instruments may not be the most
appropriate method for identifying and responding to the therapeutic needs of female offenders
(cf. Vanhooris et al 2010; Hannah-Moffat 2009). Several scholars (Morash 2009; Belknap and
Holsinger 2006; Blanchette and Brown 2006; Bloom, Owen, & Covington, 2003; HannahMoffat and Shaw 2001; Morash, Bynum, & Koons, 1998) fault risk instruments for overclassifying women, ignoring the risk factors and needs most relevant to women offenders, and
14 | P a g e
for having little regard for the validity of specific risk instruments for women 11 (Van Voorhis et
al. 2010). The predictive reliability of existing instruments for women and racialized populations
is unclear because the criminogenic factors included in generic risk tools are derived from
statistical analyses of aggregate male correctional population data and are based on male-derived
theories of crime; a gender/race problem is therefore built into the tools (Van Voorhis & Presser,
2001). Blanchette and Brown’s (2006:140) comprehensive analysis of “some of the most
commonly used mathematical risk assessment tools clearly demonstrated that the science of
recidivism prediction of girls and women lags far behind that for males.” Feminist researchers
have repeatedly critiqued mainstream criminology for treating females as “afterthoughts,” and
for the uncritical use of male norms in the management and treatment of incarcerated women;
yet, these critiques have been ignored in the development and application of risk tools. An
analogous race problem stems from a lack of attention to the racialized nature of offending and
imprisoned populations and to the specific needs of non-white offenders. The science of risk
assessment as it pertains to racialized populations is underdeveloped and, consequently, using
conventional risk tools to predict recidivism with these populations is of questionable legal and
ethical value.
Characterizing men and women as having essentially the same universal risks and needs
reproduces the male normative criteria that feminist (and critical race) researchers have
theoretically and empirically contested. 12 The ‘sameness’ argument demonstrates little regard
for, or understanding of, the intellectual breadth and depth of feminist and critical race theory,
the sophisticated feminist and critical race critiques of methodology, or the ample empirical and
theoretical academic literature documenting how women’s crime is qualitatively and
quantitatively different from men’s, even if they commit the same type of crime. Research has
demonstrated differences in the motivational factors that lead to women’s use of violence,
involvement in drug and property crimes, patterns of substance abuse, and how factors such as
drug use are connected in gender-specific ways to initial and continued prostitution and other
crimes (for example, see Blanchette and Brown, 2006; Bloom and Covington, 2003; Daly, 1992,
1994; Hannah-Moffat and Shaw, 2001; Clarke, Monahan, and Silver, 2003; Heimer and
Kruttschnitt 2006; Moretti, Odgers, and Jackson, 2004). International data also clearly shows that
women commit few violent crimes, are infrequently repeat offenders, and that when they do
15 | P a g e
reoffend, their crimes tend not to escalate in severity (Kong and AuCoin, 2008). Empirical
analyses of risk tools, including the LSI–R, show that the criteria for establishing levels of risk
routinely pay little attention to gender, racial, or ethnic differences, or to the differing social,
economic, and political contexts in which these tools are deployed (Maurutto and HannahMoffat, 2006). For example, a recent study by Reisig, Holffreter, and Morash (2006), reported
that
the LSI–R was found to correctly classify female offender only when: (1) they did not
follow gendered pathways into offending; (2) they offended in contexts similar to that of
male offending; (3) they occupied a relatively advantaged social position in society. …
The LSI–R misclassified a significant portion of female offenders who were socially and
economically marginalized and who offended within typical gendered contexts,
particularly female offenders who were drug-connected or who could be classified as
harmed and harming women. (p. 384)
Concerns about gender and risk assessment are also relevant to tools such as the PCL–R,
which has been relatively well researched on samples of women (Brown and Blanchette, 2006).
In this case, research suggests that some gender differences in the risk factors included in the
PCL–R and their salience are evident (Blanchette and Brown, 2006:73; also see Warren et al.,
2003). Researchers continue to test whether risk assessment tools developed for male offenders
can be used accurately for female offenders (Harer and Langan, 2001). Conventional risk
assessment scholars are beginning to go beyond concerns with predictive reliability and validity
to closely examine the construct validity of non-gender-specific needs for female populations
(Brennan and Austin, 1997; Brennan, 1998; Blanchette and Brown, 2006; Van Voorhis,
Salisbury, Bauman, and Wright, 2007; Van Voorhis et al. 2010; VanVoorhis, 2005). At present,
the debate on the suitability of tools such as the LSI-R for women is significantly polarized; 13
this warrants caution about their use at sentencing.
The uncritical adoption of risk instruments is also problematic from the standpoint of race
and social inequality. Risk research has not adequately vetted tools for race. Research that has
examined the suitability of available risk instruments for racialized offenders reports that racial
minorities have different needs (Gavazzi, Yarcheck, and Lim, 2005; Petersilia and Turner. 1987;
16 | P a g e
Mitchell, 2005) and that race and ethnicity can influence practitioners’ attribution of risk factors
to racialized offenders (Hudson and Bramhall, 2005). These studies raise concerns about the
extent to which “risk assessment and risk prediction is a transcultural, transracial” phenomenon
(Severson and Duclos, 2003). Race differences in exposure to risk and vulnerability are not
presently accounted for in empirical risk studies. The logic of risk assessments is premised on a
set of assumptions that underestimates and devalues social and racial inequality, while
simultaneously decontextualizing the criminal act, the offender, and his or her history, and needs
to be carefully considered before these tools become further embedded into sentencing practices
(Hannah-Moffat and Maurutto, 2010; Harcourt, 2007; Hudson and Bramhall, 2005; Pridemore,
2004; Severson and Duclos, 2003; Durrance and Williams, 2003; Bhui, 1999).
Racialized social economic structures and context are relevant to the production and
composition of the offender population. Proponents of risk instruments often overlook the fact
that both static and criminogenic risk variables cannot be easily abstracted from the sociopolitical, economic, and cultural specificity of individuals. For example, if one examines the
general risk-need factors 14 and compares these factors with the lived reality of Blacks and
Hispanics in the United States or Aboriginal people in Canada (Maurutto, unpublished), it is
clear that these marginalized groups will unavoidably score higher on risk instruments because
of their elevated exposure to risk, racial discrimination, and social inequality—and not
necessarily because of their criminal propensities or the crimes perpetrated. Marginalized
individuals’ lives tend to be mired by a range of criminogenic and other needs, and consequently
risk scores reflect systemic factors. High risk scores are associated with custodial sentences
and/or a greater number of conditions attached to their disposition, making them more vulnerable
to breach, increased surveillance, and further criminalization. 15
Race and gender are complex social constructs that cannot simply be reduced to binary
variables and tested for significance (predictive validity and reliability) in risk instruments
(Hannah-Moffat and O’Malley, 2007). An emphasis on narrowly “validating” risk tools for use
on diverse populations that were developed using theories and research mainly about Caucasian
men’s crime may produce disparity in sentencing. When the inner logic of risk is exposed,
important theoretical and methodological concerns about the relationships between race, gender,
and social inequality and specific risk factors become evident.
17 | P a g e
The science supporting risk tools is contested and insufficiently advanced to claim that
these tools do not replicate or produce forms of systemic discrimination, or worse, as Harcourt
(2007:3) theorizes “a ratchet effect,” wherein “the profiled populations become an even larger
portion of the carceral population” with highly determined consequences on their employment,
education, family and social outcomes. Individuals who are racialized, live in poverty, are
unemployed, and/or struggle with mental illness are potentially disadvantaged by risk-need
criteria. Although generic risk - need factors (i.e., substance abuse, marital family difficulties,
and employment) are presented in existing research as “relevant” or “predictive” for women and
diverse offenders (see Vose, Cullen, and Smith, 2008), these factors are experienced differently
and have different effects. As critical race theorists aptly illustrate, it is impossible to treat
individuals fairly if they are treated as abstractions, unshaped by the particular contexts of social
life. The nuances of gender, race and social inequality are of paramount importance to the
production of a just and equitable sentence. The failure to acknowledge and meaningfully
account for differences (beyond risk) among individuals is a significant jurisprudential concern.
Discrimination and the structure of “knowledge”
In a 2003 report, the Canadian Human Rights Commission 16 (CHRC) criticized the use
of third-generation risk-need instruments and their content, arguing that some common risk
factors are systemically linked to prohibited grounds of discrimination. This is concerning, as an
incorrect assessment may produce significant burdens, barriers, or missed opportunities for
offenders. 17 Ultimately, the CHRC (2003) argued that risk-need assessment technology was
important for classifying offenders for treatment, but that the instruments should be revised to
ensure their suitability for their intended subjects. The CHRC stated that “assessment and testing
processes must be responsive to the populations to which they are applied and properly crafted to
meet the purpose they are intended to achieve. Where assessment tools do not meet these
requirements, they are blunt instruments that tend to lead to unjustifiable differential treatment.
In the absence of adequate testing and modification, these instruments should not be used
[emphasis added] on women or Aboriginal offenders” (CHRC, 2003:np).
“Empirical knowledge” is rarely positioned in a broader discussion of the range of
relevant approaches to the reduction of offending available to the court. Instead, RNR and related
18 | P a g e
risk-need approaches can be misrepresented as superior to, or more scientific than, other
approaches. Feminist/race critiques undermine the “empirical” bases of many conventional risk
instruments and raise doubts about their use with certain groups. Alternate gender or race-based
holistic approaches are equally legitimate and can produce alternative understandings of and
managerial approaches to risk. For example, reforms to Canadian sentencing law in 1996, and
the Supreme Court of Canada decision R. v. Gladue (1999), opened the door for a new normative
set of practices intended to reconstitute the racial legal subject (Rudin and Roach, 2002;
Daubney, 2002; Stenning and Roberts, 2001; Vancise and Healy, 2002, Maurutto and HannahMoffat, 2010 -LSA). The court had to address systemic racial discrimination and redress the
disproportionate incarceration of Aboriginal minorities. The R. v. Gladue decision requires that
all criminal courts consider the broader collective social history of Aboriginal defendants; this
requires a different way of understanding the risk-need of Aboriginal offenders. Hannah-Moffat
and Maurutto’s (2010) article on race, risk, and pre-sentence reports argues that although riskbased pre-sentence reports recognize issues related to race, their structure and emphasis on
actuarially based risk assessments frames race and risk differently from specialized Aboriginal
reports (e.g., Gladue reports). Gladue reports use of more holistic approaches and cultural impact
factors to analyze and contextualize risk-need. The authors maintain that the conceptualization
and relevance of race is limited by actuarial risk logic. This is a significant consideration for
jurisdictions (i.e., United States, Canada, Australia) with a disproportionate racial representation
in prison populations which adopt evidence-based sentencing practices that do not reflect the
range of evidence on race (and gender), which could compromise the “objectivity” risk-need
practices.
The use of recidivism risk assessment instruments can only be as helpful as they are
accurate. More empirical research is needed about: the composition of tools; the effects of
various risk variables on specific populations; the use and impact of risk technologies on the
management of offenders; and ancillary system effects. There is a limited assessment of the
current “scientific” evidence about the validity and reliability of various tools entering into legal
discourse. Nor is there a healthy debate about the types of risk instruments that should be
integrated into sentencing practices. Even research about risk instruments with a high degree of
“capital,” such as the LSI series of tools, has reported competing findings. Yet, few scholars
19 | P a g e
have examined the impact of risk information on sentence outcomes or applied knowledge about
differential impacts of criminal histories or race and sentencing to the study of the composition
and results of risk instruments. It is critically important to do so, because if the characteristics
associated with an assessment of “risk” are not normalized for bias (i.e., in initial arrests, crime
for which no arrest was made, racial targeting, race and gender difference in crimes committed
etc.) then the prediction (risk score) is inherently flawed.
Neutrality and Transparency
The empirical evidence of “what works” that informs risk-need is presented as morally
neutral from a realistic perspective. Conversely, critical scholars and other stakeholders are
questioning the assumptions and techniques of regulation that underpin risk frameworks and
their impact on due process, justice, and governance (O’Malley, 2004; Kemshall, 1998; Hudson,
2003; Ericson and Doyle, 2003).
Risk instruments’ score and structure tend to “black-box” and obscure the subjective and
arbitrary nature of questions and judgements associated with risk-factor scoring. For example,
risk tools’ operational manuals outline normative questions designed to determine if particular
risk-need areas are criminogenic. The nature and type of questions vary across tools, but the
questions are designed to gather information about relevant relationships 18 and circumstances, so
that practitioners may assess risk of recidivism and treatment needs. For example, regardless of
the offence, the LSI–R interview guide for the risk factor “employment” asks a series of
questions about employment and school history, including whether a defendant “get’s along
with” teachers, employers, and colleagues. The assessment questions also address sources of
income, fiscal concerns, banking and credit histories (i.e., if personal cheques bounced or were
returned for non-sufficient funds), if the defendant has a budget and follows it, whether the
defendant and significant others fight about money and/or children, whether there was an
‘unwanted pregnancy’ or infidelity, if divorce has been contemplated, and a series of additional
questions about personal and emotional issues, leisure habits, and substance use (Andrews and
Bonta, 1995). Answers to these questions form the basis of the risk-need factor score, but the
only aspect of the risk-need score visible and known to the court is the score and the
recommendation derived from the score. Neutrality is assumed in the questions, the way the
20 | P a g e
questions are asked, and in the way that the responses are interpreted and recorded. However, as
my research showed (Hannah-Moffat et al., 2009), practitioners administering risk instruments
such as LSI–R “adjust the assessment of criteria in order to control the final score, rather than
relying on formal overrides, which are an option that allow for final risk scores to be adjusted for
individual cases” (405). Risk manuals acknowledge that adjustments (overrides) may be
required, but state that these should not exceed 10% of cases. When and if a practitioner elects to
use an override, they must document the reasons for not conforming to the outcome of the risk
score. This level of accountability can deter the use of overrides and favours “criteria tinkering.”
It is also clear from the nature of these questions that assessing risk through these actuarial tools
is not a neutral statistical exercise. Assessments of risk are produced using an investigative and
interpretive process that is largely concealed within the process. Practitioners still rely on their
discretion to selectively use responses to interpret, target, and isolate facts about past
experiences, and to make claims about the probability of reoffending.
In risk assessments, the simplification and organization of facts can lead to prisoner
categorization according to crude typologies; small details tend to form the foundations of the
construction of stereotyped identities and agency. Through the use of risk assessment
instruments, the potential complexities and subtleties of meaning are compacted. Actuarial
methods can accentuate prejudices and biases that are built into law and criminal law
enforcement (Harcourt, 2007) because the opaque structure of assessment instruments makes
broader discriminatory practices less visible and contestable.
Although the information acquired from risk tools is often useful to correctional
personnel, its use in court opens a legal and evidentiary “can of worms” and raises a host of
important sentencing considerations. For example, in the United States, the federal sentencing
guidelines and case law explicitly direct that
race, sex, religion, national origin, socioeconomic status, and a disadvantaged upbringing
“are not relevant” in the determination of a sentence [sic] in addition education,
vocational skills, employment record, family ties, age, mental and emotional condition,
and substance abuse are not ordinarily relevant in the determination of a sentence
(Monahan, 2006:397).
21 | P a g e
Monahan (2006:397) notes that with the single exception of criminal history, “virtually
all of the variables that potentially could be used as scientifically valid risk factors of violence
under a forward looking consequentialist ‘crime control’ theory of punishment are explicitly
excluded from consideration in federal sentencing guidelines.” Arguably, even criminal history
is gendered and racialized; it is affected by the offender’s age, as well as policing and charging
practices (Bushway and Morrison Piehl, 2007). Implicitly, risk-need instruments make certain
variables relevant to sentencing because they are the foundation of the score.
Pre-sentence reports (PSRs) frame legal subjectivities for the court; they inform the court
about the offender’s risk and treatment potential, and contribute to the type, length, and
conditions attached to the final sentences imposed. 19 Some of the data from risk assessments is
of questionable legal relevance to sentencing and are obtained via “collateral contacts,” without
the benefit of legal protections, cross-examination, or authentication. The use of such collateral
contacts (who are not subject to cross-examination), subjective interpretations, and extralegal
criteria by probation officers to compile the risk score and to formulate recommendations about
sentencing can undermine broader principles of sentencing jurisprudence (Cole and Angus,
2003:17). 20 This problem is particularly acute when practitioners’ recommendations do not
disclose the use of a risk instrument. For example, in Ontario, Canada, probation officers are
mandated by policy to use the LSI–R to complete a PSR, but they are also told to not disclose the
score or the fact that they used the tool in their court reports, an omission intended to prevent
probation officers from being called to the stand for cross-examination on the risk instruments. 21
Yet, research demonstrates that the assessor’s choice of informants and their interpretation of the
authenticity of the informants’ claims determines what information “counts” and becomes part of
the “official” record (Ballucci, 2008; Maurutto and Hannah-Moffat, 2006). At minimum, courts
should require the disclosure of the use of risk instruments in the reports and formulation of
sentence recommendations, particularly because research reports high rates of concordance 22
between the recommendations of pre-sentence reports and actual sentences (Cole and Angus,
2003; Hannah-Moffat and Maurutto, 2010).
Legal Standards and Quality of Evidence
Wellford (2007) reminds us that
22 | P a g e
“[o]ur understanding of sentencing has been limited by the narrow view we have taken of
what and who influences sentencing [...] United States v. Booker [U.S. 220 (2005)] and
Blakely v. Washington [124 S. Ct. 2531 (2004) (O’Connor, S., dissenting)] draw attention
to the fact that whatever is used in sentencing should be subject to review and debate
[emphasis added]. Although the justices think of this as a legal standard to be met, we as
researchers should recognize that our goal is the same—understanding all of the factors
and actors that actually determine criminal sentences.” (p. 400)
Courts are swayed by empirical evidence and expertise. Litigation in some jurisdictions
notes that aggravating factors that expose a defendant to higher penalties must “be proven by a
higher standard of evidence, beyond a reasonable doubt before a jury rather than a
preponderance of evidence before a judge” (Hunt, 2007:438). In a similar vein, Quigley, a
Canadian lawyer (2008:531) argues that, “[w]hile the accumulation of evidence is common to all
of these disciplines, law assesses evidence in a context bounded by values, rights (particularly
constitutional rights), rules, and principles that are broader, and, therefore, because of rules
concerning the inadmissibility of certain types of evidence, more constraining than those
governing scientific inquiry.” This rigor should apply to risk-need assessment because the score
can have significant consequences for the offender in terms of the opportunities provided, levels
of supervision, and the conditions imposed. Although contested in our adversarial culture, expert
knowledge is established and attributed authority and legal relevance. What is remarkable is that
few court actors are aware of, or concerned about, how risk scores are produced and interpreted,
or the contested science of risk prediction, especially as it applied to non-specific offending
categories and “general” recidivism. 23
The courts in various jurisdictions generally accept these technologies and are reluctant to
debate their validity. Canadian court cases that raise concerns about the LSI–R have typically
sidestepped the debate on jurisdictional grounds and have deferred to the administrative
authorities of probation departments to independently determine the suitability and type of risk
instrument to use during assessments (Hannah-Moffat and Maurutto, 2010; 2007; Cole and
Angus, 2003; Cole, 2008). Although the American court context is considerably more complex,
similar tendencies exist. For example, in June 2010 the Indiana Supreme Court ruling in
Malenchik v. Indiana 24 on the probation department’s use of the LSI–R held that “legitimate
23 | P a g e
offender assessment instruments do not replace but may inform a trial court’s sentencing
determinations and that, because the trial court’s consideration of the defendant’s assessment
model scores was only supplemental to other sentencing evidence that independently supported
the sentence imposed.” This judgement, like many others, sidesteps broader questions of the
impact and relevance of ‘general’ risk-need determinations at sentencing.
Scholars, practitioners, and law and policymakers are more regularly claiming (without
careful examination) that accurate prediction of risk can be achieved and is useful to a sentencing
decision. This infers that the use of risk assessment technologies contributes to the production of
a “just sentence.” The concept of justice is complex. Scholars on both sides of the risk and
evidence-based sentencing debates mobilize justice-based concerns to validate their positions.
More legal and empirical analysis is required to tease out how risk may enhance correctional
practices in some cases, but compromise jurisprudential ideals of justice in others. One salient
debate to consider is the extent to which differences and social context are important to
sentencing.
Few courts have entertained an analysis of these instruments or require expert testimony
on the psychometric properties of tools and their supporting research, even in cases where local
tools, which are not universally accepted or empirically vetted, are used. Too little empirical and
socio-legal research examines how factors such as race, gender, educational attainment, and
mental health impact sentencing practices, and how forms of discrimination can be reproduced
and possibly magnified through the uncontested use of actuarial risk instruments as exceptions
(Harcourt, 2007). Greater transparency and thought should be devoted to how practitioners
communicate risk information to legal decision makers. Thus far it appears that the courts and
others are accepting a relatively weak standard of scientific evidence when using risk-need
instruments.
A Piece of a Puzzle: Weighing Risk Information
Research has yet to investigate how judges and practitioners use and weigh risk
information in decision making. Interestingly, research reports that judges and legal practitioners
are generally supportive of risk instruments, but they know little about risk technologies (Bonta,
Bourgon, Jesseman, and Yessine, 2005). Research reports that judges often will use the clinical
24 | P a g e
opinions that fit their perceptions of risk (Scott, 2008), and that they use the information that is
available to them to predict risk and assign punishments (Bushway and Smith, 2007), but little is
actually known about what factors judges consider in their own professional assessments of risk
(Vigorita, 2003:361). Judicial decisions, including risk decisions, are likely based on a small
number of factors, either to simplify the decision-making process or because certain factors are
the attributes associated with the typical case (Vigorita, 2003:364). If individual factors could be
identified, it would be essential to determine how they are actually used in courts.
I have challenged the assumptions that lawyers, judges, and other relevant professionals
are familiar with the limits and strengths of actuarial risk instruments and that they can correctly
interpret and apply the findings to make recommendations for offenders and sentence them. Part
of the problem is that actuarial risk assessment instruments are created from subjective processes
that rely on a particular body of largely uncontested social scientific evidence. I am not implying
that there are no debates in the literature on risk and “effective correctional intervention,” but
rather that thus far, research on evidence-based sentencing has not closely studied the
epistemological architecture of risk-need evidence. It is clear that risk-need models conflict with
other equally important ethical and legal commitments to race- or gender-responsive correctional
treatment (Hannah-Moffat, 2009; Hannah-Moffat and Maurutto 2010).
In totality, risk instruments demonstrate an increasingly refined capacity to sort and
classify criminalized populations. Even in light of the recognition that actuarial risk assessment
instruments may be more effective than clinical assessment, there remain some unanswered
questions in the literature. There also remains a gap with regard to how much influence these
instruments actually have in determining an offender’s principle sentence. Since risk practice
should result in a redistribution of penal populations, a full appreciation of its impact would
require an integrated study of sanctions at multiple jurisdictional levels. In the United States, this
would require an examination of the use of risk-based sentencing practices in federal and state
sentencing courts (Wolfe, 2008). It would be useful to examine if the flow and outcome of cases
in various types of sanctions (probation, imprisonment, and diversion) change with the
introduction of evidence-based sanctioning. However, caution should be exercised in any
attempt to study the outcome of programs into which offenders are pre-selected through their
25 | P a g e
risk-need score, because there is an obvious selection bias created by risk-need-based selection
into programs or sanctions.
Unresolved Debates: Risk and Competing Theories of Punishment
The actuarial risk instruments are highly structured, empirical assessments that target
potential rather than actual crime and offenders. Accordingly, actuarial sentencing stresses
recidivism risk. Scholars have identified a number of important conceptual difficulties associated
with a reliance on actuarial risk at sentencing. For example, some argue that the incorporation of
risk technologies into sentencing prioritizes forward-looking theories of punishment and masks a
range of concerns about disparity, discrimination, and “just punishments.” Critical scholars have
characterized actuarial technologies as a negative development, which can result in racial
targeting, deindividualization, social exclusion, a prioritization of recidivism and future conduct,
and the devaluing of social context (Simon, 1988; Feeley and Simon, 1992; 1994; Hudson, 2003;
Hannah-Moffat, 2004; Maurutto and Hannah-Moffat, 2003; 2006; 2008; Monahan, 2006;
Harcourt, 2007; Netter, 2007)
O’Malley (2004; 2008), however, argues that the focus on risk’s negative side fails to
adequately account for how risk embodies a heterogeneous array of practices with diverse effects
and implications. Following François Ewald (1991), he argues that risk is an abstract technology,
which is always shaped and given effect by specific social and political rationalities and
environments (O’Malley, 2008:453). Extending this point, I believe that law and criminal justice
contexts play a pivotal role in shaping understandings of risk and how it should or should not be
incorporated into practices such sentencing. A quick survey of the use of risk instruments in
Canada, the United Kingdom, Australia, and the Unites States reveals a general lack of
consensus on the suitability, use, and actual role played by actuarial instruments in sentencing.
Some jurisdictions provided judges with risk instruments and explicitly required the use of risk
score to determine appropriate sanctions. In other jurisdictions, the use of risk instruments is
most pronounced in sexual or violent-offender cases where time and severity of offence both
allow for, and require, a detailed consideration of the offender’s future conduct. Risk instruments
are commonly used by probation officers to write pre-sentence reports and make sentence
recommendations, as well as to facilitate the post-sentence management of offenders. Parole and
26 | P a g e
correctional officials have a long-standing interest in the use of actuarial tools for classification
and decision making.
Actuarial risk is a seemingly progressive penal practice that operates in accordance with
the institutional conditions in which it is realized. Despite their conceptual and methodological
deficiencies, it is unlikely that the emphasis on risk will dissipate. Consequently, the appeals to
actuarial sentencing can simultaneously support two distinct utilitarian goals of sentencing:
incapacitation and reformist intervention. The later reformist logic, although not without
technical, ethical, and conceptual limits, can shift sentencing practice.
I would like to consider how the appeal of risk thinking can enable a return to a more
therapeutic approach to sentencing and possibly reduced reliance on incapacitation, which is in
danger of spiralling out of control. This is not, however, to suggest that the framing of
“treatment” in a risk model is unproblematic (see Hannah-Moffat, 2004; 2007; Maurutto and
Hannah-Moffat, 2006) or that the jurisprudential concerns about how actuarial risk undermines
just desserts and, more precisely, the principle of proportionality is insignificant. To the contrary,
there are a number of persuasive and astute arguments for limiting the role of actuarial risk logics
in sentencing decisions due to its conflict with existing principles and theories of “just”
punishments (Simon, 1988; 2005; Hudson, 2003; Maurutto and Hannah-Moffat, 2006; 2007;
Monahan, 2006; Harcourt, 2007). This is particularly the case when a jurisdiction explicitly
prioritizes proportionality in sentencing (i.e., Canadian Youth Criminal Justice Act; HannahMoffat and Maurutto, 2004; 2008).
Putting aside the importance of proportionate sentencing for a moment, 25 it is worth
considering how certain types of risk assessment logic might operate to change the type of
sanction an offender receives and the consequences of these shifts. For example, if the risk-need
principle underpinning third-generation risk assessments could be judiciously implemented, then
a portion of offenders could theoretically benefit from a “less severe sanction.” Accordingly, a
risk score could help determine who can be diverted out of the system or sent to a specialized
court or treatment program. Recall that the risk principle states that treatment services need to be
matched to an offender’s level of risk; offenders who present a high risk are targeted for the
greatest number of therapeutic interventions, and those who are low risk are targeted for the
least. Some evidence provided by proponents of the RNR suggests that too much intervention
27 | P a g e
with low-risk offenders can escalate risk (Andrews, 1991:11). Thus the low-risk offenders could
be diverted out of the system in the interest of reducing recidivism, and offenders with concrete
interventions and supervision plans are more likely to be characterised by the court as
“manageable risks.” This possibility is supported by recent evidence suggesting that judges are
more inclined to impose a community sanction when they receive information on how to manage
an offender’s risk (Dolores and Redding, forthcoming; cited in Maurutto and Hannah-Moffat,
2010). Further, Dumanis (2009:25) argues that “as a culture, we have taken the rehabilitation out
of prison and focused on punishment, creating an untenably large prison population. Now, we
are forced to rethink our way of imposing justice. Evidence-based sentencing merges punishment
with rehabilitation. Imposing a sentence with appropriate conditions based on the defendant’s
individual risk to reoffend and need for treatment or programming.” The use of risk-need can
reinvigorate rehabilitation; however, it should not be done in a vacuum. There is ample empirical
evidence of the unintended consequences of well-meaning “rehabilitation.” Risk-need informed
probation conditions and treatment requirements evoke the causal slippage discussed earlier and
make the marginalized offenders more susceptible to increased numbers of conditions and thus
surveillance and possible breaches.
To use an imperfect risk instrument to “ratchet up” sentences of high-risk offenders (i.e.,
sex offenders) or to impose indefinite sentences is considered by some to be a “misuse of
evidence-based sentencing” (Etienne, 2009:59; also see Harcourt, 2007). But, methodological
capacities aside, the use of risk instruments to seemingly customize sentences through the
provision of targeted interventions and clear strategies of risk management is persuasive. Some
evidence suggests (Dolores and Redding, 2009) that judges were more likely to release to the
community when the risk assessment included information on risk management than when it
only provided a prediction of risk level. Risk-need assessment is being popularized as a
reasonable way of restricting custodial populations, reinvigorating rehabilitation, and enhancing
public safety through “anticipated” reductions of “recidivism.”
However, is the fact that the reduction of recidivism is anticipated but not certain,
important in this context? Answering this question will require more empirical evidence to
determine if these intentions are realizable and with what effect. Given the limitations of existing
risk instruments, it is not clear whether or not the administration of risk will reproduce existing
28 | P a g e
racial inequalities, unfairly target specific groups, and result in increasingly layered and severe
penalties.
Risk can offer a politically defensible way out of mass incarceration, but part of the
empirical dilemma with risk is that it is fluid and interpretive. Risk levels and thresholds often
vary from institution to institution and sometimes indiscriminately vary over time even when the
same risk instrument is deployed (i.e., Virgina, LSI local population norms). Wormith (1997)
argues that “the number of risk levels in any scale or instrument is decided arbitrarily by the
developer or the agency using it.” As resources contract or expand, the tools can be adjusted
accordingly. Correctional management scholars (White, 2004; Kreamer, 2004; Dal Pra, 2004)
indicate that the development and adoption of particular risk tools (actuarial risk versus riskneed) are decisions that should be made in concert with an evaluation of correctional mission
statements and resources. Such claims confirm that risk technologies can be and are used
instrumentally by correctional organizations. Clearly, corrections organizations select the risk
technologies that “fit” their agency’s vision and mission (Maurutto and Hannah-Moffat, 2006;
2003). Averting disaster has become a political imperative for governments anxious to safeguard
their reputation against adverse fallout that comes if harm happens (Zedner ,2008). It is possible
that the supporters of risk approaches do not share the goal of reducing crime or of matching
offenders with the programs that “work,” but rather seek only to defensibly manage
populations. 26
Actuarial risk has both supporters and detractors who argue “against prediction”
(Harcourt, 2007), cautioning law and policy makers about some of the assumptions and
ambiguities of actuarial technologies. To what end could risk assessments be jurisprudentially
relevant to sentencing for judges concerned with crime prevention, recidivism and effective
interventions? In this context, a probabilistic statement of risk and systematic weighing of risk
factors may help a judge craft a sentence and apply meaningful conditions. New risk
configurations inherent in tools like the LSI are important not only for theorizing the concept of
risk, but also for understanding new and evolving penal strategies (Maurutto and Hannah-Moffat,
2006).
Pitfalls and Potentials: Questions Left Unanswered
29 | P a g e
Courts embracing risk are joining other institutions that believe “the world can be made more
secure by ever more perfect knowledge of risk. This leads them to search incessantly for
whatever rules, formats, and technologies will allow them to feel that they are closer to
perfection” (cf. Ericson and Haggerty, 1998 on police). But imperfect rules, technologies, and
predictions will remain. The fact that risk appears to be an efficient, empirical alternative ought
not to insulate its normative dimensions. Questioning the technical efficiency of a risk approach
is important, but it invites improvements or technical refinements and does not tackle the prior
question of whether or not it is justifiable (Zedner 2008:360).
Actuarial risk instruments systematically organize a diverse range of information about
an offender to guide practitioners 27 through a logical and simple process to itemize and score that
information. Theoretically, risk templates ensure that nothing is overlooked or missed when
reviewing an offender or case history. Risk instruments are structured to produce a managerial
form of defensible, consistent decision making. They ensure that to a certain degree all
practitioners use the same variables to make decisions. Interviews with practitioners revealed
that risk instruments are embraced because the tools standardize decision-making criteria,
enhance the defensibility of decisions, and ensure that all the players in the system are working
with the “same information,” making case files easier to transfer (Hannah-Moffat, Maurutto,
Turnbull, 2009). The use of risk instruments can have considerable cache with “elected” judges
and prosecutors who must defend their decisions to an electorate concerned with security.
Interviews with practitioners in discretionary decision-making contexts consistently showed that
they believed actuarial risk scores can neutralize politics. Institutionalization of risk insulates
practitioners who follow policy guidelines, scapegoats those who do not, and creates new forms
of organizational accountability. Risk instruments, regardless of their flaws, foster greater
professional confidence in the system because they appear objective, rational, and empirical. But
are they?
The uncritical acceptance of science and related risk technologies can jeopardize due
process, produce disparities and discrimination, undercut proportionality, escalate the severity of
sentences, and punish individuals for crimes that they have not committed. However, risk (and
some evidence-based practices) may facilitate a reduction in penal populations, and over time
lead to the application of different and perhaps more constructive interventions. Acknowledging
30 | P a g e
this possibility, I remain concerned about how introductions of risk into sentencing will shape
punishment and impact already disadvantaged and unmotivated or treatment-resistant defendants
and further how the use of actuarial methods can accentuate the prejudices and biases that are
built into law, punishment, and criminal law enforcement (cf. Harcourt, 2007).
This discussion has prompted a number of questions for which there are presently
insufficient answers:
1.
To what extent should risk predictions inform sentencing practices?
2.
How should risk predictions be balanced with wider sentencing goals?
3.
What is the impact of using risk prediction in conjunction with other sentencing
guidelines (i.e., mandatory minimums) and priorities (i.e., due process, just desserts,
etc.)?
4.
Are the risk instruments used in given sentencing jurisdiction responsive to the
populations to which they are applied?
5.
Are the instruments used properly crafted to achieve their intended purposes? This
question raises concerns about using risk instruments in sentencing without first
considering their compatibility with broader sentencing jurisprudence and guidelines.
6.
Are racialized populations disproportionately represented in risk categories?
7.
Does risk-need assessment increase sentence uniformity?
8.
How does risk assessment data compare with general population data?
9.
What are the implications of not using the risk assessment device when tools exist?
If a tool is used, then how does one reconcile conflicting understandings about how it
should be used to predict and/or sentence? For example, some of the research
supporting risk tools argues that the combination of actuarial scores with clinical
judgments inevitably produces lower accuracy than actuarial scores alone, 28 and that
“unaided” clinical judgment is less accurate than actuarially aided judgments. This
raised the interesting paradox of whether judges are, or even should be, concerned
with the accurate prediction of recidivism and the legal relevance and weight of that
information.
10. Given the technical and conceptual limits of risk, how and when can these tools be
meaningfully used?
31 | P a g e
The structural integration of risk into sentencing guidelines can restrict judicial discretion
to varying degrees. However, it also simultaneously shifts discretion to the organizations and
individuals producing risk templates and providing courts with risk scores. These organizations
are less occupied with the normative concerns of law and justice. Although law is occasionally
labelled a science, it is also seen as an “imperfect art” (Quigley, 2008:531). If sentencing is to
assume an “evidence-based” focus, then it would be prudent for legislatures and legal scholars to
integrate the range of evidence on topics such as risk and correctional intervention. I share the
scepticism of socio-legal scholars (Slobogin, 2005; Hudson, 2003; Harcourt, 2005; Simon, 2005;
Monahan, 2006) who in varying ways advocate for courts and legal scholars to pay greater
attention to how, for what reasons, and based on what “evidence” we can justify the deprivation
of liberty based on risk scores. Arguably, we should pause to reflect on the complexities of riskneed assessments and concordant calls for and against evidence-based risk jurisprudence.
References
Ægisdóttier, S., White, M. J., Spengler, P. M. et al. (2006). The meta-analysis of clinical
judgment project: Fifty-six years of accumulated research on clinical versus statistical
prediction. Counseling Psychologist, 34, 341–382.
Andrews, D. (1989). Recidivism is predictable and can be influenced: Using risk assessments to
reduce recidivism. Forum on Corrections Research, 1, 2.
Andrews, D. A. (2001). Principles of effective correctional programs. In L. L. Motiuk & R. C.
Serin (Eds.), Compendium 2000 on effective correctional programming (pp. 9–17).
Ottawa: Correctional Services of Canada.
Andrews, D. A., & Bonta, J. (1995). The Level of Service Inventory – Revised. Toronto: MultiHealth Systems.
Andrews, Don A. and Bonta, J.A. ( 2003). Psychology of criminal conduct, 3rd Edition.
Cincinnati, OH: Anderson.
Andrews, D. A., & Bonta, J. (2006). The psychology of criminal conduct, 4th Edition.
Newark, NJ: LexisNexis.
32 | P a g e
Andrews, D. A., & Bonta, J. (2007) Risk-need-responsivity model for offender assessment and
rehabilitation. Ottawa: Public Safety. Retrieved from
http://www.publicsafety.gc.ca/res/cor/rep/_fl/Risk_Need_2007-06_e.pdf
Andrews, D. A., Bonta, J., & Hoge, R. D. (1990). Classification for effective rehabilitation:
Rediscovering psychology. Criminal Justice and Behavior, 17, 19–52.
Andrews, D. A., Bonta, J., & Wormith, S. J. (2006). The recent past and near future of risk
and/or need assessment. Crime and Delinquency, 52, 7-27.
Andrews, D. A., Bonta, J., & Wormith, S. J. (2004). The Level of Service/Case Management
Inventory (LS/CMI). Toronto: Multi-Health Systems.
Andrews, D.A., & Dowden, C. (2008). The risk-need-responsivity model of assessment and
human service in prevention and corrections: Crime-prevention jurisprudence. Canadian
Journal of Criminology and Criminal Justice, 49, 4, 439–464.
Andrews, D. A., & Dowden, C. (2006). Risk principle of case classification in correctional
treatment: A meta-analytic investigation. International Journal of Offender Therapy and
Comparative Criminology, 50, 88-100.
Baker, T., & Simon, J. (2002). Embracing risk: The changing culture of insurance and
responsibility. Chicago: University of Chicago Press.
Ballucci, D. (2008). Risk in action: The practical effects of the youth management assessment.
Social Legal Studies, 17, 175.
Belknap. J., & Holsinger, K. (2006).The gendered nature of risk factors for delinquency.
Feminist Criminology, 1, 1, 48–71.
Bhui, H. S. (1999). Racism and risk assessment: Linking theory to practice with Black mentally
disordered offenders. Probation Journal, 46, 3, 171–81.
Blanchette, K., & Brown, S.L. (2006). The assessment and treatment of women offenders: An
integrative perspective. West Sussex: John Wiley & Sons.
Bloom, B., Owen, B., & Covington, S. (2003) Gender responsive strategies for female offenders:
A summary of research, practice, and guiding principles for women offenders.
Washington: National Institute of Corrections.
Blumstein, A. (1982) On racial disproportionality of the United Sates prison populations.
Journal of Criminal Law and Criminology, 73, 4, 1259.
33 | P a g e
Bonta, J. (2008a). Offender risk assessment and sentencing. Canadian Journal of Criminology
and Criminal Justice, 49, 4, 519–529.
Bonta, J. (2008b). Advances in offender risk assessment.” Research Summary, 13, 2. Ottawa:
Ministry of Public Safety. Retrieved from
http://www.publicsafety.gc.ca/res/cor/sum/cprs200803-eng.aspx
Bonta, J., Bourgon, G., Jesseman, R., & Yessine, A. K. (2005). Presentence reports in Canada,
2005. Ottawa: Public Safety, 12.
Brennan, T. (1998). Institutional classification for females: Problems and some proposals for
reform. In R.T. Zaplin (Ed.) Female Offenders: Critical Perspectives and Effective
Interventions. Gaithersburg, MD: Aspen Publishers Inc.
Brennan, T. and Austin, J. (1997). Women in jail: Classification issues. Washington DC: US
Department of Justice National Institute of Corrections.
Bushway, S.D, & Morrison Piehl, A. (2007). The inextricable link between age and criminal
history in sentencing. Crime & Delinquency, 53, 1, 156-183.
Bushway, S., & Smith, J. (2007). Sentencing using statistical treatment rules: What we don’t
know can hurt us. Journal of Quantitative Sociology, 23, 377-387.
Canadian Human Rights Commission (CHRC) ( 2003). Protecting their rights systematic review
of Human rights in correctional services for women. Ottawa: Canadian Human Rights
Commission. SEE ENDNOTE XV for link.
Chanenson, S. (2009) “Prosecutors and evidence-based sentencing: Rewards, Risks and
responsibilities”. Chapman Journal of Criminal Justice, 1, 1, 2741.
Clarke Robbins, P., Monahan, J., & Sliver, E. (2003) Mental disorder, violence and gender. Law
and Human Behavior, 27, 6, 561-5713.
Cole D. (2008). The umpires strike back: Canadian judicial experience with risk assessment
Instruments. Canadian Journal of Criminology, 49, 4, 493–519.
Cole, D., & Angus, G. (2003). Using pre-sentence reports to evaluate and respond to risk.
Criminal Law Quarterly, 47, 3, 302–64.
Covington, S.S. and Bloom, B.E. (2007). Gender responsive treatment and services in
correctional settings. Women & Therapy, 29, 3-4, 9-33.
34 | P a g e
Crow, M.S. (2008). The complexities of prior record, race, ethnicity, and policy: Interactive
effects in sentencing. Criminal Justice Review, 33, 4, 502-523.
Cullen, F & Gilbert, K. (2001). From nothing works to what works: Changing professional
ideology in the 21st century. Prison Journal, 81, 313.
Cunningham, M.D., & Reidy, T.J. (2002). Violence risk assessment at federal capital sentencing:
Individualization, generalization, relevance, and scientific standards. Criminal Justice
and Behavior, 29, 5, 512-537.
Dal Pra, Z. (2004). In search of a risk instrument. In M. Thigpen and G. Keiser (Eds), Topics in
community corrections: Issues for managers. Washington: National Institute of
Corrections.
Daubney, David (2002). “Nine Words: A Response to ‘Empty Promises: Parliament, The
Supreme Court, and the Sentencing of Aboriginal Offenders,” 65 Saskatchewan Law Rev.
35–52.
Dolores, J. C., & Redding, R. E. (2009). The effects of different forms of risk communication on
judicial decision making. International Journal of Forensic Mental Health, 8, 1–5.
Durrance, P., & Williams, P. (2003). Broadening the agenda around what works for Black and
Asian offenders. Probation Journal, 50, 3, 211–224.
Dutton, D.G., & Kropp, R. (2000). A review of domestic violence risk instruments. Trauma,
Violence, & Abuse, 1, 2, 171-181.
Etienne, M. (2009). Legal and practical implications of evidence based sentencing by judges.
Chapman Journal of Criminal Justice, 1, 1, 43-60.
Ericson, R.V. & Haggerty, K. (1997). Policing the risk society. Toronto: University of Toronto
Press.
Ericson, R.V. & Doyle, A. (2003). Risk and morality. Toronto: University of Toronto Press.
Ewald, F. (1991), ‘Insurance and Risk’, in G. Burchell, C. Gordon and P. Miller, eds, The
Foucault Effect. Chicago: University of Chicago Press.
Feeley, M., & Simon, J. (1992). New penology: Notes on the emerging strategy of corrections
and its implications. Criminology, 30, 4, 449–74.
Feeley, M., & Simon, J. (1994). Actuarial justice: The emerging new criminal law. In D. Nelken
(Ed.), The futures of criminology (pp. 173–201). London: Sage.
35 | P a g e
Fitzgibbon, D. W. (2007). Institutional racism, pre-emptive criminalisation and risk analysis. The
Howard Journal of Criminal Justice, 46, 128–44.
Fitzgibbon, D. W. (2008). Fit for purpose? OASys assessments and parole decisions. The
Probation Journal, 55, 55–69.
Gavazzi, S. M.,Yarcheck, C.M, & Lim, J-Y. (2005). Ethnicity, gender, and global risk indicators
in the lives of status offenders coming to the attention of the juvenile court. International
Journal of Offender Therapy and Comparative Criminology, 49, 6, 696 – 710.
Gendreau, P., Goggin, C., & Smith, P. (2002). Is the PCL-R really the “unparalleled” measure of
offender risk? Criminal Justice and Behavior, 29, 397-426.
Giddens, A. (1991) Modernity and Self-Identity: Self and Society in the Late Modern Age.
Stanford University Press.
Glaser, D. (1985). Who gets probation and parole: Case study versus actuarial decision making.
Crime & Delinquency, 31, 3, 367-378.
Grann, M., & Langstrom, N. (2007). Actuarial assessment of violence risk: To weigh or not to
weigh? Criminal Justice and Behavior, 34, 1, 22-36.
Gray, N., Laing, J. and Noaks. L. (2002). Criminal justice, Mental Health and the politics of
Risk. London: Cavendish Publishing.
Gross, A. (2008). History, race, and prediction: Comments on Harcourt’s Against Prediction.
Law & Social Inquiry, 33, 1, 235-242.
Haines, K., & Morgan, R. (2007). Services before trial and sentence: Achievement, decline and
potential. .In L. Gelsthrope and R. Morgan (Eds). Handbook of Probation. Cullompton:
Willan.
Haney, C. (2006). Reforming punishment: psychological limits to the pains of imprisonment.
Washington: American Psychological Association.
Hannah-Moffat, K. (2004). Criminogenic needs and the transformative risk subject:
Hybridizations of risk/need in penality. Punishment and Society, 7, 1, 29–51.
Hannah-Moffat, K. (2009) Re-imagining gendered penalities: The myth of gender responsivity.
In P. Carlen (Ed.), Imaginary penalities (pp. 193–219). London: Willan Press.
Hannah-Moffat, K. (2009). Gridlock or mutability: Reconsidering “gender” and risk assessment.
Criminology and Public Policy, 8, 1, 209–219.
36 | P a g e
Hannah-Moffat, K., & Maurutto, P. (2004). Youth risk/need assessment: An overview of issues
and practices. Ottawa: Department of Justice Canada, Research and Statistics Division.
Hannah-Moffat, K., & Maurutto, P. (2008). Understanding risk in the context of the Youth
Criminal Justice Act. Canadian Journal of Criminology and Criminal Justice, 49, 4, 465–
491.
Hannah-Moffat, K. and Maurutto (2010) “Re-contextualizing Pre-sentence Reports: Risk and
Race” Punishment and Society 12(2)
Hannah-Moffat, K., Maurutto, P., & Turnbull, S. (2009). Negotiated risk: Actuarial illusions and
discretion in probation. Canadian Journal of Law and Society, 24, 3, 391–409.
Hannah-Moffat, K. and P. O’Malley (2007) Gendered Risks. London: Routledge Cavendish
Publishing.
Hannah-Moffat, K., & Shaw, M. (2001). Taking Risks: Incorporating gender and culture into the
classification and Assessment of Federally Sentenced women. Ottawa: Status of Women
Canada.
Hannah-Moffat, K., & Yule, C. (Forthcoming). Gaining insight, changing attitudes and
managing “risk”: Parole release decisions for women convicted of violent crimes.
Punishment and Society, 13.
Harcourt, B. (2007). Against prediction: Profiling, policing and punishing in the actuarial age.
Chicago, IL: University of Chicago Press.
Harcourt, B. E. (2005), Against Prediction: Sentencing, Policing, and Punishing in an Actuarial
Age (Chi. Pub. Law & Legal Theory, Working Paper No. 94, 2005), available at
http://ssrn.com/abstract=756945.
Hardyman, P. L., & Van Voorhis, P. (2004). Developing gender-specific classification systems
for women offenders Washington, DC: USDOJ, National Institute of Corrections.
Heilbrun, K. (2009). Risk assessment in evidence-based sentencing context and promising uses.
Chapman Journal of Criminal Justice, 1, 1, 127-142.
Heimer, K., & Kruttschnitt, C. (2006). Gender and crime: Patterns in victimization and
offending. New York: New York University Press.
37 | P a g e
Hood, R., & Shute, S. (2000a). Parole decision-making: Weighing the risk to the public.
Research Findings No. 114 UK: Great Britain Home Office Research Development and
Statistics Directorate.
Hood, R., & Shute, S. (2000b). The parole system at work: A study of risk based decision
making. Home Office Research Report No. 202.
Hudson , B. (2003). Justice in the risk society. London: Sage Publications.
Hudson, B., & Bramhall, G. (2005). Assessing the other: Constructions of Asianness in risk
assessment by probation officers. British Journal of Criminology, 45, 5, 721–740.
Hudson, B., & Bramhall, G. (2007). Criminal justice and ‘‘risky’’ masculinities. In K. HannahMoffat & P. O’Malley (Eds.), Gendered risks (pp. 124–144). London: Routledge,
Cavendish Publishing.
Hunt, S. (2007). Standards of evidence. Criminology and Public Policy, 6(3):483-492.
Hutter, B. And M. Power (2005) Organizational Encounters with Risk. Cambridge: Cambridge
University Press.
Hutter, B., & Power, M. (2005). Organizational encounters with risk. Cambridge: Cambridge
University Press.
Kemshall, H. (1998). Risk in Probation Practice. Hampshire, England: Ashgate.
Kern, R.P., & Farrar-Owens (2004). Sentencing guidelines with integrated offender risk
assessment. Federal Sentencing Reporter, 16, 3, 165-169.
Kim, D-Y., Joo, H-J., & McCarty, W. P. (2008). Risk assessment and classification of day
reporting center clients: An actuarial approach. Criminal Justice and Behavior, 35, 792812.
Kleiman, Matthew, Ostrom, B.J., and Cheesman II, F.L. (2007) “Using Risk Assessment to
Inform Sentencing Decisions For Non-Violent Offenders in Virginia.” Crime and
Delinquency 53 (1), 106-132.
Kleiman, M., Ostrom, B.J., and Cheesman II, F.L. (2009 “Using Risk Assessment to Inform
Sentencing Decisions For Non-Violent Offenders in Virginia.” Crime and Delinquency
53 (1), 106-132.
38 | P a g e
Kleiman, M., Ostrom, B.J., and Kauder, N.B. “Risk Assessment: A New Approach to Sentencing
Non-Violent Offenders.” Caseload Highlights: Examining the Work of State Courts,
10(1), October 2004.
Krauss, D.A., Sales, B.D., Becker, J.V., & Figueredo, A.J. (2000). Beyond prediction to
explanation of risk assessment research: A comparison of two explanatory theories of
criminality and recidivism. International Journal of Law and Psychiatry, 23, 2, 91 -112.
Kreamer, S. (2004) ‘Quality Assurance and Training in Offender Assessment’ in Topics in
community corrections: Annual Issue 2004: Assessment Issues for Managers.
Washington: U.S. Department of Justice, National Institute of Corrections.
http://www.nicic.org
Kong, R., & AuCoin, K. (2008). Female offenders in Canada. Juristat: Canadian centre for
justice statistics (Catalog 85-002-XIE, vol. 28, no. 1).
MacLeod, S. March 1995. Youth Management Assessment: User Guide. Ministry of the Solicitor
General and Correctional Services, Bell Cairn Staff Development Centre.
Mackenzie, D. (2001). Corrections and sentencing in the 21st century: Evidence-based
corrections and sentencing. Prison Journal, 81, 299.
Manchak, S.M., Skeem, J.L., Douglas, K.S., & Siranosiam, M. (2009). Does Gender Moderate
the Predictive Utility of the Level of Service Inventory—Revised (LSI-R) for Serious
Violent Offenders? Criminal Justice and Behavior Vol. 36, 425-442.
Marcus, M. (2009a). MPC—The root of the problem: Just deserts and risk assessment. Florida
Law Review, 61, 4 , 751–776.
Marcus, M. (2009b). Conversations on evidence–based sentencing. Chapman Journal of
Criminal Justice, 1, 1, 61-125.
Maurutto, P., & Hannah-Moffat, K. (2006). Assembling risk and the restructuring of penal
control. British Journal of Criminology, 46, 3, 438–454.
Maurutto, P., & Hannah-Moffat, K. (2007). Understanding risk in the context of the Youth
Criminal Justice Act. Canadian Journal of Criminology, 49, 4, 465–491.
Ministry of the Solicitor General and Correctional Services. (1995). Level of Service Inventory –
OR – Facts behind Risk Assessment. Ontario: Ministry of the Solicitor General and
Correctional Services.
39 | P a g e
Mental Health Centre Penetanguishene. (nd). Replications of the Violence Guide or Sex
Offender Risk Appraisal Guide in assessing violence risk. Retrieved from
http://www.mhcp.on.ca/Site_Published/internet/SiteContent.aspx?Body.QueryId.Id=1673
&LeftNavigation.QueryId.Categories=62 .
Mitchell, O. (2005). A meta-analysis of race and sentencing research: Explaining the
inconsistencies. Journal of Quantitative Criminology, 21, 4, 439–466.
Monahan, J. (2010). Classification of violence risk. In R. Otto and K. Douglas (Eds) Handbook
of Violence Risk Assessment. New York: Routledge.
Monahan, J. (2006). A jurisprudence of risk assessment: Forecasting harm among prisoners,
predators, and patients. Virginia. Law Review, 92, 391.
Morash, M. (2009). A great debate over using the Level of Service Inventory–Revised (LSI-R)
with women offenders. Criminology & Public Policy, 8, 173-181.
Morash, M., Bynum, T. S., & Koons, B. A. (1998). Women offenders: Programming needs and
promising approaches. Washington, DC: USDOJ, National Institute of Justice.
Moretti, M.M., Odgers, C.L., & Jackson, M.A. (2004). Girls and aggression: Contributing
factors and intervention principles, Vol. 19. New York: Kluwer Academic/Plenum.
Murrie, D.C., & Balusek, K. (2008). Forensic assessment of violence risk in adversarial
proceedings: Pursuing objectivity and avoiding bias. Journal of Forensic Psychology
Practice, 7, 4, 141-153.
Netter, B. (2007). Using groups statistics to sentence individual criminals: An ethical and
statistical critique of the Virginia Risk Assessment Program. Journal of Criminal Law
and Criminology, 97,3, 699–730.
O’Malley, P. (1999). The risk society: Implications for justice and beyond. La Trobe University:
Department of Justice, Victoria, Australia.
O’Mally, P. (2004). Risk, uncertainty, government. London: The Glasshouse Press.
O’Malley, P. (2008). Experiments in risk and criminal justice. Theoretical Criminology, 12, 4,
451-469.
Ostrom , B., & M. Kleiman , F. Cheesman , A. M. Hansen , N. B. Kauder , M.Kleiman , R.
M. Hansen , N. B. Kauder (2002) Offender Risk Assessment in Virginia: A Three-Stage
40 | P a g e
Evaluation. Washington: US Department of Justice, document number:196815. Available
at: http://www.ncjrs.gov/pdffiles1/nij/grants/196815.pdf
Padfield, N. (2007). Who to Release: Parole fairness and criminal justice. UK: Willan
Publishing.
Padfield, N., & Liebling, A. (2000). An exploration of decision making at discretionary lifer
panels. Home Office Research Report No. 213.
Padfield, N., Liebling, A., & Arnold, H. (2003). Discretion and the release of life sentence
prisoners. In L. Gelsthorpe & N. Padfield (Eds.), Exercising discretion: Decision-making
in the criminal justice system and beyond (pp. 97–124). Portland: Willan Publishing.
Peters, T., & Warren, R. K. (2006). National Centre for State Courts. Getting smarter about
sentencing: NCSC’s Sentencing Reform Survey 2–3. Retrieved from
http://www.ncsconline.org/D_Research/Documents/Senten_GettingSmarter_SentencingR
eformSurvey_FinalPub.pdf
Petersilia, J., & Turner, S. (1987). Prediction and racial minorities. In D. M. Gottfredson and M.
Tonry (Eds.) Prediction and classification. Crime and Justice . Chicago, IL: University
of Chicago Press.
Power, M. (2004).The risk management of everything: Rethinking the politics of uncertainty.
London: Demos.
Pridemore, W. A. (2004). Review of the literature on risk and protective factors of offending
among Native Americans. Journal of Ethnicity in Criminal Justice, 2, 4, 45–63.
Quigley, T. (2008). Commentary on risk assessment. Canadian Journal of Criminology and
Criminal Justice, 49, 4, 531–539.
Redding, R. (2009). Evidence-based sentencing: the science of sentencing policy and practice.
Chapman Journal of Criminal Justice, 1, 1, 1-22.
Reichman, N. 1986. "Managing Crime Risks: Towards an Insurance Based Model of Social
Control." Research In Law Deviance and Social Control 8:151-172.
Reisig, M. D., Holtfreter, K., & Morash, M. (2006). Assessing recidivism risk across female
pathways to crime. Justice Quarterly, 23, 384-405.
Rettenberger, M., Matthes, A., Boer, D.P., & Eher, R. (2010). Prospective actuarial risk
assessment: A comparison of five risk assessment instruments in different sexual
41 | P a g e
offender subtypes. International Journal of Offender Therapy and Comparative
Criminology, 54, 2, 169-186.
Robinson, G. (2003). Implementing OASys: Lessons from research into LSI–R and ACE. The
Probation Journal, 50, 30–40.
Rudin, Jonathan, & Kent Roach (2002) “Broken Promises: A Response to Stenning and Roberts’
‘Empty Promises,’” 65(3) Saskatchewan Law Rev. 3–34.
Roehl, J., & Guertin, K. (2000). Intimate partner violence: The current use of risk assessments in
sentencing offenders. The Justice System Journal, 21, 2, 171-198.
Salisbury, E., Van Voorhis, P., & Spiropoulis, G. (2009). The predictive validity of a genderresponsive risk/needs assessment. Crime & Delinquency, 55, 550-585.
Salisbury, E. J., & Van Voorhis, P. (2009). Gendered pathways: An empirical investigation of
women probationers’ paths to incarceration. Criminal Justice & Behavior, 36, 541-566.
Sapir, Y. (2008). Against prevention? A response to Harcourt’s Against Prediction on actuarial
and clinical predictions and the faults of incapacitation. Law & Social Inquiry, 33, 1, 253264.
Schwalbe, C.S. (2007). Risk assessment for juvenile justice: A meta analysis. Journal of Law
and Human Behavior, 31, 449-462.
Scott, R. (2008). Risk assessment and sentencing of serious sex offenders. Psychiatry,
Psychology and Law, 15, 2, 188-201.
Shute, S. (2007). Parole and risk assessment. In N. Padefield (Ed.) Who to release, pp. XXX–
XXX). UK: Willan Publishing.
Silver, E. & Miller, L. (2002). A cautionary note on the use of actuarial risk assessment tools for
social control. Crime and Delinquency, 48, 1, 138–161.
Silver, E., & Chow-Martin, L. (2002). A multiple models approach to assessing recidivism risk:
Implications for judicial decisions making. Criminal Justice and Behavior, 29, 538-568.
Simon, J. (1988). The ideological effects of actuarial practices. Law & Society Review, 22, 4,
771-800.
Simon J. (2005). Reversal of fortune: The resurgence of individual risk assessment in criminal
justice. Annual Review of Law and Social Science, 1, 397–421.
Slobogin,C. (2005). A jurisprudence of dangerousness, 98 Nw. U. L. Rev. 1, 62
42 | P a g e
Severson, M., & Duclos, C.W. (2003). Assessing suicide and risk behaviors in an incarcerated
American Indian population--Investigating culturally sensitive risk assessment
instruments and procedures in a border jail: Final Report. Lawrence, KS: University of
Kansas, School of Social Work.
Stenning, Philip, & Julian V. Roberts (2001) “Empty Promises: Parliament, the Supreme Court,
and the Sentencing of Aboriginal Offenders,” 54 Saskatchewan Law Rev. 137–68.
Tata, C, (2007). Sentencing as craftwork and the binary epistemologies of the discretionary
decision process. Social & Legal Studies, 16, 3, 425-447.
Tata, C., Burns, N., Halliday, S., Hutton, N., & McNeill, F. (2008). Assisting and advising the
sentencing decision process: The pursuit of ―quality in pre-sentence reports. British
Journal of Criminology, 48, 835–855.
Vancise, William J., & Patrick Healy (2002) “Judicial Notice in Sentencing,” 65 Saskatachewan
Law Rev. 97–105.
Van Voorhis, P. (2005). Classification of women offenders: gender-responsive approaches to
risk/needs assessment. Community Corrections Report, 12, 2, 19–27.
Van Voorhis, P., Salisbury, E., Bauman, A., & Wright, E. (2007). Gender-responsive risk-needs
assessment: Final report. Prepared for the Maui CARE Project and the National Institute
of Corrections. Cincinnati, OH. University of Cincinnati: Criminal Justice Research
Center.
Van Voorhis, P., Bauman, A., Wright, E., & Salisbury, E. (2009). Implementing the Women’
Risk/Needs Assessments (WRNAs); Early lessons from the field. Women, Girls, and
Criminal Justice, 10, 6, 81-82, 89-91.
Van Voorhis, P., & Presser, L. (2001). Classification of women offenders: A national assessment
of current practices. Washington, DC: USDOJ, National Institute of Corrections.
Van Voorhis, P., Salisbury E., Bauman, A., Holsinger, K., & Wright, E. (2008). Classifying
women offenders: Achieving accurate pictures of risk and identifying gender-responsive
needs. In B. Bloom (Ed.), ICCA 2007 Plenary Addresses.Lanham, MD: American
Correctional Association.
Van Voorhis, P., Wright, E. M, Salisbury, E., Bauman, A. (2010) ‘Women's Risk Factors and
Their Contributions to Existing Risk/Needs Assessment: The Current Status of a Gender43 | P a g e
Responsive Supplement’ Criminal Justice and Behavior, vol. 37, no. 3, pp. 261-88,
March 2010
Vess, J. (2009). Risk assessment of sexual offenders for extended supervision orders in New
Zealand: Basic principles and current practice. Journal of Child Sexual Abuse, 18, 174189.
Vigorita, M.S. (2003). Judicial risk assessment: The impact of risk, stakes, and jurisdiction.
Criminal Justice Review, 14, 3, 361-376.
Viljoen, J.L., Scalora, M., Cuadra, L., Bader, S., Chavez, V., Ullman, D., & Lawrence, L.
(2008). Assessing risk for violence in adolescents who have sexually offended: A
comparison of the J-SOAP-II, J-SORRAT-II, and SAVRY. Criminal Justice and
Behavior, 35, 1, 5-23.
Vose, B., Cullen, F.T., & Smith, P. (2008). The empirical status of the level of service inventory.
Federal Probation, 72, 22–29.
Ward, T. (2002). The management of risk and design of good lives. Australian Psychologist, 37,
3, 172–179.
Ward, T., & Brown, M. (2004). The good lives model and conceptual issues in offender
rehabilitation. Psychology, Crime and Law, 10, 3, 243–257.
Ward, T., & Maruna, S. (2007). Rehabilitation: Beyond the risk paradigm. London: Routledge.
Ward, T., & Stewart, C. (2003). Criminogenic needs and human needs: A theoretical critique.
Psychology, Crime and Law, 9, 3, 125–143.
Warren, R. K. (2007). Evidence-based practices and state sentencing policy: Ten policy
initiatives to reduce recidivism, 82 IND. L. J. 1307, 1312–1314.
Wellford, C. (2007). Sentencing Research For Sentencing Reform. Criminology and public
policy 6(3)pp 399-402.
White, T. (2004) ‘IMPLEMENTING AN OFFENDER RISK AND NEEDS ASSESSMENT: AN
ORGANIZATIONAL CHANGE PROCESS’ in Topics in community corrections: Annual
Issue 2004: Assessment Issues for Managers. Washington: U.S. Department of Justice,
National Institute of Corrections. http://www.nicic.org
Wolfe, M. (2008). Evidence-based judicial discretion: Promoting public safety through state
sentencing reform. New York University Law Review, 83, 5, 1389-1419.
44 | P a g e
Wormith, J.S. (2007). Adhering to Principles of Effective Correctional Treatment: Academic
Musings of a Former Clinician and Administrator. Forum on Corrections Research 19, 4.
Zedner, L. (2008). Book review of Against Prediction. New Criminal Law Review, 11, 2, 359362.
Cases Cited:
R. v. Gladue [1999] S.C. R. 688.
R. v. B.H.D. [2006] Provincial Court Saskatchewan 32.
1
I would like to thank Linn Clark, Tony Doob, Amy Klassen, Paula Maurutto and Shawn Bushway and
reviewers and commentators for their contributions to earlier drafts.
2
Interview data.
3
There are a plethora of risk instruments available. For example, tools like the Level of Service Inventory –
Revised (LSI–R), Static 993, and more recent domestic violence risk assessment guides (Ontario Domestic Assault
Risk Assessment (ODARA) and Domestic Violence Risk Appraisal Guide (DVRAG))3, which are designed for and
administered and interpreted by non-psychological/psychiatric professionals. There are a multitude of offender and
offence (female, youth, violent) “risk instruments” that exist or are being developed. Most risk instruments share a
common structure and focus on comparable “risk” factors because they draw on the similar empirical literatures.
What varies among these tools is the degree of emphasis placed on a particular set of factors; cut-off scores for the
categorization of risk, or nee; and the targeted offender population (youth, domestic violence, violent or sexual
offenders). Consequently, the risk-assessment industry is influential and extensive, and has produced a variety of
assessment tools. In the context of sentencing, risk instruments are being built into sentencing guidelines (i.e.,
Virginia), and used in the preparation of pre-sentence reports.
4
One of the most common risk instruments for the prediction of general criminality is the Level of Service
Inventory – revised (LSI–R). LSI–R is currently being utilized within Canada and the United States to guide
sentencing decisions, placement in correctional programs, institutional assignments, and release from institutional
custody. This instrument’s widespread use and modification makes it a useful exemplar for this discussion. The LSI
was originally developed in Ontario, Canada, in the late 1970s and quickly developed international notoriety.
Currently, the tool is used in jurisdictions throughout Canada, the United States, the United Kingdom, and Australia,
among others. The tool, originally written in English, is available in Spanish, Croatian, and French (French
European and French Canadian), and it is in the process of being translated into Dutch and Icelandic. Multi Health
Systems—the company that markets the LSI–R—indicates that more than 600 agencies in the United States
currently use this risk-need tool (Lowenkamp et al., 2004) It is one of the most extensively researched offender
classification instruments; consequently, it is less vulnerable to some of the methodological critiques of less
rigorously tested local risk instruments. Nonetheless the LSI–R, similar to all risk tools, has limits, and its current
45 | P a g e
use in sentencing generates a complicated debate, as well as a need for further empirical research.
5
For a discussion of the criminogenic effects of imprisonment see Craig Haney (2006).
6
Andrews and Bonta (2006) provide a detailed description of these tools and this development.
7
Andrews and Bonta report that the research on providing any treatment to offenders as a function of risk
shows that “providing intensive services to low risk offenders may actually increase criminal behaviour and also that
these services can lead to a significant decrease in recidivism when delivered to higher risk offenders. For example,
Bonta, Wallace-Capretta and Rooney (2000) in an evaluation of a Canadian program found that low risk offenders
who received minimal levels of treatment had a recidivism rate of 15% and low risk offenders who received
intensive levels of services had more than double the recidivism rate (32%). In the same study, the high risk
offenders who did not receive any intensive treatment services had a recidivism rate of 51% but the high risk
offenders who did receive intensive services had almost half the recidivism rate (32%). The risk principle calls for
intensive treatment services to be reserved for the higher risk offender” and that “Treatment interventions that do not
adhere to any of the three principles (that is, they target the non-criminogenic needs of low risk offenders using noncognitive-behavioural techniques) are actually criminogenic! This situation is particularly exacerbated when the
treatment is given in residential/custodial settings (we presume because the offender cannot escape from the wellintentioned but poorly designed treatment)” (Andrews and Bonta 2007:15-16).
8
See Ward and Maruna (2007) for a fuller critique of RNR. The critical literatures on punishment include
a lively debate about the “what works” analyses of the logic of RNR and cognitive behavioralism (Ward, 2003;
Ward and Stewart, 2003; Ward and Brown, 2004).
9
The phrase “trust in numbers” is borrowed from T. Porter (1996) Trust in Numbers:
The Pursuit of Objectivity in Science and Public Life. Princeton: Princeton University Press.
10
See: PEW (2009) adapted from Roger Warren’s paper that was originally published in a special 2007
issue of the Indiana Law Journal entitled “Evidence-Based Practices and State Sentencing Policy: Ten Policy
Initiatives to Reduce Recidivism.” see http://www.pewcenteronthestates.org/uploadedFiles/Final_EBS_Brief.pdf
11
For a more comprehensive discussion of this debate and present developments see Van Voorhis et al.
12
For an elaboration of this argument see Hannah-Moffat (2009). Such analyses highlight that men and
2010.
women have different needs and represent different kinds of risk, and note the importance of considering “genderspecific needs” and the specific needs of minority women, (for example, see Belknap and Holsinger, 2006Blanchette
and Brown, 2006; Brennan, 1998; Hollin and Palmer, 20062007; Holsinger and Holsinger, 2005Holtfreter and
Morash, 2003; Nesbitt and Argento 1984; Reisig et al., 2006Van Voorhis and Presser, 2001Wright, Salisbury, and
Van Voorhis, 2007Some studies also demonstrate that tools such as the LSI–R function differently for men and
women. For example, a growing international scholarly body of research has been questioning the limits of risk
assessment and the potential for bias and discrimination. A number of researchers have questioned the convergent,
46 | P a g e
concurrent, and discriminate validity and reliability of established risk instruments when used on women and
racially diverse and economically disadvantaged offenders.
13
As Van Voohris et al. 2010: 262) document a number of studies have found dynamic risk assessments, such as
LSI-R “valid for women (see Andrews, Dowden, & Rettinger, 2001; Blanchette & Brown, 2006; Coulson, Ilacqua,
Nutbrown, Giulekas, & Cudjoe, 1996; Dowden & Andrews, 1999; Holsinger, Lowenkamp, & Latessa, 2003;
Simourd & Andrews, 1994; Smith, Cullen, & Latessa, 2009). Others studies have produced conflicting results (see
Blanchette, 2005; Law, Sullivan, & Goggin, in press; Olson, Alderden, & Lurigio, 2003; Reisig, Holtfreter, &
Morash, 2006; Salisbury, Van Voorhis, &Spiropoulis, 2009).”
14
That is, education/employment; family/marital relations; leisure/recreational involvement; criminal
acquaintances; attitudes toward crime; and substance abuse which are six of the eight risk/need factors in the LSI–
OR)
15
A long tradition of sociological research that demonstrates social stratification and poverty are related to
both criminality and criminalization.
16
Protecting Their Rights: A Systemic Review of Human Rights in Correctional Services for Women – Dec
2003 Canadian Human Rights Commission – Chapter 4; http://www.chrcccdp.ca/legislation_policies/consultation_report-en.asp; also see Hannah-Moffat and Shaw 2001 – Taking Risks:
Incorporating gender and culture into the classification and Assessment of Federally Sentenced women. Ottawa:
Status of Women Canada.
17
For a more comprehensive human rights analysis see CHRC (2003). http://www.chrc-
ccdp.ca/legislation_policies/consultation_report-en.asp
18
See Hannah-Moffat (2007) for a more detailed discussion of the interpretation, classification, and
treatment of women prisoners’ marital and family relationships.
19
See special issue of Punishment and Society (2010) 12 (2).
22
Although evidence on concordance between PSR and sentencing is contested, highly nuanced, and
difficult to empirically unravel (Haines and Morgan, 2007; Tata et al., 2008), existing research demonstrates that the
PSR plays a central interpretive role in sentencing andamong criminal justice professionals (see also Tata, 2010;
Wandall, 2010).
23
In cases where individuals are legally designated “dangerous offenders” there is considerable more
debate among experts on how opinions about “danger and risk” are formulated. However they are also considerable
debates in this instance about the suitability of risk instruments. For additional discussion see Monahan 2010.
24
Malenchik v. Indiana, No. 79S02-0908-CR-365 (Ind. June 9, 2010)
http://www.in.gov/judiciary/opinions/pdf/06091001bd.pdf.
25
Some authors contend that this conflict is not as problematic as it seems, see for example Marcus (2009).
47 | P a g e
26
See Hutter and Power 2005; Power 2004 for a discussion of organizational risk management in other
industries.
27
Some scholars maintain that the transition to risk-based penality has led to “deskilling,” “scientification,”
and the “erosion of professional discretion” (Robinson, 2003:33; see also Baker, 2005; Fitzgibbon, 2007, 2008:
Schneider et al., 1996), or even to the elimination of discretion among criminal justice practitioners (Hannah-Moffat
and Maurutto, forthcoming).
28
See Mental Health Centre Penetanguishene for discussion of research on the use of Violence Risk
Appraisal Guide (VRAG) and Sexual Offender Risk Appraisal Guide (SORAG).
48 | P a g e