Reconstructing the Risk–Need–Responsivity model: A theoretical elaboration and evaluation
Introduction
The Risk–Need–Responsivity (RNR) model of offender rehabilitation is deservedly the premier treatment model for offenders. It has constituted a revolution in the way criminal conduct is managed in Canada, Britain, Europe, Australia and New Zealand and led to the development of a suite of empirically derived and effective treatments for a range of crimes. The principal architects of the RNR model of offender rehabilitation are the Canadian researchers James Bonta, Don Andrews, and Paul Gendreau (e.g., Andrews and Bonta, 2003, Andrews et al., 2006, Andrews and Dowden, 2006, Andrews et al., 1990, Gendreau and Andrews, 1990). The model represents a model for the prediction of risk and classification of offenders for treatment, based on the authors’ early observations that criminal behavior can be predicted, that risk interacts with level of treatment intensity and targets in influencing treatment outcome (i.e., recidivism), and that these factors interact with offender‐based factors in influencing outcome (Andrews & Bonta, 2003). The model includes three principles that should guide intervention with offenders. In brief, the risk principle suggests that offenders at higher risk for reoffending will benefit most from higher levels of intervention, including high intensity treatment, and that lower risk offenders should receive minimal, routine, or no intervention. The need principle proposes that only those factors associated with reductions in recidivism (i.e., criminogenic needs) should be targeted in intervention. The responsivity principle states that correctional programs should be matched to offender characteristics such as learning style, level of motivation, and the individual's personal and interpersonal circumstances. The first two principles (risk and need) are used to select treatment intensity and targets, and the whole set used to guide the way practice is actually implemented. A fourth principle, that of professional discretion, states that clinical judgment should override the above principles if circumstances warrant. This principle allows for treatment flexibility and innovation under certain circumstances. We will not explicitly discuss the discretion principle in this paper, as it does not have the major role the others do in the RNR model.
An impressive body of meta‐analytic research has been developed on the RNR model. This evidence has largely been derived from meta‐analytic examination of rehabilitation evaluation research, beginning with Andrews et al.'s (1990) seminal paper. The Andrews et al. meta‐analysis, which provides the greatest empirical evidence for the RNR model, was conducted to refute the ‘nothing works’ perspective that had dominated since Martinson's (1974) famous report. Treatment that complied with the RNR principles demonstrated significantly greater effectiveness than criminal sanctions, inappropriate treatment, or unspecified treatment. Subsequent meta‐analyses by Andrews and Dowden have provided similar evidence from a variety of specific offender groups, including juvenile delinquents (Dowden and Andrews, 1999a, Dowden and Andrews, 2003), violent offenders (Dowden & Andrews, 2000), and female offenders (Dowden & Andrews, 1999b). Furthermore, adherence to the RNR principles provides a context within which increased program integrity, organizational adherence to integrity standards, and better staff practice can improve treatment outcomes (Andrews and Dowden, 2005, Dowden and Andrews, 2004). Cumulatively, this research provides a powerful empirical foundation for the RNR model.
Despite strong empirical support for the RNR model, it has been subjected to a number of critiques, primarily aimed at its underlying theoretical assumptions, their implications for practice, and lack of scope (e.g., Ward and Brown, 2004, Ward and Stewart, 2003). In summary, Ward et al. have argued that a focus on reducing dynamic risk factors (criminogenic needs) is a necessary but not sufficient condition for effective treatment (Ward & Gannon, 2006). A key component of this critique has been the argument that it is necessary to broaden the theoretical formulation, application to practice, and the scope of correctional interventions to take into account the promotion of human goods (or approach goals) in conjunction with the reduction of risk variables (or avoidance goals). Critics propose that the RNR is conceptually impoverished and is unable to provide therapists with sufficient tools to engage and work with offenders in therapy. In brief, it is claimed that (a) it is difficult to motivate offenders by focusing primarily on risk reduction (e.g., Mann, Webster, Schofield, & Marshall, 2004); (b) the RNR model does not pay enough attention to the role of personal or narrative identity and agency (i.e., self‐directed, intentional actions designed to achieve valued goals) in the change process (e.g., Maruna, 2001); (c) it works with a narrow notion of human nature and ignores the fact that as evolved, biologically embodied organisms, humans naturally seek and require certain goods in order to live fulfilling and personally satisfying lives (e.g., Ward & Stewart, 2003); (d) it pays insufficient attention to the therapeutic alliance and so‐called noncriminogenic needs such as personal distress and low self‐esteem that are important beyond their potential implications with respect to offender responsivity. For example, it has been argued that the creation of a sound therapeutic alliance requires a suite of interventions that are not directly concerned with targeting risk and it has been shown that the establishment of a good therapeutic alliance is a necessary feature of effective therapy with offenders (e.g., Marshall et al., 2003, Yates, 2003, Yates et al., 2000); (e) the RNR is essentially a psychometric model and tends to center on offender risk profiles (or traits) and downplays the relevance of contextual or ecological factors in offender rehabilitation (Ward & Brown, 2004); (f) the RNR is often translated in practice into “one‐size‐fits‐all” manner that fails to take critical individual needs and values into account. Indeed, the usual implementation of the RNR actually ignores its own principle of responsivity or at least makes it hard to accommodate the idiosyncratic features of offenders. In its most inappropriate form, the RNR is realized in a psycho‐educational format where offenders are “taught” putatively important information (Green, 1995); and (g) the RNR is not an integrated theory and the three major principles are not sufficiently theoretically grounded (Ward & Stewart, 2003). Space does not permit us to flesh out these criticisms in detail, but we make the general point that some of them are due to the application in practice of the RNR model rather than any inherent weakness in the model itself (see below).
Proponents of the RNR model have responded to these criticisms by arguing that there does exist a strong theoretical basis for this influential rehabilitation model and that once this is clearly articulated the above criticisms fail (e.g., Andrews and Bonta, 2003, Bonta and Andrews, 2003, Ogloff and Davis, 2004). While advocates of the RNR model accept that the theory is often presented purely in terms of the principles of risk, need, and responsivity, they claim that this does not mean that it lacks theoretical grounding (e.g., Bonta, 2003). In other words, it is asserted that it is a mistake to frame the RNR model purely in terms of the three rehabilitation principles and its associated program elements. Rather, it is claimed that the theory contained in Andrews and Bonta's (2003) seminal book, The Psychology of Criminal Conduct, and in accompanying articles, effectively grounds the three principles and by doing so outlines a powerful rehabilitation theory.
The trouble with this response is that at least three different theoretical models or perspectives (the two terms tend to be used synonymously) have been presented as providing an underlying theoretical justification for the RNR model. In other words, it is not clear exactly what theory is being appealed to in this debate. First, in their exposition of the RNR model, Ogloff and Davis (2004) proposed that the Psychology of Criminal Conduct (PCC) perspective outlined by Andrews and Bonta (2003) in a number of publications “provides directions for the assessment of offenders and their classification for treatment” (p. 232). Second, Andrews and Bonta affirm that a model they call a General Personality and Social Psychological Perspective on Criminal Conduct (GPSPP) is able to account for “multiple routes to involvement in illegal conduct” (p. 165). Third, Gillis (2000) asserts that the Personal Interpersonal Community‐Reinforcement (PIC‐R) perspective affords a theoretical source for predicting and explaining criminal behavior. Moreover, these three models are all to be found in Andrews and Bonta's discussion of the theoretical underpinnings of their approach to offender rehabilitation in chapters 1 and 4 of The Psychology of Criminal Conduct. For Andrews and Bonta, the three models are thought to provide theoretical support and justification for the “big” three rehabilitation principles of risk, need, and responsivity (Bonta, 2003). Thus, while all three models are discussed in Andrews and Bonta's (and other RNR proponents' writings) writings, the degree to which they collectively or individually theoretically ground the three principles of risk, need, and responsivity, is uncertain. Another area of vagueness concerns the relationship between the three models: Should they be hierarchically related or are they simply alternative conceptualizations of a psychology of criminal conduct? In other words, exactly how should the three models be configured within the RNR model of offender rehabilitation?
But why does this proliferation of theoretical bases for the RNR model matter? Surely, it could be argued that existence of multiple models is a boon to researchers and practitioners, a cornucopia of resources that can only serve to facilitate offender rehabilitation and crime prevention.
Unfortunately, this is not the case. The difficulty in having more than one theory associated with the RNR model is that it makes evaluation of the model a difficult and slippery process. It is hard to state exactly what etiological claims are being made and how the principles comprising the RNR model are derived from underlying theory and research. In the trail of such ambiguity follows problems of falsification and confirmation—how do we know whether the RNR model is an adequate rehabilitation model if we are unsure what its theoretical commitments are? Furthermore, from an intellectual point of view, it is important to coherently link the various strands of the justificatory theory to the RNR principles. Indeed, Andrews and Bonta (2003) are adamant that “psychology seeks explanations of criminal conduct that are consistent with the findings of systematic observation, rationally organized, and useful to people with practical interests in criminal behavior” (p. 4). The authors are advocates for vigorous critical debate on the theoretical, empirical, and practical aspects of offender rehabilitation. Their commitment to rational empiricism is admirable and reminds us that it is important never to simply assume the truth of our favored theories, but always to critically integrate them in the pursuit of truth and greater understanding. Finally, without a clear statement of the theoretical and methodological commitments of the RNR model, it is harder to defend it against the kinds of criticisms outlined above. In order to conclude whether or not they are reasonable, it is first necessary to delineate the assumptions of the RNR model in a systematic and coherent manner.
Our aim in this paper is to reconstruct the rehabilitation theory in which the three principles of risk, needs, and responsivity are implicitly embedded. This task is essentially an architectural one and will involve a careful reading of Andrews and Bonta and some degree of redesigning their theory in light of this process of critical reflection. We do not wish to be presumptuous or to claim that this reconstruction represents a new theory or is ours in any significant respect. Rather, our intention is to draw together the various strands of theory from Andrews and Bonta's work and additional theories, to weave them together in a more convincing way, to augment shortcomings, and to provide a more comprehensive model that can be better integrated into practice. Indeed, our hope is that the reconstructed theory will be stronger and a better and more coherent rehabilitation theory. First, we briefly discuss the nature of rehabilitation theories and their core components. Second, we review the three source theories associated with the RNR model. Third, we set out to reconstruct the RNR model in light of this analysis, essentially arguing that there are at least three components to any rehabilitation theory: (a) primary aims, values, and principles; (b) etiological and methodological assumptions; and (c) practice implications. Fourth, we evaluate the theoretical and empirical adequacy of the RNR model. The goal is not to engage in a simple dismissal of the model, but rather to elucidate the features of this important theory in order to help subsequent researchers strengthen it. Finally, we conclude the paper with a few comments on the policy, research, and clinical implications of our evaluation (and reconstruction) of the RNR model. Our reconstruction of the RNR model concentrates on treatment, but a case could be made for extending it to all interventions in the criminal justice area, which is something we intend to address in another paper (Ward & Yates, in preparation).
Section snippets
The nature of rehabilitation theory
In order to be able to adequately describe and evaluate the RNR model, it is first necessary to identify features of a rehabilitation theory. Somewhat surprisingly, we have found that the nature of rehabilitation tends to be taken for granted in the correctional field and very little has been said as to what actually constitutes a rehabilitation theory as opposed to a treatment or etiological theory. Typically, the terms “treatment”, “therapy”, and “rehabilitation” are used interchangeably as
Theoretical sources of the risk–need–responsivity model
As stated above, there are at least three related but seemingly distinct theoretical models that are associated with the RNR model: the Psychology of Criminal Conduct Perspective (PCC), the General Personality and Social Psychological Perspective on Criminal Conduct (GPSPP), and the Personal Interpersonal Community‐Reinforcement Perspective (PIC‐R). We will briefly describe each of the models before drawing from all three in our reconstruction of the RNR model within the three‐component
The reconstructed risk‐need‐responsivity model
We will now attempt to reconstruct the RNR rehabilitation theory by drawing upon the collective resources of the three models and the principles of risk, need, and responsivity (plus other elements of effective service programming). Our aim is to present the RNR model in its strongest possible form in order to more accurately evaluate its strengths and weaknesses. Note that by the term “RNR model” we are referring to the entire rehabilitation theory, not simply the three classification
Evaluative comments on the RNR model
It now remains to briefly examine the strengths and weaknesses of the RNR model in its reconstructed form. More specifically, we will revisit the criticisms outlined in the Introduction and inquire whether the version of the RNR model sketched out in this paper is indeed vulnerable to them. But before we do this, we will briefly highlight some of its most significant virtues.
In the correctional field, the shift to a risk management perspective and the focus of therapeutic efforts on the
Conclusions
The RNR model of offender rehabilitation represents an outstanding achievement. Its application by correctional services throughout the world has resulted in reduced recidivism rates and safer communities. However, despite its many virtues the model has been criticized because of what are perceived as theoretical, policy, and practice weaknesses. We have argued that one of the major reasons for these problems resides in the way the RNR model has been developed and applied. Essentially, the
Acknowledgements
We thank the following people for the excellent comments and very helpful they made on earlier drafts of this paper: Astrid Birgden, Theresa Gannon, Rachael Collie, Richard Laws, and Paul Whitehead.
References (49)
Treatment programs for offenders: Meta‐analysis, “what works” and beyond
International Journal of Law and Psychiatry
(1999)- et al.
Process variables in the treatment of sexual offenders
Aggression and Violent Behavior: A Review Journal
(2003) - et al.
Rehabilitation, etiology, and self‐regulation: The Good Lives Model of sexual offender treatment
Aggression and Violent Behavior
(2006) Principles of effective correctional programs
- et al.
The level of service inventory—revised
(1995) - et al.
The psychology of criminal conduct
(2003) - et al.
The recent past and near future of risk and/or need assessment
Crime & Delinquency
(2006) - et al.
Managing correctional treatment for reduced recidivism: A meta‐analytic review of programme integrity
Legal and Criminological Psychology
(2005) - et al.
Risk principle of case classification in correctional treatment
International Journal of Offender Therapy and Comparative Criminology
(2006) - et al.
Does correctional treatment work? A clinically relevant and psychologically informed meta‐analysis
Criminology
(1990)