You are on page 1of 23

1

Legal and Criminological Psychology (2011), 16, 123 C 2010 The British Psychological Society

The British Psychological Society


www.wileyonlinelibrary.com

The impact of nonprogrammatic factors on criminal-justice interventions


D. A. Andrews
Department of Psychology, Carleton University, Ottawa, Ontario, Canada
Purpose. Palmer (1995) drew attention to a distinction between programmatic and non-programmatic aspects of criminal justice interventions. While a considerable amount of research has accumulated on the former, the latter by comparison remains under-researched. Nevertheless some advances have been made and the present article identies the key components of this. Methods. Following analysis of the concepts forwarded by Palmer, a methodical comparison is made between his ndings on programmatic elements and those of two other major groups of meta-analytic ndings from this area. This provides further opportunity for testing of the Risk-Needs-Responsivity (RNR) model and an evaluation is offered of its current status in synthesizing relevant knowledge. A parallel set of comparisons is then drawn with respect to non-programmatic factors and the paper considers the level of agreement between separate reviews of that knowledge base. This directs attention to a number of instances of intervention failure which can be explained by insufcient attention to non-programmatic issues. Results. There is a generally high level of agreement between the three sets of data surveyed. There is not a complete consensus however, caused not by disagreement between data sets but by gaps in the types and range of evidence assembled. There are larger gaps remaining on non-programmatic factors and the nature and extent of those is described. There is also discussion of some objections and proposed alternatives to RNR, and to some conceptual confusions arising from them. Conclusions. The present state of knowledge on criminal justice interventions is a work in progress but nevertheless can provide rm guidance on the design of such interventions, highlighting areas in which much further work is needed.

The work of Ted Palmer (1995) on the programmatic and nonprogrammatic aspects of successful justice and correctional interventions was a standard in the mid-1990s

It is with very great sadness that we report the death of Don Andrews on 22nd October 2010, while this article was in the course of production. Don was Professor Emeritus and Distinguished Research Professor at Carleton University, Ottawa. Throughout an internationally successful academic and research career spanning 35 years, Don was a superb example of the scientist-practitioner in action and was always concerned both with the scientic contribution of his work and with the application of his ndings in justice policy. His work has powerfully inuenced the shape of criminal justice services in many countries. He was a remarkable man and will be a lasting inspiration to those who work in the area of applying psychology in the criminal justice eld.
DOI:10.1348/135532510X521485

D. A. Andrews

against which the performance of the emerging what works movement would be judged. Indeed, I will compare the ndings of his 1995 review with the ndings of the most recent meta-analytic reviews of the effectiveness conducted across an array of different types of programmes. Palmers summary of the knowledge in the mid1990s reected his consolidation of the ndings of 32 reviews of the relevant literature, including both narrative reviews and meta-analytic reviews of controlled studies of the effect of intervention on the criminal futures of programme participants. He was aware of the emergence of a clinically relevant and psychologically informed approach but he did not use the risk-need-responsivity model (RNR: Andrews, Bonta, & Wormith, 2009; Andrews, Zinger, et al., 1990) in the organization of his report. The RNR principles were too new and too fresh to anchor an independent review in the mid-1990s. Rather, the programmatic aspects of the intervention were described more conventionally (much as continues to be done in the Campbell meta-analytic reviews of recent years). Palmer (1995) reported that the ve most effective programmatic features were (1) behavioural, (2) cognitive-behavioural or cognitive, (3) life skills or skill oriented, (4) multimodal, and (5) family interventions. On the other hand, the ve least successful were (1) confrontation (deterrence/shock), (2) neighbourhood-wide prevention strategies, (3) diversion, (4) group counselling/therapy, and (5) individual counselling. The ve best and the ve worst programmatic factors listed by Palmer were only 10 of a set of 20 generic approaches that he identied including diversion, social casework, physical challenge, restitution, community-based approaches versus institutional intervention, and others. Ted Palmers (1995) list of nonprogrammatic factors and conditions included (1) staff characteristics such as training, orientation, and skills including the ability to enter into a high-quality relationship with clients or cases; (2) offender characteristics such as personality, interpersonal style, developmental level, and risk level (although the latter is barely mentioned in the text of his paper and surfaces only in an appendix); (3) the interaction of staff and client characteristics (such as matched according to I-level theory); (4) size of caseload, frequency of contact, individualized/exible programming, explicitness of intervention strategies; and (5) setting (community, open residential, closed residential . . . ). Integrity in service delivery was an unspecied combination of some elements from 1, 3, and 4 but a combined measure was not constructed by Palmer. The present paper will focus on nonprogrammatic factors and my list of those factors overlaps considerably with Palmers (1995) list. A shift in emphasis will be my use of the terms integrity and/or delity as key ways of summarizing the extent to which offender supervision or the delivery of services were consistent with the services that were intended to be delivered. Indicators of integrity include having an explicit and detailed model of service to begin with, along with the selection, training, and clinical supervision of staff according to that model, checks that the intended services were actually delivered to the cases for whom they were intended, and the availability of supportive resources such as training manuals. However, before reviewing nonprogrammatic factors it is important to note a problem that exists today just as it did for Palmer 15 years ago. Palmer (1995, pp. 101102 and notes 3 and 57 on p. 124) struggled with the denition of programmatic. Given denitional problems, the distinction between programmatic and nonprogrammatic was not then and is not now an obvious distinction. The 20 generic approaches that he identied were simply not mutually exclusive. For Palmer, programmatic was an uneasy mixture of ofcial sanctions, setting factors, and some conventionally described human, social, or clinical services.

The impact of nonprogrammatic factors

One improvement is to differentiate fundamentally between the ofcially dened criminal sanction and the delivery of supervision or services that are intended to contribute to crime prevention by way of reduced reoffending. Within the RNR model of correctional assessment and treatment, the introduction of human service is a fundamental principle of effective crime prevention (Andrews & Bonta, 2006, 2010). The principles of RNR are listed in Table 1 because I will be referring to them often. The human service principle (Principle 4) recognizes that there is little basis in behavioural theory and research to expect that the imposition of ofcial sanctions will reduce risk/need and thereby move risk/need factors in the direction of becoming strengths.

Table 1. The RNR model of effective correctional assessment and crime prevention services (Andrews, 1995, 2001; Andrews & Bonta, 2006, 2010; Andrews, Bonta, et al., 1990; Andrews, Zinger, et al., 1990; Bonta & Andrews, 2007; Gendreau, 1996)

D. A. Andrews

The exception to this statement is a period of incapacitation associated with institutional sanctions. Of course, antisocial conduct remains a possibility inside the prison walls. Moreover, and problematically with the Palmer classication of programmatic factors, group counselling, individual counselling, and vocational programming, family interventions, and the like all may or may not involve the use of cognitive-behavioural intervention strategies (the general responsivity principle of RNR), may or may not have been delivered to moderate and higher risk cases (the risk principle), and may or may not have been targeting criminogenic needs predominately (the need principle of RNR). Indeed, Andrews and Dowden (2007) have reported that no correctional interventions based on services no matter their generic description have yielded substantial mean

The impact of nonprogrammatic factors

reductions in recidivism without the delivery of services that are in adherence with the three core clinical principles of RNR. Table 2 illustrates the increases in mean effect sizes found when the delivery of human services is in adherence with the core clinical principles of risk, need, and general responsivity. The effects are evident across setting factors, penalty factors, case characteristics, and several conventional programme labels. The conditions described as demonstration projects, mixed, and routine corrections at the bottom of Table 2 will be dened later in this paper but that distinction will be crucial to our discussion of integrity in programming. Ted Palmer (1995) convincingly argued for increased attention to variation in the effectiveness of particular generic programmes across a variety of conditions of programme delivery. These conditions include variability in offender characteristics.
Table 2. Mean effect size (r) by Level of RNR adherence in various justice and treatment contexts including family, academic, vocational, and substance abuse programs (based on Dowden & Andrews, 2004, databank; k = number of tests of treatment)

D. A. Andrews

I go further and suggest that programmatic factors should not be dened by conventional programme labelling. The RNR model suggests that effective programming depends upon a matching of services with key characteristics of the case. As illustrated in Table 2, effectiveness increases with the matching of intensity of service with the risk level of cases (risk), matching of the intermediate targets of change with the criminogenic needs of the case (need), and matching the style and mode of service to the learning ability of cases (general responsivity). One aspect of responsivity is highly general in that learning and interpersonal inuence are best understood and facilitated through cognitive social learning theory and practices. Generally personality and cognitive social learning (GPCSL) perspectives on human behaviour provide (a) an understanding of the major risk/need factors for criminal behaviour and (b) an understanding of powerful interpersonal inuence processes and strategies of behaviour change. That combination of identifying the major risk/need factors while also identifying powerful styles and strategies for behaviour change is unique to GPCSL and not approximated by alternative psychological accounts of human behaviour. Other aspects of responsivity are more specic to case factors such as personality, motivation to participate in programming, motivation to change, and perhaps age, race, and gender. Within the RNR model, they are called specic responsivity considerations (Principle 8 in Table 1). The labelling is complex because some case-based specic responsivity may also be risk/need factors. For example, psychopathic personality is a major risk factor recognized within the broader RNR-based risk set known as antisocial personality pattern. But psychopathy may also be a specic responsivity factor. Similarly, being motivated to participate in rehabilitation programmes and being motivated to give up on crime are specic responsivity factors as well as indicators of personal cognitions and emotional states that are supportive of reduced antisocial behaviour. Still, there is a functional distinction between risk/need factors and case-based specic responsivity factors. The latter cannot be established without demonstrations of Type of Case-by-Type of Service interactions on recidivism. In the case of psychopathy and low motivation to change, the principle of specic responsivity would be that services must be made highly attractive and participation must be rewarded immediately. Still, the general principle of responsivity is paramount. General responsivity directs attention to the relationship and structuring activities of the staff who supervise offenders and the workers/clinicians/ therapists/trainers/coaches/ofcers who deliver services. Andrews and Bonta (2006, 2010) paraphrased Andrews and Kiessling (1980, pp. 462463) in order to breathe life into the phrases behavioural, cognitive-behavioural, and cognitive social learning. They were also very explicitly drawing attention to the quality of the relationship between workers and offenders:
. . . effective rehabilitative efforts involve workers who are interpersonally warm, tolerant, and exible, yet sensitive to conventional rules and procedures. These workers make use of the authority inherent in their position without engaging in interpersonal domination (i.e., they are rm but fair); they demonstrate in vivid ways their own anticriminalprosocial attitudes, values, and beliefs; and they enthusiastically engage the offender in the process of increasing rewards for noncriminal activity. The worker exposes and makes attractive the alternatives to procriminal attitudes, styles of thinking, and ways of acting. The worker does not depend upon the presumed benets of a warm relationship with the offender and does not assume that offenders will self-discover these alternatives. The alternatives are demonstrated through words and actions, and explorations of the alternatives are

The impact of nonprogrammatic factors

encouraged through modelling, reinforcement, and specic guidance (Andrews, Bonta, & Hoge, 1990, pp. 3637).

Andrews and Bonta went further:


. . . we may be pig-headed but now in the middle of the rst decade of the new millennium, we would not change a word of the above. In fact, we are inclined to underline or bold the whole paragraph. But we exercise self-control and only underline underline and bold bold.

A full understanding of adherence with the general principle of responsivity requires a detailed appreciation of effective staff practices and the relationship and structuring skills of correctional workers and others who provide correctional supervision and/or deliver human services. I want our discussion of nonprogrammatic factors to be located within the evidence base that exists today in regard to effective correctional interventions. Thus, I open with Palmers (1995) classic review of programmatic and nonprogrammatic factors and I link his ndings with the most recent generally applicable ndings of two meta-analytic research teams. The two teams of meta-analysts are ones that seek general (widely applicable) understandings of effective interventions and were identied by Mark Lipsey (2009). One team is composed of Mark Lipsey and his colleagues and the other team is composed of my colleagues and I. The latter team is known as the RNR gang, the what works bunch including but not limited to Jim Bonta and Paul Gendreau, and sometimes the team is described as the Canadians. I have heard us called other names of late but that I save that for a closing section of this paper on the limitations of RNR. That closing section is a summary of some recent highly critical reviews of RNR. Turning to programmatic factors, Mark Lipsey (and I) wanted to underscore the value of a broad meta-analytic appreciation of the elements of effective practice. This contrasts with meta-analytic reviews that are limited to narrow and conventionally labelled interventions such as family, vocational, or cognitive-behavioural therapy.

Programmatic factors in Palmer (1995), Lipsey (2009), and Andrews and Bonta (2010)
Ted Palmers (1995) conclusions regarding programmatic factors are broadly consistent with those drawn by the two teams of meta-analytic researchers. Many of the ndings of the RNR group have already been incorporated into some of the principles outlined in Table 1. Thus, I will summarize the Lipsey (2009) ndings and then compare the ndings of Lipsey with those of Palmer and of Andrews and colleagues.

The major ndings of Lipsey (2009) The rst three ndings clearly differentiated between less effective and more effective interventions with young offenders and are programmatic: (1) Therapeutic interventions (counselling and skills training) were signicantly more effective than interventions based on control or coercion (surveillance, deterrence, and discipline); (2) Effectiveness increased with the risk level of young offenders;

D. A. Andrews

(3) Cognitive-behavioural programmes were more effective than all other programmes with behavioural programmes ranked second. The remaining four ndings have to do with nonprogrammatic factors. The major nonprogrammatic ndings from Lipsey (2009) were as follows: (4) Intervention was equally effective for younger and older juveniles, for males and females, and for Whites and non-Whites. (5) Effectiveness increased with the quality of implementation (programme integrity). (6) The effectiveness of counselling programmes was reduced in samples of incarcerated young offenders. (7) The effectiveness of skill building programmes were enhanced when delivered in the community to young people not under correctional supervision.

Comparing the programmatic ndings of Lipsey (2009) with Palmer (1995) and with the principles of RNR Lipseys rst nding had to do with the relative effectiveness of therapeutic approaches relative to deterrence and control. That nding is highly consistent with Palmers review and with our meta-analytic evidence in regard to the effectiveness of human service relative to reliance on the ofcial punishment approach (and supportive of the RNR principle of human service; Principle 4 in Table 1). Similarly, and despite some differences in how adherence with the risk principle is measured, Lipseys (2009) second nding is consistent with our own meta-analytic research (and supportive of the risk principle of RNR; Principle 5 in Table 1). Palmer did not comment on the risk principle except to note that personality was important even within risk levels (which we describe as specic responsivity). Lipseys nding number three is in agreement with the principle of general responsivity of RNR (# 7 in Table 1) although Lipsey (2009) did not code for cognitive social learning strategies in programmes described as counselling. As already shown in Table 2, family approaches, educational/vocational programming, and other types of service programmes worked best when in adherence with the core clinical principles of RNR. More strongly stated, any programming is of limited value for purposes of crime prevention when it relies on ofcial punishment or when human service is delivered to low risk cases, when criminogenic needs are not targeted, when weak inuence and behaviour change strategies are employed. The principles of risk, need, and general responsivity are axiomatic without being tautological. What I mean is that the theoretical and empirical knowledge base of general personality and cognitive social learning (GPCSL) identies the risk and criminogenic need factors that allow individualized assessments of risk level and the individualized identication of particular combinations of criminogenic needs. Likewise, cognitive social learning strategies are well-established approaches to interpersonal inuence and to changes in thinking, feeling, action, and interpersonal association patterns. The dual notion of relationship skills and practices in combination with structuring skills and practices render core correctional practices very concrete (Principle 14 in Table 1). The programmatic ndings from the two meta-analytic teams and from Palmer (1995) are in total agreement in regard to the potential of human services relative to a reliance on the criminal penalty. While discussed only minimally by Palmer (1995), support for the risk principle is now also strong. Support for the risk principle perhaps constitutes one of the major steps forward in the last 1520 years. As we will also see, it is considered

The impact of nonprogrammatic factors

by some others as part of an immoral, unethical, and restrictive paradigm of effective services. Whatever the case, for now the risk nding is one of the reasons that I suggest that the programmatic elements of crime prevention intervention be based not on generic descriptions but on RNR-based matching of services and case characteristics. Indeed, following the lead of C. T. Lowenkamp (personal communication, 2010), I am inclined to conclude that the status of the human service principle, the risk principle, and the general responsivity principle should be raised from being mere principles to that of being three RNR-based laws of effective crime prevention. In that communication, Lowenkamp was expressing some frustration with the elds difculty in accepting oftreplicated ndings. With a sense of humour he was also decrying the readiness of the eld to rush to something new when the empirically well-established validity of principles such as risk, need and general responsivity had not yet been well-understood, let alone well-implemented into routine programming. This paper will close with consideration of the roots of some of the anti-RNR rhetoric that Chris Lowenkamp was reacting against. Unfortunately, the criminogenic need principle was not tested by Lipsey (2009). Palmer (1995) did support multimodal programming but it is not clear that this is the equivalent of breadth in the targeting of criminogenic needs. Thus, I will not strongly afrm a law of effective crime prevention when the principle in question has only been tested by one team of meta-analysts. Note, however, that the meta-analytic evidence in regard to the predictors of reoffending is overwhelmingly in favour of the major risk/need factors identied within the GPCSL/RNR perspectives on assessment and intervention. Certainly, Andrews and Bonta (2010) and McGuire (2004) do not hesitate to recognize the predictive power of assessments of antisocial cognitions, antisocial associates, antisocial personality pattern, a history of antisocial behaviour, and substance abuse. Predictive accuracy and understanding are additionally enhanced at least somewhat by consideration of rewards and satisfactions in the major behavioural settings of home (family/marital), school and work, and leisure/recreation. In terms of programming, the expectation is that a focus on enhanced rewards and satisfactions for noncriminal activities in those major social settings is a route to shifts in antisocial thinking and feeling and in antisocial associates. From a GPCSL perspective on human conduct, the need principle has to do with building rewarding alternatives to crime. The evidence is reasonably strong in regard to the principles of human service, risk, criminogenic need, general responsivity, and the principle of breadth. Yet, as noted, the meta-analytic base in support of need has been supplied by only one team of researchers. Aspects of specic responsivity will be considered within the discussion of offender characteristics. Explorations of strengths and the aspects of structured assessment are relatively few in number and have not been subjected to meta-analytic reviews. My comment on structured assessments refers to evidence that the adoption of an assessment system is associated with improved programming and greater reductions in recidivism rates. That evidence is limited. On the other hand, the predictive validity of risk/need assessment instruments is well-established through multiple validation studies (Andrews & Bonta, 2010) and I will not re-review that here. Lowenkamp (2004; Lowenkamp, Latessa, & Smith, 2006) systematically explored how halfway house adherence with the principles of what works was associated with recidivism rates of over 3,000 halfway house graduates. The overall pattern of results was clearly supportive of RNR. In regard to risk/need assessments in particular the important point was not that risk assessments were conducted but that the assessment actually was associated with differential treatment plans which in turn were associated with greater reductions in recidivism.

10

D. A. Andrews

As reviewed by Andrews, Bonta, and Wormith (2010), implementation of the Wisconsin case classication system was associated with signicant albeit mild reductions in revocation rates in some although not all of the geographic regions in which it was tested. It appears to have been effective in particular when the denition of the recidivistic event was under the control of persons supervising the offenders (for example with revocation measures of community supervision). The effects were not found in a study of arrest rates. Indeed, in that study, use of the Wisconsin system was associated with increased technical violations but reduced revocation behaviour on the part of case supervisors. The Wisconsin system is reasonably strong on risk assessment but relatively weak on needs assessment. It appears that training in the Wisconsin system may reduce revocation behaviour without reducing the antisocial activity of offenders. Reducing returns to prison is attractive but is not a convincing demonstration of reduced reoffending. Obviously, the inability to nd strong evidence in support of the principle of structured assessment is a serious gap in the evidential base of RNR. What about nonprogrammatic factors including the organizational aspects of stafng and management?

Nonprogrammatic factors in Lipsey (2009), Palmer (1995), and RNR


Returning to the Lipsey (2009) meta-analytic conclusions, what was found in regard to offender characteristics beyond risk and need? Regarding effectiveness across categories of age, gender, and ethnicity, nding number four is consistent with the RNR position regarding the general applicability of the theory and the principles (Principle 2 in Table 1). I do not think age, gender, and ethnicity were even commented upon by Palmer (1995). Issues regarding age, gender, ethnicity, and social class are politically controversial and some vocal proponents of gendered, racialized and stratied understandings of correctional assessment and treatment are adamant about the value of gender-responsive and culturally appropriate approaches. Indeed, as I will describe in the last section of this paper, critical criminologists in Canada and the UK appear particularly offended by GPCSL and RNR. As a co-author of the RNR-based Level of Service (LS) instruments I know that my colleagues and I were aware that gender-responsiveness and cultural appropriateness had become normative in many probation ofces and prisons in Ontario in the 1990s. Thus, LS authors have endorsed and closely followed the National Institute of CorrectionsUniversity of Cincinnati (NIC-UC: Van Voorhis, Wright, Salisbury, & Bauman 2010) effort to build a gender-responsive supplement to Level of Service Inventory Revised (LSI-R; Andrews & Bonta, 1995) risk/need assessments. That NICUC supplement is now available to LSI-R users from the ongoing project at UC but the incremental validity of the supplement has not yet been conrmed. The LS authors went somewhat further than the NIC-UC team in the mid-1990s. The latest versions of LS instruments, created in the 1990s, are already gender and culturally informed (LS/CMI: Andrews, Bonta, & Wormith, 2004; Youth LS/CMI: Hoge & Andrews, 2002). For example, weak risk/need factors and noncriminogenic needs that are considered to be gender-informed are systematically surveyed and may be entered in the case plan as specic responsivity factors or as non-criminogenic needs to be targeted for motivational purposes and/or on humanitarian grounds, and/or on other normative grounds. That does not mean however that there are gender or racial differences in the predictive validity of major risk/need factors or that the core (programmatic) principles

The impact of nonprogrammatic factors

11

of RNR services are differentially effective according to age, gender, race, or class. Two relevant papers currently in the review process for publication address some of these points. In one multi-site study, the eight subscales of the LS/CMI instruments are predictive of the recidivism of female offenders and male offenders but substance abuse is much more predictive with women than with men. In the other meta-analytic paper, factors identied as gender-informed and LS total risk-need were equally predictive of the recidivism of females and males. Additionally, relative to LS total risk-need, the incremental predictive validity of the gender-informed factors was nil-to-minimal for both females and males. As the two teams of meta-analysts with interests in general understandings of recidivism have found, there are no replicated gender differences in regard to effective correctional treatment. Indeed, Hannah-Moffat (2009, p. 216), Canadas most vocal and inuential opponent of RNR-based assessment and treatment closed a recent anti-RNR paper with a surprising conclusion. She granted that evidence in support of risk and correctional treatment was expansive. This state of affairs she contrasted with the absence of evidence in support of feminist approaches to assessment, treatment, and programming. That is exactly my position on the state of the existing evidence base but I do not doubt that arguments regarding the real meaning of terms such as expansive and absence are already underway in the feminist community. Palmer (1995) and the two meta-analytic teams agree on the importance of integrity in the implementation of services. Andrews and Dowden (2005) surveyed a greater number of indicators of integrity than did the other review teams. Yet our numerous indicators of integrity were obvious correlates of the Lipsey (2009) measure of evaluators involvement in the design and/or delivery of the service being evaluated. Indeed, Lipseys simple measure of involved evaluator is the second strongest predictor of effect size in the Andrews and Dowden (2005) dataset of 374 controlled tests of correctional interventions. Of course, the single strongest predictor of effect size was RNR adherence. Quite independently, in an impressive review and an original meta-analysis, Petrosino and Soydan (2005) also reported that an involved evaluator is associated with smart programming that is delivered with integrity and the mean effect is enhanced crime prevention. I am inclined to propose that the issue of programme integrity too be raised to the level of a law (as proposed by C. T. Lowenkamp (personal communication, 2010) in regard to the identication of all well-established ndings). In other words, considerations of integrity should be raised to the level of being a law of effective crime prevention. Row 14 of Table 2 is labelled demonstration projects. That category was dened as evaluations of programmes by involved evaluators and relatively small numbers of participants (less than 100). Row 16 of Table 2 refers to real-world programming or routine programming. These were evaluations involving large numbers of offenders and an external evaluator. The mixed programmes either involved a small number of offenders or an involved evaluator but not both. The differences in those programmes were dramatic in that the demonstration projects had (a) much higher levels of RNR adherence, (b) much higher scores on indicators of integrity, and (c) much greater crime prevention effects. Note that at low levels of RNR adherence there are no substantial differences in mean effect size associated with integrity. The strong effects of RNR adherence were particularly strong in demonstration projects. The effects of RNR adherence were apparent in routine corrections as well but the mean crime prevention effect was much greater in demonstration projects. According to Lipsey (2009), Petrosino and Soydan (2005), and Andrews and Dowden (2007), attention to

12

D. A. Andrews

integrity issues enhances the effectiveness of RNR adherence (although Petrosino and Soydan refer to smart programming). A note of caution must be introduced in regard to the ndings represented in Table 2 and indeed in the ndings of all meta-analyses and in all reviews of the correctional effectiveness literature. Look once again at Table 2. The mean effect sizes are very low when RNR adherence is scored zero or one. The mean effects suggest either signicant but mild mean increases in reoffending or nil-to-minimal reductions in recidivism. The programmes that score low on RNR constitute a collection of some truly odd ball services. The programming was not very smart if the intention was to enhance crime prevention. We know not to expect much from the sanction on its own. Yet, how strange it is for managers and service personnel to expect that offender supervision and human service correctional programmes would work well with low risk cases. Low risk cases, by denition, have a low chance of reoffending to begin with. It make no sense to offer services that may interfere with the strengths of low risk cases or increase their opportunities to interact with higher risk cases. As I used to say in training sessions, if you want to meet high-risk cases go and sit in the waiting room of a parole ofce (just as low risk parolees are required to do in some agencies). Why would intermediate targets of change be set that are unrelated to criminal futures? Why are well-established predictors of criminal recidivism not being targeted? Why would anyone interested in crime prevention deliberately choose to employ weak intervention strategies? Look at services in Table 2 that target noncriminogenic needs predominately (row 13). Only two of those services succeeded in adhering to the principles of risk and/or general responsivity. Why are such programmes and services even being offered in corrections? Did they want to fail? What were they expecting to gain from the delivery of services so completely out of touch with well-validated understandings of the criminal conduct of individuals? Perhaps, the programme designers actually did know the literature or were operating from theories of crime that have little empirical support. The latter is quite possible because as will be shown presently there are substantial portions of psychology, sociology, social work, and criminology that subscribe to models that are not GPCSL and that are even self-consciously and explicitly anti-GPCSL. There is another possibility. Some of the services that are not in adherence with RNR may well be normative in the sense that recreational programmes can be fun and a pleasant diversion for offenders. Similarly, programmes aimed at reducing emotional distress are readily acceptable on humanitarian grounds and may even reect issues of entitlement. Just as in the case of medical and dental services, the services may not be smart from the perspective of crime prevention but they are being offered for other very good reasons. One must note, however, that the other reasons had better be compelling because a distinct consequence of such programming is at least minor increases in reoffending. There is yet another possibility that I nd difcult to describe in a manner that would not be insulting to those responsible for the services and/or the evaluation of those services. It may be expected that if a service is being described then surely the intermediate and ultimate targets of change are described, along with a description of who is receiving the services, and a description of the style and mode of service delivery. As Lipsey (2009) and as my colleagues and I have found, some descriptions of the service and of the research are so weak and incomplete that there is no way for a reader to know what was really occurring in regard to RNR adherence or compliance with any other standard that might be applied. I am saying that for a service to be scored

The impact of nonprogrammatic factors

13

as being in adherence with RNR that programme had be very explicit in their reporting. As I have already indicated, RNR compliance was considerably overrepresented in the demonstration projects relative to routine corrections. Fortunately, as documented in Table 2, the effects of RNR adherence were found under all conditions tabled, but for programmes that targeted noncriminogenic needs predominately. In sum, standards for reporting on services and for evaluations of service must be improved for knowledge to advance much further; once again, echoes of Palmer (1995). The RNR principles suggest that human service programmes that are in adherence with the core clinical principles of RNR are effective in both community and residential or institutional settings but the effects are dampened within institutional settings. Lipseys ndings numbers 6 and 7, as described above, were consistent with our ndings and are supportive of the RNR principle favouring community-based service delivery. Inspection of Row 3 of Table 2 reveals that incarceration without the delivery of human service was associated with a mean increase in reoffending of 10%. The effect of RNR adherence was greater in community settings than in residential settings but RNR-based services in residential settings did reduce the criminogenic effect of being in an institution that did not offer RNR-based services. Palmer (1995) described some very complex interactions of client factors with setting factors but it does not appear to me that he took an overall position on community versus institutional settings. In terms of the meta-analytic evidence, a sound conclusion to date is that community settings are preferred over institutional settings but the moderating impact of sanctioning factors and offender factors remain to be explored and fully appreciated.

Some additional offender characteristics and specic responsivity


I am not going to attempt to summarize the differential treatment approach that is so appreciated by Ted Palmer. The task of discerning subtypes of offenders that are matched with subtypes of workers and/or with subtypes of practices and/or with subtypes of settings is almost overwhelmingly complex. It is also possible to hypothesize subtypes of worker-by-subtypes of settings interactions on recidivism or even triple interactions as in the notion that particular types of workers may work best with particular types of cases although the appropriate matching may be different in community-based programming than in institutional programming. In Palmer (1995), the number of endnotes involving qualications and clarications of the proposed interactions underscore the complexity of some models of differential treatment. I think much of the useful information is much more easily communicated by an understanding of offender risk, offender need, and general learning ability. The principle of specic responsivity picks up some additionally valuable information but to my knowledge no one has yet attempted to explore the issue meta-analytically. I see no problem with building on the strengths of a case but where is the replicated evidence of strength-by-practice interactions? Likewise, it seems quite sensible to remove or reduce barriers to full participation in service but, again, where is the replicated evidence? Who is going to argue against child care services being provided in a community-based programme for women? The lack of a cumulative summary of the extant evidence in regard to specic responsivity concerns is a weakness in the RNR approach (and for the differential treatment movement generally). It is particularly important to respond to the tendency for higher risk cases to be the very persons least likely to seek out services and most likely to terminate

14

D. A. Andrews

participation prematurely. Motivational interviewing approaches are promising but evidence on reduced reoffending is slim. Adjuncts to RNR such as the Good Lives Model (GLM) purport to have incremental value relative to RNR in the reduction of criminal recidivism. But where is the evidence? Simons, McCullar, and Tyler (2008) report substantial reductions in treatment drop-out rates but they do not yet have followup data. RNR programmers look forward to the development of effective approaches to dealing with motivational issues. Many supporters of the RNR model are certainly hoping that GLM will deliver something of incremental value in regard to enhancing motivation for participation and for change. But GLM has not delivered as yet. The last chapter of Andrews and Bonta (2006, 2010) includes a table that identies the most likely of offender characteristics to enter into important interactions with practice factors. We have presented that table over the years in the hope that some young students would become intrigued enough to actually explore the differential treatment ideas of scholars such as Ted Palmer. To my knowledge, work on specic responsivity remains stalled. In addition, considerations of motivation for change, treatment readiness and consideration of strengths and barriers have yet to yield convincing effects on recidivism. We identify a set of factors we call cognitive/interpersonal skill level. One hypothesis, reecting our admittedly uncertain understanding of Ted Palmers perspective, is that styles and modes of services that are verbally and/or interpersonally demanding and depend upon self-regulation, self-reection, and interpersonal sensitivity should be employed only with very high functioning persons. Otherwise, rely on the tried, true and concrete aspects of cognitive social learning strategies. Another hypothesis relates to services for the interpersonally anxious. The recommendation is to avoid both interpersonal confrontation and very intense interpersonal exchanges. When social support for change is low, the recommended approaches entail structuring active exposure to others who model and reinforce real alternatives to antisocial styles of thinking, feeling, and acting. To date, as already noted, no convincing replicated demonstrations of specic responsivity have been outlined meta-analytically. Consider a recent statement by Kazdin (2008). Many clinical psychologists and clinical social workers are preoccupied with exquisite matching of services to the unique characteristics of the individual case. The facts suggest something else: effective services that are sensitive to a few key characteristics of individuals for the most part are widely applicable regardless of the uniqueness of individuals. Recall our ndings and the Lipsey (2009) ndings in regard to the absence of gender differences, age differences, and racial differences in the effectiveness of correctional programming. As already noted, despite myriad differences between girls and boys and between women and men, the major risk/need factors in relation to criminal recidivism are basically the same for girls and boys and for women and men. After years of study, Van Voorhis et al. (2010) reported that a general-neutral risk/need assessment powerfully predicts the antisocial activity of women offenders (see also Smith, Cullen, & Latessa, 2009). Gender-responsive assessment and treatment have strong normative appeal but the replicated evidence in support of gender-specicity is scarce in the domain of both prediction and treatment.

Staff characteristics and staff practices


Palmer (1995) recognized that his Community Treatment Project and the CaVIC project (Andrews & Kiessling, 1980) were very similar in the identication of promising staff characteristics and effective staff practices. It should not be surprising that in 1995

The impact of nonprogrammatic factors

15

Palmer emphasized only two studies. Today, only 14 (5%) of 273 tests of human service in the Dowden and Andrews (2004) meta-analytic dataset were found to attend to the relationship skills of workers. The lack of attention to quality of relationship and the other elements of effective practice is a sad commentary on the state of the literature. A reference to structuring skills was made in only 16% of the tests. The programmes that attended to the staff factors did yield enhanced mean effect sizes. Every element of core practice but advocacy/brokerage was signicantly associated with reduced reoffending. These other elements included employment of effective reinforcement, effective disapproval, problem-solving, structured learning, effective modelling, and effective use of authority (rm but fair). There is some evidence that staff factors are being attended to in the domain of a few sexual offender treatment programmes (Simons et al., 2008; they additionally cite the work of Bill Marshall and colleagues and the work of Pamela Yates and her associates). A direct test of the effect of RNR-related training of probation ofcers is well underway in Canada and reports on recidivism rates are forthcoming. The Strategic Training Initiative in Community Supervision (STICS) is a comprehensive package that attempts to translate what works knowledge into effective and sustainable everyday practice. STICS attempted to address the many issues that threaten and potentially erode the effectiveness of community supervision to facilitate prosocial change in offenders. Preliminary results on ofcer behaviour during supervision sessions were encouraging. Data showed that STICS trained ofcers, compared to control ofcers, demonstrated signicantly more and qualitatively better core correctional practices during their interactions with clients. Once data collection is complete, we will be able to evaluate the impact of these changes in ofcer behaviour on client attitudes, behaviours, and ultimately recidivism. Overall, it is hoped that the STICS project will provide insight into how to effectively transfer empirical knowledge into the real world of community corrections (Bonta, Rugge, Scott, Bourgon, & Yessine, 2008; see also Bourgon & Armstrong, 2005). Faye Taxman and colleagues are aware of the importance of staff training but they are also aware of huge implementation issues. Managers and staff require assistance in creating and maintaining high levels of adherence with the core principles of RNR. A very concrete product was developed in Maryland by Taxman, Shepardson, and Byrne (2004). It is called Tools of the Trade: A Guide to Incorporating Science into Practice. A feature is that it incorporates some very practical aids to effective practice. Although, to my knowledge, there is no evidence as yet that distribution of the manual has been rewarded by subsequent reductions in reoffending. I continue to think that it makes sense to select, train, and clinically supervise staff with reference to the key relationship and structuring elements of the principles of cognitive social learning. In terms of representation in the literature however, there appears to be precious little of it going on in practice. Most probably, programme managers consider other matters more important. It may appear straightforward but, in fact, RNR requires major policy and organizational changes and major efforts on the part of managers and staff for adherence to be accomplished. F. S. Taxman (personal communication, 2010) alerted us to some of the many staggering barriers that exist in many settings. Andrews and Bonta (2010) asked their readers to think about it:
Many correctional agencies are viewed by their staff, managers and political bosses as the administrators (or managers) of punishment. Justice models incorporate some combination of retribution, just desert, general deterrence, and/or specic deterrence. Under justice models, the police apprehend suspects, the courts establish just sanctions, and correctional

16

D. A. Andrews or analogous agencies administer the punishment. With the addition of an expectation of crime prevention through rehabilitation, the role of correctional agencies and agents changes dramatically. Now correctional practitioners are being asked to come to see themselves as human service agencies, albeit in a justice context. What might management and staff be thinking about such a change? Our job is to administer the penalty in a just manner. We are more like the police than social workers, and we like it that way. We want to be part of public safety. Just like the police and jail personnel. We want a pay scale like the police, we want their job security, we want a health insurance plan that is as good as the one the police and jail people have. We dont want to be treated like social workers.

Such thinking in combination with inattention to the key programmatic and nonprogrammatic factors virtually guarantees a failure in crime prevention.

The failure of RNR and the implementation and policy and management issues
Andrews (2006) and Andrews and Bonta (2010) were aware that some relatively negative outcomes from presumably RNR-based programmes were being reported in the late 1990s and into the new millennium. What was interesting was how dramatic was the telling of the tales, particularly at conferences rather than published peer-reviewed papers. I have heard very senior sociological criminologists declare that RNR-based programming could not be systematically implemented. It could not be done! The implementation of RNR is impossible! I have read polemical pieces from clinical psychologists declaring that RNR programming was immoral, unethical, unworthy of serious consideration, and not up-to-date with the latest fashionable and most positive image of being human. Impossible, immoral, out-of-date, and not positive enough! Whilst writing this paper I came across another example on the web site of the publisher of what was obviously a critical criminological tract. One scholar spoke of the the now hegemonic risk factors prevention paradigm. Another made reference to assisting in bringing the whole edice crumbling down. Obviously, some scholars want to destroy RNR. The publishers web site describes what the critical criminologists nd objectionable about RNR: it is hegemonic, immoral, unethical, and a too negative model of effective crime prevention that is too deterministic and prescriptive and cannot be implemented! Make no mistake about it: crime and criminality and criminalization are social constructs governed by wider economic, structural, cultural, and political factors. Note that determinism is acceptable if it is political economy doing the governing. These criminological tracts go further according to McNeill (2009, p. 4): the search for risk factors and offender types is fundamentally misconceived in that it tends to pathologize offending by focusing on the individual offender as the main unit of analysis. I have just this moment come across another advertisement for a different book on the above-noted publishers web site. This time what works is said to limit creative work with offenders. They are not against intervention. They dont like the RNR one size ts

The impact of nonprogrammatic factors

17

all approach. A model that celebrates diversity is described as one size ts all. Now, RNR is a hegemonic, immoral, negative, out-of-date, fundamentally misconceived model of effective crime prevention that focuses on offender psychology in a too deterministic and prescriptive manner that blocks creativity, and cannot be implemented anyway! It is nothing less than the re-emergence of knowledge destruction documented by Andrews and Wormith (1989) in sociological criminology of the 60s, 70s, and 80s. It would be good to understand how scholars, managers, and practitioners interested in understanding variation in the criminal behaviour of individual human beings and in crime prevention could be so negative. How did RNR go so wrong? Why do the critical criminologists want to destroy RNR? They attack the theoretical, epistemological, methodological and ethical bases of risk-focused research. Apparently, their objections are not to RNRs empirically documented ability to contribute to reduced reoffending. They make no reference at all to the reviews of the research literature conducted by Palmer, Lipsey and my colleagues and I. I think I understand somewhat why the Marxist and Socialist criminologists are so antiRNR and anti-GPCSL. Indeed, GPCSL does reject the critical criminological perspective on the major causal signicance of age, gender, class, and race in the understanding of variation in the criminal behaviour of individual human beings. We think that the major GPCSL factors account for many of the demographic correlates of criminal conduct. The critical criminologists are also unhappy with knowledge that is based on approximations of experimental investigations of the effects on criminal behaviour of deliberate interventions. I am not certain of their ethical objections but I expect that they do not like the idea that prospective studies of the predictors of criminal conduct reveal that the favoured variables within critical criminology are of nil-to-minimal signicance relative to GPCSL factors. They seem to believe that by not following the critical party line RNR creators and users are somehow at ease with the social disadvantage experienced by girls and women, by non-Whites, and by the economically deprived. Likewise, I expect that critical criminologists cannot accept that deliberate interventions that are in adherence with the principles of RNR reduce the reoffending of females and males, Whites and non-Whites, and of the rich and the poor. They seem to believe that RNR is about xing or curing criminals. What an amazing misunderstanding of GPCSL. From a GPCSL perspective, antisocial behaviour is as normal as prosocial behaviour and as normal as socially neutral behaviour. When the evidence is overwhelming, objections revert to theory, to appeals to explanations of alternative (and often irrelevant) factors, and to assertions of moral superiority. Such is the essence of knowledge destruction in criminology that was identied over two decades ago (Andrews & Wormith, 1989). We will learn what we can from critical criminologists but I think in many ways they are in a different universe. They dislike evidence of the predictive validity of risk factors and evidence of the effectiveness of deliberate interventions. It does not serve the purpose of their critical discourse. In GPCSL terms, reading and reporting the evidence is not rewarding for critical criminologists. At times, however, they do enter our universe. McNeill (2009) proposed a model of effective practice in offender supervision. It is an attractive model and a promising one in several ways. But it is informed by positions that are oddly anti-RNR. He appears to believe that GLM applies well to the supervision of offenders while RNR does not. He seems to believe that the desistance literature and the RNR model are somehow in conict. Shadd Marunas (2001) Making good is a classic piece of cognitive social psychology and it is utterly consistent with RNR. Maruna works with higher risk

18

D. A. Andrews

persons whose thinking, feeling and actions are favourable to crime, whose associates are involved in crime, who have histories of criminal behaviour, and may engage in substance abuse. Issues in regard the sources of rewards and satisfactions at home, at school or work, leisure/recreational activities also arise. Maruna describes how changes in association patterns and changes in thinking and feeling relate to future and contemporaneous conduct. McNeill (2009) speaks the language of desistance, of personal capital, and of social capital. Fine, but he is in error if he thinks that appeals to GLM and such are going to reduce reoffending in some manner inconsistent with GPCSL and RNR. McNeill (2009) appears to be making good use of Chris Trotters prosocial

Table 3. Case studies of programme failures (from Andrews & Bonta, 2006, 2010)

The impact of nonprogrammatic factors

19

modelling approach, yet McNeill seems to think that the Trotter approach is somehow different from the principles of GPCSL on which prosocial modelling is obviously based. There is a link between the founder of GLM and Fergus McNeill and between the founder of GLM and Shadd Maruna. Perhaps, those holding the curious view that the GLM has something to offer science or criminal justice are contaminating much of what they come into contact with. If crime prevention is an objective of service delivery, it is silly to concentrate on low risk cases, to target noncriminogenic needs predominately and to eschew cognitive social learning inuence strategies. With adherence to the core principles, strengthened by attention to integrity considerations, reduced reoffending rates are almost guaranteed. The strength of the effects on crime prevention is not so strong or so certain that there is not plenty of room for creative ways of enhancing crime prevention. RNR is not a case-based cure and it is not an offender-based x. It is the delivery of services that may assist in rearranging the personal, interpersonal, and community-based contingencies of human thinking, feeling, and acting such that noncriminal alternatives are more strongly supported than are criminal alternatives. Table 3 briey examines a few examples of the failures to implement with integrity. These examples were drawn from Andrews and Bonta (2006, 2010). The brief case studies of programme failure strongly suggest that programmes that were at least partially designed with reference to what works are often not well implemented. Frankly, critics who think that those examples damn the principles of RNR are misinformed or are deliberately engaging in the dissemination of misinformation. The deliberate dissemination of misinformation is like any other form of antisocial behaviour in that misinforming is under the control of the contingencies of human action that are sometimes automatic, sometimes personally mediated, and sometimes interpersonally mediated. The eld is facing a major challenge. There are many ways of phrasing that challenge. If agencies and agents are serious about crime prevention, the available evidence suggests that routine programming has to strive for adherence with the clinical, stafng, managerial, and normative principles of the RNR model of correctional assessment and

20

D. A. Andrews

treatment. According to the available evidence, there is no alternative to RNR adherence if human and social services are involved. There is plenty of room for enhanced effects.

Conclusions Everyone with an interest in crime prevention, inside or outside of justice and corrections, should seriously consider the RNR model. The ndings of Palmer (1995), of Lipsey (2009), and Andrews and Bonta (2010; Dowden & Andrews, 2005) were very similar. In regard to programmatic factors, at least two and sometimes all three reviewers agreed with the RNR principles of human service, risk, need, breadth, and general responsivity. Two of the three reviewers also agreed on the importance of the nonprogrammatic issues of stafng and management, integrity, setting, and wider-applicability with regard to age, gender, and ethnicity. Failures to agree were not based on a disagreement but on one reviewer not having attended to a particular factor or set of factors. To my knowledge, there are no systematic reviews of the research literature in recent years that could discount the ndings regarding the promise of the RNR-based programmatic and nonprogrammatic factors that are associated with effective crime prevention. They are applicable not solely to mainstream adult male offenders but to a wide range of other groups (Dowden & Andrews, 1999; Hanson et al ., 2009; Vieira et al ., 2009), special populations (Andrews et al ., 2001), and clinical problems (Miller & Carroll, 2006). At the same time, the RNR principles of specic responsivity, strengths, structured assessment, and professional discretion have been less well researched and no meta-analytic summaries of relevant evidence are available. Before it can be considered a law of effective crime prevention, an independent meta-analysis of the need and breadth principle must be conducted. To date, only our team has conducted such a review. Integrity in service delivery is second only to adherence with core RNR principles in meta-analytic explorations of the sources of variability in effective crime prevention. It also appears to be the case that the effectiveness of interventions is maximized with RNR adherence in combination with integrity of service delivery. However, for indicators of integrity to be strongly linked with crime prevention effects, the key clinical principles of RNR must be followed. As I used to say in training sessions, if you are not in adherence with RNR you may feel free to practice in whatever way you choose: you are free to do whatever is legal, just and ethical because practices that are not in adherence with clinical principles of RNR practice are inconsistent with effective crime prevention from the start. I strongly advise potential users of RNR to place the fear-mongering and rhetoric of some critical criminologists, a few clinical psychologists and a few clinical social workers in their proper intellectual context. My colleagues and I have met many users of RNR. Many of those users and I cannot help but laugh when RNR is described as hegemonic, immoral, unethical, and a negative model of crime prevention that is too deterministic, too prescriptive, restrictive of creativity, and it cannot be implemented! Such descriptions have yet to be accompanied by evidence of the incremental crime prevention effects of the presumably attractive alternatives to RNR. We are in complete agreement with our critics if what they really wish to say is that GPCSL, RNR, and the psychology of criminal conduct are woefully incomplete and will be replaced by more powerful models in the future. All knowledge that is presented in the tradition of rational empiricism is a work in progress. My colleagues and I applaud enhanced efforts in regard to specic responsivity considerations and motivational issues in particular. Examinations of the process of

The impact of nonprogrammatic factors

21

offender change promise to advance understanding and hopefully contribute incrementally to crime prevention. Contrary to the anti-RNR rhetoric, the desistance literature and the RNR literature share the same base in personality and cognitive social psychology including cognitive social learning (see Serin & Lloyd 2009, for a balanced integration of desistance and RNR). I single out the desistance literature because, compared to GLM, it obviously has something to offer. As described, it is silly to suggest that crime prevention will be enhanced by work with lower risk cases, by work that targets noncriminogenic needs predominately, and by the employment of weak inuence strategies. That assessment applies to desistancebased services that are not in adherence with RNR. I am in no position to speak for Shadd Maruna but I cannot believe that he would argue against GPCSL-based applications of the core human service, stafng, management, and normative principles of RNR. But I am willing to learn even if I kick and scream in the process.

References
Andrews, D. A. (1995). The psychology of criminal conduct and effective treatment. In J. McGuire (Ed.), What Works: Reducing Re-offending: Guidelines from Research and Practice (pp. 3568). Chichester: Wiley. Andrews, D. A. (2001). Principles of effective correctional programs. In L. L. Motiuk & R. C. Serin (Eds.), Compendium 2000 on Effective Correctional Programming (pp. 917). Ottawa: Correctional Service Canada. Andrews, D. A. (2006). Enhancing adherence to risk-need-responsivity: Making quality a matter of policy. Criminology and Public Policy, 5, 595602. doi:10.1111/j.1745-9133.2006.00394.x Andrews, D. A., & Bonta, J. (1995). LSI-R: The Level of Service Inventory Revised . Toronto: Multi-Health Systems. Andrews, D. A., & Bonta, J. (2006). The psychology of criminal conduct (4th ed.). Newark, NJ: LexisNexis/Matthew Bender. Andrews, D. A., & Bonta, J. (2010). The psychology of criminal conduct (5th ed.). Newark, NJ: LexisNexis/Matthew Bender. Andrews, D. A., Bonta, J., & Hoge, R. D. (1990). Classication for effective rehabilitation: Rediscovering psychology. Criminal Justice and Behavior, 17, 1952. doi:10.1177/0093854890017001004 Andrews, D. A., Bonta, J., & Wormith, J. S. (2004). The Level of Service/Case Management Inventory (LS/CMI). Toronto: Multi-Health Systems. Andrews, D. A., Bonta, J., & Wormith, J. S. (2009). The Level of Service Inventory/Risk-NeedResponsivity (LSI/RNI). Toronto: Multi-Health Systems. Andrews, D. A., Bonta, J., & Wormith, J. S. (2010). A question of more complete, less confused and more balanced reviews of the evidence regarding the validity and utility of assessment and treatment models in corrections. Unpublished paper. Andrews, D. A., & Dowden, C. (2005). Managing correctional treatment for reduced recidivism: A meta-analytic review of program integrity. Legal and Criminological Psychology, 10, 173187. doi:10.1348/135532505X36723 Andrews, D. A., & Dowden, C. (2007). 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, 439464. Andrews, D. A., Dowden, C., & Rettinger, J. L. (2001). Special populations within corrections. In J. A. Winterdyk (Ed.), Corrections in Canada: Social reactions to crime (pp. 170212). Toronto: Prentice-Hall.

22

D. A. Andrews

Andrews, D. A., & Kiessling, J. J. (1980). Program structure and effective correctional practices: A summary of the CaVIC research. In R. Ross & P. Gendreau (Eds.), Effective correctional treatment (pp. 441463). Toronto: Butterworths. Andrews, D. A., & Wormith, J. S. (1989). Personality and crime: Knowledge destruction and construction in criminology. Justice Quarterly, 6 (3), 289309. doi:10.1080/07418828900090221 Andrews, D. A., Zinger, I., Hoge, R. D., Bonta, J., Gendreau, P., & Cullen, F. T. (1990). Does correctional treatment work?: A clinically relevant and psychologically informed metaanalysis. Criminology, 28, 369404. doi:10.1111/j.1745-9125.1990.tb01330.x Bonta, J., & Andrews, D. A. (2007). Risk-need-responsivity model for offender assessment and rehabilitation (User Report 2007-06). Ottawa: Public Safety Canada. Bonta, J., Rugge, T., Scott, T., Bourgon, G., & Yessine, A. (2008). Exploring the black box of community supervision. Journal of Offender Rehabilitation, 47, 248270. doi:10.1080/10509670802134085 Bourgon, G., & Armstrong, B. (2005). Transferring the principles of effective treatment into a real world prison setting. Criminal Justice and Behavior, 32, 325. doi:10.1177/0093854804270618 Dowden, C., & Andrews, D. A. (1999). What works in young offender treatment: A meta-analysis. Forum on Corrections Research, 11, 2124. Dowden, C., & Andrews, D. A. (2000). Effective correctional treatment and violent reoffending. Canadian Journal of Criminology, 42, 449467. Dowden, C., & Andrews, D. A. (2004). The importance of staff practices in delivering effective correctional treatment: A meta-analysis of core correctional practices. International Journal of Offender Therapy and Comparative Criminology, 48(2), 203214. doi:10.1177/0306624X03257765 Farabee, D., Hser, Y.-I., Anglin, M. D., & Huang, D. (2004). Recidivism among an early cohort of Californias Proposition 36 offenders. Criminology and Public Policy, 3, 563584. Gendreau, P. (1996). The principles of effective intervention with offenders. In A. Hartland (Ed.), Choosing correctional options that work (pp. 117130). Newbury Park, CA: Sage. Goggin, C., & Gendreau, P. (2006). The implementation of quality services in offender rehabilitation programs. In C. R. Hollin & E. J. Palmer (Eds.), Offending behaviour programmes: Development, application, & controversies (pp. 69111). Chichester: Wiley. Hannah-Moffat, K. (2009). Gridlock or mutability: Reconsidering gender and risk assessment. Criminology and Public Policy, 8, 209219. doi:10.1111/j.1745-9133.2009.00549.x Hanson, R. K., Bourgon, G., Helmus, L., & Hodgson, S. (2009). The principles of effective correctional treatment also apply to sexual offenders (User Report 2009-01). Ottawa: Public Safety Canada. Hoge, R. D., & Andrews, D. A. (2002). Youth Level of Service/Case Management Inventory (YLS/CMI). Toronto: Multi-Health Systems. Kazdin, A. E. (2008). Evidence-based treatment and practice: New opportunities to bridge clinical research and practice, enhance the knowledge base, and improve patient care. American Psychologist , 63, 146159. doi:10.1037/0003-066X.63.3.146 Lipsey, M. W. (2009). The primary factors that characterize effective interventions with juvenile offenders: A meta-analytic overview. Victims and Offenders, 4, 124147. doi:10.1080/15564880802612573 Lowenkamp, C. T. (2004). Correctional program integrity and treatment effectiveness: A multisite, program-level analysis Unpublished doctoral dissertation, University of Cincinnati. Lowenkamp, C. T., Latessa, E. J., & Smith, P. (2006). Does correctional program quality matter? The impact of adhering to the principles of effective intervention. Criminology and Public Policy, 5, 575594. doi:10.1111/j.1745-9133.2006.00388.x Maruna, S. (2001). Making good . Washington, DC: American Psychological Association. McGuire, J. (2004). Understanding psychology and crime: Perspectives on theory and action. Maidenhead: Open University Press/McGraw-Hill Education.

The impact of nonprogrammatic factors

23

McNeill, F. (2009). Towards effective practice in offender supervision. Glasgow: Scottish Center for Crime and Justice Research. Miller, W. R., & Carroll, K. M. (2006). Rethinking substance abuse: What the science shows, and what we should do about it . New York: Guilford Press. Palmer, T. (1995). Programmatic and nonprogrammatic aspects of successful intervention: New directions for research. Crime and Delinquency, 41, 100131. doi:10.1177/0011128795041001006 Petrosino, A., & Soydan, H. (2005). The impact of program developers as evaluators on criminal recidivism: Results from meta-analyses of experimental and quasi-experimental research. Journal of Experimental Criminology, 1, 435450. doi:10.1007/s11292-005-3540-8 Serin, R. C., & Lloyd, C. D. (2009). Examining the process of offender change: The transition to crime desistance. Psychology, Crime and Law, 15, 347364. doi:10.1080/10683160802261078 Simons, D., McCullar, B., & Tyler, C. (2008). Evaluation of the good lives model approach to treatment planning. Presented at the 27th annual research and treatment conference of the Association for the Treatment of Sexual Abusers, Atlanta, GA. Smith, P., Cullen, F. T., & Latessa, E. J. (2009). Can 14,737 Women be Wrong? A Meta-Analysis of the LSI-R and Recidivism for Female Offenders. Criminology and Public Policy, 8, 183208. Taxman, F. S., Shepardson, E. S., & Byrne, J. M. (2004). Tools of the trade: A guide to incorporating science into practice. Washington, DC: National Institute of CorrectionsRetrieved from www.nicic.org/library/020095 Van Voorhis, P., Wright, E. M., Salisbury, E., & Bauman, A. (2010). Womens risk factors and their contributions to existing risk/needs assessment: The current status of a gender-responsive supplement. Criminal Justice and Behavior, 37, 261288. doi:10.1177/0093854809357442 Vieira, T. A., Skilling, T. A., & Peterson-Badali, M. (2009). Matching court-ordered services with treatment needs: Predicting treatment success with young offenders. Criminal Justice and Behavior, 36 , 385401. doi:10.1177/0093854808331249 Received 28 June 2010; revised version received 2 July 2010

You might also like