Topic: Rehabilitation: What Works to Change Offenders

In this assignment you will apply the readings and presentations in the Reading & Study in a meaningful way to clarify your understanding of the correctional system.  

The public seems to support early intervention and rehabilitation, but many people are still in prison or being sent away.  Why do you think rehabilitation and early intervention programs are not something that is more common in everyday correctional proceedings? What are some barriers that prevent early intervention from being more widely used?

Submit your thread by 11:59 a.m. on Wednesday of June 23, 2021.

NO LATE WORK!

7 Rehabilitation What Works to Change Offenders

Paul Gendreau

University of New Brunswick at St. John

Scholar of Rehabilitation Theory and Research

In 1982, Cullen published Reaffirming Rehabilitation. At the time, not many fellow criminologists—apart from his coauthor and friend Karen Gilbert—supported offender treatment. As discussed in previous chapters, Cullen’s criminological compatriots thought that rehabilitation was a benevolent mask—an excuse—for treating offenders coercively. They did not like the discretion given to judges and parole boards that enabled them to decide who went to prison and who got out of prison. They preferred a justice model that would define in clear terms the punishment for each crime and ensure that everyone would receive the same sanction. Equal justice before the law! What criminologists did not realize was the danger of saying that the correctional system now had no obligation to provide social welfare services to offenders. Its only obligation was to do justice—to inflict pain, albeit in a supposedly fair way.

As it turns out, Cullen was a smart criminological pup. In Reaffirming Rehabilitation, he called one section “The Poverty of the Justice Model: The Corruption of Benevolence Revisited” and prognosticated that the justice model approach would backfire. Indeed, as reviewed earlier (in  Chapter 3 ), this attack on rehabilitation, although not without merit, had the unanticipated consequence of facilitating the conservatives’ takeover of corrections and of legitimating get tough rhetoric. Of course, few listened to Cullen. As his colleague Larry Travis noted upon publication of Reaffirming Rehabilitation: “Well, you are pissing in the wind.”

Because most American criminologists thought that something was wrong with Cullen, he had to find like-minded associates wherever he could. Alas, Cullen was befriended by several Canadian psychologists, most notably Don Andrews, James Bonta, and, in particular, Paul Gendreau. These remarkable scholars were strong advocates of offender treatment. It is instructive that, as Canadians, they were not enmeshed in America’s social context that led criminologists to suspect the state’s motives and power. They still embraced the belief that government should have a social welfare mission to support its citizens. And as psychologists, they were aware of a large scientific literature showing that behavior—from rats to humans—could be changed.

They were somewhat mystified by American criminologists’ conviction that offenders could not be saved from a life in crime. They saw such a view as mere ideology and, still worse, as professionally unethical. Moreover, as the get tough movement swung into full force, they were disquieted by the simplistic belief, increasingly voiced by politicians, that offenders could be induced to conform by harsh punishment rather than by planned treatment based on scientific knowledge. The Canadians observed that this wild embrace of punishment was being made by people who knew absolutely nothing about the psychology of punishment. If they had read the available literature, the get tough crowd would have realized that much of their thinking was pretty stupid, eh?

Cullen (and Jonson) finds the Canadians’ proclivity to say “eh” a rather cute cultural quirk, eh? It is said that Canada received its name when its founders were told they had to reach into a hat, pull out three letters of the alphabet, and call their country by what the luck of the draw revealed. They pulled out the three letters and, after each one, held it aloft in this way: C, eh?—N, eh?—D, eh?

Talk, especially criticism, is cheap. But Cullen’s Canadian friends did more than complain about American criminologists and politicians! Instead, they devoted their careers to developing a paradigm that could demonstrate what works to change offenders. They undertook two interrelated tasks. First, they systematically compiled evidence on the characteristics of effective interventions with offenders. Second, they developed a coherent treatment theory that spelled out the principles of effective correctional intervention (see, e.g., Andrews & Bonta, 2010; Gendreau, 1996). Cullen and Jonson—who also now hangs around occasionally with the Canadians—believe that our northern neighbors’ approach to rehabilitation comprises a powerful paradigm that should guide treatment interventions with offenders. We will return to this matter closer to the chapter’s end, eh?

Let us provide a roadmap of the trip we will take in this chapter. After briefly reminding readers of the concept of rehabilitation, we will focus in detail on the nothing works debate that was at the heart of corrections for the last quarter of the 20th century and that continues to rear its ugly head even today (Cullen, Smith, Lowenkamp, & Latessa, 2009). You will hear again the name of Robert Martinson whose 1974 essay seemingly thrust an empirical dagger into the heart of rehabilitation, killing it off once and for all. But once Martinson made the legitimacy of offender treatment an empirical issue, this opened up an important avenue for rebuttal: If researchers showed that treatment does in fact work, then rehabilitation would have to be reaffirmed, not rejected. We will show how, to a large extent, this is what has occurred. Mounting evidence now exists that treatment interventions work—and, importantly, work better than punitive-oriented programs.

As will be seen, meta-analysis has been a key statistical procedure used to assess what works and what does not work to reduce reoffending. We will show how this novel technique shaped the debate over rehabilitation. Okay, do not recoil at the thought of more statistics and of some novel statistical technique (that sounds like trouble, eh?). Cullen and Jonson are not out to inflict more academic misery on you. We will keep the discussion simple. The take-away point is that meta-analysis is capable of detecting, in a more precise way, the effects treatments have on things like offender recidivism. Meta-analysis is neither liberal nor conservative. It is agnostic about what it finds. The fact that meta-analyses of interventions showed that the nothing works doctrine was incorrect was pretty momentous. This is why we focus on this issue in some detail below.

The chapter’s punch line is that knowledge now exists as to what will and will not work to change offender behavior. Evidence-based corrections provides us with a fairly clear idea of what is a waste of time, energy, and public monies and what is the best bet for reforming offenders—and, in so doing, protecting public safety. Remember, correctional quackery—doing stupid things to offenders—not only does not save the wayward but also increases the likelihood that some innocent citizen will be victimized. Put another way: Effective corrections = public safety. Thus, we will devote the back third of the chapter to this issue, ending with a discussion of the Canadians’ principles of effective correctional intervention.

The Concept of Rehabilitation

The concept of rehabilitation was discussed way back in  Chapter 1 . So, it behooves us to take a few moments to review the essence of this correctional theory. Various definitions of rehabilitation can be used. Here is an excellent one by Francis Allen (1981):

One may begin by saying that the rehabilitative ideal is the notion that a primary purpose of penal treatment is to effect changes in the characters, attitudes, and behavior of convicted offenders, so as to strengthen the social defense against unwanted behavior, but also to contribute to the welfare and satisfaction of others. (p. 2)

This is the definition we developed and, as such, prefer:

Rehabilitation is a planned correctional intervention that targets for change internal and/or social criminogenic factors with the goal of reducing recidivism and, where possible, of improving other aspects of an offender’s life. (Cullen & Jonson, 2011b, p. 295)

Our definition contains five components. Let’s see what they are:

· The intervention is undertaken by the correctional system.

· The intervention is planned. It does not occur by chance but is designed to have specific features.

· The intervention targets for change the factors that are causing the offender’s criminality. These factors may be internal or in the offender’s social environment; regardless, the purpose of the intervention is to change them.

· The intervention’s main goal is to reduce recidivism. This should help the offender, but it also is critical to protecting the public.

· The intervention may also help to improve the offender in other ways (e.g., making him or her more educated or psychologically healthier).

Remember, rehabilitation is a utilitarian goal of the criminal sanction. Why? Because it is justified largely by its utility or benefits. It does not seek to achieve justice simply for justice’s own sake but also wishes to produce social good in terms of improving the offender, reducing recidivism, and increasing public safety. Rehabilitation also has a social welfare purpose. Similar to restorative justice but unlike just deserts, deterrence, and incapacitation, the goal is to use the correctional system to provide services to offenders that improve their lives. Finally, rehabilitation is the only correctional theory that embraces the medical model and individualized treatment. It argues that, through science, we can diagnose what is wrong with offenders and then prescribe correctional medicine (so to speak) to cure the underlying ailment.

As seen earlier in  Chapter 2 , rehabilitation also is a theory that has mattered—that has had consequences. In the first two decades of the 1900s, called the Progressive Era, the criminal justice system was reformed in hopes of achieving offender treatment. Francis Allen (1981) captures this fact:

Perhaps the most tangible evidences of the dominance of the rehabilitative ideal are found in its legislative expressions. Almost all of the characteristic innovations in criminal justice in this century [the 20th century] are reflections of the rehabilitative ideal: the juvenile court, the indeterminate sentence, systems of probation and parole, the youth authority, and the promise (if not the reality) of therapeutic programs in prisons, juvenile institutions, and mental hospitals. (p. 6; see also Rothman, 1980)

Until the late 1960s and early 1970s, the hegemony (a fancy word for widespread acceptance) of rehabilitation was virtually complete. It was seen as a sign of a modern and civilized mind to embrace offender treatment as the goal of corrections. Vengeance was dismissed as a vestige of more barbaric times when, quite embarrassingly, public hangings, whippings, and brandings were still acceptable. Only yahoos and the socially dumb held onto such an outdated belief. Again, one more insight from Francis Allen (1981):

Nevertheless, it is remarkable how widely the rehabilitative ideal was accepted in this century [the 20th] as a statement of aspirations for the penal system, a statement largely endorsed by the media, politicians, and ordinary citizens. It was in the universities, however, that the dominance of the rehabilitative ideal became most firmly established. . . . Yet even a brief glance at the college criminology textbooks in wide use at midcentury clearly reveals the importance accorded penological treatment in criminological thought, and the almost unchallenged sway of the rehabilitative ideal. (pp. 6–7, emphasis added)

Why did rehabilitation lose its grip on the minds of Americans and on crime control policy? Again, we urge you to revisit  Chapter 2  for the details. But for our purposes here, there were two types of criticisms that emerged. The first involved the criticism of state use—or misuse—of discretion. To individualize treatment, judges and corrections officials had to have the discretion—the freedom—to decide who went to prison, who was paroled from institutions, and what happened to those supervised in the community. But what if this discretion was not used for appropriate treatment goals but rather was abused? For conservatives, the abuse of discretion meant that dangerous predators were allowed to roam free and victimize. For liberals, the abuse of discretion meant that judges would discriminate in sentencing against the poor and people of color and that corrections officials could tell inmates to obey or face eternal incarceration. This is why people of all political orientations coalesced to attack rehabilitation by taking discretion away from judges and correctional officials. They did this mainly by implementing determinate sentencing (or sentencing guidelines) and by abolishing (or limiting) parole.

The attack on the abuse of discretion was a powerful one. It prompted liberals to embrace just deserts (see  Chapter 3 ) and conservatives to embrace get tough policies (see  Chapters 4  and  5 ). If rehabilitation were to guide correctional policy, then this potential abuse of discretion would be a vulnerability that advocates of offender treatment would have to address. But the debate on rehabilitation changed qualitatively in 1974 with the publication of Robert Martinson’s essay, “What Works? Questions and Answers About Prison Reform.” We discussed this study in  Chapter 2 , where we noted that the punch line to Martinson’s review of the existing evaluation evidence was that nothing works to rehabilitate offenders. Martinson’s work was quickly accepted by critics who were already prepared, due to their concern over state discretion, to believe his message. They now seemed to have science on their side. Even the data seemed to confirm that offenders were beyond redemption by any known correctional intervention.

Here is the key point: Martinson’s work reframed the debate about rehabilitation. It once was about the quality with which state discretion was used. Now, it was transformed into a debate about the effectiveness of treatment programs. Critics of rehabilitation likely embraced the effectiveness argument because it seemed so simple to make and so damning in its implications. They could ask: How can you support something that does not work? People like Cullen and Jonson would have no real response to this kind of question. The most we could say would be something like: “Oh yeah? Well, yo momma!” Growing up in Boston, Cullen found this response, delivered with a bit of a swagger, an effective rebuttal to any challenge that he did not like but to which he had no coherent response. In fact, in the city streets, attacking the sanctity of an opponent’s mother was a culturally valued, normative rejoinder. Unfortunately, much to Cullen’s dismay, in criminological circles, this retort does not usually win arguments.

But by using the effectiveness issue, critics of rehabilitation ironically have ensured their own defeat—in two ways. First, they created the opportunity for treatment to regain its credibility if its advocates could show empirically that rehabilitation worked to reduce recidivism. Second, they placed the alternatives to rehabilitation—punishment-oriented correctional interventions (e.g., boot camps, intensive supervision programs)—in the advocates’ methodological gun sights. If rehabilitation was fair game to evaluate, so too were these punishment-oriented interventions. If these programs were found wanting—as they were—then the tables could be turned: How can you support punishment when it does not work?

Science is thus a dangerous thing to use for ideological reasons. Science can be a cold partner. It can show you that your most cherished beliefs are wrong. It can turn on you and make you change your mind—if you are intellectually honest. In this case, science has indeed turned on rehabilitation’s critics. In the end, it does appear that treatment works and punishment does not (Cullen, 2005).

Knowing What Works

Let’s assume that you worked for an agency and the head administrator asked you, “Does rehabilitation work to reduce recidivism?” And then let’s say that you answered, “Well, in my experience, I think that it does. Just last week, I had met an ex-offender I once had in a program and he turned his life around.” Unfortunately, your boss responds, “I respect you and your experience. But I just read this article by Doris MacKenzie, and she said that what we need to do is to have an evidence-based corrections. So, what I am really asking you is whether the evidence shows that rehabilitation works. Or was that Martinson fellow right a long time ago?” Your boss then follows up: “What I want you to do is to write a report summarizing the research evidence on rehabilitation. Why don’t you just review the evidence for me?”

Now, what the heck does it mean to review the evidence? Your boss probably thinks that this is a pretty straightforward request. But it is actually a complicated thing to do. And how you conduct the review can influence what answer you might give in your report to your boss. There are at least four steps in a review of the evidence:

· You must find or collect all the studies you can. Ideally, you would include not only published studies but also unpublished studies. If you do not, you may bias your results. (This is sometimes called the file drawer problem; many reviews do not include studies still in the file drawer and not published.)

· You must decide which of the studies you find are methodologically sound enough to include in your review. Ideally, you would like to include program evaluation studies that have used an experimental design (in which offenders are randomly assigned to the treatment and control groups). But not all evaluations are pure experimental designs (e.g., they might be quasi-experimental designs in which a control group but not random assignment is used). Among these studies, however, some are so flawed that to include them in your study would result in misleading findings. Accordingly, you have to know and use the appropriate methodological standards to decide which studies are (or are not) of sufficient quality to be included in the review.

· You must then decide what method or way you will use to convey the information. As we will see, one method is a narrative review; another is a meta-analysis. This selection of method is critical because it can lead to different conclusions being reached by the reviewer.

· You must then interpret the review you have conducted. After all this work, what do the studies mean? What do they tell us about, in this case, the effectiveness of rehabilitation?

For the longest time, we decided what the research meant by conducting a narrative review. If you have read literature reviews in the past, this is undoubtedly the style you have encountered. It also is likely to be what you may have used for papers you have written in college. In this type of review, the scholar essentially describes the studies he or she has collected and tells a story about what they mean—thus the term narrative.

For example, let’s assume that 10 studies have been published on therapeutic communities. In a narrative review, the scholar would take each study—one at a time—and review how the study was conducted, who was in the sample, how recidivism was measured, how long the study was (i.e., the length of time the offenders were followed up), what the findings were, and so on. After going through the studies, the author also might count up how many times therapeutic communities reduced recidivism and how many times they did not. If there were 10 studies, the scholar might find that the intervention reduced recidivism half the time. The vote would be 5 to 5. Sometimes, this is called a ballot-box approach because the researcher is counting, in essence, votes or ballots for and against the treatment program’s effectiveness.

After conducting this review, the scholar would then have to interpret what all this meant. This is precisely what Martinson did in his study of 231 programs. After examining all the evidence—a pretty large task—Martinson (1974) offered this widely cited conclusion: “With few and isolated exceptions, the rehabilitative efforts that have been reported so far have had no appreciable effect on recidivism” (p. 25; emphasis in the original). In short, nothing works!

Narrative reviews are very valuable. They help us to decide what the research on a topic says. But, at times, the evidence is of such a nature that there is disagreement about what the studies actually say. Why? Because a narrative review is mostly qualitative in nature. There may be some counting of studies—such as in the ballot-box approach where the number of studies showing that a program worked or did not work is tallied. But for the most part, a narrative review relies on the scholar to make sense of what the studies mean. In this case, the scholar must use his or her judgment. Of course, this is not just any judgment; it is a judgment informed by the standards of a field of study and often by years of training. Still, in the end, there is room for two scholars to read the same evidence and reach different conclusions.

A key problem is whether a scholar sees the glass as half full or as half empty. Half-full reviewers tend to emphasize what does work and will point to the promising programs that exist. Half-empty reviewers (such as Martinson) tend to emphasize what does not work and will conclude that treatment success in programs is like flipping a coin—it’s a matter of chance. Thus, if, in evaluating an intervention, five studies show a reduction in recidivism and five do not, what should we conclude? Is the treatment glass half full or half empty? Is the program effective (“look at the five times it worked”) or ineffective (“look at the five times it did not work”)?

When it is a close call like this, all sorts of biases can creep into what should be an objective review of the evidence. Let’s say that the scholar doing the review really likes the concept of developing therapeutic communities in prison. The scholar is likely to emphasize how this program has been used in at least five prisons successfully. Alternatively, let’s say that the scholar thinks that this idea is a crock. In this case, he or she is likely to say that these programs did not work in five prisons. In this researcher’s mind, therapeutic communities fail as often as they work—and thus are unproven and a waste of the taxpayers’ money.

Not all narrative reviews are a close call. In some cases—such as the effectiveness of boot camps—the evidence is so one-sided (they do not work) that even advocates of the program cannot dispute the data. But, again, there is a looseness to narrative reviews that makes them open to different interpretations and dispute. As will be noted later, the statistical technique of meta-analysis is an alternative way to assess what works. This approach is quantitative and less susceptible to the half-full versus half-empty kind of debate.

Below, we thus look at two different responses to Martinson’s claim that nothing works in correctional treatment. The first response was presented by scholars using a narrative–ballot-box approach to reviewing the evidence on treatment effectiveness. The second response was presented by scholars using meta-analysis. Both responses reached the same conclusion that rehabilitation was an effective strategy to reform offenders.

Challenging Nothing Works: Narrative Reviews

Standing Up to Martinson: Ted Palmer’s Rebuttal

Unless one lives in a specific historical time, it is difficult to understand how ways of thinking can sweep across a field and reshape how nearly everyone thinks. Robert Martinson’s 1974 essay was that kind of pathbreaking work. He was communicating a message that, as noted, many people were prepared to hear. He was giving scientific legitimacy to a new way of thinking about corrections. For decades, the civilized or progressive view was that we should try to rehabilitate the wayward. Now Martinson was telling everyone that this was a fool’s errand. Our benevolence was misplaced. He cautioned that we were doing the objects of our assistance—offenders—no good and maybe even harming them and the society at large.

Martinson’s ideas were powerful. He was interviewed on 60 Minutes, which gave his views a national audience. Cullen was affected by his work, came to doubt rehabilitation (until he came to his senses!), and even was interviewed by Martinson for a job to do a follow-up on the 1974 article. (Cullen was distinctly unimpressive in his interview and never heard a word back from Martinson.) As Allen (1981) noted a short time after Martinson’s article, “the rehabilitative ideal has declined in the United States; the decline has been substantial, and it has been precipitous” (p. 10). A true paradigm shift occurred. One day, it seemed, everyone believed in rehabilitation; the next day, nobody did. “What is most significant about the 1970s, and what distinguishes it from the past,” observed Allen, “is the degree to which the rehabilitative ideal has suffered defections, not only from politicians, editorial writers, and the larger public, but also from scholars and professionals in criminology, penology, and the law” (pp. 8–9).

It took courage to stand up to Martinson. By Martinson, we do not really mean this one person. Rather, we mean that it took courage to stand up to all those who embraced Martinson’s nothing works ideas, who would work to discredit anything positive about correctional treatment, and who were disparaging of anyone who would still suggest that offenders could be saved (Gottfredson, 1979). But one man did quickly stand up to the nothing works crowd: Ted Palmer.

At first glance, Palmer was an unlikely gladiator in the treatment battles. He is rather thin, short in stature, and very soft-spoken. Cullen admires him, whereas Jonson thinks he is grandfatherly and cute. But Palmer also is blessed with an abiding sense of integrity, a compassionate heart, and a sharp mind (see Palmer, 1978, 1992, 1994, 2002). When in his presence, he strikes one as a special human being. Trained as a psychologist—he earned his Ph.D. from the University of Southern California in 1963—he worked for years as a senior researcher for the California Youth Authority. By 1974, he had come to know delinquent youths and to see programs that worked to reform more than a few of them. Palmer was not naïve about offenders’ criminality, but he was aware of the possibility of effective behavioral change—which, after all, is what psychologists do for a living. He also could read an article and, through diligent analysis, see that the treatment glass was, at the least, half full.

When Martinson’s study appeared, Palmer knew something was amiss. Meta-analysis was not available at the time, so he could not critique the Martinson nothing works findings using this approach. But in what is now one of the classic rebuttals in the history of criminology, Palmer (1975) took the time to go back over Martinson’s interpretation of what the evaluation studies actually found. What he discovered was rather remarkable.

In his 1974 article, Martinson cited 82 studies (of the 231 in his pool of studies), because these had recidivism data. Palmer reread these works. Based on Martinson’s nothing works conclusion, one might have expected to find that none of these studies—or virtually none of these studies—showed that treatment reduced recidivism. But this is not what Palmer discovered. In fact, he calculated that 39 of the studies—48% of the total—could be categorized as reducing recidivism. Huh? How, then, could Martinson ever conclude that treatment programs had “few appreciable effects”? Was Martinson lying or just plain stupid?

Well, actually, Martinson’s point was subtler—and not understood by the vast majority of readers who were anxious to believe that nothing works to reform offenders. Martinson, then, was saying this:

· First, there were various types of treatment programs (e.g., education, employment, counseling).

· Second, when considering the evidence, he found a consistent pattern across these treatment types or categories.

· Third, this pattern was that within any type of treatment, the program sometimes worked and sometimes did not work.

· Fourth, as a result, it was not possible to say, with any assurance, that any particular type of treatment program would consistently work to reduce offender recidivism.

· Fifth, therefore, nothing works—that is, no one specific type or category of correctional treatment works consistently or reliably to reduce offending recidivism.

Thus, let’s suppose that in 1974, someone had come up to Martinson and said: “Okay, Robert, after everything you have studied, tell me what I can do to keep offenders from going back into crime. What kind of program should I set up?” His response would have been: “I don’t know. Because treatment programs have inconsistent results, whatever I would tell you is likely to be as wrong as it is to be right.” So, here then is the key point:

· Being a half-empty reviewer, Martinson concluded that since no one type (or modality) of program could be shown to work all or most of the time, nothing works in correctional treatment.

But what kind of reviewer do you think Ted Palmer was? Right, he was a half-full reviewer! So, his conclusion was exactly the opposite. For him, there was now a body of evidence showing that about half the time, treatment worked!

Where does this leave us? We have two criminological crowds. One is chanting “nothing works” and the other is chanting “something works.” Does not work; does so! Does not, does so! Does not, does so. Does not, does so. Hmm. After a while, this does not seem to get us anywhere useful. Is there something else that might be done to resolve this criminological standoff? Well, actually, there is. Again, if we find that half the programs reduce recidivism and half do not, the next logical step would be to find out whether those programs that worked differed in important ways from those programs that did not work. There are two possibilities:

· Success or failure in rehabilitation is a random process. This, really, is Martinson’s position. In this case, we will never find features that differentiate successful from unsuccessful programs. There is no underlying pattern of effective characteristics to discover. Whether programs work or do not work is like a flip of the coin; it is a chance event.

· Success or failure in rehabilitation is patterned, not random. This, really, is Palmer’s position—and the position of the Canadian scholars (such as Paul Gendreau, James Bonta, Don Andrews) and (guess who?) of Cullen and Jonson. In this case, further study would find that there are features of programs that work that make them differ from programs that do not work. In turn, once these differences were detected, it would be possible to use them to develop principles or guidelines for conducting effective correctional interventions. Programs that follow the principles will reduce recidivism, whereas those that violate the principles will not reduce recidivism.

As it turns out, the second position is correct. Success and failure in rehabilitation is patterned and not random. Some things work better than others. We use fancy terms to describe this. We say that the effects of treatments on recidivism are not homogeneous but heterogeneous. Do not forget this fancy terminology. If you use it, it will make you seem quite erudite. Plus, we will return to these ideas shortly.

The second position here was important because it told scholars that, in the aftermath of Martinson’s study, there was work to be done. It was not the case that nothing works, but it was the case that we needed to do a lot of analysis to establish the features of effective as opposed to ineffective rehabilitation programs. Unfortunately, the nothing works message caused many scholars to abandon the study of rehabilitation programs. After all, if one were convinced that nothing works, then why would one devote one’s career to studying something that was a fruitless enterprise?

Fortunately, there was a small group of scholars who did continue to study rehabilitation (Cullen, 2005). As noted, the most prominent were Canadian scholars—Don Andrews, James Bonta, Paul Gendreau, and their colleagues. They were academics, researchers for the government, and people who set up treatment programs—and sometimes all three at various times in their careers. They really knew a great deal about programs and offenders. Eventually, the Canadians would derive the principles of effective intervention, which tell why some programs fail (they do not follow the principles) and others succeed (they follow the principles). Initially, however, the Canadian scholars saw poor Ted Palmer, like the sheriff in the movie High Noon, standing up as a lone figure facing down the nothing works crowd. They decided to protect Palmer’s back.

Thus, they migrated across the border into United States’ criminology, attending conferences and publishing in our journals. Their message was clear: Americans do not know what they are talking about. Their cynicism about treatment reigns because they are too lazy to go to the library and read the research. They are in need of bibliotherapy for cynics.

Bibliotherapy for Cynics: The Canadians Get Involved

“What was up with this Martinson guy and why were Americans snookered by him, eh?” So asked the bewildered Canadian psychologists. The Canadians had studied behavior change when still psychology graduate students (changing the behavior of a rat or two) and had changed the behavior of more than a few offenders in their careers. What did Martinson mean by suggesting that human conduct, including criminal conduct, could not be altered? Even more significant, when they searched the literature, they could find all sorts of evaluation studies showing that treatment interventions reduced recidivism (more generally, see Andrews & Bonta, 2010). Had someone ripped all these studies out of the academic journals south of their border or were Americans simply not looking for them any longer? Eh?

In this regard, Paul Gendreau and Robert Ross—two prominent Canadian scholars—compiled two of the more important and widely cited narrative reviews. One review, published in 1979, assessed 95 studies conducted between the years of 1973 and 1978. They titled this article “Effective Correctional Treatment: Bibliotherapy for Cynics.” Cullen and Jonson love the subtitle “Bibliotherapy for Cynics” because it suggests that those who were then cynical about rehabilitation—who embraced the nothing works doctrine—needed some bibliotherapy; that is, they needed to read the available literature to cure themselves of their erroneous nothing works thinking. In 1987, Gendreau and Ross published a second massive review of the evidence in an article they titled “Revivification of Rehabilitation: Evidence From the 1980s.” This was an update of their 1979 work. This second, updated review assessed 130 studies and covered the years from 1981 to 1987. Martinson’s original study had surveyed evaluation studies appearing only up to 1967; the starting date for inclusion in his assessment was 1945.

In their two reviews, Gendreau and Ross uncovered literally scores of examples of treatment interventions that were successful in reducing recidivism. For Gendreau and Ross, Martinson’s review had been incomplete and now was outdated. Indeed, the sheer number of these programs belied the idea that nothing works. In another critical conclusion, Gendreau and Ross revealed that behaviorally oriented programs (e.g., incentive systems, behavioral contracts) showed signs of being especially effective. This finding was important because Martinson’s review did not contain a category on behavioral programs—a glaring omission caused by the lack of these studies in his sample of evaluations.

Further, Gendreau and Ross noted that successful programs targeted or focused on changing what they called criminogenic needs. As used by Gendreau and Ross, the term criminogenic needs referred to two things:

· Factors or predictors that the empirical research has shown are related to offender recidivism (e.g., antisocial attitudes).

· Factors or predictors that can be changed. For example, an attitude can be changed. In contrast, past record is a predictor of recidivism but it cannot be changed.

In short, if interventions focused on factors that were shown to cause recidivism and that could be changed, they were likely to be effective. By implication, this meant that if interventions focused on factors that were not related to recidivism (e.g., self-esteem) or could not be changed (e.g., past criminal record), they were unlikely to work.

Gendreau and Ross went on to make another point that we will revisit a bit later in the chapter. They started with the observation that offenders—like other humans—are marked by individual differences (recall our previous discussion of heterogeneity). Some of these differences pertain to their criminality; offenders differ in their level of risk for reoffending. And some differences relate to their personalities and their ability to learn. Gendreau and Ross presented evidence that the effectiveness of treatment programs can vary substantially to the extent that offenders’ individual differences are measured and taken into account in the delivery of services. They suggested, for example, that high-risk offenders benefited the most from treatment interventions and that offenders with low intellectual abilities would benefit more from programs in structured learning situations. Here, they were making beginning steps at evolving principles or guidelines that, if followed, would increase the likelihood of treatment being effective.

Finally, Gendreau and Ross illuminated a major reason why correctional programs fail: They lack therapeutic integrity. For example, programs often had no underlying theory of crime, targeted for change factors that were unrelated to recidivism, used interventions that were too short or not intensive enough, employed staff that were untrained in the intervention being used, and so on. Indeed, how could rehabilitation programs ever hope to work in these circumstances? Think if medical interventions were conducted this way: no theory of why disease occurs, causes of the illness not targeted for change, hospital stays that were too short, doctors who were not trained in medical school, and so on. Again, we would call this quackery—in our case, correctional quackery (Latessa et al., 2002).

When examined objectively, the reviews of Gendreau and Ross were quite persuasive. And given the extensive evidence amassed, they should have transformed the thinking of those who endorsed the doctrinaire position that rehabilitation programs could not be effective. But, again, many criminologists, not to mention policy makers, already had made up their minds that treatment was ineffective. They were not good listeners to the empirical story being told.

Equally important, Gendreau and Ross had presented a narrative review. As advocates of rehabilitation, they could be accused of explicit or implicit bias. They could be accused of misreading the evidence—of seeing the glass as half full when it was really half empty. They could not point to any hard quantitative evidence showing definitively that correctional treatment worked. For many casual observers, this seemed like a loud argument in which it was the Canadians’ word versus the critics’ word. It would take another method for reviewing evaluation studies to make attacks on the effectiveness of rehabilitation more difficult to sustain. It is to this method—meta-analysis—that we now turn.

Challenging Nothing Works: Meta-Analytic Reviews

To get right to the point, a key strength of a meta-analysis is that it can take several hundred studies and tell you with one itty-bitty number what the effect of treatment is on recidivism. Martinson’s full report on his review of the studies ran to 736 pages (Lipton, Martinson, & Wilks, 1975). Even his article was 32 pages long (Martinson, 1974). Gendreau and Ross’s 1987 article extended to almost 60 pages. Whew! That is a lot of reading just to try to find out if rehabilitation works to reduce reoffending.

In many ways, going over study after study would be similar to watching a video of every at-bat a hitter had across every game of an entire season. Nobody would ever do that, of course. We would prefer to just look at the player’s batting average. Hmm. The batter is hitting .325. Hall of Fame material. Hit .275. Not bad for a shortstop. Oh, no: .225. Send the player to the minors, likely never to be seen again. Of course, this one number does not tell us everything about a player’s performance; also relevant are a player’s runs batted in, on-base percentage, slugging percentage, prowess in the field, and so on. Still, a batting average is a parsimonious way to reduce several hundred trips to the plate to an understandable number. A high batting average is good; a low batting average is bad.

During the 1990s, scholars increasingly subjected the available treatment evaluation studies to meta-analysis. The Canadians conducted one of their own (Andrews et al., 1990), as did other reputable scholars (Lösel, 1995). But the most influential work was published by Mark Lipsey (1992, 1995, 1999a, 1999b, 2009; Lipsey & Cullen, 2007; Lipsey & Wilson, 1998). Lipsey is a rigorous scholar with an impeccable reputation. Although sympathetic to rehabilitation, he was not clearly identified with the pro-treatment crowd. As they say in Tennessee, where he works at Vanderbilt University, Lipsey had no dog in this hunt. Thus, given his impeccable methodological credentials and no agenda, any meta-analysis he conducted would have credibility. When he reported the batting average for rehabilitation, it would be believed! Overall, Lipsey argued that treatment’s batting average was high enough to keep it in the correctional major leagues.

The other issue to keep in mind is that when correctional intervention programs were meta-analyzed, these included not only human services–oriented programs (those seeking to improve offenders in some way) but also punishment-oriented programs (those seeking to scare, inflict pain on, or discipline offenders in some way). We have already alluded to the results of the punishment-oriented programs in  Chapter 4  on deterrence theory. But as we will see, it is important that, as predicted by the Canadians and others, these harsher interventions have proved decidedly ineffective. Remember, this is the Bartlett Effect, named after Cullen’s late dog: Putting offenders’ noses in it does not make them less likely to recidivate.

What Is a Meta-Analysis?

What is a meta-analysis? Well, you should have some idea, since in other chapters we have alluded to this method for making sense of a bunch of studies done on a particular intervention or topic. Here, we want to explain this approach in more detail because it proved to play a central role in rehabilitation’s revitalization. So, to start with, it is another way of summarizing evaluation studies that have been done on treatment. In contrast to a narrative review, however, a meta-analysis is an attempt to use quantitative methods—or statistics—to synthesize all the studies that have been conducted. This is why the technique of meta-analysis is also referred to as a quantitative synthesis of the literature. It is not necessary to know the intricate details of this method (although readers certainly are free to learn them). Knowing the essence of what a meta-analysis involves is sufficient for our purposes.

In any evaluation study, there is a statistical relationship between (1) the treatment intervention and (2) the measure of recidivism. This statistical relationship can fall into one of three categories:

· The relationship is zero, which means that treatment has no effect on recidivism.

· The relationship is negative, which means that the treatment actually increases recidivism.

· The relationship is positive, which means that the treatment reduces recidivism.

Note that some authors reverse the meaning of the positive and negative in their studies (e.g., negative = lower recidivism whereas positive = higher recidivism). That really does not matter. The point is that a treatment can either increase or decrease recidivism—or have no effect.

The relationship between the treatment intervention and measure of recidivism is called the effect size. What a meta-analysis does is to compute the relationship between the treatment variable and recidivism for every study in the sample of studies being reviewed. When this is done, the researcher comes up with the average or mean effect size. This is treatment’s batting average (so to speak). Moreover, this average effect size is a number, which is usually expressed as a correlation coefficient (Pearson’s r).

So, what you get after reviewing tens, if not hundreds, of studies is essentially one number. It is more complicated than this, actually, because there are different ways of conveying this number (e.g., an effect size weighted by the sample sizes of the studies, presenting confidence intervals). But we are not going to bother you with this stuff. In the end, what you need to know is that there is a way to estimate what the magnitude of the relationship is between treatment interventions and recidivism. And here is a key point:

· Can you see the power or persuasiveness of this method? It’s this: You have one number that shows whether rehabilitation works and, if so, by how much!

In this regard, what would Martinson and other nothing works critics have argued should be the relationship between treatment programs and recidivism? If you answered zero, then you are brilliant! Critics might even have predicted a negative number, which would show that treatment interventions increased recidivism. Such a number would show that treatment really doesn’t work.

But what would happen if the effect size was positive? This would show that across all treatment programs, rehabilitation reduced recidivism. And what would happen if one analyzed different types of treatment programs and found that some program types really reduced recidivism while others did not? Alas, this is precisely what the meta-analyses showed in the 1990s and continue to show today!

Let’s just touch on a few more points. Let’s say that you are reading an article and the reviewer concludes that, overall, the effect size is +.20. That is, the r value is .20, which means that the correlation between treatment and recidivism is .20. What the heck does that mean?

In practical language, the r value (the effect size) can be seen as the difference in recidivism between the treatment group (the group that received the rehabilitation program) and the control group (the group that did not receive the rehabilitation program). One common approach is to assume that the average recidivism rate is 50%. If this were the case, then an effect size of +.20 would mean that the control group had a recidivism rate of 60% while the treatment group had a recidivism rate of 40%. This is a 20 percentage point difference. Similarly, a 10 percentage point difference would mean that the control group’s recidivism rate was 55% and the treatment group’s recidivism rate was 45%. As you can see, even a statistically modest effect size (e.g., +.10) can have practical policy effects. A 10% reduction in recidivism, for example, can save a lot of crime.

Although we are fans of meta-analyses as a way of synthesizing large bodies of research evidence, no method of analyzing or summarizing studies is without its weaknesses. Perhaps the largest problem is the garbage in–garbage out problem. No matter how sophisticated the statistical technique, the summary is only as good as the studies being analyzed. In practical terms, this means that if the studies available are too few in number or mostly are based on weak methodology, then the results reached in the meta-analysis are going to be open to question. Fortunately, in the rehabilitation area, the body of research studies is fairly extensive and allows for meta-analyses whose results are pretty believable.

We are now prepared to find out what the existing meta-analyses show about treatment effectiveness! We cover this research in two sections:

· What did the meta-analyses find was the overall effect of rehabilitation programs?

· What did the meta-analyses find was the effect of rehabilitation when different types of programs were assessed? Was there homogeneity or heterogeneity in the effect size?

The Overall Effect Size

The first approach that researchers took when assessing rehabilitation programs was this: Let’s take all the rigorous program evaluation studies we can find and see what the impact of rehabilitation is across all these studies (and programs). We will not make any distinctions by type of program. We just want to know whether rehabilitation works in general. We will include in this assessment punishment-oriented programs that advocates of rehabilitation would not see as treatment in the traditional sense of the word. Again, a meta-analysis is sort of like a batting average in baseball. It is not computed the same way, but what we are really asking here is how well treatment programs bat across all of the interventions that have been evaluated.

There have now been numerous meta-analyses conducted. In fact, McGuire (2013) has compiled a listing of 100 of these meta-analyses! They focus on different samples of evaluation studies and on offenders of different ages. Meta-analyses also have been undertaken by authors in other nations. Regardless, they seem to point to the same conclusion:

· The overall effect size for rehabilitation across all interventions is approximately +.10.

In layperson’s language, this would mean that the recidivism rate for the control group would be 10 percentage points higher than for the treatment group. Thus, if the control group’s recidivism was 55%, the treatment group’s recidivism would be 45%.

This conclusion is quite significant. It immediately contradicts Martinson’s nothing works view. That is, across all interventions, rehabilitation programs are effective in reducing recidivism. As McGuire (2013, p. 20) notes, “there are firm grounds for arguing that we can be more confident than ever that there is a range of methods which ‘work.’” Their effects are modest but not inconsequential. McGuire provides a useful context for comparison, noting that “the mean effect for aspirin in reducing myocardial infarctions (heart attack) is 0.04; of chemotherapy for breast cancer it is 0.8–0.11; and of heart bypass surgery in reducing coronary thrombosis it is 0.15” (2013, p. 31). “Set against this background,” he observes, “an average effect size of 0.10 cannot be dismissed as merely trivial” (p. 31).

Still, is a 10% reduction in recidivism the best that we can do? If so, then this means that rehabilitation will be a useful tool in corrections but not one that can be looked to for important savings in crime. Well, as it turns out, this is not the best we can do! It is to this issue that we now turn.

Heterogeneity of Effect Sizes

Some correctional interventions work better than others. To return to language used above (we told you we would get back to this), the effects of treatment programs are heterogeneous, not homogeneous. Recall that Martinson contended that across various treatment types or categories, some things worked and some things did not work but that nothing worked consistently or better than any other modality. Again, this is a prediction of homogeneity of effect sizes. But this is just not what meta-analyses find.

In fact, meta-analyses suggest that some programs have no effect or are iatrogenic—they increase recidivism. By contrast, other types of programs work quite well. They have effect sizes of +.25 or higher. Recall that an effect size of +.25 would translate into the treatment group having a recidivism rate of 37.5% and the control group having a recidivism rate of 62.5%.

Of course, a key challenge is to discover what distinguishes programs that produce success—high effect sizes—and those that do not. In fact, this is what the Canadians—Paul Gendreau, Don Andrews, and James Bonta—have been doing for over two decades. As promised, we will review their theory of the principles of effective intervention up ahead. They contend, with some justification, that programs that follow their principles produce meaningful reductions in recidivism.

At this point, however, let us share one important finding. Across numerous studies, it now appears that one type of intervention is the most reliable in achieving high reductions in recidivism: cognitive-behavioral programs (see also MacKenzie, 2006; Wilson, Bouffard, & MacKenzie, 2005). If Cullen and Jonson were going to implement a program with high-risk/serious offenders, we would start with a cognitive-behavioral approach and then perhaps add in some other elements (e.g., work, training, education).

If you would like to learn more about cognitive-behavioral programs, some good books can be consulted (see, e.g., Spiegler & Guevremont, 1998; Van Voorhis, Braswell, & Lester, 2009). But we can give you a brief overview. Thus, cognitive-behavioral programs focus on doing two key things:

· They try to cognitively restructure the distorted or erroneous cognitions of an individual. These are sometimes called thinking errors.

· They try to assist the person to learn new adaptive cognitive skills.

In the case of offenders, existing cognitive distortions are thoughts and values that justify antisocial activities (e.g., aggression, stealing, substance abuse) and that denigrate conventional prosocial pursuits regarding education, work, and social relationships. Most offenders also have minimal cognitive skills as to how to behave in a prosocial fashion. In light of these deficits, effective cognitive-behavioral programs attempt to assist offenders (1) to define the problems that led them into conflict with authorities, (2) to select goals, (3) to generate new alternative prosocial solutions, and then (4) to implement these solutions.

Thus, a cognitive-behavioral program within corrections would involve the following steps:

· The predominant antisocial beliefs of the offender in question are identified.

· In a firm yet fair and respectful manner, it is pointed out to the offender that the beliefs in question are not acceptable.

· If the antisocial beliefs continue, emphatic disapproval (e.g., withdrawal of social reinforcers) always follows.

· Meanwhile, the offender is exposed to alternative prosocial ways of thinking and behaving by concrete modeling on the part of the therapist in one-on-one sessions or in structured group learning settings (e.g., courses in anger management).

· Gradually, with repeated practice, and always with the immediate application of reinforcers whenever the offender demonstrates prosocial beliefs and conduct, the offender’s behavior is shaped to an appropriate level.

The Impact of Meta-Analysis on the Nothing Works Debate

The findings from the meta-analyses of treatment programs have been critically important in rebutting the nothing works doctrine. Again, once Martinson reframed the attack on rehabilitation as an issue of effectiveness (as opposed to state discretion), then the central issue was the status of the empirical data. At first, of course, rehabilitation’s critics were in the position of citing Martinson’s review and claiming that science was on their side. Ted Palmer and the Canadian psychologists muddied the water quite a bit, but not enough to swing the tide in the other direction.

The meta-analyses conducted by Lipsey and others, however, gave rehabilitation supporters a decided upper hand. With a quantitative analysis of a few hundred studies in tow, they could now show that overall (or across all interventions) treatment programs reduced recidivism and that some programs were quite effective. Of course, as discussed previously, what scholars and others believe or are ready to believe is not simply based on the data that are presented to them. There are still many scholars—especially criminologists—who simply fight the idea that correctional programming could possibly do any good.

Again, the meta-analyses are quite difficult to dismiss. Unlike narrative reviews, there is a clear number showing the effect size of rehabilitation overall and of specific types of intervention strategies in particular. It is possible to challenge the findings on methodological grounds, but the replication of the findings in meta-analysis after meta-analysis is hard to dispute. In fact, the onus now is on the critics to produce their own meta-analyses showing that rehabilitation programs have a zero effect size. They are free to go look at the existing research studies and show where the rehabilitation scholars have gotten it wrong! Of course, they have not done so, in large part because their findings likely would not differ from those published to date.

Before the rehabilitation crowd gets too big for their britches, it is important to realize that serious obstacles remain to making treatment a guiding correctional theory. Knowledge about what works is growing, but how to disseminate this information and to ensure the implementation of quality programs remain daunting challenges. This is a time for optimism, not hubris.

What Does Not Work

Now it is time to become a bit more specific about what works and does not work to reduce reoffending. In this section, we will briefly focus on what does not work. A weakness in all these approaches is that they are based on faulty criminology. They do not target for change the most important factors—again, which Gendreau and Ross (1987) called criminogenic needs—that underlie recidivism. Four types of programs merit special attention because they seem plausible but, alas, will never work (Andrews & Bonta, 2010):

· Punishment-oriented programs. Correctional interventions that are based on deterrence theory have not proven effective. Scared straight programs, where juveniles visit prison and are warned in an intimidating fashion about what life behind bars would entail, have either no impact or a slight criminogenic effect. As we saw in  Chapter 4 , control programs, such as intensive supervision, have produced dismal results.

· Character-building programs. This is really a subcategory of punishment-oriented interventions because they emphasize taking a tough and demanding approach with offenders. If a difference exists, it is that the underlying goal is not so much to scare offenders straight (as in traditional deterrence) but to break the offender down and then to rebuild him (or her). Exactly what breaking an offender down actually means is never specified, but it seems to imply stripping away the offender’s defenses and seeming bravado to reach some inner core of vulnerability that will then make the offender open to a new way of life. In any case, such programs would include boot camps and wilderness programs. The research does not show that these approaches are effective.

· Boosting self-esteem. There is nothing wrong with boosting a youth’s self-esteem, especially if it is based on realistic accomplishments. After all, we all like to feel good about ourselves! Still, self-esteem apparently has little to do with crime; it is a weak predictor of recidivism. Programs that focus exclusively on self-esteem are unlikely to work. In fact, they may risk creating confident criminals!

· Client-centered, non-directive counseling. When you read meta-analyses, they will often include a category called individual and/or group counseling. The problem with analyzing studies this way is that what is done in a counseling session can vary widely depending on the treatment approach taken by the counselor. In this context, it appears that one general approach to counseling does not work with offenders: counseling that is non-directive, allows inmates to set the agenda, and is not structured systematically to change antisocial values and ways of thinking. That is, if the approach is too client-centered and non-directive, the counseling sessions may be too poorly focused to change the factors central to recidivism.

What Does Work: Principles of Effective Intervention

If you were to read a meta-analysis by Lipsey and others (but not the Canadians), you would see variation in which treatment types or modalities work the best. You might be told, for example, that for youths in the community, individual counseling, interpersonal skills training, and behavioral programs reduce reoffending consistently whereas deterrence and vocational programs do not (Lipsey, 1999a, p. 150; see also MacKenzie, 2006). This guidance certainly is valuable and should be heeded. Its weakness, however, is that such advice is not a complete theory of correctional intervention. These data on effectiveness hint at which programs might work but they do not explain why and to whom they should be applied.

In this context, Cullen and Jonson believe that a better and more systematic approach is to implement programs based on the principles of effective intervention (Andrews, 1995; Andrews & Bonta, 2010; Gendreau, 1996). We alluded to core elements of these principles in our discussion of Gendreau and Ross’s (1987) review of studies revealing effective treatment programs. Although some dispute exists (Porporino, 2010), there is now growing empirical evidence that the principles approach comprises the most powerful, empirically justified treatment paradigm available (Andrews & Bonta, 2010; Andrews et al., 1990; Gendreau, Smith, & French, 2006; McGuire, 2002; Ogloff & Davis, 2004; Smith, Gendreau, & Swartz, 2009).

Quite simply, for scholarship on correctional rehabilitation to move forward, it is essential that it go beyond the nothing works debate that is now more than four decades old and growing sterile. The dichotomous question—does or does not rehabilitation work?—is no longer productive. The terms of the debate, in short, have changed. We now know that certain kinds of programs have little prospect of ever working and we know that other kinds of programs achieve frequent success. When this occurs, the logical approach is to study systematically what it is about effective programs that is at the root of their success. Again, this is the approach taken by the Canadian psychologists, led most prominently by Don Andrews, James Bonta, and Paul Gendreau. Don Andrews passed away in 2010, but his ideas and contributions are enduring.

In developing their principles of effective intervention, the Canadian psychologists did not approach the topic from an armchair. Over the years, they had accumulated plenty of experience in the field with programs. They had established programs and evaluated many more. Still, they realized that personal experience is no substitute for data if one wishes to have a science of correctional treatment that is evidence-based. As a result, from the inception of their work, they rooted their understanding of criminal conduct and its change in the experimental research of psychology generally. They conducted meta-analyses on the main predictors of recidivism so as to know what to target for reform (Gendreau, Little, & Goggin, 1996). They examined what treatment interventions could change or be responsive to these risk factors. They developed instruments to classify offenders by risk level and to assess whether correctional agencies were capable of treatment integrity.

No one approach should be seen as sacrosanct—as so sacred that it is above criticism or above improvement. Cullen and Jonson believe that the Canadians’ principles paradigm is awfully persuasive. We see it as the most criminologically sound, empirically based, and demonstrably effective treatment paradigm available (Cullen & Smith, 2011; Smith, 2013). But again, skepticism is a healthy part of science. Organized skepticism is the duty of any community of scholars (Merton, 1973). We should perhaps use the Canadians’ approach but not be awed by it. Blind allegiance can be a dangerous thing (Cullen, 2012).

The RNR Model.

So, what are these principles of effective intervention? The Canadians have listed them in various forms and numbers—an issue we return to below (see Andrews, 1995; Gendreau, 1996). However, three principles are at the core of the model: risk (R), need (N), and responsivity (R). Given the centrality of these principles, the perspective is now known by its acronym, the RNR model. Accordingly, we review each of the three principles in some detail. Note that this discussion is a bit out of order (need, responsivity, and risk or NRR rather than RNR) because it is easier to explain that way! We then add two more points at the end of this section.

· The first principle is that interventions should target the known predictors of crime and recidivism for change. This is called the needs principle.

This principle starts with the assumption that correctional treatments must be based on criminological knowledge—what the Canadian scholars call the social psychology of criminal conduct. They distinguish between two predictors that place offenders at-risk for crime:

· Static predictors—such as an offender’s criminal history—that cannot be changed.

· Dynamic predictors—such as antisocial values—that can potentially be changed. In the Canadians’ perspective, these dynamic predictors or risk factors are typically referred to as criminogenic needs. (We should warn that no one else uses this terminology in this specific way. This does not make the Canadians’ usage faulty; it only means that other scholars have not read their work and/or adopted their terminology. This is really no big deal, except to the extent that you might get confused or use this terminology with the assumption that it is commonly understood.)

Importantly, when investigating risk factors or predictors of crime, it is possible that the research could have indicated that the major predictors are static. If so, then the prospects for rehabilitation would have been minimal (you cannot change something that is static and thus, by definition, unchangeable). But this did not turn out to be the case! Meta-analyses reveal that many of the most salient predictors are dynamic (they are criminogenic needs) and thus can be changed!

Again, this is a critical point. Let us assume that the only thing that predicted recidivism was the offender’s past criminal record. If this were the case, what would this tell treatment scholars? It would mean that we could identify nothing that we could change at this time in an offender’s life that could reduce recidivism. We cannot, after all, change a past record, since this is over and done with. It would be like telling a doctor that the only predictor of a patient’s sickness is whether the person has been sick in the past. Without more knowledge of what could be changed about the patient here and now, the doctor would have no intervention to use.

But, as we have just said, there are dynamic risk factors that are closely associated with recidivism. According to Andrews and Bonta (2010), research reveals that four are most important. They call these the “Big Four” (2010, pp. 58–59). They explain them in this way (2010, p. 500):

· “History of antisocial behavior. Early and continuing involvement in the number and variety of antisocial acts in a variety of settings.”

· “Antisocial personality pattern. Adventurous, pleasure-seeking, weak self-control, restlessly aggressive.”

· “Antisocial cognition: Attitudes, values, beliefs, and rationalizations supportive of crime and cognitive and emotional states of anger, resentment, and defiance. Criminal/reformed criminal/anti-criminal identity.”

· “Antisocial associates: Close association with criminal others and relative isolation from anti-criminal others, immediate social support for crime.”

These factors should be given priority in any treatment intervention—again, because they are robustly related to offending and are amenable to alteration. Thus, if an offender is taught to embrace prosocial values, has antisocial friends replaced by “good people,” and learns to think before acting, then this person is much less likely to return to crime. Obviously, it is not possible to change a history of antisocial behavior, but it is possible to “build up noncriminal alternative behavior in risky settings” (Andrews & Bonta, 2010, p. 500). Further, Andrews and Bonta (2010, p. 59) identify four other risks that they term the “Moderate Four.” These include: “family/marital circumstances” (focus on quality of interpersonal relationships among family members); “school/work” (focus on relationships and on performance and accompanying rewards); “leisure/recreation” (low involvement and satisfaction); and “substance abuse.” Together, the “Big Four” and “Moderate Four” are called the “Central Eight risk/need factors” (p. 58).

It is important to reemphasize a point made right above. Andrews and Bonta believe that treatment must be based on an accurate understanding of the causes of recidivism. This is why they devote so much attention to documenting the Big Four and Central Eight. They want to make sure that treatments are targeting the correct sources of crime (“criminogenic needs”). This insight would be commonsensical in medicine. When you go to the doctor, you expect the physicians to diagnose the cause of your illness. You know that if this is not the case, then the treatment you are given will not work. Effective medicine thus depends on an accurate scientific understanding of diseases. This reality also occurs in corrections. Alas, many interventions do not have a strong underlying criminological foundation. As a result, these programs focus on factors that their advocates believe cause crime, such as low self-esteem, which actually are unrelated or only weakly related to recidivism. Thus, targeting these factors for intervention will produce little, if any, change in an offender’s conduct.

· The second principle is that treatment services should be behavioral, social learning, and cognitive-behavioral in nature. This is called the responsivity principle.

In general, behavioral or, as we have seen, cognitive-behavioral interventions are effective in changing an array of human behaviors. With regard to crime, they are well suited to altering the Big Four criminogenic needs—antisocial attitudes, cognitions, personality orientations, and associations—that underlie recidivism. Thus, Andrews (1995) notes that these interventions would “employ the cognitive behavioural and social learning techniques of modelling, graduated practice, role playing, reinforcement, extinction, resource provision, concrete verbal suggestions (symbolic modelling, giving reasons, prompting) and cognitive restructuring” (p. 56). Reinforcements in the program should be largely positive, not negative. And the services should be intensive, lasting three to nine months and occupying 40% to 70% of the offenders’ time while they are in the program. Remember, the goal is to have offenders learn a set of behavioral and cognitive skills—such as how to think differently, how to control anger and impulsivity, how to avoid criminal associates, and how to respond in prosocial rather than antisocial ways when in risky situations (e.g., insulted in a bar, seeing an unguarded computer).

In contrast, many other interventions are ineffective not only because they do not address the factors that cause recidivism, but also because they are not delivered in a way that can change criminogenic needs; that is, they are not “responsive” to them. They are providing the “wrong medicine.” Thus, Andrews and Hoge (1995) contend that less effective treatment “styles” have these characteristics: They “are less structured, self-reflective, verbally interactive and insight-oriented approaches” (p. 36). Punishment approaches also do not target criminogenic needs or target them responsively. Not surprisingly, they are among the most ineffective interventions with offenders (see Lipsey, 2009; McGuire, 2013).

· Third, treatment interventions should be used primarily with higher-risk offenders, targeting their criminogenic needs (dynamic risk factors) for change. This is called the risk principle.

It is often said that interventions should mainly be given to low-risk offenders because they are less hardened and thus open to change. Implicit in this view is that high-risk offenders are beyond redemption. As it turns out, however, higher-risk offenders are capable of change. And, more noteworthy, the most substantial savings and recidivism are acquired by providing them with treatment services. (Recall our discussion of this issue in  Chapter 6 .)

In part, this is because higher-risk offenders have more to change about them. Thus, when resources are scarce, it appears that it is this group of offenders that should be targeted for change. In fact, it appears that less hardened or lower-risk offenders generally do not require intervention because they are unlikely to recidivate. Subjecting them to structured, intrusive interventions is not a wise use of scarce resources and, under certain circumstances, may increase recidivism.

Note that the most effective strategy for assessing the risk level of offenders is to rely not on the clinical judgments of counselors (who they “think” are the worst cases). Clinical judgments should not be totally disregarded, but they are open to personal bias. Remember, they are a form of insider knowledge!

Not only in corrections but in other realms of human behavior, it appears that the best predictions are made through the use of actuarial-based assessment instruments (validated instruments that use largely quantitative scores to inform decisions) (Ayres, 2007). In corrections, one of the best instruments for classifying offenders—one that the Canadians developed to implement their treatment theory—is the Level of Service Inventory. Research has shown this instrument has strong predictive validity; that is, those offenders it says are high risk (i.e., score high on the LSI) are indeed more likely to offend (Smith, Gendreau, & Swartz, 2009; Vose, Cullen, & Smith, 2008; see also Andrews & Bonta, 2010).

Two More Points.

We want to end this discussion with two further observations. First, the three principles just described form the core of the Canadians’ perspective and are now widely known the field of corrections. “Everyone” in the field—including Andrews and Bonta (2010)—now uses the acronym RNR to refer to these three core principles. And as noted, “everyone” also refers to this correctional theory as the Risk–Need–Responsivity model.

However, the Andrews, Bonta, and their fellow compatriots north of the U.S. border have long understood that a range of other considerations can either boost or weaken the treatment gains achieved from the application of the core RNR principles. The Canadians thus developed a comprehensive treatment paradigm whose full statement requires 15 principles. These are laid out in the Canadians’ correctional bible—Andrews and Bonta’s (2010, pp. 46–47) The Psychology of Criminal Conduct. Here we will list a few of the major considerations that, if addressed, have the potential to increase treatment effectiveness.

· When possible, conduct an intervention in the community as opposed to an institutional setting.

· Ensure that the program uses staff who are well-trained, are interpersonally sensitive, are monitored, and know how to deliver the treatment service.

· Follow offenders after they have completed the program and give them structured relapse prevention (or aftercare).

· Specific responsivity: To the extent possible, match the way a treatment service is delivered to the learning styles of the offender. Factors that might be taken into account in service delivery are the offenders’ lack of motivation to participate in the program, feelings of anxiety or depression, and neuropsychological deficits stemming from early childhood experiences (e.g., physical trauma). For example, offenders with low IQs might not respond as well to interventions that are verbal but would do better with interventions that emphasize more extensive use of tangible reinforcers and from repeated, graduated behavioral rehearsal and shaping of skills.

Second, the RNR model was invented to guide the delivery of treatment programs within correctional agencies, whether these are prisons, halfway houses, or community based facilities. More recently, however, James Bonta has been instrumental in developing a way to use RNR principles to guide the interactions of probation and parole officers with their supervisees. He came up with another clever acronym for this initiative—STICS, which stands for the “Strategic Training Initiative in Community Supervision” (Andrews & Bonta, 2010). The development of STICS made sense because his meta-analyses of existing studies found that traditional community supervision had very little impact on recidivism (Bonta, Rugge, Scott, Bourgon, & Yessine, 2008; see also Schaefer, Cullen, & Eck, 2016). This represents a colossal missed opportunity! On any given day, recall that 1 in 51 adults in the United States is under community supervision; the raw number exceeds 4.7 million (Herberman & Bonczar, 2014).

The challenge was whether RNR principles could be applied productively in a supervision meeting that may last less than a half hour. Bonta thought that it could be so used. Perhaps the key to the initiative was training officers how to recognize “expressions of antisocial attitudes in the clients, and how to use cognitive-behavioral techniques to replace these cognitions and attitudes with prosocial ones” (Andrews & Bonta, 2010, p. 415). During the office meeting, supervising officers would use about 15 minutes in the middle of the session to target risk factors for change, which might involve “teaching the cognitive-behavioral model or doing a role-playing exercise” (p. 416). Offenders might also be assigned homework—such as trying a new behavior (e.g., recognizing and replacing risky thinking)—that would then be discussed and reinforced in the next meeting.

Notably, a two-year evaluation found that compared with the control group, those receiving the RNR intervention had a recidivism rate that was 15 percentage points lower (Bonta et al., 2011). Promising evaluation results also have been reported from two similar programs conducted independently (Robinson et al., 2012; Smith, Schweitzer, Labrecque, & Latessa, 2012). More broadly, this research suggests the potential benefits of training all correctional personnel how to teach offenders prosocial cognitive-behavioral skills. Such a continuum of treatment—whether during probation or incarceration—would provide a consistent, integrated approach to offender rehabilitation.

What Else Might Work?

Desistance-Based Rehabilitation

Whenever you are the big dog, you are an inviting target. In this sense, Andrews, Bonta, Gendreau, and colleagues’ RNR model is a victim of its own success. As the dominant paradigm in the field (Cullen, 2013), it is increasingly criticized either explicitly or implicitly. In general, Cullen and Jonson are not so enamored with these alternative perspectives (Cullen, 2012). Part of the problem is that they rarely start with a sound theory of criminal conduct that is based on the empirically established predictors of recidivism. They often sound plausible, but it is not always clear that their proposed treatment intervention would target the most important risk factors or be capable of changing them (i.e., be “responsive” to them). Just keep these ideas in mind as we proceed.

Perhaps the most important alternative to the RNR model is what might be called desistance-based rehabilitation (for examples, see Brayford et al., 2010; Veysey et al., 2009; see also Raynor & Robinson, 2009). As life-course theory and research has proliferated in criminology (Benson, 2013; Cullen, 2013)—an issue we return to when we discuss early intervention programs in  Chapter 9 —it became clear that, eventually, almost all offenders stop breaking the law (Laub & Sampson, 2003). We call this desistance from crime.

One possibility is that they just get old, and like old people (such as Cullen) just give up and retire. They may still be a jerk—argue a lot, get drunk a lot—but they stop burglarizing and robbing because it takes too much effort (see Gottfredson & Hirschi, 1990). Most criminologists, however, do not like this retirement theory, because they want to think that most things are caused. Explaining things is what makes them famous! One popular desistance theory is that offenders get lucky and meet, and then marry, a wonderful woman. Sampson and Laub (1993) describe this as acquiring a quality social bond. (Cullen, who is a father, is not so sure how many dads want their cherished daughter to be a social bond for some predator!) Another theory, this one by Maruna (2001), is that some offenders develop a “redemption script,” sort of a new identity and story about their lives and its possibilities. They relinquish a “condemnation script” that tells them that they are “doomed to deviance” and instead redefine themselves as a fundamentally good person who can serve a higher purpose in life (2001, p. 74).

Okay, here is the link to corrections. Desistance is a form of rehabilitation, for these life-course-persistent offenders stop committing crimes. The key thing, though, is that reform is not due to a planned treatment program run by a professional but rather takes place in the real world. Offenders’ desistance is called “naturalistic” because, again, it happens sort of “naturally” in the course of their lives. Notably, some scholars believe that important lessons can be learned from naturalistic desistance and imported into correctional treatment (Raynor & Robinson, 2009). One insight is that offenders often desist not because some deficit (or “criminogenic need”) is fixed—as the RNR model contends—but rather because they draw on some strength that enables them to leave crime. This might be a new social bond (not Cullen’s daughter) or a new prosocial identity or script. Another insight is that offenders who desist become motivated to change and exercise human agency through which they “will” their reformation. Desistance-based rehabilitation programs tend to emphasize these positives, arguing that correctional interventions should be strength based and prioritize efforts to inspire in offenders the motivation to change.

Cullen and Jonson suspect that desistance-based rehabilitation is popular because it is ideologically pleasing to a lot of criminologists. Rather than portray offenders as having deficits that make them predatory or at least chronically bothersome, a desistance perspective gets to see offenders as aspiring to “make good” and to live a “good life.” This is a chance for everyone to feel good! The task of rehabilitation is now clear: help offenders to build upon their valued talents and traits—their strengths—which will also inspire them to become motivated to act like “the rest of us.” Maybe this is true or true to a degree. But the other reality is that the research underlying desistance theories is limited in scope and far from definitive (see, e.g., Sullivan, 2013). For example, Skardhamar, Savolainen, Aase, and Lyngstad’s (2015) recent review concludes that those who are married are less involved in crime. Good news for correctional desistance theory? Well, not really. A closer inspection of this literature leaves more questions about the effects of marriage unanswered than answered:

Critical scrutiny of the evidence regarding the causal nature of the reported associations suggests, however, that claims about the restraining influence of marriage are overstated. None of the studies demonstrates evidence of direct (counterfactual) causality; no study has served a causal estimate unbiased by selection processes. Moreover, only a few studies address time ordering, and some of those show that desistance precedes rather than follows marriage. Evidence in support of the theoretical mechanisms responsible for the marriage effect is also mixed and insufficient. The criminological literature has been insensitive to the reality that entering a marital union is increasingly unlikely to signify the point at which a committed, high-quality relationship is formed. (Skardhamar et al., 2015, p. 385)

None of this is to say that desistance-based rehabilitation is fatally flawed. Rather, the point is just that those erecting rehabilitation models on the shaky foundation of desistance theories should exercise a lot of caution. As we have argued repeatedly, targeting for change factors unrelated or only weakly related to recidivism will consign a treatment modality to the dustbin of ineffectiveness. A major advantage of the RNR model is that Andrews and Bonta do not make this mistake. Just go and read their 672-page scientific tour de force, The Psychology of Criminal Conduct!

The Good Lives Model

Importantly, one rehabilitation theory has emerged from this general perspective that is gaining adherents cross-nationally: the Good Lives Model, also known by its acronym of the GLM. Cullen and Jonson must start with a confession: We are RNR model advocates and thus not terribly enamored with the Good Lives Model (GLM). That said, we are in the business of science, not religion; there is no correctional heresy! The weaknesses of the RNR model should be laid bare so, if possible, it can be improved. If we can develop a rival treatment technique with equal efficacy, that would be great! It would be like having two effective medicines that can treat the same disease (Cullen, 2012).

The GLM is a formidable opponent. Cullen and Jonson do not like the fact that the GLM has been put forth as a reaction to and critique of the RNR model. As a result, the advocates of these two perspectives have been in a bit of a tussle. It is sort of fun to watch academics fight it out, and we can learn from vigorous debate (see, e.g., Andrews, Bonta, & Wormith, 2011; Ward, Yates, & Willis, 2012). Still, at this point, Cullen and Jonson prefer to see each perspective as its own theory. For the GLM, criticizing the RNR perspective does not make the GLM more true. Only marshalling lots of data that demonstrates the perspective’s effectiveness will do that.

The GLM was first developed by Tony Ward in a founding article published in 2002, and then was elaborated in an influential book coauthored with Shadd Maruna in 2007 and in numerous articles (see, e.g., Ward, Mann, & Gannon, 2007; Ward & Marshall, 2007; Whitehead, Ward, & Collie, 2007; Willis & Ward, 2013). Part of Cullen and Jonson’s trepidation about the GLM is that it is based on a plausible but ultimately unverified theory of offending. For those familiar with strain theory, the GLM shares the assumption that people pursue goals and can achieve these through either legitimate or illegitimate means. There is an assumption that if legitimate means to cherished goals are blocked, individuals will likely turn to illegitimate means (see Cloward, 1959). Now, let’s put things in their language.

The overarching goal of all humans is, or should be, to live a “good life”—ergo, the “good lives” model! You do this by achieving positive goals through positive means. In GLM language, goals are called “primary goods”—sort of an awkward phrase but then that’s what the theory chooses to say! There are 11 categories of primary goods, which are listed in Willis and Ward (2013):

(i) life (including healthy living and functioning); (ii) knowledge; (iii) excellence in play; (iv) excellence in work (including mastery experiences); (v) excellence in agency (i.e., autonomy and self-direction); (vi) inner peace (i.e., freedom from emotional turmoil and stress); (vii) friendship (including intimate, romantic, and family relationships); (viii) community; (ix) spirituality (in the broad sense of finding meaning and purpose in life); (x) happiness; and (xi) creativity. (p. 307)

People, including offenders, weigh or prioritize these goals or primary goods differently. In fact, what primary goods are desired define their “sense of who they are and what is really worth having in life” (Willis & Ward, 2013, p. 307). The challenge, however, is that not everyone can gain access to their most cherished primary goods through prosocial means. Such goal blockage is the major source of crime.

In the GLM model, means are given another odd name: “secondary goods.” Secondary goods are the “concrete means of securing primary goods”; they are best seen as “the specific roles, practices and actions that provide the routes to primary goods” (Willis & Ward, 2013, p. 307). Criminogenic needs—the N in the RNR model—are seen as inappropriate secondary goods—the wrong way to achieve primary goods. Thus, antisocial peers might be a means to secure the goal of friendship and intimacy. Or antisocial attitudes might justify using violence as a means of displaying autonomy. Whereas Andrews and Bonta (2010) would see criminogenic needs as targets for treatment in and of themselves, Ward and his colleagues would see them in relation to the primary goods or goals that offenders hope to achieve.

Criminal behavior is caused directly or indirectly. “The direct route,” note Willis and Ward (2013, p. 307), “is evident when primary good(s) are explicitly sought through offence-related actions.” For example, an offender blocked from intimacy with an adult may seek this good through sexual victimization of a child; or a youth seeking happiness might steal a car for joyriding. Thus, in such cases, the offense is the illegitimate means used to reach the desired goals. By contrast, the “indirect route occurs when the pursuit of a good or set of goods creates a ripple effect in the person’s personal circumstances and these unanticipated effects increase the chances” of offending (Ward et al., 2007, p. 92). In an example given by Willis and Ward (2013, p. 308), let us say that a husband so valued excellence at his job that he worked excessive hours, which in turn led to the deterioration in his relationship to his wife. This relationship, however, was the key means (secondary good) used by him to achieve another valued goal—intimacy. To cope with the emotional distress of a looming breakup of his marriage, the man might then use pornography—a decision that reflects an inability to adequately manage moods. Using “sexual arousal as a distraction” might then result “in the entrenchment of deviant sexual feelings and, ultimately, to his sexually assaulting a woman” (2013, p. 308).

All this makes Cullen and Jonson concerned—for two reasons. First and foremost, where is the empirical evidence that offenders commit crimes because they cannot reach primary goods? Andrews and Bonta link recidivism to the “Big Four” and “Central Eight” because of a wealth of meta-analytic data showing that these are the strongest dynamic predictors of offending. The criminological theory of the GLM is plausible but not demonstrated. Even by the advocates’ own admission, much of the model is based on “assumptions.” Thus, Ward and Maruna (2007, pp. 152–153) note that “the point of this evaluation is not to determine if various assumptions (e.g., that offenders are like everyone else and strive to live a good life) are ‘right’ or ‘wrong,’ but rather whether they are consistent, reasonable, and most importantly, therapeutically useful. We argue that they are all these things.” The risk in relying on assumptions rather than on meta-analytic data, however, is that the underlying assumptions about causes of recidivism might be wrong, or only partially correct. If so, then the capacity of the GLM to target the correct factors for treatment would be compromised.

Second, even if the GLM has merit, it would seem that therapists would have to be quite skilled to implement it with offenders. In particular, they would have to be not only emotionally mature but also able to figure out for each individual what kind of good “lives” the offender wished to live (i.e., what was the person’s set of most valued primary goods), what secondary goods were blocking the attainment of these goods, and then how to build capabilities so that the client could achieve goods through prosocial means. Whew! That’s a lot to accomplish! We suspect that Tony Ward and his collaborators have the talent needed to be effective therapists. But can these complex skills really be taught to the average correctional worker?

There is one feature about the GLM that Cullen and Jonson find attractive. Ward and his buddies are concerned about the well-being of offenders. They not only want to reduce offenders’ recidivism but also want them to live a good life. The two are inextricably mixed, because committing crimes will not produce a good life. But the goal of GLM treatment is not just to stop criminal behavior but also to leave the offender a better person capable of more personal fulfillment and better citizenship. Advocates of the RNR model do not buy this approach. They are focused like a laser on fixing criminogenic needs. They believe that attempting to improve offenders’ general well-being is likely to result in an intervention targeting inappropriate, non-criminogenic needs that will lessen the effectiveness of the treatment.

From the GLM perspective, focusing on living a good life is crucial to the therapeutic enterprise because it is what inspires offenders to want to relinquish their criminal life course. Change is difficult, and it requires a lot of motivation. But how can we expect offenders to find a rehabilitation program worthwhile if it is imposed on them and they are largely told that they need to fix up a bunch of bad things about themselves? By contrast, the GLM involves offenders from the beginning of the intervention, trying to learn what their primary goods are—the goals that, if achieved, will make them fulfilled. Then, the trick is to build in them the capabilities to achieve their valued goals in a prosocial way. This involves developing with each offender an individualized Good Lives Plan. Willis and Ward (2013) describe the plan in this way:

A key task of assessment involves mapping out an individual’s good lives conceptualization by identifying the weightings given by the various primary goods. This is achieved through (i) asking increasingly detailed questions about an offender’s core commitments in life and his or her valued activities and experiences, and (ii) identifying the goals and underlying values that were evident (either directly or indirectly) in an offender’s offence-related actions. Once an individual’s conceptualization of what constitutes a good life is understood, future-oriented secondary goods aimed at satisfying primary goods in socially acceptable ways are formulated collaboratively with the client and translated into a Good Lives (GL) treatment/intervention plan. (p. 308)

So, what then, should we make of the GLM? Remember, this book, Correctional Theory, is evidence based, so Cullen and Jonson would like to answer this question by consulting the existing studies assessing the GLM. Alas, research showing the effectiveness of this novel treatment theory is frustratingly limited. Some beginning data have been accumulated that are suggestive of the model’s capacity to rehabilitate offenders (see Willis & Ward, 2013). But this research can be summarized in a few pages and pales in comparison to the case that Andrews and Bonta (2010) make for the vitality of the RNR model. For now, Cullen and Jonson would place our bets on the Canadians’ treatment theory, although we welcome the further development and empirical testing of the GLM.

Conclusion: Reaffirming Rehabilitation

There is no panacea or cure-all for offender recidivism. Any intervention with offenders is likely to experience a fair amount of failure. Treating offenders is similar to treating cancer patients. There is no magic bullet—like a polio vaccine—that will wipe out the disease. Progress is slow. It takes painstaking research and experimentation to keep chipping away at the failure rate. Still, progress is important. Lives are at stake.

The critical question is what our best bet will be when placing offenders in correctional programs. At this stage, five conclusions can be drawn from the existing research:

· Punishment-oriented programs (deterrence, control, character building) have been extensively evaluated. There is no evidence that they work and some evidence that they increase recidivism. Of course, individual offenders might benefit from a program. But across all offenders, these programs are ineffective. As a policy, they can no longer be sustained.

· Across all programs, rehabilitation interventions (and these studies include the punishment-oriented programs) reduce recidivism about 10 percentage points.

· Programs that conform to the principles of effective intervention, including using cognitive-behavioral treatments, have the potential to reduce recidivism rates 20 to 25 percentage points (maybe more).

· The RNR model has the strongest claim to being evidence based. It should not be treated as the final word on rehabilitation; other interventions, such as the GLM, should be encouraged. However, for these perspectives to rival the RNR, far more empirical evidence demonstrating their efficacy will have to be accumulated. In the meantime, the RNR model is the safer choice to make in treating offenders.

· The special challenge for those advocating for rehabilitation is to transfer the what works knowledge to those in corrections (a process already under way) and to ensure that programs based on the principles of effective intervention are implemented appropriately (a daunting challenge).

The broader point is that when it comes to corrections, the smart policy is to correct offenders. Many offenders are within the grasp of the correctional system for years on end. If they leave our supervision with the same antisocial values, the same thinking errors, and the same social and psychological deficits, then shame on us. It is, one could argue, basic human decency to improve those under our power. If for no other reason, we have an obligation to reduce their propensity for crime so as to protect the public when these offenders are released into the community—as the vast majority will be within about two to three years. Again, hundreds of thousands of inmates—more than 600,000 annually—stroll through the prison gates and into someone’s community. Releasing this mass of unreformed humanity into our midst without subjecting them to our best efforts at rehabilitation is irresponsible. The high cost of ignoring what works in corrections is often not seen but it is nonetheless quite real (Van Voorhis, 1987).

Some commentators worry that rehabilitation will rob the law of its sting. But if there is any deterrent effect at all, it most likely lies in the certainty of arrest and not in harsh sentences or a dreary, unproductive stay in prison. Rehabilitation does not undermine arrest whatsoever. Indeed, the certainty of rehabilitation depends on it! Unless offenders are caught, they cannot be treated. Similarly, even under a treatment regime, high-risk offenders would likely be sent to prison. Accordingly, much of the incapacitation effect achieved by locking up high-risk offenders would be preserved under a system guided by rehabilitation. The difference perhaps would be that low-risk offenders would not be needlessly imprisoned.

These observations are important because they suggest that were rehabilitation reaffirmed and allowed to guide the nation’s correctional system, there would likely be little lost in terms of deterrence or incapacitation. But what would be gained? Much, Cullen and Jonson think. Most important, there would be a renewed effort to ensure that our correctional system is not designed to fail (Reiman, 1984). There would be a clear mandate not to allow offenders to sit idle in their cells or to emerge from years behind bars worse off than when they first entered prison. There would be a strong imperative to use science to evaluate all of our practices so as to root out the harmful and keep the beneficial. And above all, there would be a renewed social purpose in corrections—originally articulated by the founders of the American penitentiary—to forfeit the easy policy of warehousing the wicked in favor of the more difficult but noble policy of saving the wayward.