Does correctional rehabilitation work?

I admit. For a person who’s passionate about helping people coming out of prison, I’m pretty critical when it comes to reentry programs.

Most of what’s being done isn’t based on research. The vast majority of our efforts aren’t evaluated. And the field is fraught with implementation issues that compromise even well-funded and well-designed programs.

Most of what’s being done isn’t based on research. The vast majority of our efforts aren’t evaluated. And the field is fraught with implementation issues that compromise even well-funded and well-designed programs. It’s kind of a mess. 

Still, I’m a huge advocate of programs to help people both in and out of prison, because research shows us that, on average, rehabilitation efforts net positive results. Not only that, but they’re the only thing that has been shown to change criminal behavior. Threats don’t work. Punishment doesn’t work. Surveillance doesn’t work.1  

Therapeutic interventions are our best bet. Even with all the problems. 

How do we know for sure?  It’s an interesting story that begins with a guy who said that rehabilitation doesn’t work. 

Martinson concluded that “nothing works” in correctional treatment and that “rehabilitation is a myth.”

Robert Martinson gets credit for trashing the reputation of prison rehabilitation almost 50 years ago. He published an essay in 1974 which was based on his research of 231 studies evaluating rehabilitation’s impact on recidivism. Martinson concluded that, “the rehabilitation efforts that have been reported so far have had no appreciable effect on recidivism.” Although he didn’t come right out and say it, his message was translated into “nothing works” in correctional treatment.2 

Martinson wasn’t some guy from off the street. He was a scholar. He knew what he was doing when it came to research. His analysis was considered by many to be a state-of-the-art review of the existing literature at the time. A lot of people accepted his “nothing works” viewpoint, hook, line, and sinker. His research was one of the reasons that the conversation in offices and classrooms and courtrooms turned to punishing harder and longer.3 

But there were a handful of smart folks who didn’t buy it.

One of them was criminologist Ted Palmer. He looked at the data a different way. He used a “ballot box” approach and just straight up counted the number of studies in Martinson’s research that showed a positive impact on recidivism. Forty-eight percent of the total reduced recidivism, which proved (at least to some) that the “nothing works” position couldn’t be true.4

Ted Palmer’s research proved that the “nothing works” position couldn’t be true.

See, when Martinson said, “nothing works,” he didn’t mean that no studies reduced recidivism. What he meant was that no type or category of intervention, such as skills training or therapy, could consistently reduce recidivism across studies, settings and different types of people. He interpreted the data to mean that any positive results were random occurrences and couldn’t be intentionally replicated. 

But again, Palmer didn’t buy it. He looked for patterns of successful interventions within Martinson’s studies, and he found some. He concluded that programs were more successful in the community than in prison, for young people rather than adults and for people at a “middle risk” to re-offend.

If we want to reduce recidivism, we have to ask, “Which methods work best for what type of people under what conditions and in what type of settings?”

Way back in 1975 Palmer was helping us understand that, if we want to reduce recidivism, we can’t just ask, “what works?” We have to be more specific. We have to ask, “Which methods work best for what type of people under what conditions and in what type of settings?”

A few years later Martinson did more research and ended up agreeing with Ted Palmer. In a follow-up article, he said:

The “critical fact seems to be the condition under which the program is delivered,” and that “treatments will be found to be ‘impotent’ under certain conditions, beneficial under other ones, and detrimental under still others.” 5

But it was too little too late. The damage was done. Rehabilitation programs across the country were de-funded and cancelled.  Martinson committed suicide in 1979. He didn’t live long enough to fully recant his “nothing works” position. 

Then four Canadian psychologists came onto the scene.

Paul Gendreau and Robert Ross, along with their colleagues Don Andrews and James Bonta, came onto the scene. These guys all had experience implementing, administering and evaluating correctional programs. I affectionately refer to them as the bad boys of correctional treatment. Just kidding. I don’t call them that. But they are my heroes for sure. I stood on their shoulders to design and implement the reentry nonprofit I run, the Turning Leaf Project.

People committing crimes, like everyone else, acquire their attitudes, beliefs, and behavior through exposure, reinforcement, and punishment.

The “nothing works” philosophy just didn’t make sense to them. These psychologists were trained in learning theory and believed that criminal behavior was learned. Their position was that people committing crimes, like everyone else, acquired their attitudes, beliefs, and behavior through exposure, reinforcement, and punishment. Of course, it only made sense then, that this group of people could also re-learn and acquire new behaviors just like all of us.

Gendreau and Ross conducted two extensive reviews of the research looking at the effectiveness of correctional treatment and recidivism between 1973 and 1987.

Their work had three conclusions:

1. A major reason why programs fail is they lack therapeutic integrity.
What is therapeutic integrity? It means that we’re doing something that has the potential to get the results we want based on research. It means that what we’re doing is being done by competent people. It means that we’re doing enough of that thing, at a high enough quality, and in time frames and settings that support behavior change. 

We would not be surprised, for example, if our kids turned out to be illiterate if their teachers were untrained, had no standardized curriculum and only met their students once a week for an hour. 

That’s what’s happening in most of our rehabilitation programs. 

2. Successful programs target predictors of recidivism that are amenable to change.
The sheer number of programs that Gendreau and Ross uncovered that reduced recidivism, despite the obstacles, was further proof that something worked. They found that successful programs focused on criminogenic needs – which is a fancy way of saying the things we know predict recidivism that can change (attitude and thinking, for example).  
3. Individual differences are important in understanding criminal involvement.
Gendreau and Ross found that people differed in their level of risk for future re-offending. They also differed in their personalities and ability to learn. The evidence showed that the effectiveness of programs varied significantly to the extent that individual differences were measured and taken into account in the delivery of services. 

A recap of our story so far…

First, Martinson concluded, “nothing worked.” But Palmer found that almost half of Martinson’s studies did reduce recidivism and that successful programs shared characteristics related to the conditions under which the program was delivered. Gendreau and Ross’s research supported Palmer’s findings, confirming that a lot of programs did work, but that poor design, quality, and implementation were the major culprits of poor outcomes. Also, they found that successful programs considered individual differences in the delivery of services and focused on the predictors of recidivism that can be changed (fancy term: criminogenic needs). 

 There are just two more parts to the story. Stick with me and you’ll know more about the research of correctional treatment than 99% of the population. We’re almost there.  

The meta-analysis 

The meta-analysis was a game changer in understanding the impact of rehabilitation programs.

This method of research evaluates an entire body of literature and computes for each study the “effect size” between the treatment and the outcome variable (in this case, recidivism). The effect size could be negative, neutral or positive. Then the effect sizes are combined to get an “average effect size” across all studies. By summing the effect size across a large sample of studies regardless of how much or little recidivism was reduced, the meta-analysis allows us to capture effects that we can’t when studies are looked at one by one. 

As an average, rehabilitation programs were shown to reduce recidivism by 10%. But individually, some programs increased recidivism, some did nothing and some kicked butt.

In 1995 a guy named Losel published the most comprehensive assessment of the meta-analyses of rehabilitation programs. The average program effect size was .10. This means, for example, that the recidivism rate for the treatment group would be 45% while the control group would be 55%. Correctional rehabilitation programs, on average, were shown to reduce recidivism. 

However, within any given sample of studies, the effect size for certain programs ranged from zero (or a negative, indicating an increase in recidivism) to .40 effect or higher.

Some of the programs increased recidivism, some of them did nothing, and some of them kicked butt.

What accounts for the difference in effect sizes? 

This is another way the meta-analysis changed the game. Research by Losel and a guy named Lipsey and others organized information in a way that gave more clarity on what works, what doesn’t and what factors are important if we want to reduce recidivism. They were able to single out the “best” and “worst” programs and find out what they had in common. 

 Which takes us to the end of our story. And this is an important part so don’t fade out now. 

Back on the scene: The bad boy Canadian psychologists

Until now we’d been studying programs and interventions to determine what worked and then looking for correlates between the best and worst programs. But there was no understanding of how and why elements of successful programs were interrelated. Our Canadian psychologists changed that. They developed a theory explaining why people commit crimes and what we can do about it. 

James Bonta and Don Andrews developed a theory explaining why people commit crimes and what we can do about it.

James Bonta and Don Andrews combined everything they knew from the behavioral change literature on psychology, with what they knew from personal clinical experience, with what they found from all the research, and developed a general personality and social psychology of criminal conduct (that is, a “psychology of criminal conduct”). This theory explains why people commit crimes, which helps us then predict and influence criminal behavior. 

The RNR Model

The RNR model tells us who we should work with, what we should do with them and how we should do it.

From the psychology of criminal conduct, our bad boys then put forward a model of rehabilitation called RNR. This model includes 15 empirically tested principles to guide treatment decisions in our prison and community programs to reduce recidivism. The RNR model tells us who we should work with, what we should do with them and how we should do it. These directions are called “principles of effective correctional intervention.” 

The RNR is the only model of rehabilitation that is based in theory and supported by empirical data. 

When we follow RNR we can reduce recidivism. When we don’t, it’s unlikely that we’re doing much good and possibly doing harm.

In 1990, Andrews conducted a meta-analysis of 80 program evaluations to determine if the principles of effective interventions resulted in greater reductions in recidivism. Across all programs, the total effect size was .10 (meaning on average, programs reduced recidivism by 10 percent). But like with Losel’s study, there was a lot of variation between programs. Interventions that followed RNR reduced recidivism by 30%, programs that were unspecified reduced recidivism by 13% and programs that did not follow RNR increased recidivism by 7%. 


How do we know therapeutic interventions are our best bet? 

Because of Palmer and Losel and our Canadian rock stars and  others who have dedicated their lives to proving that rehabilitation does work. Gendreau, Ross, Bonta, and Andrews created the roadmap for those of us administering prison programs, running nonprofits working with people out of prison, crafting criminal justice policy and funding anti-recidivism efforts.

As I said, these guys are my heroes.  

But we can’t forget about Martinson. Because he made us look closer at the state of rehabilitation. He pushed us to explore what Gendreau called the “black box” of treatment programs. And while we know therapeutic interventions are our best bet, it ain’t pretty. The majority of programs are not designed or implemented in a way that will allow them to be effective.

We can get the results we want, but they’re not going to come automatically. 

Our responsibility is to design, implement and run correctional treatment programs based on what the research shows us works to reduce recidivism. That’s the only way.

I don’t know about you, but I feel lucky to have a roadmap. 

Share this post:

Leave a Reply

Your email address will not be published. Required fields are marked *