I just posted a comment on a blog post on the Freakonomics website discussing a recent trend of declining burglary rates perhaps due to either: a) more people remaining home as a result of being unemployed, or b) less to steal due to increased foreclosures. The Freakonomics folks called for criminologists to weigh in. So here's what I had to say:
Here’s some insight from a criminologist. This observation (i.e., decreasing burglaries associated with increasing unemployment) is exactly what Routine Activities Theory hypothesizes. Routine Activities Theory is a criminological theory which posits that crime happens when: 1) a motivated offender, 2) a suitable target, and 3) a lack of a capable guardian converge in time and space. For example, according to Felson and Cohen’s (1979) first exposition of Routine Activities Theory, the increase of women in the workforce since the 1960s can explain a subsequent increase in burglary rates since more women in the workforce translates into more empty houses without a “guardian” at home to deter potential offenders. So it seems to me that these statistics are quite predictable based on Routine Activities theory.
As to the question of whether burglars turn to other kinds of crime when they can no longer burgle, this is a bit more tricky to examine. What we know from a longstanding body of research (see Gottfredson and Hirschi 1990 for a good summary) is that offenders are more generalists than they are specialists. Thus, it is entirely conceivable that there could be a whole shift in offending patterns when burglary becomes no longer feasible. More recently, however, some criminologists have found more specialization in offending than was previously believed to be. This research is ground-breaking; it will be interesting to track its development. Combine this with the conclusion coming from some research within the Rational Choice camp of criminology which finds evidence of certain types of offenders tending to specialize (e.g., older offenders, property offenders), and it could also be possible that a large fraction of burglars will simply “retire” when burglary becomes no longer feasible. This is a dissertation waiting to happen. If I were doing the dissertation, I would also draw upon the “hot spots policing” literature which has examined the extent to which crime just “moves around the corner” when targeted police enforcement strategies are used. The tentative conclusion so far from the “hot spots policing” literature is that crime does not move around the corner, but instead there is typically a diffusion of crime control benefits to neighboring places.
Wednesday, November 11, 2009
Sunday, November 8, 2009
ASC 2009
Another great year at ASC! I wasn't thrilled about ASC being in Philly but I had a wonderful time. This was my sixth ASC. Here's a couple of highlights:
Best session hands down was an "author meets critic" session for Mark Kleiman's new book "When Brute Force Fails". Mark presented the book and then Al Blumstein and Jeremy Travis were the discussants. I'm all about this book! This book has been getting a lot of press coverage, from the NY Times to the Economist to Governing Magazine. The basic thesis of the book is nothing earth-shattering. We've known for a while that the certainty and swiftness of punishment are more effective than the severity of punishment. I think the timing of Mark's book is ripe though, given the correctional crisis of mass incarceration and the economic situation. It's the application of the basic principles of deterrence theory that are new and innovative to Mark's book. I see this as providing an exciting and promising new experiment for parole and community corrections to try out. I've long thought that deterrence can be made to work, especially for certain offenders. There's just so much about this book that excites me that I could go on forever. It will be interesting to see how much attention this book receives by policymakers over the next year, especially given the attention it's already received by criminologists and the media.
I went to a good session on correctional forecasting methods. Some of the big names in this field were presenting at this session. I got some good practical tips for doing projections. It was interesting too to hear a more details about the California projections model.
I went to a panel discussion on "evidence-generating" research. Akiva Leiberman presented his whole idea of "evidence-generating" practice in contrast to "evidence-based" practice. Akiva first outlined this idea in an issue of The Criminologist. I've blogged about this idea before and it's a concept I've been pushing.
Nicole Rafter's Sutherland Address was excellent. She's not a vibrant presenter but the content of her address was what I was interested in. She's one of those rare historians in our discipline. My readers know that I'm all about the history of our discipline. So I was awaiting in anticipation to hear what Rafter had to say. As might be expected, she called for more attention to our history as a way of strengthening our discipline. I couldn't agree more. I look forward to reading her address in Criminology next year when it's published.
I was at a session in which Jim Austin gave his typical critique of why treatment programming won't work for reducing the nation's correctional population. I have to say that I agree with his analysis. I thought his presentation was good.
I went to a good session on theory, with Michael Gottfredson, Robert Sampson, and John Hagan presenting. The room was packed (must have been close to 200 people crammed in the room)!! I was excited to see that even in a year where the theme of the conference was policy, that theory continues to excite folks in our field. It excites me. It's also quite relevant to policy in my opinion. Anyways, Gottfredson gave his typical defense of control theory and answered some recent critiques of the theory. It was a wonderful presentation. Rob Sampson gave a talk in which he challenged the notion that experimental design studies are the "gold standard". I somewhat disagreed with his conclusions but he was interesting to hear nonetheless. And John Hagan as the discussant posed some interesting and challenging questions to each of the presenters on the panel, but unfortunately the panel was over before each presenter could respond. I was really hoping for time for questions from the audience too, especially to hear some challenging debate between Gottfredson and those who always get fired up about his control theory.
My panel on parole violators turned out quite well I thought. This was a panel that I had been wanting to put together for the past several years. We had a decent turnout and the other presentations on the panel were interesting. I think I organized a panel of what is arguably the most important research that's being done in the area of parole violators. Maybe there's a follow-up panel to come at a future ASC??
I went to an excellent panel on risk assessment methods. Avi Bhati presented his concept of risk suppression, which is a problem with risk assessment validation that I got nailed on in a risk assessment validation paper that I did for my first stats class when I started my Ph.D. Shawn Bushway is another to point out the whole idea of "risk suppression" as a threat to the validity of a risk assessment validation study (he was the one to point this out to me in my paper for my first stats class). Bushway was the panel organizer for this panel. Also on the panel was Richard Berk. Berk is of course a statistical genius. His "random forest" approach to risk assessment is, in my opinion, going to revolutionize the risk assessment business. He's currently working in several states (including Pennsylvania and Maryland) on developing a risk assessment instrument to predict violence based on his approach. I've been working with him on the Pennsylvania project. It's an exciting endeavor. This panel of presenters on risk assessment was made up of top-notch statistical experts. I always feel dumb when I hear them present.
I'm sure that there were other sessions and highlights of the conference that I'm missing. I just hit a few of the big ones. Again, another great ASC conference. I'm looking forward to San Francisco next year (if I can find a way financially to make it out there).
Best session hands down was an "author meets critic" session for Mark Kleiman's new book "When Brute Force Fails". Mark presented the book and then Al Blumstein and Jeremy Travis were the discussants. I'm all about this book! This book has been getting a lot of press coverage, from the NY Times to the Economist to Governing Magazine. The basic thesis of the book is nothing earth-shattering. We've known for a while that the certainty and swiftness of punishment are more effective than the severity of punishment. I think the timing of Mark's book is ripe though, given the correctional crisis of mass incarceration and the economic situation. It's the application of the basic principles of deterrence theory that are new and innovative to Mark's book. I see this as providing an exciting and promising new experiment for parole and community corrections to try out. I've long thought that deterrence can be made to work, especially for certain offenders. There's just so much about this book that excites me that I could go on forever. It will be interesting to see how much attention this book receives by policymakers over the next year, especially given the attention it's already received by criminologists and the media.
I went to a good session on correctional forecasting methods. Some of the big names in this field were presenting at this session. I got some good practical tips for doing projections. It was interesting too to hear a more details about the California projections model.
I went to a panel discussion on "evidence-generating" research. Akiva Leiberman presented his whole idea of "evidence-generating" practice in contrast to "evidence-based" practice. Akiva first outlined this idea in an issue of The Criminologist. I've blogged about this idea before and it's a concept I've been pushing.
Nicole Rafter's Sutherland Address was excellent. She's not a vibrant presenter but the content of her address was what I was interested in. She's one of those rare historians in our discipline. My readers know that I'm all about the history of our discipline. So I was awaiting in anticipation to hear what Rafter had to say. As might be expected, she called for more attention to our history as a way of strengthening our discipline. I couldn't agree more. I look forward to reading her address in Criminology next year when it's published.
I was at a session in which Jim Austin gave his typical critique of why treatment programming won't work for reducing the nation's correctional population. I have to say that I agree with his analysis. I thought his presentation was good.
I went to a good session on theory, with Michael Gottfredson, Robert Sampson, and John Hagan presenting. The room was packed (must have been close to 200 people crammed in the room)!! I was excited to see that even in a year where the theme of the conference was policy, that theory continues to excite folks in our field. It excites me. It's also quite relevant to policy in my opinion. Anyways, Gottfredson gave his typical defense of control theory and answered some recent critiques of the theory. It was a wonderful presentation. Rob Sampson gave a talk in which he challenged the notion that experimental design studies are the "gold standard". I somewhat disagreed with his conclusions but he was interesting to hear nonetheless. And John Hagan as the discussant posed some interesting and challenging questions to each of the presenters on the panel, but unfortunately the panel was over before each presenter could respond. I was really hoping for time for questions from the audience too, especially to hear some challenging debate between Gottfredson and those who always get fired up about his control theory.
My panel on parole violators turned out quite well I thought. This was a panel that I had been wanting to put together for the past several years. We had a decent turnout and the other presentations on the panel were interesting. I think I organized a panel of what is arguably the most important research that's being done in the area of parole violators. Maybe there's a follow-up panel to come at a future ASC??
I went to an excellent panel on risk assessment methods. Avi Bhati presented his concept of risk suppression, which is a problem with risk assessment validation that I got nailed on in a risk assessment validation paper that I did for my first stats class when I started my Ph.D. Shawn Bushway is another to point out the whole idea of "risk suppression" as a threat to the validity of a risk assessment validation study (he was the one to point this out to me in my paper for my first stats class). Bushway was the panel organizer for this panel. Also on the panel was Richard Berk. Berk is of course a statistical genius. His "random forest" approach to risk assessment is, in my opinion, going to revolutionize the risk assessment business. He's currently working in several states (including Pennsylvania and Maryland) on developing a risk assessment instrument to predict violence based on his approach. I've been working with him on the Pennsylvania project. It's an exciting endeavor. This panel of presenters on risk assessment was made up of top-notch statistical experts. I always feel dumb when I hear them present.
I'm sure that there were other sessions and highlights of the conference that I'm missing. I just hit a few of the big ones. Again, another great ASC conference. I'm looking forward to San Francisco next year (if I can find a way financially to make it out there).
Thursday, October 8, 2009
Cain and Abel and Criminological Theory
I was doing some driving today and was listening to some old crim theory lectures on my Ipod (yeah I know, I'm a nerd). The lecture on social learning theory got me to thinking. It seems to me that the problem of some criminological theories is that they don't address a first cause (or "original cause") of crime. For example, one critique of social learning theory is that it treats humans as simply reactors and not as actors with agency and volition. Social learning theory has a particularly strong emphasis on "peer pressure" and on the learning of criminal/deviant behavior in the context of peer groups. But saying that someone has become a criminal as a result of spending a disproportionate amount of time hanging out with bad friends just pushes the problem of causal explanation one step backwards and does nothing to answer the question of original cause. Where did these bad friends learn to commit crimes or to hold to values supportive of criminal behavior? From other bad friends? If so, then where did these other bad friends learn to commit crimes or to hold to values supportive of criminal behavior? You can see how the search for an explanation can keep getting pushed backwards. If we pushed it back intergenerationally, we could push it all the way back to the first known crime- the Biblical account of the murder of Abel by his brother Cain. Where did Cain "learn" that it was acceptable to murder? According to the Biblical account, there was only one family in existence on the face of the earth at that time, so there were no peer groups. There was no media or entertainment to learn from. Why did Cain murder? Social learning theory seems to me to be a wholly inadequate theory for ultimately explaining this first known murder, and as a logical extention every murder (or crime) thereafter.To me, the best solution to this problem of "original cause" is the one taken by most of the theories that form the classical school of criminological thought. These theories would include social control, deterrence/rational choice, and routine activities theory. According to these theories, motivation towards crime (or at least towards pleasure maximimizing) is constant, is a part of human nature, and thus does not need to be explained. Therefore, the original cause of crime is something intrinsic to human nature. What needs to be explained by these theories then, is why all human beings do not act upon this human nature. Social learning can only fall back on human nature as an original cause if the only learning that occurs in life is in a positive direction, away from crime. Otherwise, learning is completely unnecessary if criminality is a part of human nature.
My favorite study in support of this viewpoint is the study by Dan Nagin and Pierre Tremblay on the shape of what I call the "age-aggression curve". I know I've blogged about this study before; it's one of my favorities to talk about. In their study, Nagin and Tremblay challenge the conventional wisdom concerning the age-crime curve peaking in the late teens to early 20's, which has been a sturdy finding in criminological research. The problem is that most criminologists have only examined crimes that are officially subjected to state sanctions. What Nagin and Tremblay find is that an aggregate plot of aggressiveness across age categories reveals that aggressive behavior actually peaks somewhere around age 2. While infants and toddlers don't go to jail for doing so, they spend a lot of time kicking, punching, hitting, pushing, taking toys, etc. And yet this happens before they even get to pre-school. It's all downhill after age 2. How can social learning theorists account for this finding? It seems that very little learning of aggressive behavior can possibly happen within the first two years of most infants' lives. Thus, this would strongly suggest to me that criminality is something we're all born with but that most of us grow out of.
Saturday, September 5, 2009
Connect The Dots
It's been quite a while since I've posted anything. Still trying to settle on a dissertation topic. I hope to get back in the swing and start posting more frequently.
2. "Fear and Trembling" by the 19th century Danish existentialist philosopher Soren Kierkegaard
In the meantime, a little fun. Here's a challenge for anyone out there who cares...who can offer up a suggestion for a possible connection between the three books I'm currently reading and what they have to say about my criminological perspective? The three books I'm currently reading are:
1. a biography of the 16th century Protestant reformer John Calvin entitled "John Calvin: A Heart for Devotion, Doctrine, and Doxology" edited by Burk Parsons (an aside: I recently discovered that Burk Parsons was one of the original members of the Backstreet Boys and was later approached by producer Lou Pearlman to be the lead member of N'Sync but instead chose to go into the ministry)
2. "Fear and Trembling" by the 19th century Danish existentialist philosopher Soren Kierkegaard
3. "Nudge: Improving Decisions About Health, Wealth, and Happiness", the New York Times bestseller by Richard Thaler and Cass Sunstein
Saturday, June 20, 2009
On Rehabilitation
The idea of offender rehabilitation was one of the primary reasons why I got into the business that I'm in. I've always believed that: a) human change is always possible, b) not much separates me from the common criminal when it comes right down to it (the old adage of "but by the grace of God there goes I"), and c) restorative justice generically speaking should be a core principle of our criminal justice system. As a young person growing up, I never wanted to follow in my father's footsteps to become a medical doctor, but what I did inherit from my dad was a passion to work towards the restoration of broken human beings. I personally believe that true human change can only come through divine intervention, but this should not stop us from working towards the restoration of broken human beings here on earth, whether it be a physical, mental, social, or spiritual type of restoration. Early in college I developed a particular vocational calling for a career within the criminal justice system, working to seek the healing of the specific forms of brokeness that result from criminal behavior. Indeed, much brokenness flows from crime in a society. Victims are broken, neighborhoods/comunities are broken, and yes even offenders are broken. As I advanced in my academic training and in my self-awareness of my abilities, I learned that the best way that I could make a difference in this area was not necessarily by being on the "front lines" but instead by using my mind/intellect to shape policy through the use of my apparent apptitude in the area of research and statistics.
This essay is not intended to be a narcissistic recounting of my career development. The previous paragraph is backdrop to the question I've tried to answer up to this point in my career, and that is "how is the brokeness that results from crime best addressed by a criminal justice system?" The most fashionable answer to this question within the context of a correctional environment is to promote offender rehabilitation. I think it's important to recognize at the forefront that nobody (or at least nobody that I know) approaches this subject neutrally. Everyone starts out wanting to believe deep down that offender rehabiliation either can work or cannot work. Everyone has a gut on this. My gut has always been that offender rehabilitation can work. But my gut matters little in terms of what science has to say about it (it may matter in terms of how I interpret science but I'll get to that later). So you can imagine my satisfaction when at the beginning of my career I learned of the supposed strong body of evidence in the research literature supporting offender rehabilitation. Ahhh, my gut was confirmed by science. But as I've learned more over the years, become more sophisticated in my thinking, and allowed my inquisitive mind to wander, I've come to a point of skepticism concerning much of the scientific evidence supporting offender rehabilitation. In short, I think offender rehabilitation has been oversold. To borrow a phrase from the title of a recent essay, I am "rethinking rehabilitation". I'm not giving up on rehabilitation. And it's certainly difficult to argue with the evidence in support of some types of well-implemented rehabilitation strategies. But again, skepticism has crept in. In certain circles within criminology, to even introducing such skepticism inevitably leads to labels such as "Robert Martison", "obstructionist", "anti-scientific", etc. This unfortunate result is yet another red flag to me that points to something else outside of or in addition to science going on. So at the risk of being labeled as the second coming of Robert Martison (who coincidentally was widely misunderstood and misrepresented), here are some of my thoughts on why we should be skeptical of the body of research promoting offender rehabilitation and why we should not oversell the utility of offender rehabiliation:
1. If offender rehabilitation is effective in reducing crime, why haven't recidivism rates dropped significantly over at least the past 20 years? Just looking at the Dept. of Justice's two published recidivism studies as well as recidivism rates within my own state, recidivism rates have been surprisingly stable despite an increased movement towards rehabilitation, evidence-based practice, and "what works". Some counter this argument by saying that recidivism rates would be even higher without evidence-based treatment, but to believe this we would logically have to conclude that either: a) institutional management and other system factors affecting recidivism have gotten worse in direct proportion to the degree that our treatment programs have gotten better, or b) we have made virtually no improvement or decline in either. In terms of aggregate recidivism rates, are our treatment programs simply balancing out the effect of an otherwise declining correctional system? Or, despite the increased emphasis on implementing evidence-based treatment programs over the past 20 years, has nothing changed? One of these cases may be a possibility. But I doubt that many who work within the corrections profession sincerely believe this to be the case. I just fail to see a strong case for why a system can supposedly increasingly support evidence-based rehabilitation programs that were found to reduce recidivism and yet not realize any significant aggregate decrease in recidivism rates. Why the homeostasis in recidivism rates if offender rehabilitation is the answer?
2. Why are crime rates relatively invariant to an emphasis on treatment? Obviously a lot of factors impact crime rates. But part of the "nothing works" mantra in the late 70's to early 80's was a result of the observation that crime rates were skyrocketing during the 70's despite a strong rehabilitation emphasis within the American correctional system during the same time. Maybe it was just that the wrong types of rehabilitation strategies were being used at the time. But this is an unknown.
3. Why haven't prison populations gone down despite an increased correctional emphasis on rehabilitation? Again, I recognize that many factors affect the size of a jurisdiction's correctional population. But if we measure recidivism in terms of reincarceration, shouldn't there be some effect over time on reducing the correctional population? Or does the relative invariance of aggregate recidivism rates, crime rates, and prison population rates despite an emphasis on rehabilitation mean that rehabilitation is just maintaining the status quo in order to keep things from getting worse? Do we really believe (or have evidence to believe) that these aggregate system measures would explode in a negative direction absent rehabilitation? If rehabilition programs can't make a dent in our aggregate crime or prison rates, what good are they ultimately to a society? It seems to me that a connection must be made at the macro level for rehabilitation strategies, moving beyond individual program evaluation results.
4. The primary methodology that has been used by rehabilitation advocates in summarizing the impact of rehabilitation across the body of evidence is the meta-analysis. Meta-analyses, as used, are problemmatic on several levels. I don't want to go into a methodological discussion here (e.g., issues of publication bias, reliance on statistical significance testing, etc.), but a recent paper by Richard Berk in the Journal of Experimental Criminology does a nice job of explaining some of these issues. One of the basic problems is that those who make use of the meta-analysis tend rely on the flawed logic that there is power in the number of studies alone. It's almost as if research becomes a popularity contest. If there are 20,000 studies/evaluations of rehabilitation interventions and 19,000 of them find that the intervention reduced reoffending then it must work. But what if the vast majority of these 19,000 studies were flawed or, to put it bluntly, garbage. The old adage of "garbage in/garbage out" applies. You can put a bunch of junk studies into a meta-analysis and find a strong aggregate effect size favoring rehabilitation but this hardly seems like the type of evidence we want to rely on. I've looked into a lot of the individual studies that make up many of these meta-analyses, and what I find is that there is a lot of garbage there. I'm not the only one to make this observation, although others haven't used my terms. I'm simply amazed when I go to some of the source studies and see the weak methodologies and flawed analysis conducted in many of them.
5. Another problem with rehabilitation is that systems tend to have a difficult time taking it "out of the laboratory" and moving it to scale. several studies have found that treatment programs demonstrate stronger effect sizes when done on a small scale and under the close eyes of evaluators than when implemented system-wide. Implementation seems to fall apart when taken to scale. So what good is it if we can't take these programs to scale? This would be akin to a scientist discovering a drug that was effective in treating laboratory rats but that had no impact when taken out of the laboratory and used on sick patients in the real world. Implementation problems are always pointed to when programs are found not to work. But at what point do we begin to doubt that we can implement any better? Maybe we just can't. Maybe we're pushing at the margins.
6. Another observation from research is that programs tend to demonstrate stronger positive effects in quasi-experiments than in true randomized controlled experiments. Couple this with the fact that the vast majority of program evaluations are indeed quasi-experiments. There is a whole host of potential issues with quasi-experiments that can cast a shadow of doubt on the results. This gets back to the issues relating to the quality of the studies going into meta-analyses. David Farrington has a nice discussion in a paper in Journal of Experimental Criminology of this observation that there are few true random experiments and that the effect sizes tend to be weaker in random experiments. This finding certainly rings true among the program evaluations conducted within my home state's correctional system. When randomized experiments were used, the results (even of supposed evidence-based program models) were mostly a null effect. Maybe so much research finds support for rehabilitation because it is weak research, and when we do it right we get a more realistic answer.
7. My observation is that there is a general liberal tendency within criminology, where our researchers come to the "laboratory" already wanting to believe and expecting to find that rehabilitation will work. This should be recognized, as it has the potential to produce strong biases. We need to recognize that ideology plays a role and that research never simply speaks for itself. The art of interpretation is where research results are often finessed. I'm not saying that the researcher should come to the "laboratory" completely disinterested. This is precisely opposite my point, since I believe that no researcher can come to the "laboratory" as a "clean slate". I'm saying that there needs to be more "putting our cards on the table" up front. As an aside, this is where theory plays an important role in program evaluation.
8. Intuitively to me, it seems more realistic to think that people will change their behavior as a result of a major life turning point such as marriage, a career, family, schooling, etc. rather than through a program (or even a stacking of programs). Criminals are not so different from non-criminals, and my observation of the non-criminals that surround my social world is that they haven't come to make major changes in their lives as a result of a program or class but more often it is as a result of major life events. I just think that this is a more realistic view of how life works and how change comes to occur. So I guess what I'm saying is that once I really started to think about it, intuitively it makes sense to expect the impact of a rehabilitation program to only be modest at best. Let's don't throw common sense out.
These are but a few of my thoughts guiding my questioning of rehabilitation. Again, I want to say that I'm not giving up on rehabilitation. I've simply come to the point where I think there needs to be realistic expectations of what it can accomplish. We try to instill realistic expectations in offenders before they are released from prison, but perhaps we could use some realistic expectations ourselves. What other opportunites might we be ignoring by over-emphasizing rehabilitation?
I would offer some preliminary suggestions for what we may be able to replace an over-emphasis on rehabilitation with. First, I think deterrence-based approaches should not be ignored. The rehabilitation folks are quick to qualify the circumstances under which a treatment program can be expected to work, and yet deterrence strategies are often summarily dismissed. As early as we are in our understanding of the impact of rehabilitation, we are even earlier in our understanding of the impact of deterrence. This presents a unique opportunity for criminologists though. Under what circumstances can we use a deterrence-based approach and see a real effect? Selective incapacitation is another goal that I believe is worth working towards as a system. Tied up in this, however, is the requirement of having good prediction tools in place. Again, a perfect opportunity for criminologists is presented. Let's don't settle for simple research, but let's instead press to advance our research to a whole new plain of understanding where possible. Let's examine if we can make selective incapacitation work. I think there are good institutional management strategies that can be focused on and improved as well. Finally, I think we can find innovative ways to support informal social control mechanisms (e.g., marriage, family, employment, etc.) for offenders once they are out of prison. We can't force ex-offenders to get married, to get a job, or to integrate into their community in a way that will provide a social control net, but to borrow the title of a recent bestselling book on reforming healthcare, perhaps we can "nudge" them in this direction. We have to ask what else is out there. I believe it was in a recent issue of The Criminologist that Akiva Liberman pointed out that instead of limiting ourselves exclusively to evidence-based practices, we should be open to evidence-generating practices. How can we experiment with new ideas and set our experimental ideas up so that they have the possibility of generating the evidence to support them if the evidence is there? Evidence-generating practices seem more important to me now, at a time when I question whether evidence-based practices are really as evidence-based as purported.
As always, many of my blogs on this site do not stray from controversy, so why should this one be any different. I'm curious to hear others' takes on rehabilitation. Have we put too many eggs in one basket? Are we over-selling? How can we best fix the array of brokenness that results from crime?
This essay is not intended to be a narcissistic recounting of my career development. The previous paragraph is backdrop to the question I've tried to answer up to this point in my career, and that is "how is the brokeness that results from crime best addressed by a criminal justice system?" The most fashionable answer to this question within the context of a correctional environment is to promote offender rehabilitation. I think it's important to recognize at the forefront that nobody (or at least nobody that I know) approaches this subject neutrally. Everyone starts out wanting to believe deep down that offender rehabiliation either can work or cannot work. Everyone has a gut on this. My gut has always been that offender rehabilitation can work. But my gut matters little in terms of what science has to say about it (it may matter in terms of how I interpret science but I'll get to that later). So you can imagine my satisfaction when at the beginning of my career I learned of the supposed strong body of evidence in the research literature supporting offender rehabilitation. Ahhh, my gut was confirmed by science. But as I've learned more over the years, become more sophisticated in my thinking, and allowed my inquisitive mind to wander, I've come to a point of skepticism concerning much of the scientific evidence supporting offender rehabilitation. In short, I think offender rehabilitation has been oversold. To borrow a phrase from the title of a recent essay, I am "rethinking rehabilitation". I'm not giving up on rehabilitation. And it's certainly difficult to argue with the evidence in support of some types of well-implemented rehabilitation strategies. But again, skepticism has crept in. In certain circles within criminology, to even introducing such skepticism inevitably leads to labels such as "Robert Martison", "obstructionist", "anti-scientific", etc. This unfortunate result is yet another red flag to me that points to something else outside of or in addition to science going on. So at the risk of being labeled as the second coming of Robert Martison (who coincidentally was widely misunderstood and misrepresented), here are some of my thoughts on why we should be skeptical of the body of research promoting offender rehabilitation and why we should not oversell the utility of offender rehabiliation:
1. If offender rehabilitation is effective in reducing crime, why haven't recidivism rates dropped significantly over at least the past 20 years? Just looking at the Dept. of Justice's two published recidivism studies as well as recidivism rates within my own state, recidivism rates have been surprisingly stable despite an increased movement towards rehabilitation, evidence-based practice, and "what works". Some counter this argument by saying that recidivism rates would be even higher without evidence-based treatment, but to believe this we would logically have to conclude that either: a) institutional management and other system factors affecting recidivism have gotten worse in direct proportion to the degree that our treatment programs have gotten better, or b) we have made virtually no improvement or decline in either. In terms of aggregate recidivism rates, are our treatment programs simply balancing out the effect of an otherwise declining correctional system? Or, despite the increased emphasis on implementing evidence-based treatment programs over the past 20 years, has nothing changed? One of these cases may be a possibility. But I doubt that many who work within the corrections profession sincerely believe this to be the case. I just fail to see a strong case for why a system can supposedly increasingly support evidence-based rehabilitation programs that were found to reduce recidivism and yet not realize any significant aggregate decrease in recidivism rates. Why the homeostasis in recidivism rates if offender rehabilitation is the answer?
2. Why are crime rates relatively invariant to an emphasis on treatment? Obviously a lot of factors impact crime rates. But part of the "nothing works" mantra in the late 70's to early 80's was a result of the observation that crime rates were skyrocketing during the 70's despite a strong rehabilitation emphasis within the American correctional system during the same time. Maybe it was just that the wrong types of rehabilitation strategies were being used at the time. But this is an unknown.
3. Why haven't prison populations gone down despite an increased correctional emphasis on rehabilitation? Again, I recognize that many factors affect the size of a jurisdiction's correctional population. But if we measure recidivism in terms of reincarceration, shouldn't there be some effect over time on reducing the correctional population? Or does the relative invariance of aggregate recidivism rates, crime rates, and prison population rates despite an emphasis on rehabilitation mean that rehabilitation is just maintaining the status quo in order to keep things from getting worse? Do we really believe (or have evidence to believe) that these aggregate system measures would explode in a negative direction absent rehabilitation? If rehabilition programs can't make a dent in our aggregate crime or prison rates, what good are they ultimately to a society? It seems to me that a connection must be made at the macro level for rehabilitation strategies, moving beyond individual program evaluation results.
4. The primary methodology that has been used by rehabilitation advocates in summarizing the impact of rehabilitation across the body of evidence is the meta-analysis. Meta-analyses, as used, are problemmatic on several levels. I don't want to go into a methodological discussion here (e.g., issues of publication bias, reliance on statistical significance testing, etc.), but a recent paper by Richard Berk in the Journal of Experimental Criminology does a nice job of explaining some of these issues. One of the basic problems is that those who make use of the meta-analysis tend rely on the flawed logic that there is power in the number of studies alone. It's almost as if research becomes a popularity contest. If there are 20,000 studies/evaluations of rehabilitation interventions and 19,000 of them find that the intervention reduced reoffending then it must work. But what if the vast majority of these 19,000 studies were flawed or, to put it bluntly, garbage. The old adage of "garbage in/garbage out" applies. You can put a bunch of junk studies into a meta-analysis and find a strong aggregate effect size favoring rehabilitation but this hardly seems like the type of evidence we want to rely on. I've looked into a lot of the individual studies that make up many of these meta-analyses, and what I find is that there is a lot of garbage there. I'm not the only one to make this observation, although others haven't used my terms. I'm simply amazed when I go to some of the source studies and see the weak methodologies and flawed analysis conducted in many of them.
5. Another problem with rehabilitation is that systems tend to have a difficult time taking it "out of the laboratory" and moving it to scale. several studies have found that treatment programs demonstrate stronger effect sizes when done on a small scale and under the close eyes of evaluators than when implemented system-wide. Implementation seems to fall apart when taken to scale. So what good is it if we can't take these programs to scale? This would be akin to a scientist discovering a drug that was effective in treating laboratory rats but that had no impact when taken out of the laboratory and used on sick patients in the real world. Implementation problems are always pointed to when programs are found not to work. But at what point do we begin to doubt that we can implement any better? Maybe we just can't. Maybe we're pushing at the margins.
6. Another observation from research is that programs tend to demonstrate stronger positive effects in quasi-experiments than in true randomized controlled experiments. Couple this with the fact that the vast majority of program evaluations are indeed quasi-experiments. There is a whole host of potential issues with quasi-experiments that can cast a shadow of doubt on the results. This gets back to the issues relating to the quality of the studies going into meta-analyses. David Farrington has a nice discussion in a paper in Journal of Experimental Criminology of this observation that there are few true random experiments and that the effect sizes tend to be weaker in random experiments. This finding certainly rings true among the program evaluations conducted within my home state's correctional system. When randomized experiments were used, the results (even of supposed evidence-based program models) were mostly a null effect. Maybe so much research finds support for rehabilitation because it is weak research, and when we do it right we get a more realistic answer.
7. My observation is that there is a general liberal tendency within criminology, where our researchers come to the "laboratory" already wanting to believe and expecting to find that rehabilitation will work. This should be recognized, as it has the potential to produce strong biases. We need to recognize that ideology plays a role and that research never simply speaks for itself. The art of interpretation is where research results are often finessed. I'm not saying that the researcher should come to the "laboratory" completely disinterested. This is precisely opposite my point, since I believe that no researcher can come to the "laboratory" as a "clean slate". I'm saying that there needs to be more "putting our cards on the table" up front. As an aside, this is where theory plays an important role in program evaluation.
8. Intuitively to me, it seems more realistic to think that people will change their behavior as a result of a major life turning point such as marriage, a career, family, schooling, etc. rather than through a program (or even a stacking of programs). Criminals are not so different from non-criminals, and my observation of the non-criminals that surround my social world is that they haven't come to make major changes in their lives as a result of a program or class but more often it is as a result of major life events. I just think that this is a more realistic view of how life works and how change comes to occur. So I guess what I'm saying is that once I really started to think about it, intuitively it makes sense to expect the impact of a rehabilitation program to only be modest at best. Let's don't throw common sense out.
These are but a few of my thoughts guiding my questioning of rehabilitation. Again, I want to say that I'm not giving up on rehabilitation. I've simply come to the point where I think there needs to be realistic expectations of what it can accomplish. We try to instill realistic expectations in offenders before they are released from prison, but perhaps we could use some realistic expectations ourselves. What other opportunites might we be ignoring by over-emphasizing rehabilitation?
I would offer some preliminary suggestions for what we may be able to replace an over-emphasis on rehabilitation with. First, I think deterrence-based approaches should not be ignored. The rehabilitation folks are quick to qualify the circumstances under which a treatment program can be expected to work, and yet deterrence strategies are often summarily dismissed. As early as we are in our understanding of the impact of rehabilitation, we are even earlier in our understanding of the impact of deterrence. This presents a unique opportunity for criminologists though. Under what circumstances can we use a deterrence-based approach and see a real effect? Selective incapacitation is another goal that I believe is worth working towards as a system. Tied up in this, however, is the requirement of having good prediction tools in place. Again, a perfect opportunity for criminologists is presented. Let's don't settle for simple research, but let's instead press to advance our research to a whole new plain of understanding where possible. Let's examine if we can make selective incapacitation work. I think there are good institutional management strategies that can be focused on and improved as well. Finally, I think we can find innovative ways to support informal social control mechanisms (e.g., marriage, family, employment, etc.) for offenders once they are out of prison. We can't force ex-offenders to get married, to get a job, or to integrate into their community in a way that will provide a social control net, but to borrow the title of a recent bestselling book on reforming healthcare, perhaps we can "nudge" them in this direction. We have to ask what else is out there. I believe it was in a recent issue of The Criminologist that Akiva Liberman pointed out that instead of limiting ourselves exclusively to evidence-based practices, we should be open to evidence-generating practices. How can we experiment with new ideas and set our experimental ideas up so that they have the possibility of generating the evidence to support them if the evidence is there? Evidence-generating practices seem more important to me now, at a time when I question whether evidence-based practices are really as evidence-based as purported.
As always, many of my blogs on this site do not stray from controversy, so why should this one be any different. I'm curious to hear others' takes on rehabilitation. Have we put too many eggs in one basket? Are we over-selling? How can we best fix the array of brokenness that results from crime?
Tuesday, April 28, 2009
U.S. News & World Report Rankings
The new U.S. News & World Report rankings for Criminology/Criminal Justice graduate school programs are out. University of Maryland holds on to its #1 ranking! It's great to be at the best school. Indulge me as I brag a little bit.Of course in briefly scanning some other CJ blogs, I noticed that everyone who's at a school that's not at the top of the list are complaining about the ranking methodology. Hey, I think U.S. News & World Report got it right. No complaints from me (ok, so I'm a little biased here). Granted there are many methods of ranking graduate school programs. Another blog site rightly points out that the Journal of Criminal Justice Education has published some good pieces ranking Criminology/CJ programs using different methodologies. The results don't substantively differ though. The same schools consistently come out on top. I can't remember if University of Maryland's program was #1 in all of the Journal of Criminal Justice Education pieces, but they were at least in the top 3.
I remember being a senior in high school and planning for where I wanted to go to college. I thought the U.S. News & World Report rankings were like the definitive source of college rankings. So maybe it's just a bit of nostalgia that I get excited about seeing my current school at the top of their list.
What do others think? Did U.S. News & World Report get it right? Who should have been ranked higher? Who should have been ranked lower? What ranking criteria do you think is most important? Should it be based on faculty publication counts? Publication counts only in prestigious journals? Faculty grants received? Peer rankings (as I believe the U.S. News & World Report rankings primarily relied on)? Service to the field (such as positions held at professional associations/societies like American Society of Criminology)? Counts of faculty citations in other publications? Or is all of this just a waste of time? Are we just massaging our collective ego's by doing these rankings? It may be difficult to obtain any high degree of objectivity in producing such rankings, but I don't tend to think that they're a waste of time. Again speaking from personal experience, I know they helped me decide on a graduate school when I was looking to get my Master's degree. I also think competition is good (stay tuned for a forthcoming posting on theory competition vs. theory integration).
Sunday, April 26, 2009
From Hellraiser to Family Man
A fair body of solid research tells us that there is a moderately strong relationship between marriage and desistance from crime. I've been thinking a lot about this issue recently, and about what the policy implications of it are. I'm even thinking this may be an area I want to tackle for my dissertation (yes, I still haven't picked a dissertation topic yet).
The obvious question that comes up when we talk about a correlation between marriage and crime is whether or not it's a causal relationship. Some hold that it's merely correlational; that those who desist from crime are also those who are more likely to get married. Thus, this underlying propensity is really predicting both factors. I tend to side more with the social control perspective, that something about marriage can actually have a causal impact upon future criminal behavior. Sampson and Laub have done some good work in this area. Their most recent paper in Criminology on the marriage-crime relationship indeed finds a causal relationship. They use what they call a counterfactual approach to control for the problem of self-selection into marriage. Obviously it would be nice if we could randomly assign a group of ex-offenders to get married or not to get married, but that's never going to happen. Absent a true randomized experiment, Sampson and Laub's paper comes the closest methodologically (I believe) to being able to determine causality. They find that marriage reduces the odds of recidivism by about 35%. Quite an impressive effect! I'm not surprised that they found this strong of a relationship though. To me it just makes intuitive sense. When those studying correctional rehabilitation find similar effects for treatment programming, why shouldn't we believe that a potentially life-changing event like getting married could have an impact in the same range. How many people experience genuine change in their life from a program vs. from a life-changing event? The potential for a stronger effect from a life-changing event just resonates with me more.
But there's many questions that remain unanswered. I'd like to raise just a few of them. First, does the marriage have to be a good marriage in order to lead to desistance from crime? Can a not-so-good or even downright bad marriage still produce desistance? My gut tells me that we might still see a significant impact on desistance from even a bad marriage, although the healthy marriage will probably generate a stronger impact.
This leads to my second question. What is it about marriage that produces desistance? Is it that marriage tends to produce a change in attitudes and values? Is is that the married person's routine activities change, thus keeping them out of trouble? Is it the "nagging of a good wife"? Is it a change in self-identity ("from hellraiser to family man")? Is it "love that conquers all"? What's in that "black box" of the marriage effect. Again, I tend to lean more towards believing the social control explanations (e.g., the "nagging of a good wife"), but I also think the routine activities explanation holds a lot of weight too. I think Sampson and Laub would agree here, and present evidence of this in their interviews with the Glueck men in "Shared Beginnings, Divergent Lives".
The third question I have about the marriage effect is whether or not one has to be legally married to benefit from this effect. Can someone in a committed yet unmarried relationship equally benefit from that relationship in terms of desistance? Can same sex partners benefit? Obviously holding to a conservative Christian worldview as I do, I would expect to see that marriage in the traditional sense will produce the strongest effect. But if I'm intellectually honest then I must admit that ultimately it's an empirical question that has not yet been answered by research. Again, falling back on theory, it just makes too much sense to me to believe that traditional marriage would have the strongest effect. Remember in Hirschi's social control theory that the four elements of control are attachment, commitment, involvement, and belief. In marriage, I think that one of the factors generating the marriage effect is commitment. Obviously marriage itself is a long-term commitment that holds a lot of serious legal implications. How much of a commitment really is a "committed unmarried relationship"? It would seem that on average you will find less real commitment in such relationships, and to the extend that you do then you would expect to find less of a controlling effect of marriage on criminal behavior. As far as same sex relationships, it would make for an interesting study but practically speaking it seems that it would be very difficult to be able to identify a large enough sample size of criminals who define themselves as being in a same sex relationship.
Maybe I'm being naive about that last point. But this leads to my fourth line of questioning. Has, as some have speculated, the institution of marriage changed so much in our culture that we should not expect to see a positive marriage effect on crime any longer? Admittedly, much of the research that finds a relationship between marriage and crime are longitudinal studies that examine a sample of individuals who came of age a decades ago. Is there a period effect related to marriage? How has cultural shifts in marriage impacted the relationship between marriage and crime? Once again going with my gut, I don't expect to find that much has changed. Think about it...marriage has been around since the beginning of time and across many cultures. I think we might be seeing an assault on traditional marriage but I think marriage and its concomitant benefits are here to stay.
Fifth, what about having children? Do we see an interaction effect or a confounding effect? Is what appears to be a marriage impact really an impact of being a parent? Or does being a parent add another protective factor in the mix, further strengthening the probability of desistance? I would love to see some research on this. We need to disentangle the impact of being a parent from the impact of being married.
Sixth, are there situations in which marriage can actually increase future criminal behavior? I'm thinking here particularly of domestic violence. I believe it was in a chapter in Blumstein's book on the crime drop (I think the chapter was by Rosenfeld) in which he finds that homicide rates have dropped because less people are getting married. He examines the types of homicides that contributed to the drop in homicides and found that domestic homicides had a particularly large effect on the drop. So how does this finding jive with other findings on the positive impacts of marriage?
These questions just touch the surface. There are many other questions that could be raised. What interests me is the policy implications of this research. I currently work for the PA Department of Corrections. What are we to do with a finding that there is a relationship between marriage and crime? We can't force our inmates to get married. But maybe we can provide informational sessions on marriage and also marriage counseling. The Oklahoma Department of Corrections is doing just that. I am very much interested to see the results of the evaluations that are being done on the Oklahoma program. John Laub has turned me on to the idea that maybe we can't force offenders to get married but maybe we can "nudge" them in that direction. He takes this concept of "nudging" from the title of a recent book by Thaler and Sunstein called "Nudge: Improving Decisions About Health, Wealth, and Happiness". I haven't read the book, but apparently in this book the authors discuss the idea that we can't force people to live healthier lives in order to reduce health care costs, but there are ways that we can nudge them in the direction of healthier living. Laub thinks there might be promise for applying this same concept to marriage.
Whatever future research may find in this area, I can certainly speak from personal experience. I know the marriage effect has worked for me. From hellraiser to family man! Better run, dinner is on the table. Coming Honey...(ha ha)
The obvious question that comes up when we talk about a correlation between marriage and crime is whether or not it's a causal relationship. Some hold that it's merely correlational; that those who desist from crime are also those who are more likely to get married. Thus, this underlying propensity is really predicting both factors. I tend to side more with the social control perspective, that something about marriage can actually have a causal impact upon future criminal behavior. Sampson and Laub have done some good work in this area. Their most recent paper in Criminology on the marriage-crime relationship indeed finds a causal relationship. They use what they call a counterfactual approach to control for the problem of self-selection into marriage. Obviously it would be nice if we could randomly assign a group of ex-offenders to get married or not to get married, but that's never going to happen. Absent a true randomized experiment, Sampson and Laub's paper comes the closest methodologically (I believe) to being able to determine causality. They find that marriage reduces the odds of recidivism by about 35%. Quite an impressive effect! I'm not surprised that they found this strong of a relationship though. To me it just makes intuitive sense. When those studying correctional rehabilitation find similar effects for treatment programming, why shouldn't we believe that a potentially life-changing event like getting married could have an impact in the same range. How many people experience genuine change in their life from a program vs. from a life-changing event? The potential for a stronger effect from a life-changing event just resonates with me more.
But there's many questions that remain unanswered. I'd like to raise just a few of them. First, does the marriage have to be a good marriage in order to lead to desistance from crime? Can a not-so-good or even downright bad marriage still produce desistance? My gut tells me that we might still see a significant impact on desistance from even a bad marriage, although the healthy marriage will probably generate a stronger impact.
This leads to my second question. What is it about marriage that produces desistance? Is it that marriage tends to produce a change in attitudes and values? Is is that the married person's routine activities change, thus keeping them out of trouble? Is it the "nagging of a good wife"? Is it a change in self-identity ("from hellraiser to family man")? Is it "love that conquers all"? What's in that "black box" of the marriage effect. Again, I tend to lean more towards believing the social control explanations (e.g., the "nagging of a good wife"), but I also think the routine activities explanation holds a lot of weight too. I think Sampson and Laub would agree here, and present evidence of this in their interviews with the Glueck men in "Shared Beginnings, Divergent Lives".
The third question I have about the marriage effect is whether or not one has to be legally married to benefit from this effect. Can someone in a committed yet unmarried relationship equally benefit from that relationship in terms of desistance? Can same sex partners benefit? Obviously holding to a conservative Christian worldview as I do, I would expect to see that marriage in the traditional sense will produce the strongest effect. But if I'm intellectually honest then I must admit that ultimately it's an empirical question that has not yet been answered by research. Again, falling back on theory, it just makes too much sense to me to believe that traditional marriage would have the strongest effect. Remember in Hirschi's social control theory that the four elements of control are attachment, commitment, involvement, and belief. In marriage, I think that one of the factors generating the marriage effect is commitment. Obviously marriage itself is a long-term commitment that holds a lot of serious legal implications. How much of a commitment really is a "committed unmarried relationship"? It would seem that on average you will find less real commitment in such relationships, and to the extend that you do then you would expect to find less of a controlling effect of marriage on criminal behavior. As far as same sex relationships, it would make for an interesting study but practically speaking it seems that it would be very difficult to be able to identify a large enough sample size of criminals who define themselves as being in a same sex relationship.
Maybe I'm being naive about that last point. But this leads to my fourth line of questioning. Has, as some have speculated, the institution of marriage changed so much in our culture that we should not expect to see a positive marriage effect on crime any longer? Admittedly, much of the research that finds a relationship between marriage and crime are longitudinal studies that examine a sample of individuals who came of age a decades ago. Is there a period effect related to marriage? How has cultural shifts in marriage impacted the relationship between marriage and crime? Once again going with my gut, I don't expect to find that much has changed. Think about it...marriage has been around since the beginning of time and across many cultures. I think we might be seeing an assault on traditional marriage but I think marriage and its concomitant benefits are here to stay.
Fifth, what about having children? Do we see an interaction effect or a confounding effect? Is what appears to be a marriage impact really an impact of being a parent? Or does being a parent add another protective factor in the mix, further strengthening the probability of desistance? I would love to see some research on this. We need to disentangle the impact of being a parent from the impact of being married.
Sixth, are there situations in which marriage can actually increase future criminal behavior? I'm thinking here particularly of domestic violence. I believe it was in a chapter in Blumstein's book on the crime drop (I think the chapter was by Rosenfeld) in which he finds that homicide rates have dropped because less people are getting married. He examines the types of homicides that contributed to the drop in homicides and found that domestic homicides had a particularly large effect on the drop. So how does this finding jive with other findings on the positive impacts of marriage?
These questions just touch the surface. There are many other questions that could be raised. What interests me is the policy implications of this research. I currently work for the PA Department of Corrections. What are we to do with a finding that there is a relationship between marriage and crime? We can't force our inmates to get married. But maybe we can provide informational sessions on marriage and also marriage counseling. The Oklahoma Department of Corrections is doing just that. I am very much interested to see the results of the evaluations that are being done on the Oklahoma program. John Laub has turned me on to the idea that maybe we can't force offenders to get married but maybe we can "nudge" them in that direction. He takes this concept of "nudging" from the title of a recent book by Thaler and Sunstein called "Nudge: Improving Decisions About Health, Wealth, and Happiness". I haven't read the book, but apparently in this book the authors discuss the idea that we can't force people to live healthier lives in order to reduce health care costs, but there are ways that we can nudge them in the direction of healthier living. Laub thinks there might be promise for applying this same concept to marriage.
Whatever future research may find in this area, I can certainly speak from personal experience. I know the marriage effect has worked for me. From hellraiser to family man! Better run, dinner is on the table. Coming Honey...(ha ha)
Subscribe to:
Posts (Atom)