What do you think?
Rate this book
Violent Offenders: Appraising And Managing Risk
Argue that community risk management can be improved by using actuarial assessment and by combining what is known about the prediction of violence, the study of clinical decision making, and the literature on treatment outcome and program evaluation. In this new edition, the authors update their review, focusing on the actuarial instruments they developed and described earlier and on the measures they have continued to develop. In their lively style, they review the commentary on risk appraisal, addressing 20 of the most common arguments against actuarial risk appraisal.
462 pages, Hardcover
First published September 1, 1998
1 person is currently reading
23 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 2 of 2 reviews
August 21, 2012
I don't have training in psychology, criminal management or treatment, or formal risk appraisal, so that's my qualification on this review. Also, the authors have a focus on the treatment center they worked out of in Ontario, Canada, although they attempt to cover Canada and the US in general where relevant.
Partway through I felt like I was reading Moneyball all over again (only better written, and about a totally different subject on which many people hold very strong opinions regardless of qualifications). This book is the result of decades of research by the writers pursuing better methods of treatment for people who commit violent crimes; what they figured out along the way was that the measurement of the quality of treatment was fundamentally flawed due to the lack of measurement of who was going into treatment in the first place. Here's where it gets to be like Moneyball: statistically, clinicians measure about as well as non-clinicians in assessing how likely someone who has committed a violent crime is to commit another violent crime after being released. On the surface there's a really clear problem here: this means that the success rate of a treatment program is based on a sampling of people including many who are unlikely to recidivate regardless of presence or quality of treatment, and if the expectation is already that these are violent criminals who are likely to recidivate, then anyone who doesn't is a success (while anyone who is likely to recidivate and is instead released sees no treatment anyway). This is the equivalent of judging a school to be successful based on the presence of some gifted students, regardless of where they were at at the beginning and end of their education or what they got out of the school itself. The problems go deeper: the treatment programs shown to decrease recidivism for non-psychopathic people actually have the opposite effect for psychopaths, who are more likely to develop social rehabilitation training into more complex and skilled social manipulation techniques; people with schizophrenia, statistically much less likely to recidivate than those without, are more likely to be erroneously considered likely to recidivate; clinical review is subject to the same cons used by people who learn how to get by while committing repeat crimes, violent and not; and so on. The authors propose replacing the clinical review process with a risk appraisal system: reviewing a selection of measurable items, from age of offense to history of abuse to presence of "both" biological parents until the age of 16, for the statistical likelihood of that person to reoffend.
My first gut response to this was that it was some essentialist, classist, sexist, horrible shit that, although it doesn't mention race as a factor to be measured, would also certainly perpetuate the same institutionalized racism in the criminal justice system, not to mention would simply pervade the punishment of people whose childhoods were on the particularly-fucked-up end of the fucked-up scale. I still feel that way but for a different reason than I expected, and find that I actually really agree with where they're going. First off, how the measurement is happening. They're very upfront in noting that their effort was to remove data that did not impact their measurements and to toggle the strength of preserved data until the predicted results match actual results. Second off, the goal of this is not punishment but treatment, and specifically to target treatment based on the person's background-impacted needs, meaning that this process would be not just useful but actually critical to accomplish the opposite of what it appears to be doing - it's applying a broad scale to a diverse array of people to do much much more than simply divy them up between likely reoffenders and likely non-reoffenders, in an attempt to meet the needs of a system that was actively trying to separate certain people from society in order to keep the rest of society feeling safe, and failing to either do so effectively or to effectively meet the needs of those separated people. Many critical fixes to broken systems have to use a broken lens. Third, the items it is measuring are, ultimately, life factors that are going to increase the likelihood that a person will have social access and motivation to commit violent crimes that are likely to be prosecuted, so its focus is simply what to do with people who have been arrested for it. They make note of the fact that their tool doesn't work when most people are at one end or another of recidivism, so it doesn't work to assess the statistical probability of violent crimes among people who haven't committed any - and it's explicitly described as not a tool for predicting recidivism, rather for looking at the statistical risk of recidivism in that particular person. Higher-risk people with specific mental health needs should get different treatment services than lower-risk people and/or people with different mental health needs. Makes sense, and undermines the way the criminal justice system works now. I like it. And it involves math and risk management!
The book starts with a brief overview of historical treatment methods for violent criminals in Canada and the US as well as Britain, and where they've fallen short, which seemed to decently cover what it needed to but I don't know the history well enough to feel I'm a sharp judge of that. I appreciated their review of the role of insanity defenses in prosecution, which presented some interesting numbers on their actual success rates (not a lot). A lot of the statistical analysis was over my head, but I was reading for the other content anyway, so that wasn't such an issue. A majority of the referenced texts were unfamiliar to me (by which I mean almost all of them) so there was plenty of new information that I enjoyed. I got a lot out of the chapter reviewing some of the most common arguments against their actuarial tool, which was good to answer some of my questions as well as to frame how they saw the tool and their publications being met by the clinical assessment community. The writing is solid most of the time, although there were a number of non-clear sentences where I couldn't tell if I lacked context in psychology, in criminal management, in risk appraisal, in statistics...or simply had run into a case of the authors had worked together for many years and understood each other and failed to catch where they wouldn't be clear to others. I suspect that the amount of that last one is minimal though. And at least one of the writers has a bitter and awesome sense of humor, which pops up a handful of times to great effect (and at least one time to the aforementioned confusing effect, although that one I managed to figure out).
It's one of the more intense and challenging books I've read in a while.
Quotes I wanted to hold on to:
"The ultimate seriousness of a violent crime is a value judgement and not a scientific question...It is clear too that opinions change; many men who have assaulted their spouses are presently serving sentences for behavior that 20 years ago would not have even resulted in criminal charges." (p175)
"How unique people are is largely an empirical question. Undoubtedly, humans exhibit great similarity in many aspects of their nature. In fact, one of the authors of this book believes that people are unique only in scientifically uninteresting ways." (p180)
Partway through I felt like I was reading Moneyball all over again (only better written, and about a totally different subject on which many people hold very strong opinions regardless of qualifications). This book is the result of decades of research by the writers pursuing better methods of treatment for people who commit violent crimes; what they figured out along the way was that the measurement of the quality of treatment was fundamentally flawed due to the lack of measurement of who was going into treatment in the first place. Here's where it gets to be like Moneyball: statistically, clinicians measure about as well as non-clinicians in assessing how likely someone who has committed a violent crime is to commit another violent crime after being released. On the surface there's a really clear problem here: this means that the success rate of a treatment program is based on a sampling of people including many who are unlikely to recidivate regardless of presence or quality of treatment, and if the expectation is already that these are violent criminals who are likely to recidivate, then anyone who doesn't is a success (while anyone who is likely to recidivate and is instead released sees no treatment anyway). This is the equivalent of judging a school to be successful based on the presence of some gifted students, regardless of where they were at at the beginning and end of their education or what they got out of the school itself. The problems go deeper: the treatment programs shown to decrease recidivism for non-psychopathic people actually have the opposite effect for psychopaths, who are more likely to develop social rehabilitation training into more complex and skilled social manipulation techniques; people with schizophrenia, statistically much less likely to recidivate than those without, are more likely to be erroneously considered likely to recidivate; clinical review is subject to the same cons used by people who learn how to get by while committing repeat crimes, violent and not; and so on. The authors propose replacing the clinical review process with a risk appraisal system: reviewing a selection of measurable items, from age of offense to history of abuse to presence of "both" biological parents until the age of 16, for the statistical likelihood of that person to reoffend.
My first gut response to this was that it was some essentialist, classist, sexist, horrible shit that, although it doesn't mention race as a factor to be measured, would also certainly perpetuate the same institutionalized racism in the criminal justice system, not to mention would simply pervade the punishment of people whose childhoods were on the particularly-fucked-up end of the fucked-up scale. I still feel that way but for a different reason than I expected, and find that I actually really agree with where they're going. First off, how the measurement is happening. They're very upfront in noting that their effort was to remove data that did not impact their measurements and to toggle the strength of preserved data until the predicted results match actual results. Second off, the goal of this is not punishment but treatment, and specifically to target treatment based on the person's background-impacted needs, meaning that this process would be not just useful but actually critical to accomplish the opposite of what it appears to be doing - it's applying a broad scale to a diverse array of people to do much much more than simply divy them up between likely reoffenders and likely non-reoffenders, in an attempt to meet the needs of a system that was actively trying to separate certain people from society in order to keep the rest of society feeling safe, and failing to either do so effectively or to effectively meet the needs of those separated people. Many critical fixes to broken systems have to use a broken lens. Third, the items it is measuring are, ultimately, life factors that are going to increase the likelihood that a person will have social access and motivation to commit violent crimes that are likely to be prosecuted, so its focus is simply what to do with people who have been arrested for it. They make note of the fact that their tool doesn't work when most people are at one end or another of recidivism, so it doesn't work to assess the statistical probability of violent crimes among people who haven't committed any - and it's explicitly described as not a tool for predicting recidivism, rather for looking at the statistical risk of recidivism in that particular person. Higher-risk people with specific mental health needs should get different treatment services than lower-risk people and/or people with different mental health needs. Makes sense, and undermines the way the criminal justice system works now. I like it. And it involves math and risk management!
The book starts with a brief overview of historical treatment methods for violent criminals in Canada and the US as well as Britain, and where they've fallen short, which seemed to decently cover what it needed to but I don't know the history well enough to feel I'm a sharp judge of that. I appreciated their review of the role of insanity defenses in prosecution, which presented some interesting numbers on their actual success rates (not a lot). A lot of the statistical analysis was over my head, but I was reading for the other content anyway, so that wasn't such an issue. A majority of the referenced texts were unfamiliar to me (by which I mean almost all of them) so there was plenty of new information that I enjoyed. I got a lot out of the chapter reviewing some of the most common arguments against their actuarial tool, which was good to answer some of my questions as well as to frame how they saw the tool and their publications being met by the clinical assessment community. The writing is solid most of the time, although there were a number of non-clear sentences where I couldn't tell if I lacked context in psychology, in criminal management, in risk appraisal, in statistics...or simply had run into a case of the authors had worked together for many years and understood each other and failed to catch where they wouldn't be clear to others. I suspect that the amount of that last one is minimal though. And at least one of the writers has a bitter and awesome sense of humor, which pops up a handful of times to great effect (and at least one time to the aforementioned confusing effect, although that one I managed to figure out).
It's one of the more intense and challenging books I've read in a while.
Quotes I wanted to hold on to:
"The ultimate seriousness of a violent crime is a value judgement and not a scientific question...It is clear too that opinions change; many men who have assaulted their spouses are presently serving sentences for behavior that 20 years ago would not have even resulted in criminal charges." (p175)
"How unique people are is largely an empirical question. Undoubtedly, humans exhibit great similarity in many aspects of their nature. In fact, one of the authors of this book believes that people are unique only in scientifically uninteresting ways." (p180)
March 25, 2008
This is the source for understanding and appreciating actuarial risk assessment for violent offenders.
Displaying 1 - 2 of 2 reviews