Q: What is "smart sentencing"?
Q: Can sentencing be "smart" without technology?
Q: Do you want computers to make sentencing decisions instead of judges?
Q: Haven't recent ballot measures and statutes eliminated the role of discretion?
Q.  How can you be so disloyal to your colleagues by blaming them for crime?
Q: How does Blakely v. Washington affect all this?
Q: What happened at the ALI annual meeting in May, 2007?
Q: Is it really a good idea to have judges try to predict which sentences will work best to reduce criminal behavior?
Q: Are you trying to replace probation officers?
Q: Isn't unrealistic to expect lawyers who practice criminal law to handle data and research about what works?
Q: Smart sentencing makes a lot of sense; what are the arguments against it?
Q: Is this all an argument not to use jail or prison?
Q: Aren't you shifting responsibility from criminals to the courts when you blame the courts (or criminal justice) for recidivism?
Q: Are you saying retribution and general deterrence should be discarded as objectives of sentencing?
Q: How can you expect providers accurately to gather and report the data you need?
Q: What makes you think this technology will serve public safety?
Q: What impact do you expect to have on recidivism?
Q: Don't we already know what works?
Q: Why can't we just exploit the research we already have?
Q: How can you evaluate programs with limited access to their internal data?
Q: Are you trying to replace research and researchers?
Q: Does this sentencing support technology have any other benefits for criminal justice?
Q: Why include ethnicity in the offender profile?
Q: Is it fair to include gender and other variables the offender cannot control in the profile?
Q: Is anyone else doing this now?
Q: Have you determined that your results are statistically significant?
Q: How do you know outcomes don't reflect accurate predictions rather than cause and effect?
Q: Doesn't plea bargaining distort the data?
Q: Isn't this just a fancy risk assessment tool?
Q: Haven't the experts concluded that nothing really works?
Q: How can you expect offenders to change until they're ready to accept change?
Q: What is that graphic you use on so many pages?
 

Q:  What is "Smart Sentencing?"
A:     Some have asked what I mean by "smart sentencing" and ask for examples.  For my purposes, "smart sentencing" is sentencing that rationally seeks to accomplish an objective -- usually reducing the likelihood that an offender will commit future crimes, but the concept can apply to any sentencing objective.  Sentencing rationally seeks an objective whenever it uses best efforts to access relevant  information about the sentencing choices available for an offender.  Sentencing support tools are but one source of helpful information.  Within limits, these tools tell us what has happened (in terms of subsequent criminal behavior) to similar offenders, sentenced for similar crimes, who received any of the dispositions available for the offender now being sentenced.  Of course, "similar" is hardly the same as identical, and depending upon the size and nature of the cohort of offenders, the precision with which we've identified truly similar offenders may vary greatly.  Even with an "identical" cohort based on the data available to the tools, variables not recognized by the tools [because they are not routinely collected and shared -- or even known to the system] may make one offender more or less likely than another within the cohort to benefit from a given disposition.
   Here are two examples.  Traditional sentencing might simply take a low level theft offender with a record of low level thefts and routinely impose a period of, say, 18 months bench probation, "theft talk," and a fine of $150 plus fees and assessments.  Smarter sentencing would look at what similar offenders, of a similar criminal background, age, and gender, did after receiving any of the available dispositions.  With no other information, and no reason to suspect some individual characteristic that makes a difference, smarter sentencing would consider doing what seemed to work best in the past on such offenders for such crimes.  For some cohorts, by the way, I've seen that anger management correlates substantially better than traditional theft sentences with reduced recidivism, and for such a cohort, I might well consider adding anger management (or substituting it for a fine). If we had the resources to do a criminogenic assessment, we might find a connection between substance abuse or a mental health issue and the criminal behavior -- if so, smarter sentencing might involve addressing the substance abuse and mental health issues with a dual diagnosis treatment.
    On the other end of the spectrum, traditional sentencing of a sex offender convicted of, say, six separate crimes, might include 120 months prison total, with most of the time running concurrently to produce that 120 months -- because ten years seems "just" in light of the crimes.  [What is "just" is ultimately a tremendously fluid concept].  Smarter sentencing would look to a psychosexual evaluation to assess the risk level and susceptibility of the offender to any available sex offender treatment.  In the case of a predatory pedophile, that process might well suggest that the risk of recidivism is extremely high and the likely impact of any treatment extremely low.  Coupled with literature suggesting that post-prison recidivism does not increase with the length of custody for high-risk offenders [the opposite seems to be true with most low and medium risk offenders], and literature suggesting that such sex crimes do not taper off with advanced age (as do most crimes), smart sentencing would seek the maximum consecutive prison sentence and probably arrive in the vicinity of over 500 months in prison.
    Sentencing support tools show that for some female drug offenders, even those for whom prison is usually employed because they have multiple prior offenses, parenting classes are apparently more successful than custody for preventing new crimes.  Looking at and considering all of the information about what works on whom and making our best efforts to do that which is most likely to work -- responsibly in light of the nature and the extent of the risk to the public, both in the short and in the long run -- may well call for parenting classes instead of prison.
    Smart sentencing is evidence-based, responsible, and accountable; it employs our best efforts to accomplish what the public expects us to accomplish: crime reduction.  At the same time, it is kinder both to the potential victims smart sentencing will prevent, and to offenders who can be diverted from criminal careers -- who, without smart sentencing, would be repeatedly cycled through the criminal justice system and punished with no benefit to them or to the community their behavior offends. return to top of page
 

Q: Can sentencing be "smart" without technology?
A:   Yes.  Sentencing support technology is an important means by which to get useful information to the process, but its ultimate purpose is to get the process to employ best efforts to achieve public safety and any other relevant and legitimate purpose of sentencing.  The Oregon Judicial Department has recently revised our Judges' Criminal Law Bench Book (which is available on the Supreme Court Library's web page) to include an expanded chapter on sentencing.  The first thirty pages or so of that chapter (starting on page 727) are devoted to practical tips to assist in making informed and effective sentencing choices.  For example, if a probation officer or district attorney is urging imprisonment to ensure that a repeat low-level offender finally gets drug treatment, the tip is that we need to gather the information about the actual availability to the offender in question of drug treatment in custody.  Testing, challenging, and questioning routine assumptions is smarter sentencing than sentencing based on guesswork and empty assumptions. return to top of page

Q: Do you want computers to make sentencing decisions instead of judges?
A:     No. The computer simply gives the decision maker vastly improved access to important information - such as how similar offenders have fared after being sent to or serving sanctions or completing programs available for this offender. This helps the decision maker make more informed decisions about how best to try to reduce this offender's future criminal behavior, but it does not even address the other components of sentencing. And even within the goal of reducing this offender's criminal behavior, there is much left to advocacy, analysis, and individual assessment.
     Sentencing support tools do not purport to display causation or to factor in such important variables as offender risk and need.  The tools' primary purpose is to raise the "what works?" question by giving easy access to outcomes for similar offenders sentenced for similar crimes -- to focus sentencing attention on outcomes.  They will have served their primary purpose when the routine discussion becomes  what is most likely to work and why, with the tools serving only their secondary function: providing accurate data on outcomes correlated with similar offenders and crimes -- so the participants can address why that information is or is not predictive for the offender in question.
    At a recent law class, a student asked what the "end stage" of this effort would look like.  Of course, the future is impossible to predict, but I envision a criminal justice system that approaches sentencing much as modern public health agencies approach disease -- cognizant of the wide range of human behaviors, qualified by the limitations of human bureaucracy and finite resources, but competently and rationally addressing a persistent and changing challenge to the quality of life in our communities with responsible use of available information and appropriate techniques. return to top of page

Q: Haven't recent ballot measures and statutes eliminated the role of discretion?
A:     No. By far the most numerous cases in our system are less serious crimes which are not covered by mandatory minimum sentences or even by the sentencing guidelines that apply to felonies.  The guidelines allow for departure to serve compelling and substantial interests, but even for those cases subject to mandatory minimum sentences, judges have tremendous discretion to impose greater sanctions than those required. They frequently have discretion whether to impose consecutive or concurrent sentences. And in the great majority of cases, judges have wide discretion whether to impose incarceration or probation with conditions, to structure conditions of probation and to recommend programs during incarceration, and whether and when to revoke probation or to continue probation with modified conditions.   Without sentencing support technology, all of these decisions are commonly based on ideology, philosophy, or faith; they are almost never made based on (or even with access to) information about how similar offenders have behaved after being subjected to similar choices in the past. return to top of page

Q: How does Blakely v. Washington affect all this? 
A:     Not much. Blakely v. Washington and its progeny have spawned expansive debate about how to accommodate a right to jury trial attached to facts essential to enhanced sentences.  United States v. Booker confirmed the fears of Professor Kevin Reitz and other advocates of a guidelines centric rewrite of the Model Penal Code's sentencing provisions: an easy fix is to deem guidelines merely "advisory."   While guidelines advocates and the other major players scramble to modify sentencing forms to respond to Blakely, each segment of the sentencing debate threatens its own objectives by pursuing a strategy that ignores crime reduction - the major function of sentencing rightly sought by the public. There are other easy fixes - including the simple provision for a jury trial where required, as urged in Justice Stevens' opinion in Blakely. Any response that merely continues the sway of guidelines that are blind to public safety will only resume the tragic cycle of misdirected sentencing decisions, avoidable victimizations, responding restrictions on sentencing discretion and expanding demand for draconian sentencing.  But any fix will probably either restore or expand sentencing discretion -- even without a fix, the only discretion lost is that to depart above a discretionary range without first affording the defendant a right to a jury trial on facts critical to sentence enhancements.  So all of the argument for using whatever discretion remains to achieve best efforts at crime reduction survives Blakely.return to top of page

Q.  How can you be so disloyal to your colleagues by blaming them for crime? 
A:    I should start by revealing that I have enormous respect for my fellow judges.  In my many years of working with them (I was appointed to the bench full time in March, 1990, and practiced before the same courts for 16 years before that -- and another four in California), I have learned that performing quality, impartial, and valuable work in our public function is the ambition of virtually all judges I know.  Many of my colleagues go the extra mile for public purposes in addition to the requirements of the cases on which we sit.  Many volunteer many hours to the administration of justice – writing articles and presenting legal education courses for law students and lawyers, serving on many committees devoted to improving the law itself or the processes of the courts, and speaking to groups of citizens who want to know about the courts and the law.  Through such efforts, and in cooperation with bar associations, legislative committees, and public and private agencies, judges I know work to reduce child abuse and domestic violence, improve the foster care and adoption systems, reduce barriers to people who cannot afford attorneys to represent them in legal proceedings, help parties settle existing cases without the emotional and fiscal expense of trial, support recovery from addiction, assist victims impact panels in their attempts to convince drunk drivers not to repeat their mistakes, and help women offenders earn back custody of their children.  A good number of judges go to great lengths to make treatment courts work, and to follow up with probationers to improve their chances of leading productive lives.  I am proud to say that my colleagues voluntarily created a fund in response to our recent budget crisis to alleviate the hardship of court staff whose mandatory furlough days put them in financial crisis.  For these and many similar reasons, I consider myself fortunate indeed to be a part of such a group of colleagues. 
     But it is in the nature of and human beings and social processes that good people, pursuing normal routines with skill and the best of intentions, sometimes persist in behaviors that tolerate or even cause harm -- yet resist scrutiny, measurement, or change that might imply that they are somehow responsible for outcomes they surely never sought.  It's easy to see this scenario in other times and places, but often difficult for any profession to contemplate such notions in their own time and place.  This institutional denial explained the expulsion of  Ignaz Philipp Semmelweis from Vienna because he learned that having his medical students wash their hands between dissections and deliveries drastically reduced maternal mortality.  His hospital employer simply couldn’t accept the message that business as usual had been killing patients.  It was not that his colleagues were not committed to the ideals of their medical calling, but that human institutions have enormous capacity for denial. 
     More recently, the Institute for Healthcare Improvement tackled the problem that was costing the loss of hundreds of thousands of lives through mistakes in hospitals (at a rate of 98,000 per year – plus millions of non fatal mistakes), and eventually produced tremendous success in reducing such errors  through procedural changes.  The present point is that neither the certainty of professionals that they are well intended and competent, nor the outrage of their defenders when others suggest that the status quo is killing people, protects those who suffer unintended harm at their hands nor answers the need for change. 
     Mainstream sentencing is governed by plea bargains that generate 90-95% of sentences with little or no judicial intervention.  Outside the treatment courts, we judges permit punishment for its own sake to dominate the discussion, and we allow the enormously elastic goal of proportional punishment alone to constitute adequate performance – largely ignoring, but surely never measuring, crime reduction or any other social objective. 
     Most offenders sentenced for most crimes offend again, and most horrible crimes are committed by offenders repeatedly sentenced in the past with no responsible effort to reduce the offender's criminal behavior.  Because we're not doing our best to focus sentencing on crime reduction -- because we're giving punishment per se complete control of the process without making any effort to ensure that even the legitimate purposes of punishment are served -- we are tolerating tremendous unnecessary brutality: avoidable victimizations smarter sentencing would have prevented, and often draconian sentences imposed on some offenders with no hope of serving any social purpose. 
     The tragedy is that good people with good intentions can produce such results.  The compounding tragedy is that judges sometimes bristle at the very goal of "improvement" because of its implications for current performance, and resist performance measures on any number of grounds if the performance measures touch on our impact on harm reduction.  The final irony is that we often expect far more insight and accountability from the offenders we sentence than we are willing to adopt for ourselves. 
     Of course, changing the status quo requires avoiding rather than confronting the denial.  Those who want to change things need to convince others of the urgency of the need for change, but once there is a sufficient number of judges and others committed to change, the trick is finding tactics that work without conveying criticism or blame, but instead tap the enthusiasm for contributing to the public good that motivates so many in our role 
      Unfortunately, there are a substantial minority of judges who actually argue that crime reduction is not our job -- but that's another story. return to top of page

Q: What happened at the ALI annual meeting in May, 2007? 
A:    The short answer is that the motions were all defeated, a not unexpected outcome given the deference that the process effectively gives to the reporter in projects, and the cultural depth of acceptance of retribution regardless of function.  It is hard to untangle the process from the result.  There was substantial testimony from people unknown to me in favor of the first motion (to restrict the role of retribution both by proportionality and by some plausible connection to any social purpose served by punishment), but the discussion soon became clouded by the reporter’s suggestion (whether or not intentional) that my motion would add roles for retribution rather than limit them as compared to his draft.  I suspect that many who voted were at least confused and at worst assumed that proportional severity serves only to limit utilitarian functions in the reporter’s draft.  Critically, that misses the point of the debate.  Under the draft, which is now set virtually in stone by the vote at the annual meeting, a sentence may properly punish for its own sake regardless of any pursuit of any purpose as long as the punishment is “proportional” by just desert standards – meaning within the range of presumptive sentences promulgated by a sentencing commission.  The point of the motion was to limit punishment per se to occasions on which it plausibly serves to promote public values in the specifics articulated in the motion: 
      The struggle for a more rational sentencing structure and practice continues.  There are provisions in the MPC yet to be refined that have a role in this.  And there are efforts underway at the national level that intend to shift the culture of sentencing away from ordered just deserts to evidence based practices.   The MPC draft will squander an opportunity to provide real leadership, but may yet be of some assistance in the right direction.return to top of page

Q: Is it really a good idea to have judges try to predict which sentences will work best to reduce criminal behavior?
A:    In my quest, I have occasionally encountered the view that judges have no business trying to adjust sentences based on their impact on the offender's criminal behavior.  I think it fair to say that on all occasions, the majority of those who heard such comments were as astonished as I was, but the view that sentencing is about something other than reducing criminal behavior is actually consistent with the almost universal behavior of participants in the sentencing process, so this question merits a response.  First, the data from Multnomah County's sentencing support technology clearly shows (as does an enormous quantity of criminological and correctional research) that different dispositions have different correlations with future criminal conduct -- that some sentencing decisions are in fact at least not preventing future victimizations by those sentenced as well as other decisions.  It is obviously foreseeable that paying attention to what works on whom and what does not can make a difference in the likelihood of future victimizations, and in most other areas of the law, consciously ignoring a substantial risk of harm is equated with recklessness and the potential for civil or criminal liability.  In other words, claiming that the public safety consequences of our decisions are not are our responsibility is simply untenable; we produce those consequences whether or not we cling to denial.  And in any logical world, continuing not to try to do a better job is obviously far more risky than making our best effort to produce better outcomes.
    Second, at least in Oregon, the 1996 amendment to Article I, Section 15 of the Oregon Constitution (quoted in 1997 Oregon Judicial Conference #1), and 1997 HB 2229 have settled the issue: public safety is a goal of sentencing.
    Third, those who argue against sentencing based on prediction of future dangerousness tout the "false positives" such predictions often produce or, from the other end of the spectrum of those suspicious of data-based sentencing, the "false negative" represented by an offender who is released on a prediction of safety which proves tragically erroneous.  This misses many points: arguing that sentencing should not even try to focus on crime reduction avoids the measure of success -- it does not improve success.  Pretending we are sentencing for some other purpose than public safety, or pretending that we are not attempting to predict future dangerousness when we incarcerate for the longest terms those with the worst crimes and the worst records, does not change the percentage of people we put in jail for longer or shorter terms than necessary for public safety.  Following this path doesn't improve our rate of false positives or false negatives based on public safety, it just justifies these errors by pretending that the purpose isn't public safety to begin with.  We should do better at using sentences intelligently if we make the effort than if we do not.  And as long as the offender has been convicted of a crime and the sentence is not otherwise disproportionately severe, that a violent offender only has a one in three chance of assaulting, raping, or murdering another victim hardly argues against his incapacitation.  As to the risk of "false negatives," it is inevitable throughout the process that the length of terms of incarceration for most offenders [the vast, vast majority] will be quite limited.  Whether it is a pretrial release decision, a decision whether to render sentences consecutive or concurrent, whether to revoke a probation, or whether and under what circumstances a correctional authority ought to allow an inmate some reduction in the duration or nature of confinement, we surely are more likely to produce more false negatives if we decide based on something other than risk prediction; surely the path to public safety lies in making these decisions as knowledgeably as we can.  Insisting that offenders not be released at all, that they be imprisoned in every case as long as the law allows, and that there be no offenders whose term is somehow ameliorated at the far end does not serve public safety as long as the reality includes such releases, sentences, and ameliorations. And almost 80% of our cases are misdemeanors, and very very few produce the maximum one year incarceration -- our jails are typically "matrixing" people simply because we have no where near enough beds to lock everyone up who has been sentenced for as long as their sentence!  As to felonies, our guidelines restrict actual custody time for the majority of cases to even less than is theoretically available for a misdemeanor; even by departure, the lower felonies cannot even exceed a six month term.  The vast majority  of felons in prison will return to their communities.  It is only when we get to the most serious crimes and criminal histories that significantly longer that a year an a half to a few years is available for incarceration.   To see the Oregon guidelines, click here.return to top of page 

Q:  Isn't unrealistic to expect lawyers who practice criminal law to handle data and research about what works?
A:    No.  The underlying concern is that lawyers, particularly criminal law practitioners, are accustomed to a high volume, fast-paced, flow of cases and hearings and are ill-equipped to master data or research about what works and to be useful to the court in the style of practice to which they've become accustomed.  There are, of course, several responses:
    Although express attention to what works is rare indeed within the style of criminal justice to which we've become so unfortunately accustomed, it is not completely missing.  In the wide range of occasions and approaches to sentencing, there are certainly those that now involve a lawyer attempting to inject what we allegedly know about what works into a hearing in an attempt to persuade the judge in favor of that lawyer's goal. At the high volume end of the spectrum, a defense attorney may assert that keeping the defendant employed will lower his risk of recidivism so that any jail sentence ought to be brief or served on weekends; the prosecutor may respond that the defendant's criminal history demonstrates that employment hasn't kept this offender from offending, so that a long sentence will at least keep him from offending while his is locked up. The judge might explore whether the need for tight supervision and the general value of continued employment militates in favor of work release.  It is a sad measure of how far we are from responsible pursuit of crime reduction that increasing the frequency of even this level of debate would be a profound improvement in sentencing culture, and likely productive of improved sentencing outcomes.  And encouraging attorneys by the frequency of this discussion, and the interest it receives on the part of the sentencing judge, gradually to increase their fluency in the data and literature which improves the accuracy of either assertion as applied to a given offender, crime, criminal history (and, for that matter, cluster of criminogenic factors) does not require a sea change in skills, intelligence, or preparation.  For high volume cases, this would require no greater effort or time on the part of practitioners and courts than the constant infiltration of new information about laws affecting suppression motions, appellate decisions and legislation affecting common crimes, and the loss, modification, or addition of sentencing options -- including, by the way, information about how long a given offender is likely to serve if sent to a "jail" sentence.  The nature of the beast is that all of us have to stay up to date on a wide range of matters to retain minimal competence at this task, and that gradually including and then upgrading competence about what works best on which offenders under which circumstances would hardly overtax the people or the system involved.
    As a practical matter, it is at least immediately unrealistic to expect high volume criminal practice attorneys to carry around notebooks (or the electronic equivalent) of studies, articles, and literature reviews to cite to each other and to the court on routine sentencing debates.  But attorneys and courts are quite adept (often too much so) at capturing, distilling, and exploiting relatively common notions that are useful for the debate in the high volume cases.  Literature and study based notions are already there:  full time employment or school favors crime reduction for many offenders; domestic violence offenders' potential lethality can be roughly guessed based on factors known to most practitioners (especially, whether the victim is attempting to get away); addicted offenders with heavy use patterns are unlikely to get anywhere for our benefit or for theirs until and unless their addiction is addressed; jail keeps people from committing crimes on the outside while they are in the inside.  These notions, though hardly a substitute for thorough criminogenic assessment and risk prediction, are also far from useless.  Again, just increasing the occasions on which wielding them may affect the result because of the court's focus will encourage the gradual improvement in their wielder's knowledge and usefulness in informing an outcome productive of public safety.
    In Multnomah County, our sentencing support tools generate data about what has or has not correlated with crime reduction by an offender cohort similar to any offender considered for sentencing, when sentenced for a similar crime.  These tools are available to judges and attorneys; using them has a relatively slight learning curve.  A moderately experienced user can come up with some  useful information for most offenders in under five minutes.  I regularly generate charts during most sentencing hearings; the ensuing discussion is what may take a few minutes of additional time.  We do not turn ten minute sentencing hearings into half day proceedings; we may push some from ten to twenty minutes if the issues are difficult.  Lawyers seem to have no difficulty analyzing and challenging the information and its relevance to the decision at hand.
    At the higher end of the spectrum, lawyers are already expected to be able to handle psychological evaluations, criminogenic risk assessment, and similar issues with competence in the death penalty and dangerous offender (and sexually dangerous offender) contexts.  This is a function they take to just fine -- as do their brethren on the civil side who come with a legal training and, typically, a liberal arts undergraduate degree, to expert disputes of the highest sophistication involving virtually every science and literature known to the wide world of malpractice, environmental, product liability, and intellectual property litigation.  The same judge you may think unsuited for risk prediction in the criminal area may well be sitting on and deciding cases on the civil side involving whether there is enough basis in the literature to let a jury hear some expert give an opinion about a medical procedure, the adequacy of error detection mechanisms to cope with the products of incineration of Sarin chemical agent, the adequacy of file encryption technology afforded under a contract for computer security services, or the propriety of treating depression with a particular mix of drugs and electro-convulsive therapy.  Indeed, the same judge may already be making risk prediction decisions based on the literature of psychopathy in dangerous offender proceedings.  And the lawyers in all of these proceedings are expected to be competent (and quite usually are).
     By the way, it is among the greatest triumphs of social evolution that court proceedings and the adversary process is fully capable of taking the most renowned and published experts in a field, pitting them against each other through examination and cross examination by counsel, and testing their wisdom in this crucible to produce a more reliable or at least a more practical application of their learning than they would ever accomplish in the journals that otherwise quietly contain their debate.  This process is not without its faults, but as many have said repeatedly, it is the best we've yet devised.  It is also fully capable of producing the best results from the debates among academics and others about what works best to reduce crimes among which offenders.
    To return more closely to the topic at hand, I think it clear that we don't need notebooks full of journal copies before high volume practitioners can be more often useful and informed about what works best in the most common cases, and that lawyers and the system already demonstrate their capacity for journals, studies, literature, and experts on occasion at the other end of the spectrum represented by dangerous offender and death penalty proceedings.  That smart sentencing would encourage more information into the process probably means that more of those involved would indeed gradually have to learn more about what works as time goes on.  There is a lot of hot air and spouting that can be displaced without any loss to anyone (except, perhaps, those who savor the tragic comedy of claptrap followed by recidivism that smarter sentencing would have avoided).  And it would be no great disruption or imposition to add to the culture of practitioners and judges, which now includes notebooks and continuing education hours covering such topics as search and seizure law, notebooks and continuing education hours on best sentencing practices.  (Judges and lawyers already have some opportunities for hours on best practices and best advocacy techniques related to sentencing hearings)
     Here, as in so many areas of objection, the bottom line is that the system could well do a better job than it now does without violating its paradigms, including those of personnel and training; that we now so rarely attempt responsible smart sentencing that there is much room for public safety improvement without significant increases (but only minor redirections) of time and energy among practitioners and the courts; and that even if and when it takes some real time and energy from our participants both the reduction in avoidable victimizations and the reduction in repetitive drains on recourses by recidivist offenders are more than enough to justify those increases. return to top of page 

Q: Smart sentencing makes a lot of sense; what are the arguments against it?
A:   It's been surprisingly hard to ferret out real arguments against the notion that we should do a more responsible job of crime reduction when sentencing people by trying to focus on what works on which offenders.  Most of the contentions are either withheld, or implied indirectly in the questions I have attempted to answer elsewhere on this page.  I have tried to elicit some direct answers from the author of the proposed revisions to the Model Penal Code, Prof. Kevin Reitz.  His proposal represents a profound retreat from the present MPC sentencing provision which emphasizes public safety, and I have published substantial comments arguing that this is the wrong decision.  Comments on the Model Penal Code: Sentencing, Preliminary Draft No. 1, 30 American Journal of Criminal Law 135 (2003) [University of Texas Law School], [Available on WestLaw at 30 AMJCRL 135.]  After a long telephone conference with Prof. Reitz, I sent him a letter outlining what I understood to be our agreements and our disagreements, and asking him for confirmation or explanation as to where and why we disagree.  A copy of that letter (sent in October, 2003), is here:   Prof. Reitz has never responded.
      I have elicited some direct contentions from various significant actors in the criminal justice field in Oregon.  Here's what I have learned so far:
      From a principal author of Oregon’s Sentencing Guidelines (and past Attorney General of Oregon): 
     1) There are other purposes of sentencing besides preventing recidivism, including punishment.
     A: Of course there are.  But the recidivism statistics make it clear that we are doing an awful job of reducing recidivism.  We don’t have to reach agreement as to the relative importance of punishment per se, restitution, or general deterrence to agree that we aren’t doing much to attempt crime reduction, and that we should be able to do a much better job if we made a responsible attempt. And smart sentencing assesses incapacitation, specific deterrence, and “rehabilitation” in looking for what works best on which offenders.
     2) Wouldn’t focusing on what works on which offenders result in disparate treatment, and wouldn’t that undermine the benefit of the guidelines in avoiding unequal treatment?
     A: First, Oregon law already recognizes that differences in amenability to rehabilitation justifies different responses to offenders.  ORS 161.025(1)(a), derived from the existing (1962) Model Penal Code, declares the purposes of the Criminal Code as including "To prescribe penalties which are proportionate to the seriousness of offenses and which permit recognition of differences in rehabilitation possibilities among individual offenders."  Equal treatment means that people similarly situated should be similarly treated.  There is no reason doing more dependably what works on the offenders it works on is inconsistent with similar treatment for similarly situated offenders, unless of course one community has something that will prevent recidivism by a given offender, when another community with an identical offender lacks that sanction, treatment, or other disposition.  I’d argue that the availability of such a sanction, treatment, or other disposition is a difference that makes the offenders not equally situated, so that the disposition is not inconsistent with similar treatment of similarly situated offenders.  In any event, I have not yet encountered anyone who would actually argue that we should deprive the public of the protection of the sanction, treatment, or other disposition that works just because we don’t have it in every community and can’t apply it to all offenders.
    A: Second, the guidelines’ achievement of equal treatment is far from ideal.  We maintain the illusion of equal treatment by refusing to acknowledge differences in offenders, criminal histories, crimes, or available sanctions to insist that we are treating equal offenders equally.  Thus a one-time remorseful assault perpetrator ends up with the same presumptive sentence range as someone who committed an identical crime with the same criminal history – but is a psychopath.  And the economic categories that separate the crime seriousness ratings for property crimes offend even just deserts – why should the theft of a Rolls Royce whose owner rarely sees it in his stable of luxury cars be punished more severely than the theft of a clunker that provides a single parent her only means of transportation?  Similarly, the guidelines ignore the tremendous variety that can exist among equally rated criminal histories (two prior person felonies by a psychopath who intended the harm he caused will be treated exactly the same as two prior person felonies by a remorseful bar brawler who barely lost self-defense arguments because his response was more than reasonable).  Of course, we can attempt to accommodate these variations with departures, but departures are exceptions to the equal treatment invoked to defend the guidelines against improvement for crime reduction.
    A: Third, given our consistently poor crime reduction performance, is it even clear that consistency is a good thing?
    From a prominent Oregon District Attorney:
   1) Some fear that smart sentencing would displace other purposes (such as punishment).
   A.  Again, to be smarter in sentencing we don’t have to abandon all objectives other than crime reduction.  Let’s be direct here: Crime reduction is usually ignored, sometimes assumed without any more support than mere ideology or guesswork, but almost never responsibly pursued in sentencing.  Whenever we do something instead of smart sentencing we are reducing the chances that we will reduce or prevent future criminal conduct by the offender. Our sentences will have a public safety outcome regardless of whether we do our best to achieve crime reduction; any argument to do something else amounts to an argument not to do our best.  To that extent, we are complicit in avoidable victimizations.
   2) Sentencing based on data and research would interfere with the flow of plea bargains.
   A.  This has at least two components: administrative efficiency and control of plea bargain outcomes.  With respect to administrative efficiency (how much time it takes to negotiate and dispose of a case), revising some of the components of the process with smart sentencing does not have to make things take longer. Just as most cases are routinized around just deserts and resources, with a discount for witness difficulties or a search and seizure vulnerable to a motion to suppress, we could make a lot of progress more rigorously routinizing cases around crime reduction.  Moreover, doing a better job of avoiding future victimizations is worth some additional time in processing cases.  Finally, the biggest waste of time and resources is the sentence that fails to divert an offender from criminal behavior which again taxes the time and resources of the criminal justice system. 304 of the 372 offenders jailed for drugs in July 2000 in Portland were in the jail within the prior year on some other occasion.
    With respect to the issue of the prosecutor’s control of plea bargaining outcomes – both with respect to the defense and to the court’s role, inserting smart sentencing has no inherent impact on that control.  In other words, were the prosecutor to negotiate around crime reduction based on data and research, the issues whether the judge would avoid tinkering with the deal or the defense avoid ameliorating it when in front of the judge vary not at all from the present situation.  That situation includes that the attorneys are free to employ a “contract plea” to give the judge the limited choice of accepting and enforcing the deal as is or rejecting the plea.  That situation also includes that some judges are far more reluctant to tinker with a non-contract plea than others.  Indeed, the more “what works” catches on with judges, the more likely they are to tinker with a plea bargain that was made without regard to what works.  In other words, if the parties present the judge with a deal justified and described in terms of what is most likely to reduce crime, a “smart sentencing” judge is more likely to leave it alone than in the case of a deal with no such justification or explanation.  I avoid in this analysis any attempt to resolve the question whether and to what extent judges should be potted plants when it comes to plea bargains.
    From an activist crime victim:  Research and data are overwhelmingly flawed.
    A: Yes, much research is flawed, and few data are completely “clean” of errors.  But here’s the thing: we must always keep the choices in mind – do we endeavor to improve, select, and analyze the data and research, or do we choose to rely instead upon the status quo of unformed, essentially a priori debates about what an “appropriate sentence” should be – mixing aggravation, mitigation, and anything else that comes to mind?  The irony here is that in the face of imperfect data and research we abandon all attempts at accuracy or evidence-based decisions, and support the opposite.
    If we could achieve candor, we’d see that much of this argument is really based on a bias for or against punishment, for or against prison.  Just as the incarcerationists bemoan “false positives” of “preventive detention” (their disparaging name for incapacitation based on risk assessments) to argue against using jail or prison for crime reduction, others who criticize data and research for its imperfections ultimately fear that it will reduce the use of incarceration because they are persuaded that jail or prison is either deserved or will best produce public safety.  Even beyond the possibility that retribution and crime reduction can be at odds in crafting a sentence, the problem with this is that the choice between prison and alternatives, or among terms of jail or imprisonment, in fact has different public safety outcomes depending on the offenders (and, often, the conditions of incarceration).  Always insisting that jail or prison is the answer, particularly as to crimes for which only relatively short periods of incarceration are available, is just as dangerous as any other faith-based approach to sentencing.  We will make more wrong decisions without good information than with it; we can as easily cause avoidable victimizations in the long run by choosing jail as by choosing probation – again, at least at the level of crimes for which jail or prison terms are relatively short.
   To test the notion that faith for or against incarceration is the real agenda of many who resist smart sentencing based on the imperfection of research, see how the rigor with which they assail research with which they disagree compares with the rigor with which they’ve scrutinized any research or data that supports their favored disposition.
   This is not to say that we should be sloppy – far from it.  We must be rigorous in seeking best efforts at crime reduction.  A good way to do that is to make the issue of what works a subject of competent, informed, and vigorous advocacy in sentencing and probation hearings.
     From an academic turned appellate public defender:  I don’t see how the guidelines could be adjusted to incorporate crime reduction.
     A: It’s always amazing to me how powerfully habits and culture limit our ability to see the obvious.  Of course, it would be a formidable undertaking to structure a set of guidelines that incorporates all we know and can learn about criminogenic and risk factors, and all the other variables that might be part of a calculus that made the presumptive sentence more likely to be predictive of public safety than under the present guidelines.  But we’re capable of formidable undertakings when the motivation is sufficient.  Nanotechnology, molecular biology, computer science, and modern medicine all regularly employ concepts far harder to grasp than the two-dimensional sentencing guideline grid.  But we don’t have to undertake a perfect product to make tremendous improvement in the existing approach, and individualization will always at least potentially improve any specific sentencing.  For starters, we could use any of the standard violence prediction instruments to make a longer period of incarceration than currently prescribed by the guidelines presumptive for violent offenders with a high score. (For a comparison of several approaches to violence and risk prediction, check this link). And one state, Virginia, has actually incorporated risk assessment into its sentencing guidelines after careful and responsible study. See Brian J. Strom, Matthew Kleiman, Fred Cheesman, II, Randall M. Hansen, Neal B. Kauder, Offender Risk Assessment in Virginia - A Three-Stage Evaluation: Process of Sentencing Reform, Empirical Study of Diversion and Recidivism, Benefit-Cost Analysis (The National Center for State Courts and the Virginia Criminal Sentencing Commission 2002), available at http://www.vcsc.state.va.us/risk_off_rpt.pdf. 
     Critics from the anti-incarceration side will again invoke false positives, and from the pro-incarceration side we will hear of offenders whose sentences weren’t enhanced who recidivated in a horrible way.  But the real test is whether we do a better job of crime reduction with or without this dimension.  How many victimizations are occurring now because we don’t do our best at crime reduction?  As to false positives, we are already working within the realm of incarceration not disproportionate for the offense already committed; assessing risk rather than demanding certainty in choosing where within that range to place a given offender is entirely appropriate. [For dangerous offender schemes that provide a longer period of incarceration than the maximum otherwise available for the latest felony, I agree with the critics that rigor is required (and now imposed by Apprendi v. New Jersey, 530 U.S. 466 (2000)), but I cannot agree that we should not consider longer sentences for such offenders, that we cannot impose such sentences unless we can achieve risk prediction with certainty, or that it is somehow better to reject data and research than to attempt to make our best use of it with knowledge of its limitations.]  Guidelines proponents do not avoid false positives by rejecting risk assessment -- they compound and rename them.
      From another Oregon trial judge:  Won’t inviting argument and evidence about what works best require lengthy court hearings and testimony?
     A: Replacing some of the components of the process with smart sentencing does not have to make things take longer.  It doesn’t necessarily take longer to argue what changes will or won’t reduce an offender’s criminal behavior and what dispositions are most likely to help [ranging from incapacitation to treatment] than it now takes to argue aggravation and mitigation, heinousness of the defendant's behavior and depravity of his childhood.  Moreover, even when we chose to hear from experts through testimony, doing a better job of avoiding future victimizations is well worth some additional time in processing cases.  Finally, the biggest waste of time and resources is the sentence that fails to divert an offender from criminal behavior which again taxes the time and resources of the criminal justice system. 302 of the 374 offenders jailed for drugs in July 2000 in Portland were in the jail within the prior year on some other occasion.
    For all of the above: None of these arguments change some basic facts.  First, every sentencing decision has a public safety outcome.  Second, although it is less threatening to avoid accepting responsibility for those outcomes than to pursue “just punishment,” “equal treatment,” or an outcome whose propriety is based on its proponent’s ideology, anything other than responsible pursuit of crime reduction inevitably causes harm that could be avoided.  And, whether or not we like it or accept it, we are responsible for the outcomes of our decisions. return to top of page

Q: Are you trying to replace probation officers?
A:     No. This technology is designed to be available to probation officers, correction counselors, pretrial release workers, and all others in the correctional system who make decisions for offenders in the system. Judges continue to exercise major control at the beginning of the correctional phase of a sentence - whether to imprison or to supervise in the community is a decision which in most cases is up to the judge to make. And it is the judge who weighs the safety risks which militate for and against incapacitation or community supervision, as well as the additional components of sentencing. Sentencing support technology does nothing to shift the distribution of correctional responsibility between judicial and correctional officers, but offers both enormously improved information which they need to do their jobs more effectively.
     An extremely positive recent development is that the Multnomah County probation department (Department of Community Justice) and the judges who regularly handle criminal cases have embarked on a project to transform the role of the probation officer in probation violation reports and hearings.  Instead of coming to court looking for vindication or disappointment in the threat of punishment to motivate a probationer, probation officers would become the advocate for "what works" in the courtroom, writing reports and arguing for outcomes based on expertise about existing alternatives and sanctions, the offender in question, and the literature of what works.  Sentencing support tools are expected to be part of this effort [probation officers who write presentence reports are being trained on the tools first]. return to top of page

Q: Is this all an argument not to use jail or prison?
A:    No.  As I note elsewhere, we have pretty good information that jail and prison (at least without unusually good programming) do not generally reduce recidivism (and may even increase it) after an offender is released, and the opponents of incarceration often cite the literature to argue that community based treatment is almost always the best disposition.  But the only fair comparison is the impact in terms of crime reduction from the time the sentence is imposed, including any period of incapacitation.  After all, incapacitation is generally the most effective thing we can do to reduce crime in the short run -- or as long as the incapacitation continues.  The trick is to compare (legally available) dispositions with and without jail or prison side by side and examine which serves crime reduction best in the long run. Our tools show that for some cohorts short or no incapacitation correlates more highly with crime reduction, and for others longer jail or prison sentences correlate more highly with crime reduction.  For those who fear that this approach inherently favors jail or prison, rest assured that for most minor offenders, it does not -- even though jail or prison presumably prevents recidivism while the offender is inside.  Again, to be responsible about public safety, we must insist at least that we know what has worked best in the past without some bias for or against jail or community based dispositions. return to top

Q: Aren't you shifting responsibility from criminals to the courts when you blame the courts (or criminal justice) for recidivism?
A:     Emphasizing the responsibility of courts (or of criminal justice) is necessary because for so long we have avoided taking responsibility for the public safety outcomes of sentencing on the theory that because criminals are to blame for committing crime, we can't be blamed for not reducing it. Taking responsibility for doing a better job with sentencing in no way diminishes the responsibility of criminals. There are all sorts of reasons people commit crimes: bad character, bad choices, bad luck, bad childhoods, bad values -- select any or all that best capture "criminals" for you. My point is that we do various things to people who are criminals with bad character, choices, luck -- etc. -- and that some of the things we do to some of them "work" better than some of the things we do to others of them. To the extent that we don't make our best effort to do to each that which is most likely to prevent the next crime, we are responsible and (if it helps) worthy of blame -- and it doesn't reduce, explain, or mitigate our responsibility (or blame) that the offender is also to blame. In short, without making a responsible effort to prevent future victimizations, we are not performing our public responsibilities as we should -- and that reality in no way diminishes the responsibility of criminals for their crimes; that criminals are responsible for their crimes in no way diminishes our responsibility to do that which is most likely to prevent crimes in the future at their hands. My point isn't that we shouldn't blame criminals for crimes; it is that blaming them does not excuse our failures to prevent future crime with our sentencing. Our legacy is to celebrate their blame to avoid our responsibility -- and that's just wrong. return to top of page

Q: Are you saying retribution and general deterrence should be discarded as objectives of sentencing?
A:   What works to reduce crime is not the only purpose of sentencing.  But first and foremost, we should do a responsible job of addressing which disposition (within those available) is most likely to reduce criminal behavior.  In assaultive or other dangerous crimes, or even in compulsive property crimes, it may be that separation/incapacitation is indeed the best result because we cannot sufficiently assure ourselves or the public that non-incarcerative options will eliminate the risk. For many minor crimes, at least with offenders with relatively few priors, treatment, counseling, or other dispositions properly designed and delivered, will be the best way to protect the public.  Only after responsibly addressing the crime reduction objective should we start going through the other possible purposes of sentencing.  When there are victims whose recovery may depend upon the satisfaction of seeing punishment delivered, and if the punishment necessary to accomplish their satisfaction is both legally available and not disproportionate to the offense, it is certainly appropriate to consider whether punishment is called for beyond (or even instead of) what is most likely to work to reduce crime.  In cases in which what works is a close call, such a consideration may push the balance in favor of a more punitive disposition than otherwise; in most cases, what works will already satisfy all other present objectives of punishment.  In relatively extreme cases (such as a negligent homicide perpetrated by an offender who needs little or no intervention to prevent future crime, with victims' family members who have real needs for "justice"), we may properly consider putting the needs of the victim ahead of the needs of the community in overall crime reduction over the course of an offender's potential career.  And with some crimes, there are vulnerable victims for whom punishment properly serves a therapeutic purpose -- such as sex crimes against children who may feel responsible for the crime (but we need to be careful here -- I try to get input from any treatment provider to avoid compounding the victimization of a child who is suffering guilt for causing the offender's punishment).
   Some victims who come to court to be heard at sentencing end up endorsing "preventing another victimization" as the highest priority, though many are simply (and understandably) angry and hurt.  It is also worth noting that restitution, restorative justice, and compensatory fines may be the best way to serve both the victim and the social need for doing that which most likely prevents future criminal behavior by the offender.  But again, I don't deny that there are occasional (and relatively rare) cases in which serving the victim's needs (or the needs of the family of a deceased victim) may properly call for something more punitive than that which is most likely to work.
    Personally, I don't think the traditional interests in what some western jurisdictions constructively call "denunciation" and  what we include within "retribution" -- or the largely preposterous argument for general deterrence -- should ever justify doing something substantially less likely to work when there is no specific victim (or victim's surviving family) needing a punitive result.  I understand that others will disagree with me here.  But this disagreement is trivial; my point does not require consensus on this issue.  My point comes down to this:
    We must stop enabling the notion that sentencing is all about just deserts and "appropriate punishment" in the sense of moral equivalency. To the extent that we have a ceremonial role (which is presupposed, by the way, by the arguments for denunciation and general deterrence), we judges must stop promoting what doesn't work by ignoring what does.  We should instead use our ceremonial leadership and role to direct the sentencing process overwhelmingly toward what works, but need not abandon the other purposes of punishment to be addressed in the typical case with the same disposition, and in the rare case by rejecting the result that would pertain if we only sought what works to reduce crime. return to top of page

Q: How can you expect providers accurately to gather and report the data you need?
A:   We are not asking providers to gather new information, and we are not relying on their information about their effectiveness. DSS relies on existing (and growing) operational data. What's new is not so much the information we collect but the ability to access and compare the information we already have even though it is spread across multiple databases. Providers used by the criminal justice system already report failures; courts already keep track of pending and closed charges and convictions, and record sentences and conditions imposed and probation violation charges and results; law enforcement already records police contacts and charges; and many of these and other databases record criminal histories and offender characteristics. Because we can link offenders in these information pools with data about what sanctions they have been subjected to (and failed or completed) and what new charges and convictions and police contacts they have generated after completion, we are already in a position to compare programs' and sanctions' graduates performance for definable categories of the population.  As we gather more and more categories of information, such as how offenders performed within jail, prison, or programs (in or out of custody) we will continue to improve our ability to exploit this information to make more accurate predictions about what is most likely to reduce a given offender's risk to the community in both the immediate and long-term future.  But as soon as the technology is available, existing data will enable us to make far more informed decisions than we now make, and should help us do a far better job of diverting offenders from criminal careers. return to top of page

Q: What makes you think this technology will serve public safety?
A:     Sentencing decisions are presently made with virtually no information about how offenders behave after being sanctioned, counseled, or treated by the various dispositions available to us. Most sentencing decisions are based on whim, rumor, folklore, and public relations, with a strong dose of myth. Although the better correctional agencies do their best to study the research which is overflowing from the shelves of academia, even they do not generally have access to actual performance data concerning the programs and sanctions they have to chose from. I submit that we cannot possibly be doing as well by accident as we would with far better information about what seems to work best on which offenders. Advantages are many: we can send offenders likely to benefit from a program (in terms of reduced criminal behavior) to the right program instead of clogging programs with offenders they can't affect; we can stop using programs that don't work on anyone; we can make wiser decisions about investing public dollars in correctional programs; we can do a better job of using incapacitative solutions (typically jail and prison) on the offenders who we can't expect to improve; and we can give programs information about their performance so they can improve their product, their "client" selection, and the basis of their competition with other programs for public dollars. return to top of page

Q: What impact do you expect to have on recidivism?
A:     In the foreseeable future, we can expect that a substantial percentage of the correctional population will continue in its criminal behavior interrupted only to the extent to which we incapacitate offenders and they show signs of aging. Some other substantial proportion of that population, however, is certainly susceptible to more improvement than we now make by accident. Once we understand that our approach to crime must transcend the focus on individual cases and appreciate that we are managing criminal careers, we can recognize that the attainable goal is to increase our success in diverting offenders from those careers. With few exceptions, the vast majority of crimes are committed by repeaters. Diverting an offender from a criminal career has a multiplier effect on crime reduction; diverting an offender from crime spares many people from future victimization and has long range economic benefits as well. An analogy is health care - where we are learning that wise management of early intervention has long-run benefits for the costs of medical care throughout the patient's life. The health industry is rapidly consuming DSS software for precisely this reason. return to top of page

Q:  Don't we already know what works?
A:    Yes and no.  There's an enormous volume of wisdom in research about what does and doesn't work.  And there is some common sense.  In general, researchers tell us that measured by reduced subsequent criminal behavior, incarceration, scare tactics, regimentation, and punishment ("scared straight," boot camps, and the like) don't "work;" that "shock" probation and parole do not "work." E.g., Oregon Department of Corrections, "What Doesn't Work to Change Behavior";  U.S. Department of Justice, Office of Justice Programs, National Institute of Justice "Research in Brief"  (July 1998).  What does work includes treatment programs that identify and address multiple criminogenic factors with a methodology that is in fact (as opposed to merely labeled) cognitive and behavioral.  E.g., National Mental Health Association, Treatment Works for Youth in the Criminal Justice System; U.S. Department of Justice, Office of Juvenile Justice and Delinquency Prevention Programs, Corrections Rehabilitation and Treatment; Mark Gornik, Moving from Correctional Program to Correctional Strategy: Using Proven Practices to Change Criminal Behavior (U.S. Department of Justice, National Institute of Corrections).  Common sense may not be the best guide on these matters, but it is persuasive (and not contradicted by literature) on this point: offenders usually cannot commit crimes on the outside while they are incarcerated.  Literature seems persuasive on these refinements:  if the object of jail is to change behavior, short sentences, even for repeated violations, work better than longer ones.  And if the object of incarceration is incapacitation, common sense is unrefuted: the longer the offender is incarcerated, the longer the offender is incapacitated.  But if the offender is going to be released, the offender's inclination to reoffend is as likely as not to be enhanced by long incarceration unless incarceration includes effective treatment with community follow-up. E.g, The Effects of Punishment on Recidivism, 7 Research Summary No. 3 (May 2002), Office of the Solicitor General of Canada, (citing Smith, P., Goggin, C., & Gendreau, P. (2002) The effects of prison sentences and intermediate sanctions on recidivism: General effects and individual differences (User Report 2002-01) Ottawa: Solicitor General Canada).  Another settled principle: on many levels, it is overwhelmingly wiser to expend correctional resources on "high risk" offenders than on "low risk" offenders, with "risk" being measured by likelihood to recidivate (regardless of the nature of the criminal conduct). Where among high risk offenders to focus limited resources may reasonably depend on the nature of the risk: it makes sense to give higher priority to preventing violent crimes than to preventing property crimes.
      There are a least three reasons why knowing these things is not enough.  First, we currently run sentencing hearings as though we either don't know these things or don't want to know:  sentencing hearings are not about what works. A major reason for the sentencing support tools is to help direct sentencing hearings (and all that precedes and follows them) to the issue of how best to reduce recidivism.  Second, most of what we "know" is highly generalized.  Most of the studies are about what "works" on offenders in general instead of what "works" on which offenders; they give us average outcomes. Some research does indeed distinguish among risk levels and some focuses on specific categories of crime; when the question is addressed, it is obvious that some things work better on some offenders than on others -- an obvious conclusion for educators who well know that different people learn differently.  Sentencing is and should be the occasion on which to apply what we know in general to what we know about an offender, and to do our best to choose (within applicable limits) that which is most likely to work on the particular offender before us.  Third, there is and has been a steady process of research and learning within corrections for years.  Yet when probation officers come to court, they -- like everyone else -- rarely address what "works."  Even if the officer in question is committed to using best practices, he or she may expect that judges don't welcome an analysis based on what works -- presiding, as we do, in what gives all appearances of a temple of just deserts.  Our challenge is to support and encourage best practices by the probation officers to whom we delegate so many decisions about how to treat offenders, and to encourage them to give us their best wisdom in the courtroom when we hear probation violations (or, in their equivalent in some jurisdictions, conditional sentence violations) or read their sentencing recommendations when we are fortunate enough to have a presentence investigation report at sentencing. return to top of page

Q: Why can't we just exploit the research we already have?
A:     I'm not sure why, but I'm sure we don't.  At the 2002 international Sentencing and Society Conference in Glasgow, Scotland, it was obvious that disciplines having to do with assessing correctional program effectiveness are entirely ignored by those that purport to study sentencing practices.  In brief, the response to popular anger at recidivism, conference participants heard analyses of how "populist punitiveness" is improperly measured by flawed polling practices, exploited by politicians, and inflamed by the media -- but no presenter mentioned that there might be a basis for popular dissatisfaction with our public safety performance.  No input was received or apparently expected from criminologists or corrections expects.  Participants received copies of the latest Annual Report of the British Sentencing Advisory Panel, whose non-binding input is prerequisite to the promulgation of sentencing guideline judgments by the Court of Appeal under the Crime and Disorder Act of 1998.  Incredibly, although the Act expressly contemplates consideration of factors including "the cost of different sentences and their relative effectiveness in preventing re-offending," the Panel apparently never considered using its resources to gather evidence on which sentences might best prevent recidivism, and instead spent its expert energy on public opinion polls. Those skeptical of this assessment of the abyss between centers of sentencing "research" and centers of criminological and correctional research are invited to read the abstracts of papers presented at Glasgow; apart from the presentation reporting on the project subject of this web site, only one paper suggested that it might make sense to try to rationalize sentencing around a new objective: public safety.
   My experience at the 40th annual conference of the Academy of Criminal Justice Sciences in March, 2003, reinforces my impression that academia is unlikely to address how public safety outcomes of sentencing might be improved.  Just as the judicial sentencing process makes no effort to benefit from what academia has learned about what works, academia makes no effort to influence the effectiveness of judicial sentencing behavior.  I left with the image of a remora suckerfish attached to a shark; it has no interest in where the shark is going or why, but can provide a great picture of a very small portion of the shark.  I suppose it would be happy to share its experience with other suckerfish, but would not show up at a conference about how to influence the shark's behavior for the benefit of the underwater community.
        Only once in my experience since taking the bench in 1990 has anyone involved in a sentencing argument invoked any study or data to attempt to convince me that one or another sentencing choice is most likely to reduce an offender's criminal behavior. Academia and the courts rarely talk, in part because so much of the data is specific to the study of some other program under some other circumstances, and in part because academia applauds studies and publication more loudly than social improvement. Academia shuns "activism" for analysis; courts shun responsibility for the outcome of sentencing choices.  When courts realize their responsibility for outcomes, I hope academia will start to talk to us, and that we will start to pay attention.  Meanwhile, the many studies which do exist have a tremendous capacity to be ignored by policy makers as well as by practitioners.
        Another piece of this is that academia does not hold the respect of most line workers in correction or criminal law -- including the great majority of prosecutors, most probation officers, many judges, and even a good portion of defense attorneys.  The problem is that the major thrust of the literature focuses on the ineffectiveness of incarceration and punishment to improve behavior after release. Because most "incarcerationists" tend to ignore the impact of incapacitation upon the criminal behavior of offenders during incarceration, they perplex all who know that locking up some persistent offenders is the obvious choice among the available options.  While "incarcerationists" celebrate the lack of correlation between mass incarceration and reduced crime rates, the "incarcerationists" tout the math: locking up persistent and prolific offenders prevents the multiple crimes they would otherwise commit.  In their turn, the few proponents of incarceration in academia largely ignore the criminogenic effects of most punishment, and refuse to assess the overall impact of punishment over an offender's lifetime.  Our answer to all of this is to insist that we compare incarceration and other responses side by side and by the same measure -- exploiting our ability through these sentencing support tools to see which really best correlates with crime reduction over a given period of time for a given offender.  Our data certainly shows that while both sides of this debate are right in the abstract, public safety is best served by incarceration for some but not for other offenders.  Enlisting the involvement of researchers and academia would also profoundly benefit academia.
return to top of page

Q: How can you evaluate programs with limited access to their internal data?
A:    Although meaningful program evaluation may someday be part of sentencing support technology, it is not an early benefit.  We will start with information that essentially shows whether an offender was sent to a program or to jail, whether he or she "completed" the jail or the program assigned, and whether and how the offender re-entered the criminal justice system.  We can classify "like" offenders based on their available data (criminal histories, age, gender, any recorded risk assessments, etc.), then compare how "similar" offenders have performed after having been sent to or having completed the sanction in question.  Without a much closer look at the details involved, the "programs" are black boxes -- we can look at their outputs and their inputs, but have precious little data about what goes on inside. (A recent exception is growing with the Department of Corrections so that judges, probation officers, correctional personnel, and the sentencing support technology itself will soon have access to tremendous detail about offenders' needs assessments and program performance, in custody and on supervised probation). While judges and others making decisions about offenders certainly should know how "like" offenders perform after being sent to these sanctions, that information is not at all the same as program evaluation.  Many pieces of the puzzle need to be added: for example, was the program getting the offenders it was suited to; did other factors in the circumstances of the offenders' lives enhance or undermine the program's efforts; did other circumstances in the offenders' lives reduce or increase their criminal behavior notwithstanding the program?  But good sentencing support technology will identify disparities in outcomes which should help isolate issues for responsible evaluation, and it will have the beneficial effect of emphasizing future criminal conduct as the ultimate test of program performance, at least from the perspective of public safety. return to top of page

Q: Are you trying to replace research and researchers?
A:     No. This technology will greatly benefit from and will greatly benefit research. The critical factor is to make "what works" regularly addressed and considered (as is the case, for example, with child support guidelines, sentencing guidelines, and any other structure which helps a practitioner predict or a decision maker to reach a result). Once that happens, advocates, practitioners, and providers will create great demand for research to challenge apparent correlations, to learn from them (why one program's 35 year old thieves stop stealing while another's do not, for example), and how to extrapolate and replicate that which makes one program work to another program. And the technology makes data far more accessible for a wide range of traditional (and even for useful) research projects.  In Multnomah County, we've recently encountered three separate national criminal justice assessment projects (evaluating victim-offender mediation, evaluating pretrial release, and evaluating an intensive driving under the influence of intoxicants program); all of the researchers involved were immediately impressed with the potential of sentencing support tools to make their work easier. return to top of page

Q: Does this sentencing support technology have any other benefits for criminal justice?
A:     My focus is making more informed sentencing decisions. Closely related functions are similarly improved: supervision decisions by correctional officers (in and out of custody, on probation and on post-prison supervision), and release decisions by pretrial release workers. The same tools of course help corrections, judicial, and policy making officials determine which programs and sanctions to enhance, modify, replicate or discard. They also support analysis of such resource allocation issues as law enforcement deployment. Just linking information systems and improving accessibility to the data we already collect can have enormous benefits just for improving operational functions. For example, PPDS collects and maintains information about addresses which are subject of neighborhood "drug house activity" complaints. With ready access, probation officers could routinely be alerted whenever an address linked to an offender is also identified as subject to such complaints. Another example: allowing police ready access to NCIC out of state conviction records for persons identified with residences about to be entered under an arrest or search warrant. With expansion, the tool can also help us more intelligently to address such questions as which law enforcement strategies appear most effectively to reduce the incidence of, say, domestic relations calls or street drug activity or fatal alcohol related accidents.
    Another area entirely is the status of criminal justice in our society.  I submit that our failures to accept responsibility for recidivism -- or at least responsibly to seek to reduce it -- have steadily eroded public respect for the courts.  This is evidenced by the wide-spread emergence of sentencing guidelines, ballot measures, three strikes laws, and mandatory sentencing provisions, all of which imply that we judges can't be trusted with much or any discretion.  While most of my colleagues seem to think the erosion of our sentencing discretion is simply evidence of lack of social wisdom, I'm not so sure.  The case can be made that we don't deserve the discretion we've lost since we have done such an abysmal job of recognizing our responsibility to reduce criminal behavior by making well informed sentencing decisions.  In any event, I expect that a major long-run benefit of sentencing support technology is to earn back the discretion we've lost, and that well informed discretion, properly directed, will better serve public safety than either the uninformed and misdirected discretion of the past or the blanket and sometimes draconian measures that have begun to displace it.
     Finally, the "analytics" of sentencing support tools also have potential application to juvenile delinquency, dependency, and family law dispositions, and the planning documents for Oregon's "eCourt" technology update contemplate precisely that breadth of application. return to top of page

Q: Why include ethnicity in the offender profile?
A:      The short answer is that we have yet to learn which differences among offenders correlate with susceptibility (or not) to reduced criminal behavior after an offender is subjected to an available sanction, program, or other disposition.  The more variables we are able to employ, the more we have the potential to learn.
    But this question is usually posed by people wary of "racial profiling" and other invidious discriminations based upon minority status.  The more compelling answers to this question, then, are these:
    First, a good number of programs are designed to address the needs of particular ethnic groups within the criminal justice system -- African American youth at risk of (or already involved in) gang violence; Hispanic offenders who need Spanish speaking program providers; Native American drug and alcohol providers who attempt to address and employ cultural aspects unique to these offenders, and so on.  In order to assess how well these programs perform in reducing recidivism, we must be able to compare their graduates with the appropriate ethnicity variable; it would do all a disservice to blind ourselves to this aspect of our data.  Variations among offenders that affect their likelihood of benefiting from a given form of intervention is called "responsivity" by treatment experts, and in an optimum world, our ideal response to offenders should be based on our knowledge of their risk, need, and responsivity.  See James Bonta & D.A. Andrews, Risk-Need-Responsivity Model For Offender Assessment And Rehabilitation  (Canada 2007), available at http://www.publicsafety.gc.ca/res/cor/rep/risk_need_200706-eng.aspx.
    Second, many of us in the criminal justice system are convinced that racial profiling (and I would add profiling based on apparent economic and social class, but it's hard to find data to support this as an element in any profile) has substantially distorted the representation of minorities in the criminal justice population.  Statistics from every major study show that police are far more likely to stop certain minorities than others, and that although there is no justification in the "hit rate" [the percentage of stops yielding evidence of a crime is not greater for the targeted minorities than for the general population], the result is overrepresentation of those minorities in the system.  This means that a minority person is often likely to have a more substantial criminal record even though his or her actual involvement in criminal behavior is equivalent to a non-minority with a smaller (or no) record.  In order to compare outcomes fairly, we need the ability to include this variable.  If the disparities do not exist, or if they have no impact on likely success for a given offender, this should be readily apparent.   And it is always a simple matter to uncheck the box (or check all boxes) and to "recalculate" to see what difference this variable seems to make.
    Finally, not all concern with our ability to see distinctions based on ethnicity are benign.  It is a sad fact of modern policy debate that some voices insist on "color blindness" to hide rather than to combat discrimination against minorities.  It is only by virtue of our tools' ability to discern cohorts based on ethnicity was I able to report to the Oregon Commission on Black Affairs at a recent (10/31/09) meeting that, for example, while for "white" offenders prison is the 13th most frequently imposed disposition in our county, while for "African American" offenders, prison is the third most frequently imposed disposition.return to top of page

Q:  Is it fair to include gender and other variables the offender cannot control in the profile?
A:  Yes.  Some argue that immutable factors such as gender cannot fairly be part of a risk assessment instrument; they would presumably have the same criticism of our tools.  Users can eliminate such variables when using our sentencing support tools, but should not do so unless they must to obtain enough data to yield displays.  Criticism of cognizance of factors such as gender may make sense in a world of just deserts, but it surely does not serve fairness in a system focused on public safety.  Ignoring the variable is as likely to result in a sanction that is unnecessarily severe as one that is fortuitously lenient as compared with one that best serves public safety.  We would unfairly ignore or underrate programs that are uniquely valuable for one gender or age group.  There is nothing fair about blinding ourselves to sentences that best serve public safety and offenders.  That kind of blindness is unfair to the victims whose crimes would be avoided without it.  If immutable factors such as gender in fact help determine what is most likely to prevent avoidable victimizations, it is no more "fair" to ignore them than it would be to ignore psychopathy in some violent offenders.  Besides, were this criticism to succeed, it would merely perpetuate uninformed sentencing practices that are inherently susceptible to the personal history, philosophy, preconceptions, and proclivities of the sentencer – i.e., to unfairness, including class, gender, and racial bias – and that apparently perpetuate otherwise avoidable victimizations. return to top of page

Q: Is anyone else doing this now?
A:     Not in sentencing. Probation managers in the District of Columbia ("SMART-STAT")  use a data warehouse "to evaluate which treatment and intervention methods work best for specific offenders" to reduce recidivism.  But as far as I can tell (by tracking down all the rumors I've encountered so far), no jurisdiction is actually using this technology to support sentencing decisions that are more likely to reduce recidivism.  Some systems exist to routinize parts of the process of gathering information thought important to sentencing.  For example, a program called "ASSYST" helps practitioners through the meticulous mysteries of the federal sentencing guideline calculus, and a program called "Judge's Apprentice" automates retrieval and weighing of "aggravating" and "mitigating" factors for sentencing rape and robbery offenders in Israel.  See HaCohen-Kerner Y. and Schild U.J., "The Judge's Apprentice"; 5  The New Review of Applied Expert Systems191-202, (2001).  But such systems do not attempt further the objective of crafting a sentence most likely to succeed at crime reduction. 
    The Midtown Manhattan Community Court project is sometimes mentioned as an example of  decision support technology applied to sentencing.  The Court's website once predicted the benefits sentencing support would bring when it was on  their "to do" list: "A new version of the Judge's Desktop will include analytic software that analyzes success rates for alternative sanctions for different kinds of defendants, and displays predictions for the judge."  Unfortunately, this item was never accomplished, and the list disappeared from the web.
    A more recent version, based on the Midtown Manhattan model, is Brooklyn, New York's, Red Hook Community Justice Center.  The Center wonderfully exploits technology to provide judges with good information about the offender and about treatment and program resources, but it makes no progress whatever toward comparing outcomes based on recidivism for like offenders.  For a good recent article on the Red Hook project by Claire Barliant in the New York Law Journal, click here. [return with your browser's "Back" button].
    Another rumor arises from the Colorado's Integrated Criminal Justice Information System (CICJIS) which again addresses operational needs and promises the benefits of linkage without offering the analytical tools characteristic of DSS: "The CICJIS network and middleware will link existing databases/platforms among the departments and agencies giving users the appearance of querying one database, when in reality they are querying five. The network features will be transparent to the user. The CICJIS network will handle real time event driven data transfers, data extracts, and queries between systems. CICJIS allows the data to be entered only once by the originating agency and to be automatically updated in the other relevant systems." This benefit would be available with DSS; the critical added part is the ability to subject the data to the sort of analysis which compares, for example, programs by graduate performance by criminal activity for selected subsets of their graduate population.  CICJIS is a data integration project in progress; it makes no attempt to link sentencing decisions to outcome-based data, though it could support such an application.
    Sentencing systems in New South Wales and Western Australia provide judges with a great deal of information about how similar offenders are sentenced under similar circumstances, and also provide useful information about both the offender and sentencing literature and trends, as well as sentencing laws and appellate decisions. The New South Wales Sentencing Information System, and a more recent Sentencing Information System developed in Scotland in 1996, strive to assist judges to do what most judges have done before with defendants convicted of similar charges with similar criminal histories. An Israeli version (apparently now defunct) gave the user the ability to screen for one of several judicial sentencing philosophies, and then to see what judges with that philosophy would have done.  The goal is consistency.  This approach, too, critically lacks any component that would enable a user to see how sentences have worked -- they do not include statistics on which sentences have and have not correlated with new criminal activity by which offenders.   Similar approaches, with similar limitations, exist in British Columbia and some provinces in the Netherlands. With respect, the notion that we can learn something by averaging or emulating the sentences historically imposed by judges presupposes that the judges are sentencing based on useful information (and that they know how to use it).  That a sentence is most likely to work hardly follows from its popularity -- particularly in a world in which recidivism is our most common crime.  For a critique of the New South Wales system, click here.
    Technicians in Scotland have built computer models designed to emulate judicial "analysis" to produce the same sort of sentences from any given set of circumstances that a typical judge would produce.  Another is proposed for the Victorian Lower Court Criminal Jurisdiction in Australia. Although some in the process have suggested a performance-based approach such as that we are employing in Multnomah County and, soon, in Oregon state-wide, Scotland has not yet attempted to move beyond regularizing what judges have always done.  Accordingly, the Scottish approach suffers from the same fundamental shortcoming that the New South Wales technology exhibits: to the extent that sentencing has historically fallen far short of its potential to reduce recidivism, these applications simply automate the status quo and offer no improvement in public safety.  More sophisticated modeling approaches (such as that proposed for Victorian criminal courts in Australia) have the advantage of assembling knowledge from experts and users about the various other ingredients in sentencing, and purport to automate and regularize the consideration of such factors in aggravation and mitigation of punishment.  But they, too, make no attempt whatever to improve users' ability to achieve crime reduction or to provide data about the public safety performance of choices available to the sentencing authority.
    Perhaps the newest innovation is that of the Missouri Sentencing Advisory commission (http://www.mosac.mo.gov/directors_corner.htm), which in June of 2005 launched a web-based automated "Recommended Sentencing Application."  The application generates a recommendation based on crime seriousness, offender criminal history, some factors in aggravation or mitigation, and available resources. The program amounts to an automated advisory sentencing guidelines system which, like Virginia's (Richard P. Kern, Meredith Farrar-Owens, Sentencing Guidelines with Integrated Offender Risk Assessment, 16 Fed. Sent. Rptr. 165 (2004); see http://www.vcsc.state.va.us/risk_off_rpt.pdf ), incorporates risk assessment into the process (http://168.166.76.135/RSWeb/JSP/RSAINFO40.html).  These are valuable steps in the right direction, but Missouri's system is hobbled by the circumstance that it makes no efforts to steer judges towards dispositions based on their likely recidivism-reduction value; instead, like the New South Wales and UK versions, they report what other judges have most commonly done in similar situations -- which, unfortunately, emulates the mistakes of sentencing made without information about outcomes and without clear direction at crime reduction. return to top of page

Q: Have you determined that your results are statistically significant? 
A:     We have frequently discussed our approach with statisticians and less frequently with mathematicians.  "Statistical significance" is a concept that relates to the reliability of sampling -- where we attempt to predict totals based on examination of a relatively small portion of a universe of data.  The formulae that express the level of confidence with which our results for the sample will hold true for the whole yield a number which researchers and statisticians invoke when they say that a result is or is not "statistically significant."   Although this process holds the appearance of precision because it is derived from equations and expressed numerically, it actually represents routinized convention.  In any event, our tools run against all available data, and hence do not call for any expression of the reliability with which a sample reflects the whole.  That is not to say that there are no issues as to the "purity" of the data and the extent of error introduced by defects in linking records or data entry in the source data sets, but the error rate related to that set of issues is different from what is normally invoked by the concept of statistical significance.  The relevant issue is not statistical significance, but the extent to which correlations displayed represent causation (i.e., that different recidivism rates reflect different levels of effectiveness in the sentences received by different offenders) or merely description (i.e., that something other than disparate efficacy is responsible -- such as variables other than those exploited by the tools, or the ability of judges to predict or not success when selecting a disposition).  See the next answer.
     The results of sentencing support tools are not presented as reflecting causation.  The tools are not intended to yield a sentence with mathematical precision, or to displace discretion or analysis or advocacy.  Rather, they are intended to encourage and inform the discussion and analysis by giving participants good information about outcomes for similar sentences imposed on similar offenders for similar crimes in the past.  They amount to an ongoing study from which users can pull data relevant to their sentencing.  It is up to all to make the best use of the information; the highest objective of the tools is to focus the sentencing process on crime reduction. return to top of page

Q:  How do you know outcomes don't reflect accurate predictions rather than cause and effect?
A:    When we look at a cohort that shows that short jail sentences or bench probation correlate with reduced recidivism better than longer jail or formal probation, what we may be seeing is that there was something about the higher risk offenders -- not captured in the limited variables now accessible to sentencing support tools -- that led the sentencing judge or the prosecutor to conclude, or the defense attorney to concede, that the offender was more likely to reoffend than others.  So it certainly may be that heavier sanctions for some cohorts show worse outcomes because the process somehow correctly identified offenders at higher risk to sentence more severely.  The short answer is that we cannot be sure the tools show us causation rather than reflecting differences.  But the tools do our best to compare offenders who are alike in terms of age, gender, ethnicity (optionally), and six flavors of criminal history, and who have been sentenced for similar crimes.  The closer the cohort and crime of sentencing match the offender, the more likely the results will be predictive of outcome rather than reflective of differences not captured by the variables.  In practice, users must often widen the default variables to gather enough data to provide any useful information, but we also encourage them to narrow the variables whenever there is sufficient data to permit choosing a cohort even more like the offender than the defaults generated by the tools.  Read the user manual for more information.  For example, here's what the user manual says:

These tools are designed to give as useful a glimpse as possible at aggregate data.
They do not dependably display causal relationships. Although the point is to do our best
to point sentencing arguments and decisions toward public safety, the tools display
correlations only. It is up to advocates and judges to evaluate when and to what extent
these correlations merely reflect descriptions or actually help make responsible
predictions of what sentence is most likely to work.

Although these tools profoundly improve our ability to compare sentences and outcomes
for categories of defendants, they surely do not reach all of the variables that may affect
the outcome of a given sentence in a given case. It is up to advocates and judges to
examine and consider whether and to what extent specific circumstances affect how well
the tools can help us select the best sentence for the offender before the court.

Q:  Doesn't plea bargaining distort the data?
A:     It's actually more likely to be the other way around - that trial results "distort" the data, as plea bargains account for the great majority of convictions.  There are countless individual variations whether or not we limit our criminal history data to trial results.  Even if we only draw data from trial results, we still do not distinguish, for example, accomplice or accessory liability from principal liability.  The same is true of "culpability" facts which do not rise to a successful defense (duress or mutual combat or hungry children or addiction), and a case can be made for the notion that plea bargaining is better than verdicts at sorting these out.  At this point, there's no reason to assume that the variations in how a given fact scenario translates into a criminal history element skews the history differently in one slice of offenders than in another -- the range of variations is presumably largely the same for all.  Of course, we have much to learn, and we may discover distortions and respond to them with our analysis. Meanwhile, the simple fact that we would have to discard roughly 90 to 95 percent of the data to exclude plea bargained convictions is more than enough reason to include all convictions.
    We are nowhere near (and may well never attain) perfection in capturing any offender's profile or in assembling "like" offenders for comparison.  Wholly beyond the limited meaning of the data we do have [and we even might find some distinction worth isolating between convictions by plea and those by trial, for example], there is the tremendous range of probably important data we do not yet include: mental health status, present addiction if any, prior treatment history, personality assessment, and so forth.  We hope to improve in all these areas.
    But the point is that even with these major limitations, we should be able to do a far better job than we now do of analyzing and deciding the public safety implications of our sentencing choices if we have the access to what seems to correlate with success for roughly similar offenders convicted of roughly similar crimes and sentenced in the past.  We need to see the correlations, then consider individual circumstances that might vary the predictive value of the correlations in the case before us, but most important of all we need to have this discussion in sentencing decisions and in plea negotiations.
    In any event, we generally work now with no data whatever, and we generally never have the "what works" discussion.  That there are variations in how plea negotiations and trials translate behaviors into convictions is no reason to opt for continuing on our present path.  Besides, it is our responsibility to encourage plea bargaining to incorporate intelligent discussion about what is most likely to work. return to top of page

Q:  Isn't this just a fancy risk assessment tool?
A:     Risk assessment does not assess recidivism, it predicts it; it amounts to either a device to identify offender characteristics correlated with failure or to identify offenders likely to reoffend.  A “higher risk score” may detect that an offender is less amenable to treatment.
     A “risk score” cannot by itself determine whether an offender is appropriate for a given sanction or treatment provider.  We may find that some programs are more appropriate than others for offenders of a given risk, and a valid risk assessment may be critical in accurate measures of a program's impact on criminal behavior: in fairly comparing programs’ graduates’ performance, it is necessary to compare results of graduates with comparable risk scores.  It may be that experience will teach that matriculation from a given program lowers the risk of recidivism, but we would not have used the risk score going in as a “baseline,” we would have learned that the program reduces risk by accurately analyzing data about that program's graduates and deducing the risk impact.
     Several existing risk instruments do a surprisingly good job of predicting violence, recidivism, or other negative outcome.  They work on a distinct plane, however, from offender profiling, and typically include such variables as substance abuse, age, gender, family composition, employment history, criminal history.  But in reducing the offender to the single variable of a risk score, such assessments lose the flexibility of dynamic analysis: for a specific species of criminal behavior, how does an offender cohort, composed of all of the data captured in the risk assessment plus the specific contour of criminal history and whatever other variables are relevant but not captured by the risk assessment instrument, fare after which responses.
     For example, we may have offenders of a given risk score who differ in other respects that correlate with different likelihoods of success for different sentencing dispositions for the same criminal behavior.  We can have two defendants of identical risk scores facing sentencing for Theft II.  They can differ in age, ethnicity, gender, drug of choice, species of criminal behavior (one might get to a level 5, or whatever, by committing two DUIIs, a PCS, and a domestic assault, while the other has a string of UUMVs), and they might be susceptible to improvement (at least in terms of correlation) in response to very different assignments to custodial or noncustodial programs.
     So risk score will always be no more than one component of an analysis. return to top of page

Q:  Haven't the experts concluded that nothing really works?
A:   Early in the 1970s, academics came up with the conclusion that nothing really works to modify criminal behavior.  This position was enormously convenient to some budgetary and philosophical agendas, and it took hold.  But by the end of the 1970s, even the principal author of this "nothing works" movement began to recant, and the enormous bulk of subsequent literature compels the conclusion that the proposition that "nothing works" was simply wrong at the time and has been completely discredited. Though there are certainly some offenders on whom nothing works, there are also far more on whom some responses, including correctional responses, have demonstrated potential for reducing their criminal behavior. See, e.g., Harland, A. T. (ed), Choosing Correctional Options That Work: Defining the Demand and Evaluating the Supply, Thousand Oaks, Cal: Sage Publications (1996); McGuire, J. (ed), What Works: Reducing Re-offending: Guidelines from Research and Practice. Chichester: John Wiley & Sons (1995) ; Palmer, T., The Re-Emergence of Correctional Intervention. Newbury Park, Cal: Sage Publications (1992); Sherman, L., Gottfredson, D., McKenzie, D., Eck, J., Reuter, P. and Bushway, S., Preventing Crime: What Works, What Doesn’t, What’s Promising. Washington, DC: Office of Justice Programs (1997).  See also McGuire, James, What Works in Reducing Criminality (2000), http://www.aic.gov.au/conferences/criminality/mcguire.pdf. return to top of page

Q:  How can you expect offenders to change until they're ready to accept change:
A:   "You can't do any good until the offender is ready to change his life" is a fairly common litany. In practice, it is a complete abdication of responsibility for outcomes.  The vast majority of offenders are not "ready to change" in this sense, and the rest of the criminal justice community (probation and corrections officials) are keenly aware of the need for "stage of change analysis" -- which identifies and responds appropriately to the relevant level of an offender's awareness that there is a "problem" from none (denial or "precontemplation") through "contemplation," "preparation," "action," and "maintenance."  Although it is certainly true that a sincere recognition of the need for change is a hopeful sign, it doesn't follow that nothing will work until the offender demonstrates this level of self-awareness or -- more importantly -- that we are not responsible for outcomes with offenders who don't seem "ready to accept their need for change."  Most offenders we sentence don't recognize their need for change; our public safety depends upon  making the best choices (within those available) to improve their chances of avoiding further crime.  Besides, if there are offenders who have at some point recognized their need for change, surely some of them did so after being sentenced a few times.  Even considering what sentence might hasten an offender's arrival at this point is an improvement over abdicating responsibility for outcomes for all who are not yet "ready." return to top of page

Q:  What is that graphic you use on so many pages? <- click on image for another view
A:   The graphic is Jeremy Bentham's "Panopticon," a design for the modern prison, and dates from the late18^th century.  Bentham's notion was that this architecture would help reform prisoners through solitude. Bentham wrote "that in regard to persons of the description of those to whom punishments of the nature in question are destined, solitude is in its nature subservient to the purpose of reformation, seems to be as little disputed, as its tendency to operate in addition to the mass of sufferance." Bozovic, Miran, ed., Jeremy Bentham: The Panopticon Writings (© London: Verso, 1995), Letter VIII.  Although the notion that the solitude of prisons would produce penitence and good citizenship has long been discredited, Bentham's design for what became the modern prison (considered a great reform, championed by Quakers, and first built in Philadelphia) has quietly but relentlessly fulfilled a purpose originally explicit but now quite covert: "the great load of trouble and disgust which it takes off the shoulders of those occasional inspectors of a higher order, such as judges and other magistrates, who, called down to this irksome task from the superior ranks of life, cannot but feel a proportional repugnance to the discharge of it." The Panopticon Writings, supra  (Letter VI).  The modern prison was the last great leap forward in correctional policy; data-based sentencing responsibly aimed at crime reduction should be its next. return to top of page