Sentencing Support Technology
The Judges Journal, Winter 2002
[the Journial of the American Bar Association Judicial Division]

The public is frustrated with the perception of the "revolving door" of criminal justice--the specter of repeated criminal behavior by those we process repeatedly in the criminal justice system. Grasping for solutions, voters and legislatures across the nation have resorted to mandatory minimum sentencing provisions and three-strikes laws--measures that substantially limit judicial discretion in sentencing.

Most sentencing decisions are profoundly ineffective in diverting offenders from criminal careers. For instance, in the most recent month for which Portland, Oregon, police released such statistics, twenty-two of the twenty-three persons jailed for robbery had been in the same jail within the previous year, as had twenty-two of the thirty-two jailed for burglary, twenty of the twenty-six jailed for theft in the first degree, 304 of the 372 jailed on drug charges, and thirty-two of the thirty-nine jailed for vehicle theft. National statistics show seven of ten jail inmates have prior incarcerations, and over 62 percent of those released from prison are arrested again within three years. (1)   These numbers surely reflect some victimizations that we should have prevented, misdirected correctional resources, and spiraling waste of law enforcement and criminal justice dollars. More broadly, these failures contribute to erosion of respect for the judiciary and the rule of law itself.

Any candid examiner can easily uncover the reasons for these statistics: the legal system has not truly made public safety its ultimate objective. Participants in sentencing hearings almost never negotiate or argue based on what available disposition is most likely to prevent new crimes by the offender. After accepting or contesting any constraints such as sentencing guidelines, the participants generally argue what the defendant "deserves," or why we should be sympathetic to or mad at the offender, or, occasionally, what works for the offender. Advocates virtually never initiate a discussion about what will keep the offender from reoffending, and they never come armed with citations to literature supporting why a given disposition is likely to work (or not work) for a given offender. We may send thieves to "theft talk," drunk drivers to alcohol treatment, drug offenders to substance abuse counseling, and sex offenders to behavior modification programs, but we do this largely out of a sense of symmetry rather than anything approaching social science. After all, we may ask whether offenders have completed the program but we never ask whether program graduates reoffend.

Corrections professionals occasionally give us compilations of information presumed useful in making informed sentencing decisions--criminal, social, and employment history; substance abuse and mental health status; and performance on probation. But they always stop short of connecting that information to a reasoned analysis of which disposition is most likely to prevent future criminal conduct by the offender and why. We persist in a just deserts ritual to the exclusion of responsible consideration of public safety, and our results reflect that persistence.

Frustrated with offenders who repeat their criminal behavior, Oregon voters and the Oregon legislature have declared public safety the primary goal of sentencing in the state. A 1996 citizens' ballot measure amended the Oregon Constitution to make "protection of society" a primary purpose of sentencing. (2)  Legislation enacted in 1997 declared reduced criminal behavior the dominant performance measure for criminal justice agencies (and their private partners) for juvenile and adult offenders; required that agencies share data (including juvenile and adult data) to reveal the impact of correctional efforts on future criminal behavior; and directed that criminal justice agencies use the data to see correlations between correctional responses and reduced criminal behavior. (3)  The Oregon Judicial Conference in the same year resolved that in considering the public safety aspect of sentencing, "judges should consider and invite advocates to address the likely impact of the choices available to the judge in reducing future criminal conduct."(4)

In pursuit of these policy objectives, we have begun to develop and use technology to improve our public safety performance. The sentencing support tools developed in Multnomah County run against a data warehouse. Data warehouse technology collects operational data from law enforcement and criminal justice agencies, including the courts, and transforms the data to produce a pool of information susceptible to analysis. This allows us to see what happened to any category of offenders after they were sentenced for a particular category of crime.

A user enters a defendant's case number and selects the charge before the court for disposition. The program then assembles data for "similar offenders" previously sentenced for "similar crimes," and displays outcomes for each disposition element imposed at least thirty times on such offenders for such crimes. Disposition elements reflect all of the actions taken against a particular offender--jail terms of various lengths, formal or bench probation terms, treatment programs, community service, prison, and so forth. The result is displayed in a bar graph (Table 1), with the most frequently imposed disposition element represented by the first bar with subsequent bars displayed to the right in decreasing order of their frequency. The height of each bar reflects the proportion of such offenders who avoided "new criminal behavior" after receiving that disposition element. Below the bar chart is a table listing each element, the percentage of offenders who received that element and remained crime free, and the number of applications of that incident on such offenders for such crimes. Although no bar appears for any element imposed fewer than thirty times, the table includes all results, displaying numbers lower than thirty in red. The screen also sets out the variables represented by the graph and table.

Table 1 here
[omitted in favor of more complete graphics - click here]
{Return with your browser's BACK button}

By default, defendants are sentenced for "similar crimes" if they were sentenced to a crime that is in the same category as that selected by the user for sentencing (except that we default to driving under the influence of intoxicants if that is the crime for which the offender is being sentenced).

Also by default, offenders are "similar offenders" if they share criminal history ratings in each of six categories and they share the same age range, gender, and ethnicity. Criminal history ratings range from none, low, and moderate to major or severe in each of the following crime categories: violent, sex, drug, property, DUI/major traffic, and domestic violence. The ratings are based on data rules set by a user group consisting of a judge, prosecutor, defense attorney, and project head. By default, offenders avoid "new criminal behavior" if they have not received a new conviction within three years after being sentenced.

The user (judge, prosecutor, or defense attorney) is able to access and modify all of the variables, and generate a new display based on those variables. For instance, if there are enough data to allow the judge to do so, he or she can narrow "similar crimes" to the same crime for which the offender is being sentenced, or the user can broaden "similar crimes" to "all crimes," or select any of the many variations in between, such as felony theft crimes. The user can adjust the settings for "similar offenders" to accommodate additions or corrections to the offender's criminal history, age, or gender, and broaden the category by including more than one age group, crime rating, ethnicity, or gender. Moreover, the user can also compare results with and without the ethnicity variable, or as between or among various ethnic groups--all important capabilities where criminal histories may be inflated for minorities, or where some programs are designed with specific ethnic groups as target populations.

Finally, the user can redefine "new criminal behavior" by category of behavior, incident of behavior, and time of behavior. The user can look for behavior constituting "any crime" instead of only the category that includes the crime for which the offender is being sentenced. The user can also select any other crime category, any felony, or any misdemeanor, and can include arrests instead of convictions, and six months, five years, or all time known to the available data instead of the default three years.

Although these tools cannot and should not replace judicial discretion by determining sentences, they greatly enhance our ability to craft a disposition that is most likely to serve public safety by preventing repeated criminal behavior. At least as important is the impact of the use of these tools on the entire sentencing process; advocates, participants, and observers will understand that what works now matters. Oregon lawyers are beginning to inject bar charts into plea negotiations. Multnomah County judges recently added a check box to pre-sentence investigation orders asking pre-sentence investigation writers to include "[a]nalysis of what is most likely to reduce this offender's future criminal behavior and why, including the availability of any relevant programs in or out of custody."

Oregon has a long way to go in enhancing the utility of these tools, but the promoters of sentencing support technology hope to include additional criteria that will define "similar offenders," such as substance abuse assessments, mental health assessments, educational assessments, and vocational assessments. These judges hope to attract the attention and the assistance of academia in refining these tools, and welcomes the addition of corrections and criminology expertise. They also anticipate that the criminal justice system will adapt the technology to support decision making related to probation violation hearings and release decisions.

The promise of this early application of technology to support sentencing decisions is as profound for criminal justice as Kitty Hawk was for transportation. And, by accepting responsibility for seeking the best public safety outcomes for sentencing decisions, the judiciary will begin to earn back a good deal of the public respect that it has lost in recent decades.


1. Allen J. Beck and Bernard E. Shipley, Bureau of Justice Statistics, Bureau of Justice Statistics Special Report, available at

2. Oregon Constitution, Article I, section 15.

3. 1997 Oregon Laws, Chapter 433 [1997 House Bill 2229].

4.  1997 Oregon Judicial Conference Resolution #1.