January 2, 2005
IDEA LAB

Sentencing by the Numbers
By EMILY BAZELON

Imagine that you could stop a crime before it happened. Not by zapping a murderer seconds before he bloodied his victim, like the future cop played by Tom Cruise in ''Minority Report,'' but by sitting calmly on the bench in judge's robes and perusing a single sheet of data. Armed with the stats, you could predict the likelihood that a convicted drug dealer or thief standing before you would be arrested again in the near future if you let him go free.


For decades, the science of predicting future criminality has been junk science -- the guesswork of psychologists who were wrong twice as often as they were right. But today, the detailed collection of crime statistics is beginning to make it possible to determine which bad guys really will commit new offenses. In 2002, the Commonwealth of Virginia began putting such data to use: the state encourages its judges to sentence nonviolent offenders the way insurance agents write policies, based on a short list of factors with a proven relationship to future risk. If a young, jobless man is convicted of shoplifting, the state is more likely to recommend prison time than when a middle-aged, employed woman commits the same crime.


Virginia's new sentencing method was born of a budget crunch. Faced with the prospect of building new prisons after passing a tough-on-crime measure in 1994, the Legislature asked the state sentencing commission to figure out which nonviolent offenders could be kept out of prison without posing a risk of committing new crimes. The commission's director, Richard Kern, and his staff members tracked 1,500 nonviolent drug, larceny and fraud offenders for three years after their release from prison. The researchers found that men were 55 percent as likely to be rearrested as women, and that offenders in their 20's were a much higher risk than those older than 40. Being unemployed made offenders more likely to commit another crime. So did being single.


Using these factors and a few others, including a defendant's adult and juvenile criminal records, Kern designed a simple 71-point scale of risk assessment as an aid for judges. If he scores 35 points or less, a defendant who would have otherwise gone to prison under Virginia sentencing guidelines is recommended for an alternative sanction like probation or house arrest. Anything above 35 means a recommendation of jail time. ''Judges make risk assessments every day,'' Kern said. ''Prosecutors do, too. Our model brings more equity to the process and ties the judgments being made to science.''


Kern tested his model on prisoners released five years earlier and found that his ratings correctly predicted who would be reconvicted in three out of four cases. Of the felons who scored at or below the 35-point cutoff, 12 percent committed new crimes, compared with more than 38 percent for those who scored higher. After calculating that only a slight increase in recidivism would result, the state raised the 35-point cutoff to 38 points last July. Meanwhile, the growth of the state's prison population -- which used to be more than twice the national average -- has slowed nearly to a halt.


But if Virginia is pleased with itself, those who think punishment should reflect blameworthiness are not. They argue that by penalizing offenders differently for the same crimes, for reasons that have nothing to do with moral culpability, the state has abandoned the idea that punishment is a form of ''just deserts'' for wrongdoing. ''If you're punishing people because of a bunch of factors that have nothing to do with blame, well, you're not in the business of doing justice anymore,'' said Paul Robinson, a law professor at the University of Pennsylvania. As he and like-minded legal thinkers see it, a woman in her 40's who deals drugs hasn't done anything more to earn trust or deserve a break than a male dealer in his 20's charged with the same offense. She has just gotten lucky, by falling into a group whose other members have generally proved a good public-safety bet. Meanwhile, jobless single men in their 20's start with 36 points on Virginia's risk scale, putting them on the cusp of going to prison before the crime they committed is even taken into consideration.


It's not a foregone conclusion that Virginia's method of sentencing is permissible under the Constitution, though no young male offender has brought a court challenge so far. Age and sex are what the law calls ''immutable characteristics,'' and it's a fundamental principle of antidiscrimination law that the government has to tread carefully when it treats people differently because of qualities that are beyond their control. (Being married or holding a job are different, but not entirely so, since these statuses reflect people's opportunities as well as their preferences.)


Still, if the state has a compelling reason to discriminate based on age and sex, it sometimes can. And public safety is almost always deemed to be such a reason. The moral and legal picture would be cloudier, of course, if Virginia's sentencing recommendations were based on race. There's a statistical basis for doing that; Kern's study found that African-American felons committed new crimes at higher rates than whites. But Kern and his commission advised the state to stop short of taking race into account at sentencing, reasoning that it was merely serving as a proxy for socioeconomic disadvantage.


Virginia has tried to deflect criticism by emphasizing that judges are free to ignore the sentencing recommendations if they so please. By retaining such flexibility, Virginia has also reduced the danger that the predictions might have counterproductive consequences. This potential pitfall is what economists call a moral-hazard problem: drug dealers with low scores will realize that they're likely to avoid prison, and so have a greater incentive to sell more cocaine. (If that sounds unlikely, consider that some dealers already use child couriers to shield themselves.) If Virginia was to notice a jump in crime among middle-aged and elderly women, the state could adjust its predictions, Richard Epstein of the University of Chicago Law School points out. But allowing judges to send a little old lady to prison for a long time, if necessary, helps prevent the moral hazard in the first place.


Predicting risk can lead to harsh results. At the Legislature's request, Kern's commission tracked 579 sex offenders for an average of eight years after their release from prison. The small minority who scored above 43 on a 61-point scale were all arrested again, either for sex offenses or other crimes that harmed someone. So the sentencing commission recommended tripling the sentences for offenders in this group. For rape, a 13-year sentence would soar to 39 years.


For the sex offender who would have stayed clean despite the odds, that heavy sentence would seem unjust. But for states faced with overcrowded prisons and limited budgets, it may not be irrational.

As long as the coin is two headed.

So, if gay marriage is not legal in Virginia, then all gay men under 40 start out with 36 points.

Won't be long before the ACLU jumps all over that little 'system'.

This "system" is perfect for liberals and the ACLU. It completes the separation of any concept of morality from the law. Decisions are made solely on the basis of predicted behavior. No longer do we teach that you do good because it is a universal law. You do good to avoid confinement.

Of course, once that "system" is in place, the debate will be redirected to remind all of us just how little we actually know about predicting behavior. Remember the attacks on psychiatrists predicting future dangerousness in capital cases?

This approach is likewise consistent with the ongoing effort to divorce the Biblical concept of punishment from the secular laws. Any reasonable history of the law would recognize the significance, for example, of the Ten Commandments. But, apparently, we are not allowed to include that lesson anymore. (Let me be clear that I strongly support the First Amendment's separation of church and state but feel like the current Supreme Court formula doesn't fairly enforce that constitutional principle.)

Will be interesting to see if the Supreme Court, which has a case on the issue, will speak clearly on the issue. Particularly given the presence of the Ten Commandments in their own building.

It seems to me that this another attempt to validate common sense.

OK, so guy kills his wife. We determine he will never kill again. So, we put him on probation, right?

Not in my book. If we fail to punish in recognition that a life was taken (at great human value), then we diminish the value of life and abandon the moral element of crime and punishment.

I am not willing to measure the value of punishment in pure fiscal terms. Sure, there are limits to what we can do within a given governmental body, but the approach suggested above is purely utilitarian.

As the other post have suggested, not only does such a scheme remove the penalty aspect, it also completely ignores the impact of the offense on victims and the community at large. Surely our system should have consideration for "fixing" the offender, but the offender is not the only party in interest. We talk about future danger (which is vital)but do we ignore the empirically established past danger and harm inflicted?

The concepts of punishment, general deterrence, and yes even retribution can not be ignored by a criminal justice system that will be trusted by the members of the society it serves.

Perhaps for centuries we underutilized rehabilitation, we should not err now in making it the only goal of criminal justice.

This system would not take into account that young people addicted to drugs or alcohol do usually make great improvements in their lives as they mature, IF they can stay out of the prison system. A perfect example is our own president. Some are lucky enough to get through their crazy years without too much damage to the rest of their lives, or have the means to pay their way out of the majority of their problems, and go on to be good productive citizens.

In order for this "point" system to be remotely fair, it would have to start out on the assumption that all people have the same opportunities and chance at the American dream. and that will never be possible.

As I recall, the President was arrested and convicted of a DWI. But I do agree that some people are treated better because of their backgrounds or status. That's part of the real world.

But, probation does get used far more often than the media reports. Prosecutors resolve 95% of their cases by plea agreements. Defendant largely agrees he is guilty and prosecutor largely recommends probation.

Likewise, the majority of defendants on probation successfully complete the probation and have successful lives. Something must be working.