Keeping track of sex offenders Critics say sex offender screening tool doesn't work Minnesota's probation system overloaded One probation program that works A new approach to treating sex offenders American RadioWorks' Hard Time: Life After Prison Audio Critics charge sex offender screening tool doesn't work (story audio) Photos Dru Sjodin Alfonso Rodriguez, Jr. William Grove View full slideshow (4 images) Resources Minnesota Department of Corrections See the Minnesota Sex Offender Screening Tool Your Voice Join the conversation with other MPR listeners in the News Forum.

	E-mail this page			Print this page

Critics charge sex offender screening tool doesn't work
by Dan Gunderson, Minnesota Public Radio
April 19, 2004

The disappearance of Dru Sjodin in northwest Minnesota in November 2003, and the arrest of a sex offender in the case, has brought new scrutiny on the state's handling of sex offenders. (Photo courtesy of finddru.com)

Minnesota uses risk levels to let people know which sex offenders are the most dangerous. But some experts charge the state uses a flawed tool to help determine that risk level.

Every sex offender in a Minnesota prison is given a risk level; 1, 2, or 3. Level 3 offenders are considered the most dangerous. The Minnesota Sex Offender Screening Tool is the first step in setting those risk levels. But some scientists contend the tool is wrong more often than it's right.

Moorhead, Minn. — Shock and anger drive the debate about what to do with sex offenders.

This year the name on everyone's mind is Dru Sjodin, the college student who disappeared in November 2003, and whose body was found near Crookston, Minn., Saturday.

Sjodin's father, Allan, said it had been "a devastating day."

"We were waiting for that call and when that call came we all stopped living for a second," he said at a Saturday news conference.

Alfonso Rodriguez Jr. has been charged with kidnapping in the disappearance of Dru Sjodin, and may face more charges now that her body has been found. Rodriguez is a convicted sex offender who was released from prison about six months before Sjodin's disappearance.

Fifteen years ago the shock and anger was the same, but the names were different. The victims then were Carrie Coonrod and Mary Foley. Both were young women raped and killed by sex offenders.

	Alfonso Rodriguez, Jr.

Lawmakers and the public demanded action to stop violent sex offenders. Several task forces were appointed to study the issue. They suggested a variety of solutions, including civil commitment.

Commitment allows the state to confine the most dangerous offenders indefinitely in a treatment facility after they serve their prison sentence.

But state officials knew they needed a better way to predict which offenders were most dangerous. Research showed psychiatrists were nearly always wrong when predicting how sex offenders would act in the future.

So the Department of Corrections worked with Iowa State University professor Doug Epperson to develop the Minnesota Sex Offender Screening Tool.

It's an actuarial tool, the same kind of statistical tool insurance companies use to decide who pays the highest premiums. Actuarial tools are complicated, but here's a simple description of how the Minnesota Sex Offender Screening Tool works.

Researchers identified 16 factors, or characteristics, common to dangerous sex offenders. All sex offenders who go to prison in Minnesota are compared against those common characteristics.

Then a complex scoring system assigns a number to the offender. That number puts the sex offender in risk category 1, 2 or 3, indicating low, medium or high risk. Those who fall in the high risk category are considered for civil commitment.

	William Grove

William Grove, a psychology professor at the University of Minnesota, believes the results are useless. He says the sex offender screening tool inflates the rate of recidivism -- the likelihood that an offender will commit another crime.

If the screening tool was being used to decide which offender would be civilly committed, and it identified five men as the most dangerous offenders, only one would be likely to re-offend, according to Grove.

"If this test is -- as I believe it is -- not valid, you'd be putting four people into sex offender treatment, probably indefinitely, who wouldn't recidivate, for every one you'd put in that would (recidivate)."

William Grove studies the science of prediction. He's examined several sex offender screening tools currently used around the country. He spent months taking apart the Minnesota sex offender screening tool to see how it worked. It's wrong more often than it's right, according to Grove's research.

Grove is not a lonely voice of dissent. A handful of researchers have come to the same conclusion. Richard Wollert is a clinical psychologist in Portland, Oregon, who has worked with sex offenders for three decades.

Wollert published a paper contending the Minnesota Sex Offender Screening Tool is flawed. He argues those design flaws significantly inflate its effectiveness. Wollert says MnSOST-R needs to be subjected to further scientific scrutiny.

But there are also studies supporting the Minnesota screening tool, says its primary developer. Iowa State University professor Doug Epperson was lead researcher on the project to develop the Minnesota Sex Offender Screening Tool in the early 1990s. He also revised the tool in the late '90s. It's now called the Minnesota Sex Offender Tool-Revised, or MnSOST-R.

There are several actuarial tools for evaluating the risk of recidivism among sex offenders. The Minnesota tool is one of the three most commonly used by officials across the country.

	Doug Epperson

Epperson says statistical prediction tools are not perfect, and future behavior by sex offenders is among the most difficult to predict. But Epperson says the Minnesota Sex Offender Screening Tool is one of the most accurate ways to identify sex offenders at risk of recidivism.

The dispute about MnSOST-R centers on accuracy. Sex offenders are assigned a score while in prison. The score is between -5 and 17. The higher the number, the more dangerous the offender. Epperson says those who score eight or higher are classified as Level 3 offenders.

But a high score doesn't necessarily mean the person will commit another crime. Consider this hypothetical group of 10 convicted rapists. All 10 scored an eight on the screening tool. They're Level 3 offenders.

MnSOST-R developer Doug Epperson says statistically, six of the 10 would commit another crime. Four would not. So the tool would be right about 60 percent of the time. Not a perfect prediction, but better than flipping a coin, and research shows it's a better predictor than a psychiatrist's clinical judgement.

But the U of M's William Grove says the tool is actually much less accurate. He says it's wrong more often than it's right. Grove is concerned state lawmakers put too much faith in the prediction of future behavior.

"They can make the best decision if they have the most accurate data about -- among other things -- how accurate these predictions are," says Grove. "It's my concern (the predictions) are being overrepresented in various venues as to how accurate they are, so people have the idea they're performing better than they very likely are."

Department of Corrections officials say they have not exaggerated the ability of the Minnesota Sex Offender Screening Tool to predict risk. But Department of Corrections Deputy Commissioner Harley Nelson believes lawmakers and the news media have attached too much importance to individual offender scores.

"People were starting to get excited if a score was an 11 versus a 12 or a 12 versus a 13. There's no tool that's going to give you that kind of specifics -- that two points is going to dramatically change it," says Nelson. "That doesn't mean it's a bad tool. It's an instrument to help us do it. But it's not the Holy Grail that's going to tell you exactly how they're going to act when they're released, or their absolute risk of re-offending."

It's an instrument to help us ...but it's not the Holy Grail that's going to tell you exactly how they're going to act when they're released, or their absolute risk of re-offending. - Harley Nelson, Minnesota Department of Corrections

Nelson is quick to point out the sex offender screening tool is not the only thing officials consider when deciding risk levels or requesting civil commitment. However, it is a key part of assigning risk, and deciding which offenders are considered for civil commitment.

The sex offender tool is not perfect, according to Nelson, but he has no doubt it is the best available tool for categorizing sex offenders.

Nelson says the state is looking for a better way to identify dangerous sex offenders. That may involve refining the current screening tool, or combining the Minnesota tool with other sex offender screening tools being using around the country.

He says it's not likely any screening tool can achieve perfect prediction of future human behavior.

Some critics contend anything less than perfection is unacceptable. They say it's wrong to lock an offender away for life, based on an inaccurate prediction of future behavior.

Minnesota State University professor Mark Hansel teaches criminal justice. The current system clearly keeps dangerous sex offenders off the street, he says. But he says lawmakers should consider whether it's fair to lock away even one person who would not commit a new crime, based on a less than perfect prediction.

"This person has done awful things, they've spent a period of time in prison. Has this person lost all dignity? Is this person human garbage? And even though he will do no harm in the future, (is he) worth simply throwing on the garbage heap and forgetting about, in order to achieve a real and measurable social benefit?" Hansel says. "But you have to treat this person as garbage in order to achieve that social benefit. Is that where we've gotten as a society?"

Other critics are more pragmatic. They say the state simply can't afford to continue putting sex offenders in treatment centers.

Civil commitment of sex offenders is expensive. The bill is nearly $23 million this year. It's estimated the cost could triple in five years as more sex offenders are committed.

Most experts agree a few very dangerous sex offenders cannot be rehabilitated, and must be kept off the street. But they also point out sex offenders in general are not at a high risk to re-offend. Corrections Department Deputy Commissioner Harley Nelson says among all levels of sex offenders, about one in 20 will re-offend. The recidivism rate for high-risk offenders is about one in 10.

The problem for state officials is there's no surefire way to know who will re-offend. So they identify a range of people considered most likely to be a risk to public safety.

It's a delicate balance. They don't want to lock up people unnecessarily. But they also don't want to release someone who will make headlines.

And that's where politics can become entangled with science.

Harley Nelson of the Corrections Department is concerned about political pressure on the people deciding which sex offenders are dangerous.

"That can have a chilling effect on people having to make decisions, and not just in that area. From judges to probation officers, there's many different people working the front lines with these people," Nelson says. "They have to make decisions, and you don't want them so intimidated it hampers their judgment."

Some experts contend political passion can lead to expensive mistakes.

John LaFond holds the endowed chair in Law, the Constitution and Society at the University of Missouri Law School. LaFond spent years researching effective treatment for sex offenders. He's published numerous articles and books about the topic. He's also challenged civil commitment laws in court.

LaFond believes commitment is not the best alternative for sex offenders. After an offender gets out of prison, a combination of probation and treatment is successful with most, according to LaFond, and the cost is a fraction of the cost of long term confinement.

A "lock 'em up and throw away the key" solution is good politics, but bad public policy, according to LaFond.

"These are difficult and complex public policy issues, and cannot be solved with sound bites or simplistic solutions," says LaFond. "Public policy needs to be based on what we know about sex offenders and sex offending, and not tomorrow's headline."

But public policy is often driven by fear and anger.

Minnesota officials say they expect a significant increase in the number of sex offenders committed to state treatment facilities. As a result of the Dru Sjodin case, Gov. Pawlenty ordered all Level 3 offenders be considered for civil commitment.

News Headlines

Related Subjects