When New Hampshire Attorney General Kelly Ayotte asked lawmakers to approve tougher penalties against sex offenders, she provided a familiar argument: Sex criminals are destined to strike again.

"I don't want you to be fooled," Ayotte told House legislators in January. Sex offenders "are committing the same type of offense over and over."

Such claims are typical in the public debate over whether to toughen penalties for those convicted of sex crimes. But they also distort a complex reality, according to experts who study how often sex offenders commit new crimes.

Recidivism rates for convicted sex offenders vary greatly, ranging from less than 5 percent to more than 50 percent, based on such factors as the victim's gender, the offender's relationship to the victim and whether the offender has received treatment, multiple studies have shown. In general, younger men who abuse boys they don't know have the highest risks of re-offending.

Taken as a group, about 14 percent of convicted sex offenders committed new sex crimes over a five-year period and about 20 percent did so over a 10-year period, according to Canadian researcher Karl Hanson's widely cited review of studies involving more than 4,700 offenders.

Recidivism rates for sex offenders drop for those who receive treatment to help them control their impulses and steer clear of dangerous situations, recent studies have shown.

In Vermont, for instance, correctional officials tracked 195 adult male sex offenders over a six-year period. The sexual re-offense rate for those who completed treatment was 5.4 percent, versus a 30 percent rate for those who refused treatment or did not complete it.

Dr. Fred Berlin, assistant professor of psychology at Johns Hopkins University and the founder of a sexual disorder clinic there, likened sex offenders to alcoholics. Alcoholics can be treated and learn to manage their disease.

Sexual offenders have "a craving disorder," he said. "It's not like an infection that can be cured." While treatment is not effective on everyone, he said, it appears to be more effective than the public may think.

"It's really a judgment where you have to look at the individual," said Berlin.

Four hours a week, convicted sex offenders at the Vermont state prison in Springfield talk about the crimes they committed and the steps they can take to keep them from happening again.

The program is like others in Vermont and New Hampshire, giving criminals deemed at risk to re-offend the tools to keep themselves out of trouble.

Which sex criminals are most likely to re-offend?

Studies that tracked groups of sex offenders over their lifetimes found that 52 percent of diagnosed pedophiles who molested boys committed another sex crime while nearly 40 percent of rapists who targeted adult women offended again, according to the Association for the Treatment of Sexual Abuse.

At the other end of the recidivism scale were the vast majority of sex criminals — 97 percent or more of offenders in New Hampshire and Vermont — who knew their victims.

Multiple research studies have found that between four and 10 percent of offenders who targeted children in their families committed new sex offenses after their convictions, the association says. By contrast, a 1995 study found that offenders who targeted boys unrelated to them had recidivism rates of 35 percent.

All this research comes with important caveats, most significantly that official recidivism rates in sex crimes are likely to be artificially low because a significant number of offenses are believed by experts to go unreported.

Getting at why some types of sex offenders are more likely to commit new crimes than others is difficult, said David Burton, an associate professor at Smith College who researches and treats sex offenders.

"Different sex offenders are driven by different things," said Burton. "Some of those things are deeper and more difficult to change."

Those offenders with antisocial tendencies, for instance, would be more likely to re-offend and more difficult to treat than those whose offenses were related to stress and inability to express themselves, he said.

The minority of child molesters who fit the psychiatric diagnosis of pedophilia are also more likely to re-offend, said social worker Mary Jane Egerton, who leads sex offender groups with West Central Services, an Upper Valley mental health agency.

Pedophiles, whose primary sexual attraction is toward children, have urges that are entrenched and difficult to change, she said. But most child abusers are not pedophiles, experts say. They are adults who are primarily attracted to other adults but who, for a variety of reasons, have used children to fulfill their needs.

And many in that group are adults who have used opportunity and access to abuse children who are family members, said Jill Levenson, a researcher at Lynn University in Boca Raton, Fla.

Such offenders do not face urges as ingrained as a pedophile's, said Levenson, making it easier for them to resist the impulse the next time around. Their re-offense rates also may be lower for a practical reason, she said: After they are convicted, they likely to lose easy access to their victims.

She added that even the subgroup of offenders who target minors to whom they are not related is diverse: Those offenders may include young men convicted of statutory rape for having sex with underage girlfriends, or men who have had sex with physically mature teenage girls whom they do not view as children.

"All sex offenders are not the same," Levenson said.

Sex offenders, taken as a whole, may be more "pro-social" than other types of criminals and embarrassed by people learning of their crimes, said Georgia Cumming, director of Vermont's sex offender program.

Given the widespread underreporting of sex crimes, she said, a significant number of sex offenders have likely had multiple victims by the time they finally get arrested. But she said that an arrest — and the public shame that comes with it — is for many offenders the "line in the sand."

How officials view a particular sex offender's risk of re-offending can have an impact on how much, if any, prison time he may get as well as the type of treatment program he may be required to complete.

The case of child molester Mark Hulett illustrated just how crucial the classification can be in some cases. Hulett, who knew his victim and had no previous convictions, was ranked as a low risk of re-offending in the pre-sentencing report that corrections officials performed. That designation made him ineligible at the time for in-prison treatment.

Judge Edward Cashman, however, said he believed Hulett needed treatment as soon as possible in order to not re-offend. In early January, he handed down the controversial 60-day minimum sentence, saying Hulett could then return to the community and get the treatment he needed. (Cashman later changed the minimum sentence to three years after the state announced plans to provide in-prison treatment to low-risk offenders.)

To determine the probability of offenders committing new crimes, experts use a combination of statistical models.

By those measures, sex offenders are assigned certain scores based on a number of risk factors, a method that is similar to the actuarial models long used by the insurance industry to determine, for instance, which drivers pose the biggest risk on the road.

The factors that lead to an offender being deemed "high-risk" include:
• Male victims;
• Stranger victims;
• Prior sex offenses;
• Lack of a cohabitation history with adult lovers.

"Can we perfectly predict?" said Levenson, the Florida researcher. "No, but we can do a pretty good job assessing the likelihood" of re-offending.

One state, Virginia, took into account such statistical predictions when it came up with sentencing recommendations for certain crimes. The Virginia Sentencing Commission, which proposes the optional sentencing recommendations to judges, used statistical models to predict the threat each offender poses once he or she is released.

Based on the statistical probability a sex offender will re-offend, the commission advocates increasing the recommended sentence. Under the commission's proposal, sex offenders who pose the biggest statistical risks would see their sentences triple, say, from the 10 years in the guidelines to 30 years.

Critics say Virginia judges are imprisoning people because of the probability that something might happen based on their statistical classification. Commission director Richard Kern said the statistical models give judges "tools based on research."

In both Vermont and New Hampshire, statistical models predicting offenders' risk levels can assist judges in making sentencing decisions. Experts conduct psychosexual examinations, for instance, to determine how much of a public safety risk the offenders would pose if they were to be placed on probation and receive treatment in the community.

Consider the case of Frederick Stebbins, a married, 58-year-old Quechee man who admitted recently to sexually assaulting a female minor whom he knew.

Egerton, the Lebanon clinical social worker, performed an evaluation at the request of the defense. Stebbins' scores on the statistical tests showed he has a low likelihood of re-offending, making him an appropriate candidate to receive treatment in the community, she testified in court.

Once offenders are sentenced to prison, corrections officials in both states again assess them to determine which treatment program they should participate in. The more intensive programs are reserved for the higher risk inmates.

No one claims the statistical measures are a guarantee, said Daniel Millis, the interim director of New Hampshire's sex offender treatment program. However, he said, "I think they're effective enough to give us a good idea."

And that's important, says Egerton, because most sex offenders do leave prison. Having an idea of their risk levels helps the probation and parole officers and community treatment providers get a sense of how best to help the offenders manage their risks.