Scoring sex offenders to predict future risk
- By Matt Leonard
- Mar 14, 2018
To conserve money and resources when monitoring released sex offenders, police departments want to keep a closer watch on those more likely to reoffend. Determining which offenders those are, however, has been difficult.
Now predictive software, first tested by the Baltimore Police Department and now being used by the Vanderburgh County Sheriff’s Office in Indiana, can help police prioritize their monitoring.
OffenderWatch's Focus generates a risk score meant to predict an individual offender’s likelihood of reoffending. It uses Census Bureau data, including crime rates and local income levels, along with information related to the individual offender like employment status, living situation and relationship status in its prediction scores, OffenderWatch President Mike Cormaci told GCN.
Other variables considered by the copyrighted algorithm include the individual’s proximity to highways, access to vehicles and the dates of the charge, according to Joe Gauthier, VP of sales and marketing for Watch Systems, the parent company of OffenderWatch. The company works with more than 3,500 law enforcement agencies across the country, facilitating the sharing of offender records among all law enforcement users on the network.
With data from the OffenderWatch system, Focus analyzes more than 100 different risk factors found in a registered sex offender’s record, as well as other state, federal and commercial data sources, to assign a score to a registered offender. It doesn’t, however, take into account physical descriptors like height, weight or race [that] can introduce bias, Gauthier told GCN. And none of the information is stored locally; users access it through a thin client.
What makes Focus different from other predictive policing tools is that it updates frequently and ranks a given pool of offenders.
“The only reason the predictive analytics works in the sex offender world" is because released offenders are required to provide frequent updates to law enforcement on their personal life, such as if they move or change their physical appearance. This, Cormaci said, results in more accurate analysis.
Focus scores offenders on a 10-point scale and then evenly divides the population into each of the 10 levels, ranking offenders likelihood to reoffend, Cormaci said. “So if you have 100 offenders there’s going to be 10 in the high-risk '10' category and 10 in the 'one' category,” he said. “If you have 300 it will be 30 in each one.”
“What we find is that because of the resource demand, all offenders are treated equally,” Cormaci said, so the ranking helps law enforcement agencies prioritize which offenders to check on.
Deputy Mike Robinson manages offender monitoring for the Vanderburgh County Sheriff’s Office. Before using Focus, he prioritized offenders by the state’s three classification levels: sex offender, offender against children and sexually violent predator.
“The Focus score looks beyond what the classification is,” Robinson said.
To evaluate the Focus scoring, Robinson went back and looked at people who reoffended in Vanderburgh County, and he saw their risk scores were higher.
One reoffender had no suspicious interactions with law enforcement, but when officers retroactively checked his risk score, it was a 10. “He was a compliant registrant, he always lived where he was supposed to be, he always reported, but that doesn’t always mean” the offender is remaining compliant, Robinson said.
Robinson uses the scores to determine which offenders he should prioritize for check-ins and if he should bring another officer along for the task. The officers only get the score -- they don't see any detail on how the score was determined.
“What we’re finding is the Focus scores give us a little bit more clarity as to some of these people who may go out and harm another individual,” he said.
Vanderburgh County has been using Focus since last year, but it wasn’t fully operational until January, Robinson said. His office is still working out exactly how the tool will change its workflow, but so far Robinson is optimistic about what he is seeing.
“We’re pretty new to getting this thing fielded,” Robinson said. “But … I can see how it can be used to prioritize our verifications.”
Matt Leonard is a reporter/producer at GCN.
Before joining GCN, Leonard worked as a local reporter for The Smithfield Times in southeastern Virginia. In his time there he wrote about town council meetings, local crime and what to do if a beaver dam floods your back yard. Over the last few years, he has spent time at The Commonwealth Times, The Denver Post and WTVR-CBS 6. He is a graduate of Virginia Commonwealth University, where he received the faculty award for print and online journalism.
Leonard can be contacted at firstname.lastname@example.org or follow him on Twitter @Matt_Lnrd.
Click here for previous articles by Leonard.