Researchers at the University of Nebraska at Omaha have built a new model of using a criminal charge to predict a future criminal offense.
In a new study published in the journal “Criminology,” they found the risk for a reoffense drops exponentially each year after someone is criminally charged, returning to the same level as the rest of the population within about a decade.
“This study bridges a gap between prior knowledge of the age-crime curve and practical utility or criminal history in probation, parole, pretrial supervision, as well as employment background checks,” Zachary Hamilton, a professor of criminology and criminal justice and lead author of the study, wrote in a press release this week.
While current risk-assessment systems assign a fixed weight to a previous charge, the researchers said new versions of the tools should instead let the weight of a criminal record change over time.
“Essentially, we found a way to make criminal history dynamic, identifying when a prior criminal charge depreciates, until a point that it loses utility and can be eliminated,” Jennifer Tostlebe, an assistant professor at UNO and a co-author of the study, added in the press release.
The research looked at more than 16 million people from Alaska, Arizona, Florida, Texas and Washington. The people were tracked during a 40-year period from 1980 to 2020, resulting in one of the largest datasets of its kind.
The team found that contrary to common belief, up to two-thirds of people charged with a crime were never charged again.On a webpage laying out their findings, the researchers instead noted that a handful of people are charged again multiple times, “inflating the apparent share of the whole population that reoffends, and slows the apparent decay of risk over time.”
The misconception is traced to previous models that considered reoffenses as independent events — treating one person with five charges the same as five people with one charge.
To account for this factor, the UNO team built its statistical model to “cluster” reoffenses committed by the same person. Doing so recalculates the probability of a reoffense in the first year alone from 22% to 5% — and sees the probability of an offense drop to 1% within nine years of the initial charge.
Additionally, the study found that the predictive ability of charges is changing over time; people charged between 2001 and 2020 were at a much lower risk of reoffending than those charged between 1980 and 2000. The study’s authors wrote that this means, “estimates built on more contemporary data are likely a more accurate picture,” of the probability of a reoffense.
The full study is available to read through the journal’s open-access page.