Risk Assessment Tool Led to Harsher Sentences for Young and Black Defendants


An analysis of sentencing in Virginia showed that the state’s risk assessment system increased sentences for Black and young defendants, The Washington Post reported.

Researchers at George Mason University and Texas A&M examined tens of thousands of felony convictions in Virginia, focusing on the period between 2000 and 2004, to measure how judges used a risk assessment algorithm adopted statewide in 2002.

The study revealed that defendants younger than 23 were 4 percentage points more likely to be incarcerated after risk assessment was adopted, and their sentences were 12% longer than their older peers.

Risk scores are largely based on age—in Virginia, 13 points are added to the score for being younger than 30. (Having five or more previous incarcerations as an adult adds only 9 points.) A 2018 analysis showed 58% of the popular COMPAS algorithm’s score can be traced to age.

“People are getting this very stigmatic label—high risk for violent recidivism—largely because they’re 19 years old,” researcher Megan Stevenson said.

Racial disparities also increased in the Virginia circuits that used risk assessment most. The algorithm doesn’t use race in its calculations, but Black defendants were 4 percentage points more likely to be incarcerated after risk assessment was adopted, compared with otherwise equivalent white defendants. Black defendants’ sentences were also 17% longer.

The risk assessment system was intended to divert from prison people convicted of nonviolent offenses to free up space for people convicted of violent offenses, who were expected to serve longer terms after Virginia abolished discretionary parole, Meredith Farrar-Owens, director of the Virginia Criminal Sentencing Commission, told The Post.

But the algorithm did not change the rate at which people were incarcerated or the length of their sentences, in part because judges did not follow the algorithm’s suggestions in most cases.

Judges were supposed to impose shorter sentences on people with low risk scores, or send them to probation or substance abuse treatment instead of prison. Some judges reported they weren’t trained to use the risk assessment tool; others said no diversion programs were available. Some didn’t use the risk scores at all in their sentencing decisions.

If judges had fully complied with the algorithm’s recommendations, researchers found, disparities based on race and age would have been even greater.

“Virginia’s nonviolent risk assessment reduced neither incarceration nor recidivism; its use disadvantaged a vulnerable group (the young); and failed to reduce racial disparities,” the study authors concluded.

Similar algorithms are used in 28 states.