A new study by UC Berkeley professor of social welfare Jennifer Skeem debunks the assumption that risk assessments of the likeliness that an offender leaving prison will be rearrested are racially biased.
Published in the November issue of Criminology, the study found that rearrest risk predictions were accurate regardless of race, despite critics claiming that scores were affected by race and contribute to the issue of mass incarceration. The study was co-authored by Skeem and Christopher Lowenkamp, a social science analyst at the Administrative Office of the U.S. Courts.
Skeem said the primary motivation behind the study was to determine the validity of risk assessment tools as it relates to race, which can be used to identify intervention methods to reduce subsequent returns to prison.
To identify the effect of race on risk assessment, approximately 33,000 Black and white offenders were randomly matched with one another and monitored for at least a year. Results from the study concluded that race was not a major factor in the prediction score of rearrest.
“Although a lot of (risk assessment scores) are being used in different ways to try … to reduce prison populations, very few people had been paying attention to how they were related to race,” Skeem said. “A number of scholars (were) raising questions about whether the use of the instruments could increase disparity about incarceration.”
The Post Conviction Risk Assessment, or PCRA, was created to improve supervision programs for released convicted offenders. According to a report by the Administrative Office of the U.S. Courts, the PCRA consists of two sections: one filled out by the probation officer and the other by the offender in question, answering questions in seven domains, including criminal history, education, substance abuse and social networks.
“Looking at the scores that a group of Black offenders get and white offenders get, the overlap is about 73 percent … even though there’s a tendency for the Black group to have slightly higher scores,” Skeem said.
According to the study, most of the racial difference in risk assessment scores can be traced back to criminal history.
The study comes after a major ProPublica article in May implied that software automatically assigning risk scores discriminates against people of color. After analyzing data from Broward County, Florida, the authors of the ProPublica story concluded that Black offenders were more likely to receive a higher risk assessment score than their white counterparts — meaning they could receive more intense services in order to prevent future misdemeanors.
“Our article prompted a nationwide discussion of the appropriate use of criminal risk scores,” said Julia Angwin, a senior reporter at ProPublica and a co-author of the article, in an email.
Skeem said the United States has an exceptionally high incarceration rate. Research conducted by the Bureau of Justice Statistics found that about 67.8 percent of the prisoners tracked by the study released in 2005 were rearrested within three years and about 76.6 percent were rearrested within five years.
“I think the most important movement that’s taking place is trying to reduce how big the population is of people who are incarcerated,” Skeem said. “One of the ways to do that is (to) use risk assessment to help inform decisions about who can be (appropriately) incarcerated.”