It’s been nearly a decade since state lawmakers, eyeing a prison population that had quadrupled in size and reflected stark racial disparities, decided to take action. They ordered the Pennsylvania Commission on Sentencing to create a risk-assessment tool that would give judges the information they needed to divert low-risk individuals from prison into alternative sentencing programs.
Nine years and five drafts later, the commission is scheduled to vote on a proposal Thursday. The problem is, critics say, the risk assessment they’ve developed would reinforce racial and gender biases, could actually increase incarceration, and would likely invite a legal challenge on constitutional grounds.
Even former State Sen. Stewart Greenleaf, the Montgomery County Republican who was behind the 2010 legislation, has rejected the proposal.
“I urge the commission to adopt a risk assessment only if it will significantly reduce incarceration and racial disparities," he wrote to the commission in a memo. “Otherwise, we run the risk of reinforcing the drivers of over-incarceration for decades to come.”
But the 11-member commission is poised to move forward, potentially approving a plan in its final session before a shake-up engineered by state lawmakers takes effect, removing Gov. Tom Wolf’s three appointees and replacing them with legislative picks.
“There are a lot of people on the commission who are tired of talking about it,” said State Sen. Sharif Street, a Philadelphia Democrat who sits on the commission and expressed grave concerns about the most recent iteration of the risk-assessment instrument. “I think the commission members who have worked on it for a while would like to see it through to completion.”
State Rep. Todd Stephens, vice chairman of the commission and a Montgomery County Republican, noted that the commission has responded to feedback and addressed many concerns, improving the instrument in response to several rounds of public testimony. He said continuing on the current path, sentencing without in-depth assessments in many cases, represents an ongoing injustice.
“This mandate is nearly a decade old now," he said. "It’s time to put something up for a vote and get it out into use in the criminal justice system in Pennsylvania.”
Under the most recent plan, defendants would be scored based on their prior criminal history, age, and gender. Those found to be either high or low risk would be flagged for a further assessment of their risk for recidivism and need for treatment or support. For that second-level review, the commission proposed putting those individuals through whatever assessment is used by the county probation department. In Philadelphia, that’s particularly controversial, as the risk-assessment instrument used by the probation department is a complex “black box” algorithm, meaning that it’s unclear how it weighs various factors in assessing any given individual.
Formerly incarcerated people, civil-rights advocates, lawyers and officials lined up at public hearings in August to express concern about the proposal and to pledge to continue fighting the risk-assessment tool they’ve unaffectionately nicknamed the "RAT.”
In a letter to the commission, the County Chief Adult Probation and Parole Officers Association also warned the plan could slow case processing, calling it “an unnecessary burden.”
Sonja Starr, a University of Michigan law professor who has studied risk assessments at sentencing, warned that combining multiple risk assessments introduces new sources of inequality and undermines transparency. In a letter to the commission, she also condemned the plan on legal grounds. “Use of gender as a risk factor plainly violates the Constitution," Starr warned, adding that the Equal Protection Clause “forbids the government from treating people differently on the basis of sex even when there is a statistical basis for doing so.”
In an interview, Stephens acknowledged that the commission may need to re-evaluate what happens once the risk tool identifies an individual for further review.
A team from Carnegie Mellon University that analyzed the proposal urged policymakers to restrict usage to identifying low-risk individuals who could be diverted from prison — noting that when the tool was used to identify those at high risk of re-offending, it was accurate only about half the time. However, the commission rejected that finding.
“It was envisioned that this was going to reduce costs, reduce mass incarceration,” Street said. “Now, we’re in a place talking about it may increase mass incarceration.”