Md. Bail Assessment Tool Differs From Other States
ANNAPOLIS, Md. (AP) - Gov. Martin O'Malley has said he may issue an executive order creating a pilot program for reforms to criminal bail procedures.
It would be the next chapter in Maryland's efforts to comply with a Court of Appeals ruling that defendants have a right to legal counsel at every bail hearing. Legislators failed to agree on a solution this past session.
The order would create a pilot program that uses data on each defendant, such as criminal history, to assess the person's risk of committing more crimes if released from jail.
Some lawmakers have argued such assessment tools essentially turn criminal justice decisions over to computers. They also worry the tools haven't been properly tested in the way Maryland may start using them, and they worry about potential bias against black defendants, saying blacks are statistically more likely to have prior arrests on their records.
Maryland leaders have discussed using a risk assessment tool that is simply a mathematical formula. In other states, after the risk assessment is done, a judge or court employee makes the final decision on each defendant
Sen. Brian Frosh, D-Montgomery, who pushed this year for a data-based assessment program, has suggested using a tool like the one developed by a nonprofit group called The Arnold Foundation.
Kentucky uses the foundation's tool statewide, and jurisdictions in in Arizona, California, Colorado and North Carolina recently announced plans to use it as well.
The foundation collected data from 750,000 cases across the country and analyzed common traits between defendants who skipped court dates or committed more crimes after being released from jail. Besides criminal history, they looked at traits like residence, employment status, education, drug use habits and mental illness.
Ultimately they identified nine factors that were most predictive. All were based on criminal history and the person's newest criminal charge. They worked these into a formula that assigns each defendant a risk score.
Anne Milgram, a former New Jersey attorney general now working with the Arnold Foundation, said criminal justice is far behind industries like health care and education in reliance on data.
"There is no area in America I can think of where it's more important to do it," Milgram said.
Milgram said computers aren't necessary to run the Arnold tool - it can usually be done on paper within 30 seconds. So the process doesn't need to shut down when technology malfunctions.
As for prior arrests, Milgram said some risk assessment tools do account for them, but the Arnold Foundation's does not.
However, naysayers in Maryland are right concerning a key point: In Maryland, policymakers have discussed keeping human discretion out of the bail process, while in other states a court official makes the final decision.
Milgram said every jurisdiction using the Arnold tool still has a judge or court employee making the final decision about each defendant. The risk score is just a guide.
Maryland officials have been focusing on using the formula by itself, as a way to cut costs.
The state's highest court has ruled that any time a judge or court worker is deciding whether to hold a person in jail, the defendant has the right to a defense attorney. Providing attorneys for all these hearings would cost millions of dollars annually, which legislators have been trying to avoid.
Some Maryland lawmakers suggest that if the decision isn't left to human discretion, the right to defense counsel likely won't apply.
Seventeen Kentucky counties are trying a program where defendants with low risk scores are automatically released. But even there, employees can ask a judge to interfere if a defendant seems especially dangerous, said Tara Klute, manager of the state's pretrial services.
Frosh said that "in a perfect world," Maryland would imitate this approach.
"But it's not a perfect world," he said.
However, a pilot program would let legislators compare results between states that use data-based risk assessment and those that don't. Frosh said this creates ideal scientific conditions for evaluation.