COMPASS provides little direction for defense: Implementation of program has been postponed pending further discussion


 By John Minnis

Legal News

After being “trained” for two hours recently on the “Use of COMPAS,” defense attorneys present found little, if anything, they liked about the proposed evidence-based risk assessment program being proposed by the Michigan Department of Corrections.

“This is a little scary to me as a defense attorney,” said John English, of Lansing. “You’re telling me this data has nothing to do with my client, but it will be attached to my client’s PSI (presentencing investigation). This looks like a setup to me.”

English viewed a simulcast of the May 2 training seminar, “Use of COMPAS: What Every Defense Attorney Needs to Know,” conducted by the Michigan State Appellate Defender Office at Cooley Law School’s Auburn Hills campus. It was simulcast at Cooley’s campuses in Ann Arbor, Grand Rapids and Lansing. Two classrooms of 50 attendees each were occupied at Auburn Hills alone. 

“We’re very grateful to Cooley Law School for letting us be here today,” said Marla McCowan, assistant defender and manager of SADO’s Criminal Defense Resource Center. “We’re just about at capacity at all our locations.”

Anticipating objections such as those from English, McCowan added, “We are defense attorneys doing this training and we are unapologetically defensive.”

According to SADO, COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) is a computerized assessment and case management system created by the Northpointe Institute for Public Management. Michigan sought to begin using the risk assessment program on June 1; however, implementation has been postponed pending further discussion.  

When COMPAS is implemented, according to SADO, all PSIs must include risk assessment pursuant to a new Michigan Department of Corrections policy. The results are intended to inform sentencing discretion in terms of the rehabilitation and supervision needs of an offender in the community. The training seminar, “Use of COMPAS at Sentencing: What Every Defense Attorney Needs to Know,” was intended to be a primer with an emphasis on the meaning of COMPAS scores, reliability of the scores, limitations on their use and the constitutional and state-law objections to raise at sentencing hearings.  

“What evidence-based sentencing is about is not the evidence of your case,” said Jacqueline McCann, assistant defender, author of the “Sentencing Book” and the “Defender Sentencing Guidelines Manual Annotated,” and the primary trainer for SADO who conducted the Cooley training session on COMPAS. “What evidence-based sentencing refers to is defendant recidivism. It’s not about your particular client. It is about the average recidivism of offenders.”

All state prisoners have been COMPAS-screened since 2009. “As defense attorneys, you are seeing COMPAS come up at parole hearings,” McCann said. However, “the report is not intended to determine the length of sentences.”

The COMPAS questionnaire contains 137 questions, covering areas such as criminal history, gang association, crime in the offender’s neighborhood, access to and perceived need for guns, income, education and so on. Offenders can answer the questions themselves or they can be administered by trained personnel using scripts or they can be presented more “dynamically.” 

The answers are entered into the COMPAS database to determine the offender’s “risk assessment” from 1 to 10, where 1 is low risk and 10 is high risk. Areas of risk include violence, recidivism, noncompliance and failure to report. Further, populations are rated differently. A woman defendant will always be rated a lower risk than a man even though they answered the questions identically. 

“COMPAS explicitly considers gender, age and socioeconomic factors in evaluating risk,” McCann said. “This flies in the face of Michigan law concerning individual versus group sentencing. We’re doing our utmost to challenge this.”

She again noted that while Northpointe did not design COMPAS for sentencing, “it appears the DoC may want to use this for sentencing. On individual cases, you may want to ask the judge to strike it.”

Evidence-based risk assessment of offenders is being done in other states, McCann said, and there have been challenges. The Indiana Supreme Court has sanctioned it, she said. 

The defense attorneys present questioned the reliability and accuracy of the COMPAS reports, citing “garbage in, garbage out.” McCann urged defense attorneys to obtain their clients’ answered questionnaires to look for inaccuracies. Others wondered whether offenders would have the right to have an attorney present when answering the questionnaire. “Generally not,” the SADO assistant defender said, “but they can’t not be there either.”

McCann said the accuracy of COMPAS-type risk assessments is said to be 76 percent, 64 percent for African-Americans. “Foremost, we can argue COMPAS results are not reliable enough,” she said. “We should at least consider the inadmissibility of the results.”

A defense attorney from Genesee County opined that an offender can refuse to answer questions on Fifth Amendment grounds, but McCann countered, “I’m sure they are not going to write it up favorably.”

The leading scholar in the field of evidence-based risk assessment is Sonja B. Starr of the University of Michigan Law School, author of “Evidence-Based Sentencing and the Scientific Rationalization of Discrimination.” She was in attendance at the Cooley simulcast in Ann Arbor.

 “Sonya Starr’s research in evidence-based sentencing is tremendous,” McCowan of SADO said, “and basically what this is based on today.”

Starr called evidence-based assessments such as COMPAS “statistical generalizations.” Most of the legal objections she enumerates have been found in the parole process, she said.

“There really haven’t been constitutional challenges on this,” Starr said. 

She also questioned the reported accuracy of evidence-based tools such as COMPAS. “The 76 percent accuracy figure makes these instruments seem more accurate than they are,” she said. “Perfect chance accuracy is 50 percent. A coin flip will get it right half the time.”

Another attorney questioned whether it would be in the offender’s best interest to not answer the COMPAS questions under Fifth Amendment grounds, especially if answering them would result in a higher risk assessment.

“We’re certainly in a brave new world with this stuff,” public defender McCann said. “Any objection you can make is worth trying.”