Adding COMPAS assessment to sentencing report stirs controversy



By Cynthia Price
Legal News

A training held by the State Appellate Defender Office (SADO) last Friday turned at times into a forum to air concerns about the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS).

The trainers took advantage of Thomas M. Cooley Law School’s ability to simulcast presentations in their four Michigan campus locations to offer the free training, entitled “Use of COMPAS at Sentencing: What every Defense Attorney Needs to Know.” Presenter Jacqueline McCann, assistant defender with SADO, was at the Auburn Hills campus. Marla McCowan, SADO’s Criminal Defense Resource Center Manager, opened up the session from Grand Rapids, while others facilitated the venues in Lansing and Ann Arbor.

This meant that there was a fairly sizeable group of defense attorneys in attendance to find out more about Department of Corrections use of the COMPAS assessment tool in pre-sentencing investigation (PSI) reports.

Many of those attorneys registered opposition to such a use.

According to McCowan, McCann is SADO’s expert on “everything that has to do with sentencing law.” She has extensive experience arguing at the Court of Appeals and the Michigan Supreme Court, particularly about interpretations of statutory sentencing guidelines, and wrote the Defender Sentencing Book and the Defender Sentencing Guidelines Manual Annotated.

Also present at the Auburn Hills site was Sonja B. Starr, a University of Michigan Law School professor whose article, “Evidence-Based Sentencing and the Scientific Rationalization of Discrimination,” is just a few weeks away from publication in the Stanford Law Review. McCowan confessed, “We have pretty brazenly stolen from that paper.”

McCann noted at the outset that the term “evidence-based” in reference to such assessment tools may be counterintuitive. “What it is not is about the evidence in your case,” she said. “What it does refer to is the empirical research on factors predicting criminal recidivism or violent behavior.”

COMPAS is a computer software tool, involving a set of questions (137 in the case of adult males) that a probation agent must ask each offender as part of completing the PSI. The answer to those questions are then processed through a computer by the for-profit company who licenses COMPAS, Northpointe Inc.

According to Northpointe’s website, “COMPAS is composed of 22 different scales that empirical research has identified as predictive of future behavior. The 22 scales are grouped into five main categories: criminal involvement, relationships/life-styles, personality/attitudes, family, and social exclusion. COMPAS assesses three categories of risk: recidivism, violence, and failure to appear at a court hearing.”

As McCann pointed out Friday, even  Northpointe never claimed that COMPAS was intended for use in guiding sentencing. Its main purpose, as is clear from its title, is to help corrections personnel with decision-making regarding how to handle offenders.

Corrections departments, both nationwide and in Michigan, have been using similar assessments for years. The Michigan Prisoner Re-Entry Initiative, in 2004, started administering such instruments to guide decisions on what prisoners were likely to succeed.

Where such a tool becomes problematic is when it is applied to decisions that are, by statute, supposed to be decided on a case-per-case basis.

While the use of the COMPAS results is specifically precluded from consideration in determening the length of a sentence, the fact that the data will be included in every PSI raises red flags for SADO?and defense attorneys.

The Criminal Law Section of the State Bar of Michigan has also registered  is concerns about COMPAS. In a letter to the presidents of the Michigan Judges Association and the Prosecuting Attorneys Association of Michigan, the section chair, Judge David Hoort of Ionia County, expressed the section’s  misgivings about the reliability of COMPAS and possible constitutional and state-law objects.

Whether the PSI will include the information in its conclusions and recommendations or the probation agent will simply attach the results is as yet unclear. McCann said that she had witnessed one instance in a pilot project where the agent summarized the results in the body of the report, but the DOC has not issued any official rules yet.

Regardless, the SADO presentation called out a number of issues.

People v. Eason states, “[D]ue process is satisfied so long as the information the sentencing judge considers has sufficient indicia of reliability,” and SADO questions whether that is the case with COMPAS. Northpointe claims it is 64 to 80 percent accurate, which its evaluators determined by comparing, for example, an individual who fails to appear in court to COMPAS?predictions of whether he or she would fail to appear.

Michigan sentencing guidelines mandate individualized sentencing, not sentencing based on generalizations, and the “evidence-based” approach categorizes people by generalized groups, making it in appropriate for judicial decision-making, several present felt.

Moreover, a different test is given to different offender groups, pulling out women and youth. This, as well as the implicit biases involved in asking questions about whether the offender comes from a background of poverty, raises constitutional issues.

Professor Starr said that the notion that the tests encroach on people’s right to equal treatment and due process under the law was a significant finding of her research. “There really haven’t been constitutional challenges brought yet,” she said, “mostly because the 20states that have this have adopted it

so recently.”

The summary of her highly critical article states, “This paper critiques, on legal and empirical grounds, the growing trend of basing criminal sentences on actuarial recidivism risk prediction instruments that include demographic and socioeconomic variables. I argue that this practice violates the Equal Protection Clause and is bad policy: an explicit embrace of otherwise-condemned discrimination, sanitized by scientific language.”

Starr also wondered if the questionnaire impinged on the freedom from self-incrimination.

The presenter and participants alike surmised that one of the reasons the Department of Corrections has changed its original effective date, from June 1 to one as-yet-undetermined, was the negative feedback it received so far.

One man said he felt that attorneys should be allowed to be present while the questionnaire is administered. McCann said attorneys should at the very least be able to access the questions prior to their being asked. 

The training was made possible by a grant from the Michigan Commission on Law Enforcement Standards.