For all the talk in this nation about bail reform the past few years, the move to the use of risk assessment as a one-size-fits-all solution has been touted and embraced as the cure to the supposed ills of cash bail. At first finding favor with advocacy groups, its reach has extended to key decision makers, finally reaching the highest levels of our court system.
Five years ago, the Conference of Chief Justices (CCJ) formally adopted a policy supporting the nationwide use of evidence-based pretrial risk assessment. They also endorsed a policy of the Conference of State Court Administrators, arguing that use of risk assessment tools to make judicial decisions to release or not based on its results, “fits within a well-functioning case management regimen.”
CCJ’s statement is disturbingly sweeping because it includes the use of a pretrial risk assessment algorithm not only to set a bail, but also to inform a decision of preventative detention where allowed. Of the latter, it calls for its use under any attendant conditions, which includes liberty-restricting conditions including GPS monitoring, drug screening, check-ins, etc.
It is important to keep in mind that “attendant conditions” are often paid by defendants themselves. In effect, these individuals must pay for their own personal criminal dragnet.
Lawyers who have not practiced criminal law in a major urban jurisdiction, may not be familiar with pretrial risk assessment tools, which are sold to score the risk of a defendant of no-showing for court or being rearrested for a new crime. Those that do, know how they’ve been forced to hone their arguments based on the binary manner in which the tool has dehumanized their client, herded him or her into a group based on data, and ultimately, seared a criminal scarlet letter number onto their forehead.
When the score is favorable to one’s client, attorneys vociferously argue that the tool is scientific. When it is unfavorable, suggesting they will not show up to court or will commit a new crime, counselors will claim it is not scientific and is merely one instrument in the toolbox. These premises must surely be behind a favorite legal colloquialism, “Everyone loves judicial discretion…until they lose.”
As it turns out, risk assessments are not scientific and likely may magnify or at least cement existing bias against protected classes within the criminal justice system.
When it was recently suggested to a Spokane, Washington judge that risk assessments were only 66 percent accurate, the defensive response was that the Public Safety Assessment tool built by the Arnold Foundation was actually 72 percent accurate and “more useful in setting bail than flipping a coin.” Notwithstanding that the statement did not exactly inspire confidence, could the term “predictive” be used in any other field of science when it is wrong 28 percent of the time? How could such error-riddled labeling even be called justice?
This is not an isolated view. Over the past two years, civil rights and digital technology groups, along with numerous prominent academics from across the nation have come out in opposition to the use of these tools. This includes the national ACLU and NAACP; Google, IBM, Apple and Amazon; and researchers from the nation’s top universities, including Harvard and Princeton.
One of the central arguments against the use of risk assessment tools is that they may increase racial bias. The standard response from advocates, including the Arnold Foundation, is that the tools do not make existing racial bias worse, and, in any event, judges can always simply ignore the results, despite the fact that they are sold as scientifically validated. Conveniently forgotten in this claim, however, is the original promise made by John Arnold, as he money-balled the criminal justice system, building a public safety risk assessment, specifically designed to reduce racial bias and increase objectivity. In fact, the Pretrial Justice Institute — which has now reversed course 180 degrees and opposes pretrial risk assessment — in 2017, deemed both the Arnold Foundation and Virginia’s Pretrial Assessment tools as race neutral.
Evidence against risk assessment tools has continued to mount. Peer-reviewed research by Megan Stevenson of the University of Virginia School of Law debunked the Arnold tool’s performance in Kentucky, which years earlier had been heralded as a success. Results from other studies are nearly identical to one another, finding that the tools do not generally make a given situation better and often make it worse.
Last year, a study of the Virginia Pretrial Risk Assessment tool, used in Palm Beach County, Florida, found that it was race and gender-biased. Conducted by Florida State University in 2019, it determined that its risk levels were “not similarly indicative of pretrial risk for defendants of color and whites or for female and male defendants.” It also concluded that “defendants of color are being assessed as higher risk,” despite the fact that their risk of reoffending was low to being with and consistent with then-existing pretrial risk assessments.
Meanwhile, Mesa County, Colorado, doubled its jail population after initiating bail reform, which included the use of the Colorado Pretrial Assessment Tool, begun roughly a decade ago despite flat population growth.
Despite all of this, the Conference of Chief Justices — our nation’s highest judges — are moving forward with a continuing effort to force every jurisdiction to use pretrial risk assessments on all criminal defendants. Efforts from my office, through formal letters sent on behalf of the American Bail Coalition, urging CCJ to rescind and revisit this destructive policy, were met with no response. In 2018, we wrote, “The damage that the Conference of Chief Justice’s policy [in favor of pretrial risk assessment] has done to the rule of law has been immeasurable. It has shaken confidence in the independence of the judiciary and politicized the judicial branch.”
A second letter was sent from my office to the Conference of Chief Justices in 2020, following the landmark joint statement by 27 prominent academics from Harvard, Princeton, MIT, Columbia, NYU and UC Berkeley, stating that, in their professional opinion, pretrial risk assessments were fatally flawed and beyond repair. Again, CCJ did not even acknowledge receipt of the communication.
Despite these efforts and mountains of evidence piling up against risk assessment tools, the policies of the Conference of Chief Justices remains in force and is still being used by those pushing for their use. This includes the chief justices of Colorado, Montana, Ohio, Utah and Texas, who have used their judicial power in recent years to advocate for pretrial risk assessments. As advocacy groups continue to call into question their increasingly dubious validity, its proponents have still been able to counter with the support of the Conference of Chief Justices, which remains a powerful force in favor of these pretrial algorithmic tools.
Let there be no doubt, the public trial on pretrial risk assessments is over. The jury has come back with a verdict and declared that it is against the policies of the Conference of Chief Justices, which do not provide for equal justice under law. Ironically, CCJ was responsible for their default in the debate — a finding they themselves would never overturn on appeal.