“Weapons of Math-Destruction”: A Pathway to Criminal Justice Reform is Threatened by Fear of Risk As
“These advanced mathematical models — or “weapons of math destruction” as data scientist Cathy O’Neil calls them — appear colorblind on the surface but they are based on factors that are not only highly correlated with race and class, but are also significantly influenced by pervasive bias in the criminal justice system.” -Michelle Alexander
I disagree with Michelle Alexander. Risk assessment is more reliable than the decision making of a prosecutor or judge. As you can imagine when professionals are allowed to use unfettered discretion in deciding the fates of individuals, especially those with whom they cannot relate, due to race, gender, class or other easily identified differences, they rely on stereotypes and generalizations which may or may not be confirmed by their personal experiences or observations. In psychology we call these attribution biases. In other circles, we simply call it prejudice. We like to believe that professionals and clinicians are trained to put aside those errors in discernment, and objectively assess the individuals before them, but unfortunately human bias is a persistent bugger in the human subconscious. Historically, this has led to widespread disparate treatment of marginalized populations in the justice system, specifically for African American men and increasingly African American women. African American men and women are five and two times as likely as their White counterparts to be arrested. While most attempts to address these disparities focus on reducing the offending behavior of African Americans; criminologists privately acknowledge differential treatment by justice systems is just as much a factor in minority over-representation in the prison-industrial complex. Michelle Alexander called this form of institutionalized racism the New Jim Crow. I agree with Michelle Alexander. Many do. In fact, mass incarceration of African Americans has emerged as the signature Civil Rights issue of this generation, giving rise to mass mobilization efforts like Black Lives Matters. Black Lives Matter ultimately challenged the Democratic National Party’s unwillingness to offer a concrete and specific policy agenda to bring an end to mass incarceration. That challenge, leading up to the 2016 Presidential election, factored in the disillusionment of minority youth voters and their abandonment of Presidential Candidate Hillary Clinton at the polls. While remaining stalwarts of the Civil Rights Movement era have criticized BLM and other nascent Black activist groups for breaking with the traditional Black Voting Bloc that overwhelmingly supports Democratic candidates, the calculation on the part of young African American strategists by simply not voting or choosing third party candidates appears to be bearing fruit. Conservatives, presenting themselves as Trump populists have seized on mass incarceration as a wedge issue for Democrats and proposed the first major justice reform bill in nearly two decades, calling it The FIRST STEP Act.
The issue is so powerful that Midterm elections boasted two African American gubernatorial candidates in the battleground states of Georgia and Florida; who eventually lost by margins that could easily have been overcome by a simple Black voting bloc. This voting bloc never emerged with their Republican opponents earning double digit support from Black voters in the face of massive African American turnout. While both candidates appealed to the rightful ire of Black voters having to confront widespread voter suppression instigated by republican state officials, neither offered a concrete agenda for significant justice reform regarding police brutality or mass incarceration in their states, where disproportionate minority contact ranks high in comparison to other states. This failure to acknowledge mass incarceration was highlighted when Donald Trump signaled his support of The FIRST STEP Act to reform federal corrections in the wake of the so-called Blue Wave that failed to sweep Congress into the hands of progressives. Not to be outdone, the establishment within the DNP quickly turned to its old guard of professional Civil Rights organizations, producing a letter from the Leadership Conference on Civil and Human Rights with “108 undersigned” groups demanding that advocates “Vote No on The FIRST STEP Act”. In a scathing policy critique that echoed the sentiments of Michelle Alexander, they argued that in a legislation that does not go far enough, the policy is doubly harmful because it institutionalizes risk assessment, which to their understanding is discriminatory against African Americans. The critique rightly points to cases where risk assessments incorrectly classified offenders as future repeat offenders, who did not in fact re-offend. It also rightly points out that risk assessments that rely on faulty correlations between race and crime are also problematic. Where they erred is in failing to explain that modern risk assessments no longer rely on fixed characteristics like race and it is protocol to “validate” risk assessments with the offender population they are to be used on before relying on their recommendations in case-management. The argument goes to great length depicting risk assessment as cold and robotic, given to failure to account for context and essentially amounting to an attempt to hide racial profiling behind an “algorithm”.
As I said, I disagree with Michelle Alexander and the old guard of the Democratic Party establishment, not in their justified resistance of a Trump Presidency that has proven itself duplicitous and opportunistic in its use of populist sentiments regarding immigration, reproductive rights and now the mass incarceration of African Americans to secure their hold on power. But in their taking advantage of the public’s general lack of knowledge about risk assessment’s present and long-standing central role in case-management, and ultimately its implications for reducing the footprint of mass incarceration on Black Communities. The insidious relationship between Black Leadership and their allies within the Democratic National Party, has created a huge policy blind side for marginalized Black people, which emerges in contexts like the 1994 Crime Bill or the Obama era Blue Lives Matter law, where African American members of Congress supported the very measures that excuse police brutality and give rise to mass incarceration and the larger prison-industrial complex that require reform today.
Enter Recidivism risk-based adjudication… My seminal work was on the utility of risk assessments in reducing disproportionate minority contact. I conducted a ten-year study of a juvenile justice system that incorporated risk assessment in its case-management and disposition decision. The use of risk assessment resulted in over 30% of young offenders being diverted from further contact with the justice system based on their low likelihood of recidivism. That was nearly 3,000 youth in a single system. In tracking the outcomes of these diverted youth in comparison to youth with similar risk levels that were steered deeper into the system, I found no differences in offense rates. The community at large mirrored larger national trends with juvenile offending declining over the decade. On average risk assessment beat prosecutors and judges by fifteen percent in predicting recidivism. Bias is so strong that human judgment is no better than flipping a coin in predicting whether an offender will recidivate. By demonstrating that offenders with similar risk levels had similar recidivism outcomes whether they were diverted from the justice system or received intensive supervision, this seriously begs to question the worthwhileness of expending resources that make no difference other than profiting the prison-industrial complex. A question only posed when made apparent by the implementation of risk assessment in justice systems, rather than relying on eye-ball judgments of practitioners.
Risk assessment is premised on a little-known fact outside of criminologist circles. A small group of offenders commit the majority of crime through repeat offending (i.e., recidivism). This small group can be as few as six percent of offenders yet commit over half of the crime in a community. These repeat offenders have characteristics in common, known as risk factors that are associated with delinquency, crime and recidivism. To qualify as a risk factor, the attribute must be amenable to intervention, like education or drug involvement, not things like age, race or gender. When these characteristics are combined into a psychological measure, they are called risk assessments and are often used to both predict criminal behavior of individual offenders and determine services that directly address the need areas of the offender. In theory, using this approach reduces the likelihood of low risk and high-risk offenders being mixed, which increases offending on the part of minor offenders. Low risk offenders can be diverted from formal contact with the system. High risk offenders can be targeted for programming. Coincidentally, the majority of most offenders in the US are low risk non-violent offenders and thus technically eligible for diversion based on their recidivism risk.
That said this model is very much dependent on a number of factors and trends in criminal justice. First and foremost, this system is predicated on the belief that diagnostic tools like risk assessments are accurate predictors of human behavior. Secondly, to the extent these diagnostic tools represent an improvement over human judgement (ie., probation officers, prosecutors, psychologists, judges) in predicting future criminal behavior; practitioners must ensure that the specific risk assessments used in their communities are accurate with African Americans and females who are disproportionately represented in the legal system. Finally, while the previous passage takes it as a matter of fact that risk assessments are replacing human judgement in the adjudication of offenders; this is a tremendous paradigmatic shift in the legal system that to date has not been adequately brought before the public for open discourse.
At its core the central issue then is not whether risk assessment is harmful to the interests of African Americans, but whether it’s in the interest of African Americans to defend a status quo that places the power of confinement in the hands of the state through unfettered discretion on the part of prosecutors and judges, whether progressive of conservative. Risk assessment in its modern form simply steers most first time and non-violent offenders back to the communities from which they came to be managed as the community sees fit, as long as they are not likely to reoffend.
Dr. Eyitayo Onifade is the Director of the Whitney M. Young, Jr., School of Social Work Center for Children and Families at Clark Atlanta University. His primary areas of scholarship are restorative justice and justice system reform; publishing widely on disproportionate minority contact, dual-status offenders, and Black empowerment.