"Beyond the Algorithm Pretrial Reform, Risk Assessment, and Racial Fairness"

The title of this post is the title of this notable new report released by the Center for Court Innovation and authored by by Sarah Picard, Matt Watkins, Michael Rempel and Ashmini Kerodal. Here is its introduction: 

Pretrial detention, often resulting from a defendant’s inability to afford bail, is one of the primary drivers of incarceration nationwide. The Bureau of Justice Statistics estimates that two out of three people in local jails in 2016 were held while awaiting trial, having not yet been convicted of a crime.  Jurisdictions looking to safely reduce their use of bail and pretrial detention have increasingly turned to automated or actuarial risk assessments.  These tools employ a mathematical formula, or algorithm, to estimate the probability of a defendant incurring a new arrest or failing to appear in court.  Typically, in a risk assessment, defendants’ criminal history, criminogenic needs, and/or basic demographic information, such as age and gender, are weighted and combined, generating a score which can be used to group defendants into risk categories ranging from low to high.

With the aid of better information about the defendants who appear before them, judges, in theory, can make more consistent decisions regarding pretrial release and bail.  For example, jurisdictions that use risk assessments may be more likely to consider pretrial release for defendants in lower-risk categories, or pretrial supervision in the community for higher-risk defendants.  In cases where victim or community safety is a concern, risk assessment may provide guidance regarding the need for bail or detention hearings.

The appeal of pretrial risk assessment — especially in large, overburdened court systems — is of a fast and objective evaluation, harnessing the power of data to aid decision-making.  Research suggests that actuarial risk assessments are more accurate than decisions made by criminal justice officials relying on professional judgment alone.  By intervening in a process historically driven by subjective decisionmaking, risk assessments arguably act as a corrective to a system plagued by bias, as witnessed in the racial disparities long seen in incarceration rates across the country.

That said, important objections have been raised that, far from disrupting racial biases in the criminal justice system, risk assessments unintentionally amplify them, only this time under the guise of science.  The debate is still unresolved, but from a justice system practitioner’s perspective — let alone that of a defendant — the stakes are urgent.

What follows are the results of an empirical test of racial bias in risk assessment and, based on an original analysis, a consideration of whether there are policy-level solutions that could conserve the benefits of risk assessment, while also addressing valid concerns over racial fairness.