AuthorMeyn, Ion

TABLE OF CONTENTS INTRODUCTION 222 I. RISK ASSESSMENT TOOLS IN THE CRIMINAL SYSTEM 226 A. The Promise and Perils of Risk Assessment Tools 226 1. Criticism That Tools Attribute Risk to Individuals 228 2. Criticism About What Tools Measure 229 3. Criticism About What Type of Recidivism Is Being Detected 230 4. Criticism About Lack of Transparency 230 5. Criticism That Tools Insulate Unfair Practices 231 6. Criticism About Racial Harms 231 B. A Case Study: Milwaukee County 234 C. Potential Approaches to Remove Racial Bias from RAIs 244 D. A Proposed Remedy: The Racial Disparity Cap 247 II. RACE-BASED REMEDIES, EQUAL PROTECTION, AND THE CRIMINAL LAW 256 A. Compelling Interest 260 1. What Is the Compelling Interest of the Criminal System That Is at Play, and What Is the Basis for That Interest? 262 a. A Close Fit Between the Stated Interest and the Institution's Core Expertise and Mission 267 b. Supplemental Constitutional Concerns Amplify the Compelling Nature of Institutional Needs 269 c. The Existence of Current or Past Discrete Acts of Racial Discrimination Can Justify and Make More Urgent the Institution's Stated Interest 270 B. Narrowly Tailored 272 1. Prior Attempts to Use Race-Neutral Remedies 273 2. The Remedy Responds Directly to the Need 275 3. The Remedy Is Proportional to the Harm 276 4. A Remedy That Does Not Unduly Burden Members of a Racial Group 282 5. The Remedy Is Subject to Periodic Reevaluation 283 CONCLUSION 284 INTRODUCTION

The criminal system is a site of alarming, pervasive racial harms. (1) Twenty-four years ago, Professor Paul Butler proposed race-based interventions to address these harms. (2) In the absence of a race-informed approach, he warned that race-neutral reform would fail to disrupt these conditions. (3) Twenty-four years later, the disparities Butler sought to address persist. (4) Scholars continue to identify the inefficacy of race-neutral interventions and the need for race-informed approaches. (5) And yet, the feasibility of race-conscious strategies is broadly dismissed. (6) Looking out over the treacherous waters of equal protection jurisprudence, criminal law scholars conclude that race-based remedies face insurmountable obstacles. (7)

The conclusion is understandable. Since Butler's call for race-informed reform, Supreme Court pronouncements have only become more sweeping and definitive; for example, "the Constitution is not violated by racial imbalance" and any institutional attempt to simply "achieve racial balance" would be "patently unconstitutional." (8) The Court made clear that whenever a public or federally funded institution implements a race-based policy, strict scrutiny applies. (9) Criminal system actors are not excepted from this requirement; whether it is a police department attempting to diversify officer ranks or a warden attempting to reduce the threat of prison riots, these decisions are subject to the most searching constitutional review. (10)

But a close look at strict scrutiny as applied in the criminal law context reveals a pattern of judicial deference that is uncharacteristic in, for example, higher education and employment spheres. (11) Unlike the mixed record of educational institutions' efforts to survive strict scrutiny, (12) efforts of criminal system actors to implement race-based remedies survive strict scrutiny challenges on a comparatively weaker showing. (13) These successes within the criminal law arena remain uncontroversial, if unnoticed, while battles over affirmative action in education and employment draw oxygen. (14) Among the institutions that have considered race-based remedies, criminal system actors appear more favorably positioned to weather constitutional scrutiny. (15) What ultimately accounts for the "as applied" exceptionalism in the criminal law context is unarticulated, but the broadly observed judicial deference to the exercise of police and prosecutor discretion seems to be at play, even within strict scrutiny review. (16)

This Article seeks to unearth and clarify interests within the criminal system that courts find compelling, and examines features of criminal law that provide favorable conditions within strict scrutiny review. To anchor this constitutional analysis, this Article addresses ongoing efforts to improve risk assessment tools. There are a number of reasons to do so. Risk assessment instruments (RAIs) are increasingly used by prosecutors, judges, and correctional personnel in making pretrial detention, charging, sentencing, and supervision decisions. (1)' Intended to mitigate decision maker bias, RAIs have been found to reproduce racial disparities. (18) To better appreciate challenges that jurisdictions face in addressing these racial disparities, this Article presents a case study of Milwaukee County, which has made nationally recognized efforts to address racial inequities in its criminal system. (19) Based on multiple studies and stakeholder interviews, the challenges in Milwaukee County put in full relief the importance of the option to implement race-based remedies. (20)

This Article also proposes a race-based remedy particularly well-suited to address racial disparities reproduced by RAIs. The proposed remedy emerges from the rich scholarship discussing RAIs and the insights of Professor Sandra Mayson, who recognized that because RAIs are data-informed, they are susceptible to a statistical analysis that permits racial disparities to be numerically expressed. (21) The proposed remedy can be described as a "racial disparity cap": a jurisdiction would assess the mean risk score of racial groups and subtract any differential from the scores of individual members of burdened racial groups. Clearly, the "racial disparity cap" is race-based. Virtually all scholars in this area fear that such attempts would fail under a strict scrutiny analysis. (22) This Article, however, finds reasons to be optimistic about how the remedy fares under equal protection constraints.

This Article proceeds in two parts. In Part I, it provides essential background to understanding RAIs, presents a case study of a jurisdiction that has struggled to use race-neutral approaches to mitigate racial disparities reproduced by its RAI, and proposes a race-based remedy that would significantly reduce racial harms associated with the use of RAIs. In Part II, this Article articulates a compelling interest available to criminal system stakeholders that, applied within existing constitutional constraints, provides a way forward for jurisdictions to survive equal protection challenges. I. RISK ASSESSMENT TOOLS IN THE CRIMINAL SYSTEM

Racial disparities remain in jurisdictions that have launched race-neutral reform. (23) This dynamic is particularly on display in jurisdictions that adopt RAIs to mitigate racial bias, and then seek to mitigate the racial disparities that RAIs reproduce. (24) Illustrating this dynamic, Milwaukee County adopted a nationally prevalent RAI to reduce racial bias in pretrial detention hearings, (25) a significant decision point; studies show outcomes are significantly worse for detained defendants, compared to similarly situated defendants who are released. (26) Before examining Milwaukee's particular experience with RAIs, it is worth assessing the promise and perils of RAIs generally.

  1. The Promise and Perils of Risk Assessment Tools

    Increasingly, Big Data influences criminal system outcomes. (27) Algorithms burrow through data to predict recidivism--how likely is it that a criminal defendant will re-offend? (28) To predict that likelihood, risk tools rely on data points, like a person's criminal history, age of first arrest, affiliation with felons, education level, work history, frequency of moving residences, and missed court hearings. (29) Some tools use a constellation of data points, some use just a few. (30) Different tools might weigh similar factors differently. (31) Some tools require personal interviews, while some rely on pre-existing data. (32) Some tools are specifically designed to aid in pretrial decision-making, while others might be designed to assist a judge during the sentencing phase. (33) Algorithms reduce information to numerical values, but the tools' results are commonly expressed in terms of low, medium, or high risk. (34) The meaning of these terms can vary across jurisdictions; for example, low risk to one jurisdiction might be medium risk in another. An aspiration is that a validated, algorithmic analysis will mitigate decision maker bias. (30) These tools also promise to reduce recidivism by assessing "risk, need, and responsivity" to determine the appropriate "level of correctional intervention." (36)

    As jurisdictions increasingly adopted RAIs, an impressive body of criticism also emerged. (37)

    1. Criticism That Tools Attribute Risk to Individuals

      For the typical decision maker, the RAI answers, "what is this defendant's risk score?"; however, tools ultimately predict the average likelihood of recidivism for individuals who share a characteristic. (38) Data used by the RAI also reflects characteristics that are attributed to a defendant and not reflective of a defendant's exercise of agency. (39) Risk levels are determined by "characteristics and circumstances statistically associated with an increased chance of recidivism." (40) In doing so, RAIs often consider factors over which a defendant has no control, for example, having family members with a felony, a socio-economic status, or being part of a particular racial group. (41) Reliance on RAI factors also limits the universe of risk factors; an RAI may not measure "lack of self-esteem," but "self-esteem issues can and do occasionally lead to serious violence." (42) A tool can assign risk that, upon closer evaluation, is not there because information is stale or incorrect. (43) Even if a decision maker is made aware of such an error, making an adjustment is not possible. A decision maker has no way of considering...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT