Racial bias as a multi‐stage, multi‐actor problem: An analysis of pretrial detention

• Published date: 01 March 2023
• Author: Joshua Grossman,Julian Nyarko,Sharad Goel
• Date: 01 March 2023
• DOI: http://doi.org/10.1111/jels.12343

ORIGINAL ARTICLE

Racial bias as a multi-stage, multi-actor problem:

An analysis of pretrial detention

Joshua Grossman

1

| Julian Nyarko

2

| Sharad Goel

3

1

Stanford University, Stanford, California, USA

2

Stanford Law School, Stanford, California, USA

3

Harvard University, Cambridge, Massachusetts, USA

Correspondence

Joshua Grossman, Department of Management Science and Engineering, Stanford University, 475 Via Ortega,

Stanford, CA 94305, USA.

Email: jdgg@stanford.edu

Abstract

After arrest, criminal defendants are often detained before trial to mitigate potential risks to

public safety. There is widespread concern, however, that detention decisions are biased

against racial minorities. When assessing potential racial discrimination in pretrial detention,

past studies have typically worked to quantify the extent to which the ultimate judicial deci-

sion is conditioned on the defendant’s race. Although often useful, this approach suffers from

three important limitations. First, it ignores the multi-stage nature of the pretrial process, in

which decisions and recommendations are made over multiple court appearances that influ-

ence the final judgment. Second, it does not consider the multiple actors involved, including

prosecutors, defense attorneys, and judges, each of whom have different responsibilities and

incentives. Finally, a narrow focus on disparate treatment fails to consider potential disparate

impact arising from facially neutral policies and practices. Addressing these limitations, here

we present a framework for quantifying disparate impact in multi-stage, multi-actor settings,

illustrating our approach using 10 years of data on pretrial decisions from a federal district

court. We find that Hispanic defendants are released at lower rates than white defendants of

similar safety and nonappearance risk. We trace these disparities to decisions of assistant US

attorneys at the initial hearings, decisions driven in part by a statutory mandate that lowers

the procedural bar for moving for detention of defendants in certain types of cases. We also

find that the Pretrial Services Agency recommends detention of Black defendants at higher

rates than white defendants of similar risk, though we do not find evidence that these recom-

mendations translate to disparities in actual release rates. Finally, we find that traditional dis-

parate treatment analyses yield more modest evidence of discrimination in pretrial detention

outcomes, highlighting the value of our more expansive analysis for identifying, and ulti-

mately remediating, unjust disparities in the pretrial process. We conclude with a discussion

of how risk-based threshold release policies could help to mitigate observed disparities, and

the estimated impact of various policies on violation rates in the partner jurisdiction.

DOI: 10.1111/jels.12343

©2023 Cornell Law School and Wiley Periodicals LLC.

86 J Empir Leg Stud. 2023;20:86–133.

wileyonlinelibrary.com/journal/jels

INTRODUCTION

An extensive body of work investigates racial disparities in the criminal justice

system and its causes, and there is ample evidence of discrimination in different

parts of the process, including policing and arrests (Antonovics & Knight, 2009;

Fryer, 2019; Pierson et al., 2020), charging (Chohlas-Wood et al., 2021;

Rehavi & Starr, 2014), bail and pretrial detention (Arnold et al., 2018,2022;

Grau & Vergara, 2020; Kutateladze et al., 2012), plea bargaining (Metcalfe &

Chiricos, 2018), conviction and sentencing (Anwar et al., 2012), and incarcera-

tion (Abrams et al., 2012). Among the different decision points in the criminal

process, whether or not to detain a defendant until their trial is of particular

importance, in part because it can significantly influence downstream outcomes.

Evidence suggests that defendants subjected to pretrial detention are more likely

to plead guilty (Leslie & Pope, 2017; Sutton, 2013) and receive harsher sentences

(Didwania, 2020; Dobbie et al., 2018; Heaton et al., 2017; Oleson et al., 2016;

Spohn, 2008; Stevenson, 2018; Sutton, 2013). Furthermore, when found guilty,

defendants detained pretrial may be more likely to recidivate following their

sentence (Heaton et al., 2017; Lowenkamp et al., 2013). Meanwhile, defendants

released pretrial can engage in programs and activities giving them the opportu-

nity to demonstrate that a shorter or noncustodial sentence is warranted, in turn

reducing costs imposed by longer sentences on defendants, their communities,

and the carceral system (Carr, 2016; Oleson et al., 2016).

Although previous studies have examined discrimination in the pretrial pro-

cess (Arnold et al., 2018,2022; Ayres & Waldfogel, 1993;Demuth,2003;

Demuth & Steffensmeier, 2004; Grau & Vergara, 2020;Hull,2021;Lynch&

Patterson, 1991; Schlesinger, 2005,2007; Spohn, 2008; Sutton, 2013;Turner&

Johnson, 2007), they are subject to significant shortcomings. First, past studies

have conceptualized pretrial detention as the result of a unitary decision process

involving only a judge. But pretrial detention, like other outcomes in the criminal

justice system, results from an interrelated series of decisions, and studying one

decision point does not always capture the root causes of discrimination, as dis-

parities may accrue or subside as a case proceeds from one stage to the next

(Schlesinger, 2007). For instance, the typical federal pretrial process consists of at

least three decision points with direct carceral consequences for the defendant.

First, there is an initial hearing during which it is determined whether the defen-

dant is “held for detention”or released. Then, if held for detention, there is a

detention hearing to determine whether the defendant will be detained. Finally,

between hearings, defendants have the possibility to consent to detention.

Given these complexities, one might be inclined to follow the prescriptions

of a more recent strand of the literature and focus on cumulative disparities

(Arnold et al., 2018,2022; Kurlychek & Johnson, 2019; Kutateladze

et al., 2014; Omori & Johnson, 2019; Stolzenberg et al., 2013; Sutton, 2013).

But an exclusive emphasis on cumulative disparities, too, can mask important

ANALYSIS OF PRETRIAL DETENTION 87

costs for the criminal defendant. For instance, even though a cumulative

analysis may, hypothetically, suggest that there are no disparities in final deten-

tion decisions, there could still be disparities at the initial hearing, with minority

defendants spending additional time in jail. Given that costs are disproportion-

ately borne at the very beginning of incarceration, such a pattern would mean

that marginalized groups face significant additional burdens that are not

reflected in analyses that are limited to the final outcome of the pretrial process.

Furthermore, by examining aggregate disparities across the process as a whole,

a cumulative analysis makes it difficult to design targeted interventions to reme-

diate problems at specific stages. Thus, instead of embracing either the individu-

alized or the cumulative view, it appears appropriate to examine disparities both

at individual decision points and cumulatively across decision points.

A second way in which the pretrial detention process defies typical assump-

tions made in the literature concerns the number of actors involved in the

decision-making process. Rather than being an isolated decision of a judge,

pretrial detention decisions result from a complex interplay between at least

three actors in any case: (1) an assistant US attorney (AUSA), who decides

whether to move for detention; (2) a pretrial services (PTS) officer, who prepares

a bail report and recommends detention or release; and (3) a judge, who makes

the detention decision subject to certain statutory constraints. This interplay

suggests that disparities can enter the detention process not only at separate

decision points, but also through several different actions taken by numerous

actors (Bohren et al., 2022). In order to formulate effective policy proposals

aimed at reducing disparities, it is important to take these nuances into consider-

ation and to identify the concrete source of disparities. At the same time,

obtaining data that allow for a detailed analysis of relevant decision makers can

be difficult, given that information on AUSA motions for detention, PTS bail

reports, and intermediate judicial decisions are often not publicly available.

Finally, a third shortcoming of past studies of pretrial detention is that, with

some notable exceptions (Arnold et al., 2018,2021,2022), they have taken a

narrow view of what constitutes discriminatory conduct.

1

In particular, they

employ a methodology that seeks to assess whether detention decisions are

implicitly or explicitly conditioned on race. The primary statistical concern of

1

The Arnold et al. (2021) measure of disparate impact is based on outcomes that are only observed after a judicial

decision is made. Importantly, and in contrast, our definition of risk is based on information available at the time

of the decision. In our setting, and under the conceptualization of disparate impact articulated by Jung et al.

(2019), it is critical to consider such ex ante measures of risk in order to determine whether similarly risky people

are treated similarly. Both measures may be potentially useful ways to characterize disparate impact, but we

believe that the approach we take is more closely aligned with traditional legal doctrine. In particular, co urts have

deemed legally permissible policies that first rank individuals by ex ante risk and then distribute resource s

(e.g., loans) to those above a threshold risk level. Such threshold policies would likewise not be found to have

disparate impact under our definition but would in general violate definitions based on ex post outcomes (Corbett-

Davies & Goel, 2018). Of course, aside from these legal considerations, there are larger normative concerns about

how one should assess the equity of different policies, which we leave to future work.

88 ANALYSIS OF PRETRIAL DETENTION