Content uploaded by Justin D. Levinson
Author content
All content in this area was uploaded by Justin D. Levinson on Jan 29, 2014
Content may be subject to copyright.
Electronic copy available at: http://ssrn.com/abstract=1601615
307
DIFFERENT SHADES OF BIAS:
SKIN TONE, IMPLICIT RACIAL BIAS, AND
JUDGMENTS OF AMBIGUOUS EVIDENCE
Justin D. Levinson* & Danielle Young**
I. INTRODUCTION ....................................................................................
308
II. SCHOLARSHIP ON IMPLICIT BIAS AND RACE IN LEGAL DECISION-
MAKING ............................................................................................... 311
A. Legal Scholarship ................................................................... 311
1. Non-Empirical Work on Implicit Bias in Society ..... 312
2. Non-Empirical Work on Implicit Bias in the Legal
System ................................................................................... 315
3. Empirical Legal Scholarship ..................................... 319
B. Mock-Jury Research on Racial Bias ...................................... 323
III. ACTIVATING POWERFUL RACIAL STEREOTYPES ................................. 326
A. Racial Stereotypes are Primed Easily .................................... 327
B. Primed Racial Stereotypes Affect Decision-Making............... 329
IV. TESTING FOR BIAS — THE EMPIRICAL STUDY .................................... 331
A. Methods .................................................................................. 331
B. Limitations of Study and Future Directions ........................... 334
C. Demographics of Study Participants ...................................... 335
D. Results — Skin Tone and Racially Biased Judgments ............ 336
1. Skin Tone Affects Judgments of Ambiguous
Evidence ................................................................................. 337
2. Skin Tone Affects Judgments of “How Guilty Is the
Defendant” .............................................................................. 337
3. Evidence Judgments Predict Guilty Verdicts ............ 337
4. Evidence Judgments Unrelated to Explicit Racial
Preferences ............................................................................. 338
5. Stimuli Recall, Priming, and IAT Results ................. 338
E. Summary of Results and Implications .................................... 338
Copyright © 2010 by Justin D. Levinson and Danielle Young
* Associate Professor of Law and Director, Culture and Jury Project, William S. Richardson
School of Law, University of Hawai`i at Mānoa. The authors would like to thank Susan Serrano,
Kapua Sproat, Galit Levinson, and Dina Shek for their input on previous drafts. Sara Ayabe
contributed outstanding research assistance. Dean Aviam Soifer provided generous summer re-
search support to the first author. The authors would also like to thank Dr. Huajian Cai for adapt-
ing the empirical study for computerized testing.
** Department of Psychology, University of Hawai`i at Mānoa.
Electronic copy available at: http://ssrn.com/abstract=1601615
308 WEST VIRGINIA LAW REVIEW [Vol. 112
V. DEVELOPING A NEW MODEL OF IMPLICIT RACIAL BIAS AND DECISION-
MAKING ............................................................................................... 339
A. The Story Model of Decision-Making ..................................... 340
B. Memory Errors, Biased Evidence, Implicit Associations, and
More ....................................................................................... 342
VI. CONCLUSION: SETTING A RESEARCH AGENDA .................................. 345
APPENDIX A ...................................................................................................... 348
APPENDIX B ...................................................................................................... 350
The Robbery: 11:00 p.m., December 18, 2008. The owner of the Quick
Stop Mini Mart has just closed the store for the evening when two armed men
barge in. One of the men points a gun at the owner while the other walks be-
hind the counter to the cash register. The men take the money from the register
and escape with $550 in cash.
The Security Camera: A Mini Mart security camera captures the rob-
bery in progress, generating a clear photo of one of the masked perpetrators.
The perpetrator’s face is totally hidden by the mask, but the camera loosely cap-
tures his body type and his short sleeved shirt reveals his dark skin tone.
The Arrest and Trial: A suspect is arrested and goes on trial for the
robbery. Evidence presented at trial includes the following: the defendant was
a youth Golden Gloves boxing champ in 2006, the defendant purchased an un-
traceable handgun three weeks before the robbery, the defendant is a member of
an anti-violence organization, and the defendant had a used movie ticket stub for
a show that started 20 minutes before the crime occurred.
The Empirical Question
I. INTRODUCTION
: Does seeing the perpetrator’s dark skin tone in
the security camera photo elicit implicit racial bias that affects the way jurors
evaluate ambiguous trial evidence?
Before scholars discovered the link between social science research on
implicit racial bias and legal theory, the notion that people act automatically and
unintentionally in racially biased ways was entirely outside the scope of legal
discourse.1
1 See Jerry Kang, Trojan Horses of Race, 118 HARV. L. REV. 1489, 1497–1539 (2005); Linda
Hamilton Krieger, The Content of Our Categories: A Cognitive Bias Approach to Discrimination
and Equal Employment Opportunity, 47 STAN. L. REV. 1161 (1995); Charles R. Lawrence III, The
Id, the Ego, and Equal Protection: Reckoning With Unconscious Racism, 39 STAN. L. REV. 317,
331–36 (1987) [hereinafter Id, Ego, and Equal Protection]. These articles, though groundbreaking
in legal scholarship, did not discover the psychological phenomena they discussed. In fact, psy-
chological research for decades has examined the way racial schemas work, often unintentionally.
For example, in a famous study by Gordon Allport and Leo Postman, “participants viewed a pic-
ture of passengers on a streetcar (one of whom was Black). In the picture, one White passenger
holds a razor blade and the Black passenger is empty-handed. After viewing the picture, partici-
pants were then asked to describe the picture to other participants who did not see the picture. As
participants told and retold the story to others, the story changed. After the story had been retold
Since this meaningful interdisciplinary breakthrough, many com-
2010] DIFFERENT SHADES OF BIAS 309
mentators and judges have come to accept the changing reality of racial discrim-
ination — discrimination that has largely shifted from overt and intentional to
covert and unintentional.2 Despite this scholarly progress, evidenced largely by
commentators’ willingness to consider the implications of complex scientific
evidence as a possible pathway to legal reform,3 the dearth of empirical studies
testing implicit bias within the legal system is surprising.4 In an effort to begin
filling the empirical research gap, this Article proposes and tests a new hypothe-
sis called Biased Evidence Hypothesis. Biased Evidence Hypothesis posits that
when racial stereotypes are activated, jurors automatically and unintentionally
evaluate ambiguous trial evidence in racially biased ways. Because racial ste-
reotypes in the legal context often involve stereotypes of African-Americans
and other minority group members as aggressive criminals, Biased Evidence
Hypothesis, if confirmed, could help explain the continued racial disparities that
plague the American criminal justice system.5
Social science research, and in particular, social cognition
6 research on
the phenomenon of “priming”7
several times, some participants reported that the Black passenger — not the White passenger —
held a razor blade.” Justin D. Levinson, Forgotten Racial Equality: Implicit Bias, Decisionmak-
ing, and Misremembering, 57 DUKE L.J. 345, 381 (2007) [hereinafter Levinson, Forgotten Racial
Equality] (citing GORDON W. ALLPORT & LEO POSTMAN, THE PSYCHOLOGY OF RUMOR 65–68
(1965)).
demonstrates that even the simplest of racial
2 See generally Jerry Kang & Mahzarin R. Banaji, Fair Measures: A Behavioral Realist Revi-
sion of “Affirmative Action”, 94 CAL. L. REV. 1063 (2006); Krieger, supra note 1. Despite this
progress, legal reforms recognizing implicit bias have been slow. See Justin D. Levinson, Race,
Death and the Complicitous Mind, 58 DEPAUL L. REV. 599 (2009) [hereinafter Levinson, The
Complicitous Mind].
3 Implicit bias scholarship is by no means the only example of this progress. Scholarship in
law and economics, behavioral law and economics, and more all rely upon scientific research
methods. Even new journals, peer reviewed by law professors trained in empirical methods, have
emerged. For example, the Journal of Empirical Legal Studies was launched in 2004.
4 A few research projects have begun to examine how implicit biases may or may not func-
tion in legal decision-making. See, e.g., Jeffrey J. Rachlinski et al., Does Unconscious Bias Affect
Trial Judges? 84 NOTRE DAME L. REV 1195 (2009) [hereinafter Trial Judges]. These studies are
discussed infra Section II.A.3. See infra notes 50–76 and accompanying text.
5 One source of information on racial disparities in the criminal justice system is the Depart-
ment of Justice Bureau of Justice, which publishes statistics on crime and race. See BUREAU OF
JUSTICE STATISTICS, JAIL INMATE CHARACTERISTICS, available at
http://bjs.ojp.usdoj.gov/index.cfm?ty=tp&tid=122 (last visited Jan. 16, 2010) (indicating that 60%
of inmates in local jails in the United States are minorities). For more on the continued racial
disparities in the criminal justice system, see generally RANDALL KENNEDY, RACE, CRIME, AND
THE LAW (1997); MARC MAUER, RACE TO INCARCERATE (2d ed., 2006).
6 Generally, the field of social cognition studies how people think about themselves and oth-
ers, “often using methods from cognitive psychology to investigate how the human mind works.”
Levinson, Forgotten Racial Equality, supra note 1, at 354 (citing SUSAN T. FISKE & SHELLEY E.
TAYLOR, SOCIAL COGNITION 2, 19 (2d ed. 1991)).
7 Priming describes “the incidental activation of knowledge structures, such as trait concepts
and stereotypes, by the current situational context.” John A. Bargh et al., Automaticity of Social
310 WEST VIRGINIA LAW REVIEW [Vol. 112
cues introduced into a trial might automatically and unintentionally evoke racial
stereotypes, thus affecting the way jurors evaluate evidence. To examine this
possibility (considered in the context of the armed robbery case outlined at the
beginning of this Article), we designed an empirical study that tested how
mock-jurors judge trial evidence.8 As part of an “evidence slideshow,” we
showed half of the study participants a security camera photo of a dark-skinned
perpetrator and the other half of the participants an otherwise identical photo of
a lighter-skinned perpetrator.9
This Article examines how exposing jurors to simple racial cues can
trigger stereotypes and affect how they evaluate evidence in subtle but harmful
ways. The Article is organized as follows. Section II first contextualizes this
investigation in light of legal scholarship on implicit bias and interdisciplinary
scholarship on race and legal decision-making and concludes that legal scholar-
ship has generally succeeded in understanding how implicit bias may affect so-
cietal actors, but has been less successful in empirically testing specific hypo-
theses of how implicit bias affects decision-making. Section III sets the stage
for our empirical study by explaining the scientific underpinnings of Biased
Evidence Hypothesis. Social cognition research, particularly on the phenomena
of priming, has demonstrated that the human mind responds quickly and auto-
matically to racially stereotypic information, and that these automatic cognitive
responses can have harmful effects on decision-making.
The results of the study supported Biased Evi-
dence Hypothesis and indicated that participants who saw a photo of a dark-
skinned perpetrator judged subsequent evidence as more supportive of a guilty
verdict compared to participants who saw a photo of a lighter-skinned perpetra-
tor.
10
Section IV presents the empirical study we conducted. After informing
participants of the basic facts surrounding an armed robbery, we showed them a
series of crime scene photographs as part of an evidence slideshow. We ran-
domly assigned the participants into two experimental conditions. Participants
in each condition saw identical photos except in one key respect: half of the
participants saw a surveillance camera photo of a dark-skinned perpetrator and
half of the participants saw a surveillance camera photo of a lighter-skinned
perpetrator.
11
Behavior: Direct Effects of Trait Construct and Stereotype Activation on Action, 71 J.
PERSONALITY & SOC. PSYCHOL. 230, 230 (1996).
We then presented participants with evidence from the trial, and
asked them to judge how much each piece of evidence tended to indicate wheth-
er the defendant was guilty or not guilty. Results of the study showed that par-
8 This empirical study is described in detail in Section IV, infra.
9 For a description of the methods of the study, see infra notes 121–132 and accompanying
text. The Photos are reproduced in Appendix B.
10 Priming thus helps explain the cognitive processes underlying Biased Evidence Hypothesis.
11 The photos were identical except that the skin tone was lightened using computer software.
Thus, the only difference in the security camera photo of the perpetrator was the skin tone of the
perpetrator.
2010] DIFFERENT SHADES OF BIAS 311
ticipants who saw the photo of the dark-skinned perpetrator were more likely
than participants who saw the photo of the lighter-skinned perpetrator to judge
the evidence as tending to indicate criminal guilt and were also more likely to
believe that the defendant was guilty of armed robbery.
Section V explores the results of the study as part of an amplification of
the Story Model, an acclaimed model of jury decision-making. This model,
which provides a step-by-step explanation of how jurors make decisions, has yet
to consider the potentially pervasive impact of implicit racial bias in decision-
making. Using the Story Model as a guide and considering the study results
together with other emerging research on implicit bias, we deconstruct the mul-
titude of ways that implicit racial bias can affect jury decision-making. We then
call for the development of a complete model of implicit racial bias in jury deci-
sion-making. Section VI concludes by proposing a research agenda for contin-
ued interdisciplinary investigation on implicit bias in the law.
II. SCHOLARSHIP ON IMPLICIT BIAS AND RACE IN LEGAL DECISION-MAKING
When considering whether simple racial cues can trigger stereotypes
and cause jurors to evaluate evidence in racially biased ways, it is helpful to
contextualize this examination first in light of existing legal scholarship on im-
plicit bias, and second in light of interdisciplinary scholarship on race in legal
decision-making. These areas of scholarship have brought social science and
legal scholarship quite close together, but few scholars have employed empirical
studies designed to test how implicit bias affects jury decision-making.
A. Legal Scholarship
Scholarship on implicit bias has emerged rapidly since the 1990’s, and
has made quite a splash in legal discourse.12
12 For reviews of implicit bias scholarship, see Kang, supra note
This scholarship, which has led
many commentators to reconsider laws in light of the scientific reality of racial
bias, can best be reviewed in three separate categories, based both upon the sub-
stance of the scholarship and whether or not the projects are empirical in nature.
These categories are: (1) non-empirical work dealing with the law’s reaction to
the implicit bias of societal actors; (2) non-empirical work dealing with implicit
bias in legal decision-making or legal policies; and (3) empirical work examin-
ing implicit bias in the legal system. As this subsection will demonstrate, al-
though scholarship on implicit bias has paved an important path for social jus-
1; Kristin A. Lane et al.,
Implicit Social Cognition and Law, 3 ANN. REV. L. & SOC. SCI. 427, 444 (2007); Levinson, For-
gotten Racial Equality, supra note 1; Justin D. Levinson, Culture, Cognitions, and Legal Deci-
sion-Making, in HANDBOOK OF MOTIVATION AND COGNITION ACROSS CULTURES 423, 423–39 (R.
Sorrentino & S. Tamaguchi eds., 2008).
312 WEST VIRGINIA LAW REVIEW [Vol. 112
tice scholars and scholar-advocates who continue the fight for racial justice,13
1. Non-Empirical Work on Implicit Bias in Society
legal scholars have yet to consider fully the specific mechanisms whereby im-
plicit bias manifests within the legal system. Similarly, they have yet to conduct
(or collaborate with social scientists to conduct) more than a few empirical in-
vestigations of implicit racial bias in the legal context.
A great deal of legal scholarship on implicit bias discusses the implicit
bias of decision-makers in the everyday world. Bias by these decision-makers
might manifest in a variety of ways, ranging from a school board’s decision to
redistrict school boundaries, to an employer’s decision about whether to hire a
worker, to a fire department’s decision to use a particular test to determine pro-
motion. In an article widely credited as introducing the concept of unconscious
bias14 to legal scholarship,15 Charles Lawrence criticized the Supreme Court’s
failure to understand the true, unconscious nature of discrimination.16 Law-
rence, who famously declared, “we are all racists,”17 wrote the article in re-
sponse to the then recent decision in Washington v. Davis,18 which established a
difficult to demonstrate intent-based threshold for proving discrimination. Law-
rence sought to “expose and challenge the way that the Court had, with this sin-
gle opinion, declared the reconstructive work of the 13th, 14th, and 15th
amendments accomplished.”19
13 For more on scholar advocates and political lawyers, see Eric K. Yamamoto, Critical Race
Praxis: Race Theory and Political Lawyering Practice in Post-Civil Rights America, 95 MICH. L.
REV. 821, 833–34 (1997).
Through a critical cultural lens that called for
14 Legal scholarship often refers to this bias as “unconscious bias,” while social scientists
typically refer to it as “implicit bias.” Social scientists generally prefer to use the term “implicit”
instead of “unconscious” because there are cases in which people may have some awareness of
their own implicit biases, and therefore they may not be entirely “unconscious.” See Russell H.
Fazio & Michael A. Olson, Implicit Measures in Social Cognition Research: Their Meanings and
Use, 54 ANN. REV. PSYCHOL. 297, 303 (2003).
15 Representing the impact of Lawrence’s article, the Connecticut Law Review published a
symposium issue containing various reflections on the legacy of Lawrence’s article. See, e.g.,
Avital Mentovich & John T. Jost, The Ideological “ID”? System Justification and the Uncons-
cious Perpetuation of Inequality, 40 CONN. L. REV. 1095 (2008); Gowri Ramachandran, Antisu-
bordination, Rights, and Radicalism, 40 CONN. L. REV. 1045 (2008). Lawrence also reflected on
his own piece. See Charles Lawrence, Unconscious Racism Revisited: Reflections on the Impact
and Origins of “The Id, The Ego, and Equal Protection,” 40 CONN. L. REV. 931 (2008) [hereinaf-
ter Unconscious Racism Revisited].
16 See Id, Ego, and Equal Protection, supra note 1, at 331–36.
17 Id. at 322.
18 426 U.S. 229 (1976). Washington v. Davis, which was based on a challenge to allegedly
racially discriminatory hiring procedures, was and is particularly controversial because of the
Court’s focus on proving not just a discriminatory impact, but discriminatory purpose.
19 Lawrence describes his famous piece as primarily being “concerned with exploring how
white supremacy is maintained not only through the intentional deployment of coercive power,
but also through the creation, interpretation, and assimilation of racial text.” Unconscious Racism
2010] DIFFERENT SHADES OF BIAS 313
shared societal responsibility for racism, Lawrence introduced legal scholars to
a psychologically informed critique of race and power.20 Lawrence relied on
both Freudian and cognitive psychology to discuss Americans’ pervasive un-
conscious racial bias, and called for the Court to employ a “cultural meaning”
test (rather than an intent based one) for racial discrimination.21
Following Lawrence’s lead in examining the connection between evi-
dence on unconscious bias and anti-discrimination law, employment discrimina-
tion scholars have argued that laws requiring proof of an employer’s intentional
discrimination, such as Title VII, fail to understand the true non-conscious and
automatic nature of discrimination.
22 Initiating this scholarly progress in the
employment discrimination arena, Linda Hamilton Krieger deconstructed social
cognition work in light of Title VII’s subjective intent standard.23 Krieger relied
upon empirical social cognition studies to demonstrate that employment dis-
crimination is not always intentional and to argue that legal standards should
match psychological reality.24
Revisited, supra note
Specifically, Krieger argued that courts should
15 at 939. Lawrence is quite critical of scholarship that has focused on the
science of implicit bias without connecting it to a shared societal responsibility for racism. Id. at
956–66.
20 See generally Id, Ego, and Equal Protection, supra note 1.
21 Id. at 324. Lawrence explained this test:
It suggests that the “cultural meaning” of an allegedly racially discriminatory
act is the best available analogue for, and evidence of, a collective uncons-
cious that we cannot observe directly. This test would thus evaluate govern-
mental conduct to determine whether it conveys a symbolic message to which
the culture attaches racial significance. A finding that the culture thinks of an
allegedly discriminatory governmental action in racial terms would also con-
stitute a finding regarding the beliefs and motivations of the governmental ac-
tors: The actors are themselves part of the culture and presumably could not
have acted without being influenced by racial considerations, even if they are
unaware of their racist beliefs. Therefore, the court would apply strict scruti-
ny.
Id. (citation omitted).
22 See generally Samuel R. Bagenstos, The Structural Turn and the Limits of Antidiscrimina-
tion Law, 94 CAL. L. REV. 1 (2006) [hereinafter The Structural Turn]; Melissa Hart, Subjective
Decisionmaking and Unconscious Discrimination, 56 ALA. L. REV. 741 (2005); Krieger, supra
note 1; Audrey J. Lee, Unconscious Bias Theory in Employment Discrimination Litigation, 40
HARV. C.R.-C.L. L. REV. 481 (2005); Ann C. McGinley, !Viva La Evolución!: Recognizing Un-
conscious Motive in Title VII, 9 CORNELL J.L. & PUB. POL’Y 415 (2000); Deana A. Pollard, Un-
conscious Bias and Self-Critical Analysis: The Case for a Qualified Evidentiary Equal Employ-
ment Opportunity Privilege, 74 WASH. L. REV. 913 (1999) (arguing for a qualified evidentiary
privilege to encourage unconscious-bias testing).
23 See Krieger, supra note 1.
24 Id. at 1186–1211. Krieger made her claim in light of evidence of three broad themes from
social cognition research: “(1) normal cognitive processes automatically trigger stereotyping, (2)
stereotypes and biases operate absent an explicit intent to use them, and (3) ‘people’s access to
their own cognitive processes is in fact poor.’” Levinson, Forgotten Racial Equality, supra note 1,
at 354 (citing Krieger, supra note 1, at 1188).
314 WEST VIRGINIA LAW REVIEW [Vol. 112
change the focus from “an intent standard (whether an employer intended that
race make a difference in an employment decision) to a causation standard
(whether race or group status ‘made a difference’ in the decision).”25
Krieger’s work, combined with an impressive batch of social cognition
studies that emerged in the late 1990’s and early 2000’s,
26 helped to ignite de-
bate about employment discrimination law, an area in which many scholars
have asserted that implicit bias information should be considered in determining
employment discrimination.27 Furthermore, it stimulated discourse in other
areas of law, where scholars have often referred to implicit bias as triggering
inequities.28
Post-2006 scholarship has broadened the discussion of how societal ac-
tors may automatically and unintentionally propagate racial bias.
These projects have sought to understand the way societal actors
may unconsciously perpetuate inequality.
29
25 Levinson, Forgotten Racial Equality, supra note
Robert
1, at 366 (citing Krieger, supra note 1, at
1242).
26 For a description of many of these studies, see Kang, supra note 1; Levinson, Forgotten
Racial Equality, supra note 1; Antony Page, Batson’s Blind-Spot: Unconscious Stereotyping and
the Peremptory Challenge, 85 B.U. L. REV. 155, 236–57 (2005). See also Bargh et al., supra note
7; Patricia G. Devine, Implicit Prejudice and Stereotyping: How Automatic are They? Introduc-
tion to the Special Section, 81 J. PERSONALITY & SOC. PSYCHOL. 757 (2001); Anthony G. Green-
wald et al., Targets of Discrimination: Effects of Race on Responses to Weapons Holders, 39 J.
EXPERIMENTAL SOC. PSYCHOL. 399 (2003); B. Keith Payne, Prejudice and Perception: The Role
of Automatic and Controlled Processes in Misperceiving a Weapon, 81 J. PERSONALITY & SOC.
PSYCHOL. 181 (2001); Laurie A. Rudman, Social Justice in Our Minds, Homes, and Society: The
Nature, Causes, and Consequences of Implicit Bias, 17 SOC. JUST. RES. 129 (2004).
27 See The Structural Turn, supra note 22; Ivan E. Bodensteiner, The Implications of Psycho-
logical Research Related to Unconscious Discrimination and Implicit Bias in Proving Intentional
Discrimination, 73 MO. L. REV. 83 (2008) (examining existing proof schemes in light of social
cognition research); David L. Faigman et al., Symposium, A Matter of Fit: The Law of Discrimi-
nation and the Science of Implicit Bias, 59 HASTINGS L.J. 1389 (2008) (focusing on gender dis-
crimination); Hart, supra note 22; Christine Jolls & Cass R. Sunstein, The Law of Implicit Bias, 94
CAL. L. REV. 969 (2006); Audrey J. Lee, supra note 22; McGinley, supra note 22; Pollard, supra
note 22; see also Franita Tolson, The Boundaries of Litigating Unconscious Discrimination:
Firm-Based Remedies in Response to a Hostile Judiciary, 33 DEL. J. CORP. L. 347 (2008) (suggest-
ing that firms should have the burden of reducing unconscious discrimination). But see Amy L.
Wax, The Discriminating Mind: Define It, Prove It, 40 CONN. L. REV. 979 (2008) (arguing that
implicit bias research may not warrant changing legal standards).
28 In the criminal law realm, for example, these projects assert or hypothesize that unconscious
bias leads to racial or other disparities in the criminal justice system. See, e.g., Julian A. Cook, Jr.
& Mark S. Kende, Color-Blindness in the Rehnquist Court: Comparing the Court’s Treatment of
Discrimination Claims by a Black Death Row Inmate and White Voting Rights Plaintiffs, 13 T.M.
COOLEY L. REV. 815 (1996); Sheri Lynn Johnson, Comment, Unconscious Racism and the Crimi-
nal Law, 73 CORNELL L. REV. 1016, 1016–17 (1988); Cynthia Lee, The Gay Panic Defense, 42
U.C. DAVIS L. REV. 471, 479 (2008).
29 Even scholarship outside of the law has begun to consider the ramifications of implicit bias,
such as in medical treatment decisions and public health. For example, a study by Alexander
Green and colleagues found that implicit racial bias predicted the medical treatment decisions
made by doctors. Alexander R. Greene et al., Implicit Bias among Physicians and its Prediction
of Thrombolysis Decisions for Black and White Patients, 22 J. GEN. INTERNAL MED. 1231 (2007).
2010] DIFFERENT SHADES OF BIAS 315
Schwemm, for example, relied on social cognition research to argue that lan-
dlords may discriminate unintentionally and suggested that implicit bias in
housing discrimination may be more widespread even than implicit bias in em-
ployment discrimination.30 Rigel Oliveri alleged that implicit biases may dis-
proportionately harm illegal immigrants.31 And Tristin Green and Alexandra
Kalev broadened the discussion of implicit bias in employment law by focusing
on how implicit bias may affect relational aspects of the employment situation.32
Scholarship examining the law’s response to implicit bias in society has
thus been provocative and influential and continues to develop.
2. Non-Empirical Work on Implicit Bias in the Legal System
Several commentators have considered the way implicit biases are ei-
ther facilitated by the law itself or how legal decision-makers may unintention-
ally propagate these biases. These projects can be distinguished from studies of
implicit bias in society because instead of considering how law should react to
the implicit biases of societal actors, they consider how the law itself may prop-
agate bias.33
30 Robert G. Schwemm, Why Do Landlords Still Discriminate (And What Can Be Done About
It)?, 40 J. MARSHALL L. REV. 455, 507 (2007).
31 See Rigel C. Oliveri, Between a Rock and a Hard Place: Landlords, Latinos, Anti-Illegal
Immigrant Ordinances, and Housing Discrimination, 62 VAND. L. REV. 55 (2009).
32 Tristin K. Green & Alexandra Kalev, Discrimination-Reducing Measures at the Relational
Level, 59 HASTINGS L.J. 1435 (2008) (considering implicit bias on a relational level in the
workplace).
33 The following projects, many of which focus on racial disparities in the death penalty, rec-
ognize the ways in which implicit bias may manifest in legal processes. However, most of them
do not specifically connect their hypotheses to legal theories. See, e.g., Lucy Adams, Comment,
Death by Discretion: Who Decides Who Lives and Dies in the United States of America? 32 AM.
J. CRIM. L. 381, 389–90 (2005) (stating that “a white prosecutor may — consciously or subcons-
ciously — perceive a crime to be more ‘outrageously or wantonly vile, horrible, or inhuman’ if it
is alleged to have been committed against a white victim” (quoting GA. CODE ANN. § 17-10-
30(b)(7) (1994))); Andrew Elliot Carpenter, Note, Chambers v. Mississippi: The Hearsay Rule
and Racial Evaluations of Credibility, 8 WASH & LEE RACE & ETHNIC ANCESTRY L. J. 15, 15
(2002) (arguing that unconscious racial bias affects judicial determinations of witness credibility);
Scott W. Howe, The Futile Quest for Racial Neutrality in Capital Selection and the Eighth
Amendment Argument for Abolition Based on Unconscious Racial Discrimination, 45 WM. &
MARY L. REV. 2083, 2094–2106 (2004) (considering the extreme deference given to prosecutors);
Rory K. Little, What Federal Prosecutors Really Think: The Puzzle of Statistical Race Disparity
Versus Specific Guilt, and the Specter of Timothy McVeigh, 53 DEPAUL L. REV. 1591, 1599–1600
(2004) (addressing “unconscious race empathy” that white prosecutors may have with white de-
fendants or white victims); Jeffrey J. Pokorak, Probing the Capital Prosecutor’s Perspective:
Race of the Discretionary Actors, 83 CORNELL L. REV. 1811, 1819 (1998) (alluding to uncons-
cious biases produced due to similarities between prosecutors and victims); Yoav Sapir, Neither
Intent nor Impact: A Critique of the Racially Based Selective Prosecution Jurisprudence and a
Reform Proposal, 19 HARV. BLACKLETTER L.J. 127, 140–41 (2003) (proposing that it “is likely
that unconscious racism influences a prosecutor even more than it affects others.”).
316 WEST VIRGINIA LAW REVIEW [Vol. 112
In a comprehensive article that connected lawmaking to the propagation
of racial bias, Jerry Kang argued that the Federal Communications Commis-
sion’s policies encouraging the broadcasting of local news served to increase
television viewers’ implicit racial biases.34 In making the connection between
local news and increased societal bias, Kang relied on social cognition studies to
demonstrate that the media continually reinforces racial biases within the Amer-
ican public.35 After detailing a plethora of social cognition studies on implicit
bias and outlining the ways even seemingly minor racial cues can lead to racial
biases,36 Kang questioned whether the local news’ consistent broadcasting of
racialized and violent images might have devastating effects.37 Kang’s detailed
and thoughtful inquiry forced scholars to confront the notion that well-
intentioned laws and policies can unintentionally lead to disastrous results.38
Justin Levinson relied on implicit bias research to propose that death
qualification, the process of qualifying jurors to sit on capital cases, might unin-
In
doing so, he set the stage for examinations of implicit bias in other areas of law.
34 See generally Kang, supra note 1.
35 See id. at 1449–53 (citing James M. Avery & Mark Peffley, Race Matters: The Impact of
News Coverage of Welfare Reform on Public Opinion, in RACE AND THE POLITICS OF WELFARE
REFORM 131, 136 (Sanford F. Schram et al. eds., 2003); Lori Dorfman & Vincent Schiraldi, Off
Balance: Youth, Race & Crime in the News 4–5, 7 (2001), available at
http://www.buildingblocksforyouth.org/media/media.pdf; Robert M. Entman, Modern Racism and
the Images of Blacks in Local Television News, 7 CRITICAL STUD. MASS COMM. 332, 335 (1990);
ROBERT M. ENTMAN & ANDREW ROJECKI, THE BLACK IMAGE IN THE WHITE MIND 49 (2000);
Franklin D. Gilliam, Jr. et al., Crime in Black and White: The Violent, Scary World of Local News,
1 HARV. INT'L J. PRESS/POL. 6, 7 (1996); Mark Peffley et al., The Intersection of Race and Crime
in Television News Stories: An Experimental Study, 13 POL. COMM. 309, 315 (1996); Nicholas A.
Valentino, Crime News and the Priming of Racial Attitudes During Evaluations of the President,
63 PUB. OPINION Q. 293, 301–02 (1999)).
36 Kang, supra note 1, at 1491–95. For example, Kang described a study by Frank Gilliam
and Shanto Iyengar that examined whether showing a Black suspect’s mugshot photograph (com-
pared both to a White suspect’s mugshot and to no mugshot at all) in a local news broadcast af-
fected participants’ subsequent judgments of punishment. As Kang explained, “[h]aving seen the
Black suspect, White participants showed 6% more support for punitive remedies than did the
control group, which saw no crime story. When participants were instead exposed to the White
suspect, their support for punitive remedies increased by only 1%, which was not statistically
significant.” Id. at 1492 (citations omitted).
37 Kang asked: “[i]f subliminal flashes of Black male faces can raise our frustration . . . would
it be surprising that consciously received messages couched in violent visual context have impact,
too?” Id. at 1551. In referencing subliminal flashes and frustration, Kang was referring to a study
by John Bargh and his colleagues. Id. (citing John A. Bargh et al., Automaticity of Social Beha-
vior: Direct Effects of Trait Construct and Stereotype Activation on Action, 71 J. PERSONALITY &
SOC. PSYCHOL. 230, 239 (1996)). Bargh and colleagues’ study found that subliminally flashing
images of Black male faces during a computerized task heightened participants’ visible frustra-
tions after participants subsequently were led to believe that the study’s computer had crashed and
that the experiment would have to be re-started.
38 See also Reshma M. Saujani, “The Implicit Association Test:” A Measure of Unconscious
Racism in Legislative Decision-Making, 8 MICH. J. RACE & L. 395, 413–15 (2003) (also focusing
on the potential implicit bias of policy-making).
2010] DIFFERENT SHADES OF BIAS 317
tentionally trigger implicit racial biases.39 The hypothesis, which Levinson
termed Death Penalty Priming Hypothesis, proposes that the (supposedly race
neutral) process of death qualifying jurors actually elicits racial stereotypes in
the final jury panel.40
According to Levinson,
when jury venire members are ‘death qualified,’ the supposedly
race-neutral line of questioning acts as an indirect prime that
triggers stereotypes of African Americans, including criminali-
ty, dangerousness, and guilt. These largely implicit stereotypes,
which most Americans likely possess, become activated during
the death qualification process, and subsequently affect the way
jurors process information, deliberate, and render verdicts when
African-American defendants are on trial.41
Death Penalty Priming Hypothesis, similar to Kang’s proposal relating
to local news, suggests that commentators should look more critically at the
legal system and its processes.
Like Kang and Levinson, other scholars have looked within the legal
system to discover how implicit biases may manifest.42 Unlike Kang and Le-
vinson, however, most of these scholars have focused not on legal policies or
procedures that result in bias, but on the effect of implicit bias on attorneys,
judges, jurors, and even police officers. For example, Antony Page relied on
social cognition research in proposing that prosecutors and defense attorneys
unintentionally rely on implicit racial biases when using peremptory chal-
lenges.43 Connecting this hypothesis to Supreme Court jurisprudence, Page
critiqued Batson v. Kentucky’s44
39 Levinson, The Complicitous Mind, supra note
three-step procedure for attacking peremptory
2, at 619–32.
40 Id. Levinson’s second hypothesis considered in the same article, Racial Bias Masking Hy-
pothesis, focuses not on how the legal system triggers biases, but on how racially biased legal
treatment of defendants may evade scientific detection. This hypothesis posits that sophisticated
studies examining racial disparities in criminal convictions and sentencing may actually cover up
such racial disparities because those studies rely on already biased case facts. It is worth noting
that Racial Bias Masking Hypothesis, which specifically focused on studies of race in the capital
setting, might also apply to studies of race in the non-capital setting. Id. at 632–43.
41 Id. at 619 (citation omitted). Levinson did not test these hypotheses empirically, and urged
“caution in relying upon the accuracy of these hypotheses until they have been explored more
systematically.” Id. at 644.
42 A few scholars have begun expanding on the ways in which the law itself may lead to bias.
See Dale Larson, Unconsciously Regarded as Disabled: Implicit Bias and the Regarded-As Prong
of the Americans with Disabilities Act, 56 UCLA L. REV. 451 (2008); Lee, supra note 28.
43 Page, supra note 26, at 236–57.
44 Batson v. Kentucky, 476 U.S. 79 (1986). In Batson, a case in which a Black man was on
trial for burglary and receipt of stolen property:
the prosecutor used his peremptory challenges to construct an all-white jury
by striking all four of the black people on the venire. The defense counsel
sought to discharge the jury as violative of Batson’s Sixth and Fourteenth
318 WEST VIRGINIA LAW REVIEW [Vol. 112
challenges45 and concluded that the Batson test fails to protect against the harm-
ful effects of implicit bias.46
In a project that focused more broadly on the criminal justice system,
Alex Geisinger critiqued the role of implicit bias in police racial profiling.
47
Claiming that profiling necessarily includes automatic and implicit cognitive
processes, Geisinger argued that racial stereotypes necessarily become intro-
duced into police profiling.48 According to Geisinger, the likelihood and power
of racial stereotypes in profiling means that “the use of racial information in the
process of policing likely never will be defensible.”49
These studies demonstrate the breadth of projects examining implicit
bias in the legal system and also highlight the many areas that have yet to be
investigated. Although most of the scholarship examining implicit bias in the
legal system has not employed empirical methods to test their hypotheses, these
works set the stage for future collaborations.
Amendment rights to a jury ‘drawn from a cross section of the community’
and guarantees to equal protection of the laws. . . . [T]he Batson Court held
that ‘a defendant may establish a prima facie case of purposeful discrimina-
tion in selection of the petit jury solely on evidence concerning the . . . [use]
of peremptory challenges at the defendant’s trial.’. . . [T]he Court recognized
that it is governmental racial discrimination when black citizens are excluded
from jury service.
Camille A. Nelson, Batson, O.J., and Snyder: Lessons from an Intersecting Trilogy, 93 IOWA L.
REV. 1687, 1693–97.
45 Page outlined the Batson standard:
In step one, the defendant must raise an inference that the prosecutor used a
preemptory challenge to exclude the person from the jury on account of her
race. In step two, the trial court judge, in order to determine whether the pe-
remptory challenge was exercised unconstitutionally, asks the prosecutor to
supply a race-neutral reason. If the prosecutor meets this burden, in step three
the judge decides whether the prosecutor exercised the peremptory challenge
with the requisite purposeful discrimination.
Page, supra note 26, at 158 (citing Batson, 476 U.S. at 93–98).
46 Id. at 245 (suggesting that the best option is to eliminate peremptory challenges). Providing
empirical support for Page’s hypothesis, Samuel Sommers and Michael Norton conducted an
empirical study that tested whether mock lawyers selected jurors using race as a determinant.
Samuel R. Sommers & Michael I. Norton, Race-Based Judgments, Race-Neutral Justifications:
Experimental Examination of Peremptory Use and the Batson Challenge Procedure, 31 LAW &
HUM. BEHAV. 261, 269 (2007). Keeping the profiles of the potential jurors identical in all other
respects, they then manipulated the racial identity of the potential jurors. They found that the
racial identity of the prospective jurors significantly affected the participants’ jury selections, and
also found that participants gave race-neutral explanations for their jury selections.
47 Alex Geisinger, Rethinking Profiling: A Cognitive Model of Bias and Its Legal Implications,
86 OR. L. REV. 657, 670–72 (2007).
48 Id.
49 Id. at 672.
2010] DIFFERENT SHADES OF BIAS 319
3. Empirical Legal Scholarship
A few legal scholars, including the authors of this Article, have empiri-
cally examined implicit bias related hypotheses. Although these studies have
been rare, they demonstrate the potential for testing directly the ways implicit
racial bias manifests in the legal system. In one such study of legal decision-
making, Levinson proposed that judges and jurors may misremember case facts
in racially biased ways.50 In forming this hypothesis, Levinson relied on a varie-
ty of social cognition studies that demonstrated: first, that people’s memories
are quite faulty; second, that stereotypes are a key ingredient in the way people
remember information; and third, that people have little ability to identify their
own memory errors.51 In the empirical component of the project, Levinson pre-
sented stories of a fight to study participants, distracted them briefly, and then
tested how well they recalled elements of the stories.52 One third of the partici-
pants read about an African American actor, one third of the participants read
about a Native Hawaiian actor, and one third of the participants read about a
Caucasian actor.53 Results indicated that participants who read about an African
American actor remembered his aggressive actions better than participants who
read about the other actors.54 In addition, Levinson found that participants in
some instances possessed false memories of the African American actor acting
aggressively.55
50 Levinson, Forgotten Racial Equality, supra note
1, at 345. Levinson also conducted anoth-
er empirical study that blended social cognition with cultural psychology. In that study, Levinson
proposed that jury duty acts as a cognitive prime that can introduce bias into decision-making.
Justin D. Levinson, Suppressing the Expression of Community Values in Juries: How “Legal
Priming” Systematically Alters the Way People Think, 73 U. CIN. L. REV. 1059 (2005) [hereinafter
Levinson, Suppressing Community Values]. Because of the complex sets of both implicit and
explicit knowledge structures related to law that most Americans possess, Levinson proposed that
when citizens are called for jury duty, these knowledge sets (which might include biases and other
negative stereotypes) are triggered. Id. at 1065–69. Levinson conducted an empirical study
where he asked participants to make various judgments in several hypothetical stories. Half of the
participants were told that they were jurors in the case, and the other half were told they were
reading facts from a newspaper story. Id. at 1075. Levinson found that the mock-jury participants
gave significantly different responses on several measures, compared to the other participants,
even for measures (such as judgments of an actor’s intentionality) for which lay judgments should
have been similar to legal judgments. Id. at 1075–78. Interestingly, the direction of the results
indicated that participants in the mock-juror condition treated defendants more harshly than the
participants in the lay judgment condition. Id. Other results indicated, although less clearly, that
perhaps these decisions worked to the detriment of out-group members. Id.
51 Levinson, Forgotten Racial Equality, supra note 1, at 373–81.
52 Id. at 390–96.
53 Id. at 394.
54 Id. at 398–401.
55 Id. at 400–02. Participants who read about a European American or Native Hawaiian actor
often did not display these false memories. Some of these false memories yielded statistically
significant results. Although the results of the empirical study do not prove conclusively that
320 WEST VIRGINIA LAW REVIEW [Vol. 112
A few research teams have become particularly interested in running
studies using a specific social cognition measure, the Implicit Association Test
(IAT). The IAT is typically administered as a computerized, timed test that
“pairs an attitude object (such as a racial group) with an evaluative dimension
(good or bad) and tests how response accuracy and speed indicate implicit and
automatic attitudes and stereotypes.”56 As psychologists Nilanjana Dasgupta
and Anthony Greenwald explain, “[w]hen highly associated targets and
attributes share the same response key, participants tend to classify them quickly
and easily, whereas when weakly associated targets and attributes share the
same response key, participants tend to classify them more slowly and with
greater difficulty.”57 Results of race and skin-tone IATs have consistently
shown implicit associations between Black (or dark skin) and Bad, compared to
White (or light skin) and Good.58 Thus, as Jeffrey Rachlinski and his colleagues
summarize, “[t]he prevailing wisdom is that IAT scores reveal implicit or un-
conscious bias.”59
Most legal scholars have discussed the IAT simply as one measure of
implicit racial bias, albeit a captivating measure.
60 However, a few research
teams since 2004 have employed the measure to test various hypotheses empiri-
cally.61
judges and jurors misremember information in racially biased ways, they lend support to that
contention and set the stage for future investigations.
Theodore Eisenberg and Sheri Lynn Johnson were interested in whether
56 Id. at 355.
57 Nilanjana Dasgupta & Anthony G. Greenwald, On the Malleability of Automatic Attitudes:
Combating Automatic Prejudice with Images of Admired and Disliked Individuals, 81 J.
PERSONALITY & SOC. PSYCHOL. 800, 803 (2001).
58 Brian Nosek et al., Pervasiveness and Correlates of Implicit Attitudes and Stereotypes, 18
EUR. REV. SOC. PSYCHOL. 36 (2008).
59 Trial Judges, supra note 4, at 1201 (citing Samuel R. Bagenstos, Implicit Bias, “Science,”
and Antidiscrimination Law, 1 HARV. L & SOC. POL’Y REV. 477 (2007) [hereinafter Implicit Bias,
“Science” and Antidiscrimination Law]).
60 See, e.g., Saujani, supra note 38; see also Kang, supra note 1; Levinson, Forgotten Racial
Equality, supra note 1. There is considerable debate about whether legal reforms should be based
on the IAT. See Implicit Bias, “Science” and Antidiscrimination Law, supra note 59; Adam
Benforado & Jon Hanson, Legal Academic Backlash: The Response of Legal Theorists to Situa-
tionist Insights, 57 EMORY L. J. 1087, 1135–43 (2008); Gregory Mitchell & Philip E. Tetlock,
Facts Do Matter: A Reply to Bagenstos, 37 HOFSTRA L. REV. 737 (2009); Gregory Mitchell &
Philip E. Tetlock, Antidiscrimination Law and the Perils of Mindreading, 67 OHIO ST. L.J. 1023,
1028, 1032–33 (2006); Wax, supra note 27.
61 Other than the studies discussed below, two studies, one unpublished, and another published
in German, have tested whether the IAT was able to predict legal decisions in mock cases. See
Arnd Florack et al., Der Einfluss Wahrgenommener Bedrohung auf die NutzungAutomatischer
Assoziationen bei der Personenbeurteilung [The Impact of Perceived Threat on the Use of Auto-
matic Associations in Person Judgments], 32 ZEITSCHRIFT FU¨R SOCZIALPSCYHOLOGIE 249
(2001)); Trial Judges, supra note 4, at 1204 (citing ROBERT LIVINGSTON, WHEN MOTIVATION
ISN’T ENOUGH: EVIDENCE OF UNINTENTIONAL DELIBERATIVE DISCRIMINATION UNDER CONDITIONS
OF RESPONSE AMBIGUITY 9–10 (2002) (unpublished manuscript, on file with the Notre Dame Law
Review)).
2010] DIFFERENT SHADES OF BIAS 321
implicit racial biases played a role in capital cases. 62 They conducted an expe-
riment in which capital defense attorneys, a group who might be expected to
resist racial biases, took a Black-White/Good-Bad IAT.63 Eisenberg and John-
son found that the majority of the study participants displayed implicit racial
biases and noted that capital defense attorneys appeared to display the same
racial biases as the rest of the population.64
Jeffrey Rachlinski later teamed up with Johnson and others to test
whether a Black/White IAT could predict racial bias in judicial decisions.
Although they found compelling
results consistent with social science data on the IAT that raise concern particu-
larly in the capital context, Eisenberg and Johnson did not test whether the IAT
predicted actual behavior or decision-making in the legal context.
65 The
researchers were interested not just in what results an IAT might yield when
given to judges, but also whether judges’ IAT performance affects their deci-
sions.66 Rachlinski and his colleagues took advantage of a unique empirical op-
portunity. Recruiting judges from several judicial educational conferences,67
62 Theodore Eisenberg & Sheri Lynn Johnson, Implicit Racial Attitudes of Death Penalty
Lawyers, 53 DEPAUL L. REV. 1539, 1539 (2004).
the
63 Id. at 1544. The researchers also tested a group of first year law students, and found similar
results in both samples. Unlike many of the computerized IAT’s discussed in legal and popular
literature, Eisenberg and Johnson used a paper and pencil version of the IAT. Id. at 1543. Eisen-
berg and Johnson explained:
In the paper and pencil version, subjects are faced with a column of words and
faces, which he or she is asked to categorize "as quickly as possible without
making too many mistakes" in twenty seconds. . . . [T]he subjects are in-
structed to go down the column checking the items that are "white or good" on
the left of the item and items that are "black or bad" on the right of the item.
After permitting questions, the subjects are told that when they turn the next
page, they will be asked to check white faces or good on the left, and black
faces or bad on the right, completing as many as possible in the allotted
time . . . . After completing this task, the subjects are asked to turn the page,
and the new pairing of black with good and white with bad is explained. Sub-
jects then complete the same task with the new pairing . . . . The number of
items correctly completed on each test is then counted; it is not the number of
items a particular subject can complete that is of significance, but the differ-
ence in the number of items he or she completes when white is paired with
good and black with bad, as contrasted with the number completed when
black is paired with good and white with bad.
Id. at 1543–45.
64 Id. at 1553.
65 See generally Trial Judges, supra note 4.
66 Id. at 1211–21.
67 This recruitment methodology is rare in empirical legal studies and provides unique infor-
mation about one of the most important groups of legal system participants. For studies employ-
ing a similar participant recruitment methodology, see id. at 1205 (citing Chris Guthrie et al.,
Blinking on the Bench: How Judges Decide Cases, 93 CORNELL L. REV. 1, 13 (2007); Chris Guth-
rie et al., Inside the Judicial Mind, 86 CORNELL L. REV 777, 814–15 (2001); Jeffrey J. Rachlinski
et al., Inside the Bankruptcy Judge’s Mind, 86 B.U. L. REV. 1227, 1256–59 (2006); Andrew J.
322 WEST VIRGINIA LAW REVIEW [Vol. 112
researchers ran White-Black/Good-Bad IAT’s and asked the judges to make
decisions in a series of short mock cases.68 Results of the study first showed
that the vast majority of the Caucasian judge participants exhibited an implicit
preference for White over Black.69 Furthermore, the results showed that, in two
of the three mock cases (those in which race was primed subliminally), the Race
IAT scores predicted the judges’ decisions.70 Yet, the results also demonstrated
that in a mock case where the defendant’s race was specifically identified, the
Race IAT did not predict decisions.71 The researchers claimed that this result
indicated that motivated judges may sometimes be able to resist the effects of
implicit bias on judicial decisions.72
Levinson, Huajian Cai, and Danielle Young also were interested in the
IAT in the legal context.
73 Like Rachlinski and colleagues, Levinson and his
colleagues were particularly interested in the predictive validity of the IAT in
the legal setting. Unlike other projects, however, which relied exclusively upon
existing IAT measures, Levinson and his colleagues designed their own IAT. In
addition to running an IAT similar to those run by previous legal scholars, they
created a Black/White Guilty/Not Guilty IAT and tested it empirically. Results
of the study showed first that participants held a strong implicit association be-
tween Black and Guilty compared to White and Guilty, raising concerns about
implicit racial bias and the presumption of innocence.74
Wistrich et al., Can Judges Ignore Inadmissible Information? The Difficulty of Deliberately Dis-
regarding, 153 U. PA. L. REV. 1251, 1323–24 (2005)).
Next, specifically ex-
amining predictive validity, they found that the IAT scores predicted partici-
68 Trial Judges, supra note 4, at 1209. In two of the three vignettes used by Rachlinski and his
colleagues, the race of the defendant was not indicated. However, using a priming technique
similar to that used by Sandra Graham and Brian Lowery in Sandra Graham & Brian Lowery,
Priming Unconscious Racial Stereotypes about Adolescent Offenders, 28 LAW & HUM. BEHAV.
483 (2004), the researchers primed half of the participants subliminally with words related to
African Americans. The researchers treated this priming technique, in which participants were
primed with Black or neutral words (not White words), as being similar to a subliminal racial
identification of the defendant.
69 Trial Judges, supra note 4, at 1210. 87.1% of the Caucasian judges exhibited this prefe-
rence. Id.
70 This finding carried marginal statistical significance. Specifically, Rachlinski and his col-
leagues found that IAT score “had a marginally significant influence on how the [racial] prime
influenced their judgment. Judges who exhibited a white preference on the IAT gave harsher
sentences to defendants if they had been primed with black-associated words rather than neutral
words . . . .” Id. at 1214–15.
71 Id. at 1218.
72 Id. at 1223.
73 Justin D. Levinson, Huajian Cai & Danielle Young, Guilty by Implicit Bias: The Guilty-Not
Guilty Implicit Association Test, 8 OHIO ST. J. CRIM. L. (forthcoming 2010). This IAT study was
initially conducted as a companion study to the empirical project described in this Article, but due
to the unique nature of the IAT study and the study’s empirical results, it is considered in a sepa-
rate article.
74 Id. at 17.
2010] DIFFERENT SHADES OF BIAS 323
pants’ evidence judgments.75 Participants who implicitly associated Black and
Guilty were more likely to make harsher judgments of ambiguous evidence.76
Legal scholarship has thus evolved considerably since Lawrence’s in-
troduction to implicit racial bias and the law. Yet there are still only a few em-
pirical studies that explore the role of implicit racial bias in the law. In Section
IV, we detail the empirical study we conducted. Next, however, we review so-
cial science work broadly investigating the role of race in legal decision-
making.
Thus, not only did the study challenge the supposed racial equality underlying
the law’s presumption of innocence, but also it connected this racial bias to
judgments of trial evidence.
B. Mock-Jury Research on Racial Bias
Similar to legal scholars discussing implicit bias, social scientists have
yet to examine fully how implicit biases may affect the way jurors evaluate evi-
dence. However, social scientists have also been long interested in understand-
ing racial disparities in the criminal justice system. They have thus conducted
empirical investigations designed to test whether jurors decide cases differently
based upon the defendant’s race.77 Although they have not tested the role of
implicit bias in decision-making, these studies have laid the foundation for un-
derstanding how race might affect the way jurors think and make decisions.78
75 Id. at 18. These evidence judgments were the ones described in this Article.
We therefore briefly describe two large-scale projects that reviewed, summa-
rized, and evaluated dozens of studies that have attempted to investigate the role
of the defendant’s race in mock-jury decisions.
76 Id. at 19.
77 These studies have tended not to focus on jurors’ cognitive processes, such as memory and
evidence evaluation, instead focusing on outcome measures such as guilt and punishment. See
Tara L. Mitchell et al., Racial Bias in Mock Juror Decision-Making: A Meta-Analytic Review of
Defendant Treatment, 29 LAW & HUM. BEHAV. 621 (2005). This does not mean, however, that all
studies ignore the role of evidence in decision-making. James Johnson, for example, found that
mock jurors were more likely to disregard instructions to ignore evidence when that evidence
tended to indicate the guilt of a Black defendant (compared to a White defendant). James D.
Johnson et al., Justice is Still Not Colorblind: Differential Racial Effects of Exposure to Inadmiss-
ible Evidence, 21 PERSONALITY & SOC. PSYCHOL. BULL. 893 (1995).
78 See Samuel R. Sommers & Phoebe C. Ellsworth, How Much Do We Really Know about
Race and Juries? A Review of Social Science Theory and Research, 78 CHI.-KENT L. REV. 997
(2003). Considering the amount of discussion in legal scholarship about race and decision-
making, Sommers and Ellsworth found the lack of social science studies “surprising.” Id. at 1005.
See also Denis C. Ugwuegbu, Racial and Evidential Factors in Juror Attribution of Legal Re-
sponsibility, 15 J. EXPERIMENTAL SOC. PSYCHOL. 133 (1979) (varying the strength of the evidence
provided, and finding that White mock jurors required less evidence to convict a Black defen-
dant).
324 WEST VIRGINIA LAW REVIEW [Vol. 112
In 2003, Samuel Sommers and Phoebe Ellsworth comprehensively re-
viewed a variety of studies that tested the role of race in mock-jury decisions.79
The researchers found mixed results — some studies purported to find racial
bias against African American defendants,80 and other studies purported to find
either no bias or a pro-African American bias.81
[N]o consensus has been reached regarding the influence of a
defendant’s race on White mock jurors. Some studies have
suggested that White jurors are biased against Black defendants,
others have yielded no evidence of bias, and a few researchers
have found that White jurors are biased against White defen-
dants. But substantial evidence exists to support the conclusion
of many legal scholars that, at least under some conditions,
Summarizing the results of
their review, Sommers and Ellsworth stated:
79 Sommers & Ellsworth, supra note 78. Most of the studies Sommers and Ellsworth re-
viewed primarily measured outcome variables, such as guilty/not guilty verdicts and length of
sentence. This methodological decision to focus on verdicts and punishment has both advantages
and disadvantages. The primary advantage is obvious: it presumably measures the most important
legal outcome measures. The disadvantages are somewhat less obvious: first, focusing on guilt
and punishment judgments may overlook the way implicit racial bias truly functions. Measuring
verdicts and punishment judgments without also measuring cognitive processes might cover up
the most meaningful part of the jury decision-making story. And second, testing verdicts and
punishment judgments in a mock trial setting may actually heighten differences between decision-
making in real trials (with real consequences) and mock trials (with no consequences). For earlier
comprehensive discussions of race and decision-making, see Sheri Lynn Johnson, Black Inno-
cence and the White Jury, 83 MICH. L. REV. 1611 (1985); Nancy J. King, Postconviction Review
of Jury Discrimination: Measuring the Effects of Juror Race on Jury Decisions, 92 MICH. L. REV.
63 (1993).
80 For example, Sommers and Ellsworth discussed a study by Klein and Creech that showed
participants a short video summary of a rape trial. Sommers & Ellsworth, supra note 78, at 1006
(citing Kitty Klein & Blanche Creech, Race, Rape, and Bias: Distortion of Prior Odds and Mean-
ing Changes, 3 BASIC & APPLIED SOC. PSYCHOL. 21 (1982)). As Sommers and Ellsworth describe
the study, some participants watched a video depicting a Black defendant and others watched a
video depicting a White defendant. The researchers found that participants who saw the film
depicting a Black defendant were “more likely to believe he was guilty than jurors who saw the
same trial video with a White defendant.” Id.
81 Id. at 1008 (citing Ronald L. Poulson, Mock Juror Attribution of Criminal Responsibility:
Effects of Race and the Guilty But Mentally Ill (GBMI) Verdict Option, 20 J. APPLIED SOC.
PSYCHOL. 1596 (1990)). Poulson presented participants with an audio summary of a murder trial.
Watching an evidence slideshow, half of the participants saw a Black defendant and the other half
saw a White defendant. Paulson found that White participants were more likely to acquit Black
defendants by reason of insanity compared to White defendants. Sommers and Ellsworth point out
that these results are difficult to generalize to other scenarios, because “[t]o the extent that White
jurors view mental illness or ‘insanity’ as more consistent with their stereotype of Black versus
White defendants, Poulson’s finding of same-race bias becomes less surprising.” Id. at 1009.
2010] DIFFERENT SHADES OF BIAS 325
White jurors exhibit racial bias in their verdicts and sentencing
decisions.82
A 2005 meta-analysis conducted by Tara Mitchell and her colleagues
followed up Sommers and Ellsworth’s discussion by quantitatively testing race-
effects on mock-jurors.
83 Aggregating and analyzing verdict data from thirty-
four studies and 7397 participants, and sentencing data from sixteen studies and
3141 participants, the researchers analyzed whether mock-jurors across these
studies demonstrated racial bias in decision-making. The researchers found
small but significant differences in race effects on both verdicts and sentences,
indicating that mock jurors were biased in favor of defendants of their own
race.84 However, this significant effect was quite small, and the statistical signi-
ficance disappeared if the experimenters eliminated certain types of studies.85
Sommers and Ellworth’s review and Mitchell and her colleagues’ meta-
analysis demonstrate that although scholars have continued pursuing the hypo-
thesis that jurors harbor racial biases in decision-making, more research, in-
formed by implicit social cognition, is needed. Existing research often fails to
give a theoretical explanation for how and why these biases might occur, and
thus far has often assumed that the bias is conscious and intentional.
86
82 Id. at 1010. Sommers and Ellsworth relied on social cognition theory to propose one poten-
tial explanation for the conflicting results they discussed. Specifically, they proposed that
“Whites are less likely to demonstrate racial bias when concerns about prejudice are salient.” Id.
at 1012–13 (citing Samuel L. Gaertner & John F. Dovidio, The Aversive Form of Racism, in
PREJUDICE, DISCRIMINATION, AND RACISM 61, 69 (John F. Dovidio & Samuel L. Gaertner eds.,
1986)). Sommers and Ellsworth also discussed a study they conducted that supported that hypo-
thesis. In the study, participants in the race-salient condition read a story of a crime that included
the following sentence: “You know better than to talk that way about a White (or Black) man in
front of his friends.” Id. at 1015 (citing Samuel R. Sommers & Phoebe C. Ellsworth, Race in the
Courtroom: Perceptions of Guilt and Dispositional Attributions, 26 PERSONALITY & SOC.
PSYCHOL. BULL. 1367, 1373 (2000)). Participants in the non-race-salient condition read the same
sentence, except there was no reference to race at all (“You know better than to talk about a man
that way . . . .”). Id. When race was not salient, White mock jurors gave higher guilt ratings and
punishment judgments to Black defendants. Id. But when race was made salient, there was no
racial bias. Id. Priming, for example, which is frequently subliminal and not even noticed by the
participant, would presumably not trigger a desire to avoid bias.
In addi-
tion, most research to date, including several meta-analyses on the topic (leading
83 See Mitchell et al., supra note 77.
84 Id. at 627. Mitchell and her colleagues looked at how study participants made decisions
when the mock-defendant was of a different racial group. This methodology thus was not limited
to examining race effects on Black defendants. Id. at 624–25.
85 For example, the researchers found that studies using dichotomous (guilty/not guilty) va-
riables did not have the same race effects as studies using continuous scale variables (e.g. on a
scale of 1–10, how guilty is the defendant). Id. at 630. The researchers also found that communi-
ty samples displayed greater race-based sentencing bias than college student samples. Id. at 631.
86 Sommers and Ellsworth note this assumption and briefly discuss social cognition literature.
Sommers & Ellsworth, supra note 78, at 1011–12.
326 WEST VIRGINIA LAW REVIEW [Vol. 112
up to and including Mitchell and her colleagues’ study),87
III. ACTIVATING POWERFUL RACIAL STEREOTYPES
demonstrates that
although quite a bit of research has attempted to test race effects on guilty and
not-guilty verdicts, little research has looked at race effects in the way jurors
evaluate evidence. Therefore, the hypothesis that racial cues lead to biased eval-
uations of trial evidence has yet to be fully examined. The next section thus sets
the stage for our empirical test of Biased Evidence Hypothesis by explaining the
ways that simple racial cues can activate powerful racial stereotypes.
Social science research on implicit bias has opened up new doors for
investigating how implicit bias affects jurors.88 Of the numerous ways in
which social cognition researchers have found that people harbor implicit bi-
ases,89
87 Mitchell et al., supra note
research on priming is particularly relevant in considering how racial
stereotypes might affect the way jurors evaluate evidence. Priming describes
“the incidental activation of knowledge structures, such as trait concepts and
77 (citing Ronald Mazzella & Alan Feingold, The Effects of Phys-
ical Attractiveness, Race, Socioeconomic Status, and Gender of Defendant and Victims on Judg-
ments of Mock Jurors: A Meta-Analysis, 24 J. APPLIED SOC. PSYCHOL. 1315 (1994)); Laura T.
Sweeney & Craig Haney, The Influence of Race on Sentencing: A Meta-analytic Review of Expe-
rimental Studies, 10 BEHAV. SCI. & THE LAW 179 (1992).
88 See, e.g., Levinson, Forgotten Racial Equality, supra note 1.
89 For example, research demonstrates that people automatically associate African Americans
and guns. See Payne, supra note 26, at 185–86 (2001). Related studies include “shooter bias”
studies, which test how fast video game participants can “shoot” armed perpetrators and refrain
from shooting unarmed innocents. These studies show that people are faster to shoot dark-
skinned perpetrators than light-skinned perpetrators, but are faster to hit a “safety” button when
seeing light-skinned innocents compared to dark-skinned innocents. See Joshua Correll et al.,
Event-Related Potentials and The Decision to Shoot: The Role of Threat Perception and Cognitive
Control, 42 J. EXPERIMENTAL SOC. PSYCHOL. 120, 122 (2006); Joshua Correll et al., The Police
Officer’s Dilemma: Using Ethnicity to Disambiguate Potentially Threatening Individuals, 83 J.
PERSONALITY & SOC. PSYCHOL. 1314, 1321 (2002).
Research also demonstrates that people hold strong implicit associations between African
Americans and negative attitudes and stereotypes. See Anthony Greenwald et al., Measuring
Individual Differences in Implicit Cognition: The Implicit Association Test, 74 J. PERSONALITY &
SOC. PSYCHOL. 1464, 1478 (1998) [hereinafter Measuring Individual Differences]; Anthony
Greenwald & Mahzarin Banaji, Implicit Social Cognition: Attitudes, Self-Esteem, and Stereotypes,
102 PSYCHOL. REV. 4 (1995); Brian Nosek et al., Harvesting Implicit Group Attitudes and Beliefs
from a Demonstration Website, 6 GROUP DYNAMICS 101, 102 (2002)).
Finally, research demonstrates that racial stereotypes can be activated in milliseconds and
can affect people’s judgments and actions. See Patricia G. Devine, Stereotypes and Prejudice:
Their Automatic and Controlled Components, 56 J. PERSONALITY & SOC. PSYCHOL. 5 (1989) [he-
reinafter Stereotypes and Prejudice]; Payne, supra note 26. See generally Anthony G. Greenwald
& Linda Hamilton Krieger, Implicit Bias: Scientific Foundations, 94 CAL. L. REV. 945 (2006);
Kang, supra note 1 (explaining a variety of social science research on implicit attitudes and race);
Levinson, The Complicitous Mind, supra note 2.
2010] DIFFERENT SHADES OF BIAS 327
stereotypes, by the current situational context.”90 Priming research demon-
strates first, that stereotypes are activated easily, automatically, and often un-
consciously, and second, that once people have been primed, it affects the way
they make decisions in racially stereotyped ways.91
This section explains social science research on priming, with the par-
ticular purpose of considering whether exposing jurors to simple racial cues
(such as showing them a security camera photo of a dark-skinned perpetrator)
can incorporate stereotypes into the way they evaluate ambiguous trial evidence.
In light of this research, and
considering the various ways racial stereotypes might become elicited in a crim-
inal trial, one can hypothesize that jurors may unintentionally evaluate trial evi-
dence in racially biased ways.
A. Racial Stereotypes are Primed Easily
In order to investigate the hypothesis that jurors’ racial stereotypes can
be primed easily during a trial, it is important to examine first how stereotypes
are primed. Research confirms that racial stereotypes can be activated quickly
and often without a person’s conscious awareness.92 Keith Payne, for example,
hypothesized that showing participants a photograph of a Black face for a mere
200 milliseconds could activate racial stereotypes associated with African
Americans.93
90 Bargh et al., supra note
Payne quickly flashed photos of Black or White faces on a screen,
7, at 230. As Levinson has described, “priming studies show how
causing someone to think about a particular domain can trigger associative networks related to
that domain.” Levinson, The Complicitous Mind, supra note 2, at 608.
91 In the context of jury decision-making, priming may be relevant in a variety of ways. See
Levinson, The Complicitous Mind, supra note 2 (introducing Death Penalty Priming Hypothesis,
which posits that the supposedly race-neutral death qualification of jurors can act to automatically
trigger racial stereotypes in capital cases); Levinson, Suppressing Community Values, supra note
50 (arguing that simply placing citizens on juries primes them to think in ways that might be
concerning). One particularly concerning possibility is that jurors may be primed with racial cues
during a trial and that these cues may activate a broad range of racial stereotypes relating to the
crime. There are a number of ways that jurors might be primed during a trial. For example, jurors
might be primed by the race of the defendant, a piece of evidence, a witness, the victim, or co-
defendants. These activated stereotypes may then influence the way jurors evaluate trial evidence
and make decisions.
92 See Stereotypes and Prejudice, supra note 89, at 8 (citing John A. Bargh & Paula Pietromo-
naco, Automatic Information Processing and Social Perception: The Influence of Trait Informa-
tion Presented Outside of Conscious Awareness on Impression Formation, 43 J. PERSONALITY &
SOC. PSYCHOL. 437 (1982)); Levinson, The Complicitous Mind, supra note 2, at 606–08. Accord-
ing to Devine, Bargh and Pietromonaco’s study showed that “even when subjects were unaware of
the content of the primes, priming increased the likelihood that the primed category was used to
interpret subsequently presented ambiguous category related information.” Stereotypes and Pre-
judice, supra note 89, at 8.
93 See Payne, supra note 26, at 184–85. Much of the remaining language of this paragraph,
including citations, is taken more or less verbatim from an earlier work. See Levinson, The Com-
plicitous Mind, supra note 2, at 607.
328 WEST VIRGINIA LAW REVIEW [Vol. 112
and then immediately displayed images of guns or tools.94 Participants were
told to identify the guns and tools as quickly as possible.95 Results of the study
indicated that when participants saw photos of Black faces immediately before
photos of guns, they were significantly faster at identifying the guns than when
they saw photos of White faces before photos of guns.96
Other studies employing different priming techniques confirm that ra-
cial stereotypes can be activated easily. In a study that used music to prime
participants, Laurie Rudman and Matthew Lee had study participants listen to
either violent rap or contemporary pop music songs for approximately thirteen
minutes.
Payne’s study shows
that racial stereotypes can be elicited automatically through visual stimuli in a
number of milliseconds, and that these stereotypes can affect the speed and ac-
curacy of meaningful object classification tasks.
97 The researchers then examined whether the music primed partici-
pants’ implicit and explicit racial stereotypes98 and found that participants who
listened to the rap music songs exhibited greater implicit stereotypes99 than par-
ticipants who listened to the pop tunes.100 The researchers also found that the
stereotype-music prime activated participants’ stereotypes irrespective of the
participants’ self-reported prejudice levels, indicating that the stereotype net-
works elicited by priming often operate implicitly and without people’s en-
dorsement or even awareness.101
94 See Payne, supra note
This study illustrates a basic principle underly-
ing priming: simple primes that implicate racial stereotypes (such as a clip of
violent rap music) can automatically activate a vast network of associated impli-
cit racial stereotypes. As we will discuss, it should not be surprising then if see-
ing a security camera photograph of a dark-skinned perpetrator automatically
26, at 184–85.
95 Id. The objects consisted of guns and non-gun objects (the non-gun objects were hand tools,
such as a socket wrench and an electric drill). Payne also told participants that the quickly flash-
ing photographs of faces, which appeared long enough that participants might notice them gener-
ally, but short enough so that they would not recognize them, only served to signal the participant
that a photograph of an object was about to appear. See id.
96 Id. at 185. Similarly, when participants saw photos of White faces immediately before
photos of tools, they were significantly faster at identifying the tools than when they saw photos
of Black faces before photos of tools.
97 See Laurie A. Rudman & Matthew R. Lee, Implicit and Explicit Consequences of Exposure
to Violent and Misogynous Rap Music, 4 GROUP PROCESSES & INTERGROUP RELATIONS 133
(2002). Participants were led to believe that they were participating in a marketing study.
98 Id. at 138–39.
99 The researchers used a stereotype IAT test to measure stereotype activation. This IAT tests
how quickly people associate Black men and negative attributes, like “hostile” and “criminal.” Id.
See supra notes 56–59 and accompanying text for a detailed explanation of the IAT test.
100 The contemporary pop tunes were sung by both White and Black artists. Note that the ste-
reotype IAT was administered after participants believed that the music study had ended. Partici-
pants were told that they were taking a pilot test of a separate study. Id. at 136.
101 Id. at 142.
2010] DIFFERENT SHADES OF BIAS 329
activates a network of racial stereotypes.102
B. Primed Racial Stereotypes Affect Decision-Making
Research on priming thus confirms
that racial stereotypes can become activated quickly and easily. We must next
investigate the effects of activating these implicit racial stereotypes.
In the courtroom, it is important to consider not only whether elements
of a trial can prime jurors’ racial stereotypes, but also whether the priming mat-
ters. After all, if the prime has no influence on decision-making, then the prim-
ing phenomenon would not be particularly concerning. Research has con-
firmed, however, that priming can automatically affect a broad range of deci-
sions and behaviors, a finding that supports the hypothesis that priming matters
in jury decision-making.
In one well-known study, Patricia Devine found that priming stereo-
types of African Americans affected the way people judged ambiguous beha-
viors involving African American actors.103 Devine primed participants by
flashing words such as “poor,” “athletic,” and “Black” so quickly that partici-
pants saw them but did not consciously recognize their content.104 After prim-
ing the participants, Devine asked them to read a paragraph in which a person
behaved in an ambiguously hostile way. For example, participants read about a
person who “demands his money back from a store clerk immediately after a
purchase and refuses to pay his rent until his apartment is repainted.”105 De-
vine then asked participants to make judgments about the person.106 She found
a direct relationship between the priming and the judgments participants made:
“participants who were primed with more stereotyped words judged the actor’s
ambiguous behavior as more hostile than participants who were primed with
fewer stereotyped words.”107
102 These networks connect related stereotypes with one another. For example, priming some-
one to think about African Americans and athleticism will simultaneously and automatically cause
people to also think about African Americans and aggressiveness as well as African Americans
and inferior intelligence. See Stereotypes and Prejudice, supra note
To put it more simply, subliminally priming par-
ticipants with flashing words like “lazy” and “Black” affected the way partici-
pants later judged the hostility of African Americans in racially stereotyped
ways.
89.
103 Id.
104 This non-conscious priming activated racial stereotypes of African Americans. Id. at 9–10.
Participants in one condition saw these racial stereotype words more frequently than participants
in the other condition (Eighty percent versus twenty percent).
105 Id. at 10. This paragraph had been used in previous research. Id. (citing Bargh & Pietro-
monaco, supra note 92; C.S. Carver et al., Modeling: An Analysis in Terms of Category Accessi-
bility, 19 J. OF EXPERIMENTAL SOC. PSYCHOL. 403 (1983)).
106 Id.
107 Levinson, The Complicitous Mind, supra note 2, at 625 (citing Stereotypes and Prejudice,
supra note 89 at 11–12).
330 WEST VIRGINIA LAW REVIEW [Vol. 112
Following Devine’s work, researchers have continued to investigate
how primed stereotypes affect decision-making and behaviors. In a follow-up
study to their music priming study, Rudman and Lee tested how briefly listening
to violent rap music (compared to non-violent pop music) affected participants’
unrelated behavioral judgments of Black and White actors.108 As in their first
study, Rudman and Lee played audio clips for participants. This time, however,
the researchers rigged the audio player to break down during the sixth song.
Participants were then lead to believe that the study, which had been described
as a “marketing” study, had to be discontinued. Participants were told, howev-
er, that they could complete a separate, unrelated study questionnaire on “person
perception”109 in order to complete their obligation of participating in a study.110
As a result of this study design, Rudman and Lee were able to test
whether the rap music prime affected entirely unrelated judgments of ambi-
guous behaviors. Participants read a story about ambiguously sexist behaviors
(such as a man refusing to tip a female waitperson or a man refusing to let a
female door-to-door salesperson inside the house).
111 Half of the participants
read about Donald (whom the researchers believed participants would consider
White) and half of the participants read about Kareem (presumably Black).112
Participants were then asked to rate the story actor’s (Donald or Kareem’s) level
of hostility, sexism, and intelligence.113 Results of the study corroborated the
researchers’ predictions. Participants who previously listened to rap music
(compared to participants who listed to pop music) made greater hostility and
sexism ratings for Kareem (compared to those who read about Donald), and
even rated Kareem as having lower intelligence (also compared to those who
read about Donald).114
Demonstrating the dangers of priming racial stereotypes in the criminal
justice context, one research team examined how subliminal racial priming af-
fected judgments related to juvenile offenders. Testing a participant group of
police officers and juvenile probation officers, Sandra Graham and Brian Lo-
This study shows that a simple prime activation (here,
hostile and misogynistic rap music) can influence judgments of seemingly unre-
lated ambiguous traits and behaviors (such as intelligence) in racially biased
ways. It could follow then that racial stereotypes primed in the courtroom might
affect the way jurors interpret and evaluate trial evidence.
108 Rudman & Lee, supra note 97, at 140–41.
109 The experimenters verified that participants did not suspect that the music study was related
to the person perception/ judgment study. See id.
110 All participants agreed to complete the study. To increase the believability of the story that
the second questionnaire was unrelated to the audio/music study, participants received a separate
consent form. Id. at 141.
111 Id. at 140.
112 The researchers later confirmed that these beliefs were true. Participants did consider gen-
erally Donald as White and Kareem as Black. Id. at 141.
113 Id. at 140.
114 Rudman & Lee, supra note 97, at 142.
2010] DIFFERENT SHADES OF BIAS 331
wery employed a priming scheme based on Devine’s methodology, flashing
content-coded words at a high speed such that participants would be primed but
would remain unaware of the content of the prime.115 They primed half of the
participants with words related to African Americans, such as “Harlem”, “ghet-
to,” and “dreadlocks,” and primed the other half of participants with non-racial
content words such as “sunset,” “mosquito,” and “toothache.”116 They then
presented participants with two hypothetical crime reports detailing juveniles
(whose race was not identified) engaging in misbehavior and measured whether
the priming affected judgments of those behaviors.117 The results of the study
confirmed that the priming activated racial stereotypes of African Americans,
and demonstrated that priming affected the way the participants made judg-
ments.118 Both police officers and juvenile probation officers who had been
primed with African American words made harsher decisions of the juveniles.
The biased decisions included a broad range of judgments, including the juve-
niles’ traits,119 culpability, and likelihood of reoffending.120
These studies demonstrate powerfully that racial stereotypes are acti-
vated easily and that they can affect a broad range of decisions. They, thus, set
the stage for an empirical examination of whether racial cues can automatically
affect the way people evaluate trial evidence.
IV. TESTING FOR BIAS — THE EMPIRICAL STUDY
In light of the still evolving stage of legal scholarship on implicit bias,
and building upon the social cognition theory of priming, we examined whether
altering the skin tone of a perpetrator in a security camera photo affected the
way participants judged various pieces of trial evidence.
A. Methods
Sixty-six jury eligible students at the University of Hawai`i participated
in the study.121
115 Graham & Lowery, supra note
Participants were seated in separate cubicles, each of which
68.
116 Id. at 489.
117 Id. at 490.
118 Id. at 493–94.
119 Examples of traits tested were mature-immature and violent- nonviolent. Id. at 490.
120 Id. at 496.
121 Participants received extra credit for participating in the study. All participants were 18
years old or older and none had been convicted of a felony. All but seven of these participants
were jury eligible in the state of Hawai`i. Seven participants were residents of other states. Sta-
tistical tests were conducted in order to test for any significant differences between Hawai`i resi-
dents (n = 58) and non-Hawai`i residents (n=7). No significant differences were found, and there-
fore we combined Hawai`i residents and non-residents for the purpose of data analysis. There
were originally sixty-seven participants, but one participant was removed due to missing data.
332 WEST VIRGINIA LAW REVIEW [Vol. 112
contained a Dell computer.122 After participants gave informed consent, the
study began. The first task began by describing an armed robbery. Participants
read the following information:123
The defendant has been charged with armed robbery. The inci-
dent occurred at 11pm on December 18, 2008, when the Quick
Stop Mini Mart was robbed by two armed men wearing masks.
According to the police report, the owner of the Mini Mart had
just closed the store when two armed men barged into the store.
One of the men pointed the gun at the owner while the other
walked behind the counter to the cash register. The owner ob-
eyed all of the men's commands and was not injured. The men
left the store with approximately $550 in cash. They fled in a
dark blue 4-door full sized sedan.
Participants were then instructed that they would view several photo-
graphs of the crime and crime scene.124 After receiving this instruction, partici-
pants viewed five photographs in succession. The photograph slideshow was
pre-timed such that each photo was displayed for four seconds each. All partic-
ipants viewed the photos in the same order. The third slide was the subject of
the experimental manipulation. This photo contained an image of a gunman
wearing a mask. The gunman holds a gun in his left hand and is reaching over
the counter with his right hand, presumably in an attempt to collect money.125
After participants viewed the slideshow, they were informed that a sus-
pect was arrested and charged with armed robbery. They were then told that
they would be asked to evaluate certain pieces of evidence from witnesses who
would testify at the robbery trial.
Based upon the experimental condition they were randomly assigned to, partici-
pants viewed a photo of either a dark-skinned perpetrator or a lighter-skinned
perpetrator. The photos were identical in all other respects. Because the only
skin visible in the photo was the perpetrator’s forearms, experimental manipula-
tion was accomplished without regard for any race-identifiable facial features.
126
122 There were two cubicles in the lab. Thus, no more than two participants took the study at
any given time. The entire study was computerized.
The participants then read and responded
123 The instructions for the first task were: “The first task is about how jurors evaluate evidence
for criminal trials. Please carefully read the following description of a crime as if you were a jury
member.”
124 They were instructed: “You will now see a slideshow of photographs of the crime and
crime scene. Please study the photos carefully. After seeing the photographs you will be asked
some questions.”
125 These photos are attached as Appendix B.
126 They were instructed:
You will read one piece of evidence at a time. We are trying to determine
which pieces of evidence are most important to the trial and verdict. In partic-
ular, we want to know whether each specific piece of evidence tends to indi-
2010] DIFFERENT SHADES OF BIAS 333
individually to twenty pieces of evidence, which were presented in randomized
order.127
The defendant used to be addicted to drugs.
A few examples are:
The defendant has been served with a notice of eviction from
his apartment.
The defendant is left-handed.
The defendant was a youth Golden Gloves boxing champ in
2006.
The defendant is a member of an anti-violence organization.
The defendant does not have a driver’s license or car.
The pieces of evidence were designed to be ambiguous to varying de-
grees. Certain pieces of evidence tended to indicate that the defendant may be
guilty (e.g. the store owner identified the defendant’s voice in an audio line-up).
Others tended to indicate that the defendant may be not guilty (e.g. the defen-
dant had a used movie ticket stub for a show that started 20 minutes before the
crime occurred.) Still others tended to be more neutral (e.g. the defendant was a
youth Golden Gloves boxing champ in 2006). Based on social cognition re-
search,128
cate that the defendant is Guilty or Not Guilty. On each screen that follows,
you will see one piece of evidence listed. For each piece of evidence you see,
select one of the numbered responses.
we predicted that priming would affect evidence evaluation so long as
Participants were asked to respond on the following 1–7 scale:
1 = very strongly tending to indicate Not Guilty
2 = strongly tending to indicate Not Guilty
3 = somewhat indicating Not Guilty
4 = neutral evidence
5 = somewhat tending to indicate Guilty
6 = strongly tending to indicate Guilty
7 = very strongly tending to indicate Guilty
127 These items of evidence are listed in Appendix A. The order was randomized to eliminate
order effects. The order that evidence is presented has been shown to effect the way it is judged.
See Kurt A. Carlson & J. Edward Russo, Biased Interpretation of Evidence by Mock Jurors, 7 J.
EXPERIMENTAL PSYCHOL: APPLIED 91 (2001).
128 See Eisenberg & Johnson, supra note 62, at n.43 (citing Sheri Lynn Johnson, Black Inno-
cence and the White Jury, 83 MICH. L. REV. 1611, 1626–34 (1985)) (noting that stereotypes are
more likely to affect a decision when that decision is difficult). See generally Stereotypes and
Prejudice, supra note 89 (testing how priming race affects judgments of ambiguously hostile
behaviors).
334 WEST VIRGINIA LAW REVIEW [Vol. 112
the items were at least slightly ambiguous, such that activated stereotypes would
serve to influence the way participants perceived and interpreted information.
For example, if a racial stereotype of aggression were activated by viewing a
security camera image of a dark-skinned perpetrator, participants might be more
likely to believe that the defendant who was seen shopping at the store two days
prior was “casing the joint,” rather than simply shopping (which might indicate
a case of mistaken identity129
Participants were next asked to decide whether the defendant was guilty
or not guilty. This measure allowed us to test whether the participants’ evidence
judgments predicted their guilty and not guilty verdicts. To add more meaning
to this measure by providing a wider range of responses, we also asked partici-
pants to answer “on a scale of 0 (definitely not guilty) to 100 (definitely guilty),
how guilty is the defendant?”
). Each piece of evidence was chosen so that mul-
tiple interpretations of that evidence would be possible, although this was par-
ticularly true for the more neutral pieces of evidence.
After completing the evidence measurement task, participants then
completed several other measures, including the Modern Racism Scale,130 “feel-
ing thermometer” measures,131 and two IATs designed to test implicit associa-
tions underlying the concepts of race and criminal guilt.132 Including these
measures allowed us to examine whether the participants’ evidence judgments
were related to other measures. For example, if racially biased evidence judg-
ments were related to negative explicit racial preferences, one would see a cor-
relation between evidence scores and scores on the Modern Racism Scale.133
B. Limitations of Study and Future Directions
Because Biased Evidence Hypothesis is a new hypothesis that had not
previously been tested, our study was necessarily limited, particularly in that it
129 Cases of mistaken identity, often based upon the memory-driven phenomenon of “uncons-
cious transference,” have been reported. The human memory can sometimes transfer the source
of one memory to another context. See Francis A. Gilligan et al., The Theory of “Unconscious
Transference”: The Latest Threat to the Shield Laws Protecting the Privacy of Victims of Sex
Offenses, 38 B.C. L. REV. 107, 123–24, 142 (1996); Levinson, Forgotten Racial Equality, supra
note 1, at 382; Richard L. Marsh et al., Gender and Orientation Stereotypes Bias Source-
Monitoring Attributions, 14 MEMORY 148, 157–58 (2006).
130 The Modern Racism Scale purports to measure racial beliefs by asking self-report questions.
Researchers use it to evaluate explicit racial preferences and to compare responses of implicit
measures on race to explicit measures on race. See J.B. McConahay, Modern Racism, Ambiva-
lence, and the Modern Racism Scale, in PREJUDICE, DISCRIMINATION, AND RACISM 91 (John F.
Davido & Samuel L. Gaertner eds., 1986).
131 Feeling thermometers also measure explicit racial preferences by asking how warm or cool
a person feels towards a certain group. It is an alternative measure to the Modern Racism Scale.
See Measuring Individual Differences, supra note 89.
132 For a complete discussion of the IATs, see Levinson et al., supra note 73.
133 Based on the results of other social cognition studies, we did not expect that Modern Racism
Scale scores would correlate with skin tone-based judgments of evidence.
2010] DIFFERENT SHADES OF BIAS 335
was not tested in real trial conditions. Future studies should build upon this
paradigm, and do so in a more trial-like manner. Several elements of this study
could be expanded in future studies. First, participants could be drawn from a
broader population. Our study participants were jury-eligible University stu-
dents, but they were not under legal obligation to act as jurors at the time of the
study. Next, in future studies, evidence could be presented in a more trial-like
manner. In our study, evidence was presented from only one perspective, rather
than from multiple perspectives. Furthermore, the amount of evidence that was
presented was limited. Participants read only a short description of the crime
and then evaluated twenty pieces of evidence from the trial.134 The amount of
evidence we presented is thus less substantial than the amount of evidence a
typical juror would hear.135 Finally, future studies could expand upon the prim-
ing task we used. Our study limited the priming task to the security camera
photo. It would be important to test how priming in various contexts (e.g. eth-
nicity of defendant, witnesses, victim, etc.) might implicate evidence evalua-
tion.136
C. Demographics of Study Participants
Study participants came from several different ethnic backgrounds.
Twenty-five participants were Japanese American, eighteen were European
American, and five were Chinese American. Other participants were Native
Hawaiian, Pacific Islander, Korean American and Latino.137 Nineteen of the
participants were male and forty-seven were female.138 The average participant
age was 21.85.139
134 Future studies might also consider using evidence collected from real trials.
Despite the diversity in the participant pool, there were no
135 It is possible that as evidence complexity and a juror’s cognitive load increases, the oppor-
tunity for bias may also increase. Some studies have suggested that increasing cognitive busyness
increases stereotype consistent cognitive errors. See Levinson, Forgotten Racial Equality, supra
note 1, at 374 (citing C. Neil Macrae et al., Creating Memory Illusions: Expectancy-Based
Processing and the Generation of False Memories, 10 MEMORY 63 (2002)).
136 In our study, participants saw a photograph of a perpetrator, and were subsequently in-
formed that a suspect was arrested and charged with the robbery. We did not ask participants
whether they believed that the defendant was in fact the perpetrator in the photo. Future studies
should seek to identify the exact nature of the prime and examine its relationship, if any, to the
defendant.
137 In addition, three participants identified themselves as multi-racial, and five participants
indicated “other.”
138 Differences between the male study participants and female study participants were not
significant, possibly due to the low number of male participants. Future studies should examine
whether there are significant gender differences in responses.
139 Participant’s ages ranged from eighteen to forty (SD=3.95). There are demographic advan-
tages to conducting the study in Hawai`i. Levinson has argued that researchers can “examine
African-American bias and stereotypes in a community with a historically small African-
American population (finding implicit biases against African Americans in Hawai`i might tend to
illustrate the power of implicit racial bias and its propagation through external sources such as
336 WEST VIRGINIA LAW REVIEW [Vol. 112
statistically significant differences based on the ethnicity of the participants.
For example, Asian American and Caucasian participants did not differ signifi-
cantly in the way they displayed evidence evaluation bias. This result indicates
that Biased Evidence Hypothesis, if confirmed, would not be limited to Cauca-
sian jurors.
The diversity of the participant pool is also notable because of Ha-
wai`i’s unique cultural community, where racial stereotypes exist in a complex
historical environment.140 The history of racial discrimination in Hawai`i is
different from many locations in the continental United States.141
D. Results — Skin Tone and Racially Biased Judgments
One might
expect that the biased evidence evaluations we predicted might be even stronger
in locations where there are more salient stereotypes of African Americans.
We tested our hypotheses using two separate statistical models, a
MANCOVA142 (a multivariate analysis of variance test) and a logistic regres-
sion.143 A MANCOVA was performed to investigate the effect of the experi-
mental condition (dark skin tone versus light skin tone of the perpetrator) on
total evidence judgments144
media).” In addition, researchers can “test biases and stereotypes among a diverse population,
which would indicate that implicit biases are manifested in the legal setting by a broader (not just
Caucasian) juror population.” Levinson, Forgotten Racial Equality, supra note
and on judgments of how guilty the participant per-
1, at 396.) One
drawback of our sample is that none of the participants were African American. (Although the
African American community in Hawai`i is small, there is a rich history of African Americans in
Hawai`i. See THEY FOLLOWED THE TRADE WINDS: AFRICAN AMERICANS IN HAWAI`I (Miles Jack-
son ed., 2004)). Thus, we cannot test whether African American participants would judge evi-
dence in the same stereotype-consistent way. However, research from social cognition sugges-
tions that such a result is entirely possible. For example, research has found that members of
some groups harbor implicit biases towards their own groups. See Mahzarin R. Banaji & Anthony
Greenwald, Implicit Gender Stereotyping in Judgments of Fame, 68 J. PERSONALITY & SOC.
PSYCHOL. 181, 194 (1995) (finding that both men and women harbored negative implicit gender
stereotypes of women); Brian Nosek et al., Pervasiveness and Correlates of Implicit Attitudes and
Stereotypes, 18 EUR. REV. SOC. PSYCHOL. 36 (2007) (indicating that participants over sixty years
old often have an implicit bias in favor of young over old).
140 Eric Yamamoto, The Significance of Local, in SOCIAL PROCESSES IN HAWAI`I 138, 138–49
(Peter Manicus ed., 1974).
141 See Levinson, Forgotten Racial Equality, supra note 1, at 395–96.
142 MANCOVA, a form of multivariate general linear modeling, is particularly appropriate in
this case because we predicted differences in multiple dependant outcome variables. This statis-
tical analysis allows the researcher to preserve power while performing multiple analyses. The
overall multivariate test (Wilk's Lambda) was significant for the skin tone of the photo show
(F=3.31, p<.043).
143 We controlled for sex and age. We implemented these control measures to allow us to see
the effect that skin tone priming had above and beyond any effects that may be due to age or sex.
144 In order to view racial priming’s effect on a trend of judging evidence as opposed to its
effect on individual pieces of evidence, we summed judgments of evidence.
2010] DIFFERENT SHADES OF BIAS 337
ceived the suspect to be. A logistic regression was performed to ascertain the
affect of these two variables on the dichotomous decision of guilty or not guilty.
1. Skin Tone Affects Judgments of Ambiguous Evidence
The perpetrator’s skin tone in the photo significantly affected evidence
judgments.145 Participants who saw the photo of the perpetrator with a dark skin
tone146 judged ambiguous evidence to be significantly more indicative of guilt
than participants who saw the photo of a perpetrator with a lighter skin tone.147
2. Skin Tone Affects Judgments of “How Guilty Is the Defendant”
The perpetrator’s skin tone also significantly affected judgments of how
guilty the defendant was (on a scale of 0–100). Participants who saw a darker-
skinned perpetrator judged the defendant as more guilty148 than participants who
saw a lighter-skinned perpetrator.149 This indicates that simply being primed
with darker skin tone not only affected the way participants judged evidence,
but also led them to perceive the defendant as more guilty.150
3. Evidence Judgments Predict Guilty Verdicts
The logistic regression analysis confirmed that participants’ evidence
judgments also predicted their guilty/not guilty verdicts.151 In addition, scaled
0–100 ratings of the defendant’s guilt statistically predicted dichotomous guilty/
not guilty verdicts.152
145 F=4.835, p=.032. The statistics we report in this subsection are multivariate tests of analy-
sis, F-test values, p values, and Beta coefficients. Multivariate tests, which are used in MANCO-
VAs and other analyses with more than one dependent variable, show the overall appropriateness
of conducting a multivariate analysis. F-test values consider the distribution of variance between
groups and within groups to help determine the source of variance. P values report the statistical
significance levels of the tests performed. In other words, P values report how likely it is that
results were found by “chance” as opposed to measuring an actual effect. The lower the p value
is, the more significant the result. P values of .05 or less are generally considered to be statistical-
ly significant. Beta (B) coefficients in logistic regression represent a parameter estimate that
predicts the logit odds of our dependant variable being “Guilty” or “Not Guilty”. In this case, a
higher beta score predicts higher logit odds of participants finding the defendant guilty.
146 Mean (M)= 86.23.
147 M= 80.49.
148 M=66.97 on a scale of 0–100.
149 M=56.37 on the same scale of 0–100.
150 F= 4.40, p= .034.
151 Beta (B)= -.101, p =.05.
152 B= -.152, p<.001.
338 WEST VIRGINIA LAW REVIEW [Vol. 112
4. Evidence Judgments Unrelated to Explicit Racial Preferences
We were particularly interested in whether the skin tone effects we
found were implicit in nature.153 In order to test whether explicit racial prefe-
rences were playing a role in evidence judgments, we calculated correlation
coefficients between the explicit measures of racial preferences (Modern Rac-
ism Scale and feeling thermometers) and the other dependent variables (total
evidence judgments, and how guilty 0–100 judgments). All correlations be-
tween explicit measures of bias and experimental measures were non-
significant.154 Furthermore, when we included them in regression models, none
of the explicit measures of prejudice emerged as significant predictors.155
5. Stimuli Recall, Priming, and IAT Results
In order to further determine whether the priming task had functioned
on an implicit level, we asked participants at the end of the study to recall the
race of the perpetrator in the security camera photo.156 The results for this ques-
tion, although somewhat difficult to interpret, tended to indicate that many par-
ticipants did not recall whether they had seen a perpetrator with the dark or ligh-
ter skin tone. Importantly, regardless of the accuracy of their memory, partici-
pant responses did not differ significantly based on the race of the perpetrator
that they reported seeing.157 This result suggests that the skin tone of the perpe-
trator was not being considered consciously as part of judgments based on the
skin tone in the photo. In addition, the results of the IATs we ran predicted evi-
dence judgments, strengthening the claim that the evidence judgments were
implicit in nature.158
E. Summary of Results and Implications
The study we conducted yielded several interesting results. First, we
found that simply showing participants a photo of a dark-skinned perpetrator
introduced racial bias into a crucial jury function — evaluating evidence. Next,
153 Unlike in some of the priming tasks we discussed infra Section III, participants saw the
photos for four seconds. Thus, the priming was not subliminal.
154 However, there was a marginally significant relationship between total evidence scores and
feeling thermometer scores (which were scored by subtracting feelings towards African Ameri-
cans from feelings towards European Americans). r=.226, p=.068.
155 p>.5.
156 For this question, we were unable to ask participants whether they saw a dark-skinned or
light-skinned suspect, particularly because skin tones are relative. We thus decided to ask partici-
pants to identify the race of the perpetrator in the photo, and then to consider their responses in
light of the skin tone of the perpetrator they saw.
157 All p’s >.05.
158 For more on these IATs, see Levinson et al., supra note 73.
2010] DIFFERENT SHADES OF BIAS 339
we found that these biased evidence judgments mattered; they predicted guilty
and not guilty verdicts. Finally, we found that the evidence-based racial cues
likely were implicit in nature and may have activated stereotypes even without
the participants’ awareness.
These results should raise concerns about jurors’ ability to meet out jus-
tice objectively. If simple racial cues have the ability to automatically incorpo-
rate racial stereotypes into evidence evaluation, particularly without the jurors’
awareness, the implications of this bias could be far reaching. First, proof of
unintentional racial bias in evidence evaluation would contradict legal assump-
tions that verdicts are determined based upon an objective weighing of the evi-
dence. Second, confirmation of implicit bias in evidence evaluation would pro-
vide empirical support for what many scholars have argued for years — the law
sometimes acts as a tool to subordinate already disadvantaged groups.159
In view of the need to consider implicit bias in decision-making in the
context of racial justice, the next section more broadly considers the results of
our study together with other proof of implicit racial bias in decision-making. It
presents a well-known model of jury decision-making and argues that implicit
bias has the potential to affect not only evidence evaluation, but also nearly
every key element of jury decision-making.
As
Biased Evidence Hypothesis is investigated further, the connection between
biased evidence evaluation and racial justice should also be explored.
V. DEVELOPING A NEW MODEL OF IMPLICIT RACIAL BIAS AND DECISION-
MAKING
Biased evidence evaluations may well lead to racial disparities in the
criminal justice system. Yet broader legal scholarship on unconscious racism,
as well as scientific evidence documenting the pervasiveness of implicit bias in
American society, suggests that biased evidence evaluations are unlikely to be
the sole point of infiltration of implicit bias in decision-making. This section
describes how implicit bias has the potential to wreak havoc not just on evi-
dence evaluation, but on each key element in the juror decision-making
process.160 This exploration is necessarily nascent, and as our brief examination
reveals, there are some areas where implicit bias in decision-making has been
examined only partially, and others where it has not been examined at all. Nev-
ertheless, considering several emerging projects on implicit bias,161
159 See generally DERRICK BELL, RACE, RACISM, AND AMERICAN LAW (4th ed. 2000).
there is
160 Because the decision-making model we rely upon in this section focuses mostly on the
decision-making of individual jurors, so will this section. As Lora Levett and her colleagues
summarize, “many studies have demonstrated that the best predictor of postdeliberation verdicts is
individual jurors’ predeliberation verdicts.” Lora M. Levett et al., The Psychology of Juror and
Juror Decision-Making, in PSYCHOLOGY AND LAW: AN EMPIRICAL PERSPECTIVE 365, 370 (Neil
Brewer & Kipling D. Williams eds. 2005).
340 WEST VIRGINIA LAW REVIEW [Vol. 112
enough evidence to propose that implicit bias may disrupt the entire decision-
making process.162
An acclaimed model of decision-making, Nancy Pennington and Reid
Hastie’s “Story Model” of decision-making,
163 provides proper context for the
consideration of how implicit racial bias may broadly affect decision-making.
This section explains how Biased Evidence Hypothesis and other emerging re-
search on implicit bias fit together with the stages of the Story Model.164
A. The Story Model of Decision-Making
It first
describes the three major components of the Story Model. It then explains how
scholarship on implicit racial bias shows that each Story Model component is
particularly susceptible to the introduction of implicit bias. Finally, it calls for
the pursuit of a research agenda that will lead towards a complete implicit bias
model of decision-making.
The Story Model proposes that jurors in criminal cases develop expla-
nation-based stories describing “‘what happened’ during events testified to at
the trial.”165
161 Other projects we have conducted on implicit bias in the law claim that these biases affect
jurors in multiple ways: the way they remember and misremember case facts, see Levinson, For-
gotten Racial Equality, supra note
According to Pennington and Hastie, the particular story a juror
1; the way they hold implicit associations between race and
criminal guilt, see Levinson et al., supra note 73; the way supposedly neutral laws and legal
processes can unwittingly prime racial biases in jurors, see Levinson, Suppressing Community
Values, supra note 50; Levinson, The Complicitous Mind, supra note 2; and, as demonstrated in
our study, the way that jurors perceive and weigh evidence. These studies are just the first step in
developing a complete and interdisciplinary model of the way racial stereotypes operate in crimi-
nal law decision-making.
162 Beginning to develop a model now, when research is still emerging, will eventually help to
consider whether it is possible to employ change processes or legal changes that will at least par-
tially combat the harm being done. We thus view our analysis as only the first step in developing
such a model.
163 Nancy Pennington & Reid Hastie, Practical Implications of Psychological Research on
Juror and Jury Decision Making, 16 PERSONALITY & SOC. PSYCHOL BULL. 90, 95 (1990); see
generally Nancy Pennington & Reid Hastie, A Cognitive Theory of Juror Decision Making: The
Story Model, 13 CARDOZO L. REV. 519 (1991); Nancy Pennington & Reid Hastie, Explaining the
Evidence: Tests of the Story Model for Juror Decision Making, 62 J. PERSONALITY. & SOC.
PSYCHOL. 189, 189–90 (1992) [hereinafter Explaining the Evidence].
164 It also considers other decision-making research that supplements the Story Model. See,
e.g., Dan Simon, A Third View of the Black Box: Cognitive Coherence in Legal Decision Making,
71 U. CHI. L. REV. 511 (2004). Although this section investigates specific projects on implicit
bias primarily in the context of the story model of decision-making, one might consider the possi-
bility that implicit bias operates as a heuristic, or cognitive shortcut, that affects decision-making.
See Jolls & Sunstein, supra note 27, at 7–10 (considering implicit bias as a heuristic). This ap-
proach is consistent with the “heuristic-systematic model” of decision-making, which proposes
that “people process information along a heuristic/systematic continuum.” Ryan J. Winter &
Edith Greene, Juror Decision-Making, in HANDBOOK OF APPLIED COGNITION 739, 744 (Francis
Durso ed., 2007).
165 Explaining the Evidence, supra note 163.
2010] DIFFERENT SHADES OF BIAS 341
constructs ultimately “determines the juror’s decision:”166 “story construction
enables comprehension and organization of the evidence so that evidence can be
meaningfully evaluated against multiple verdict judgment dimensions.”167 The
stories that jurors construct are dependent upon three cognitive processing com-
ponents that explain the way jurors interpret information.168 These components,
and the Story Model generally, are developed and implemented both through “a
mixture of conscious and nonconscious processes.”169
The first component of the Story Model, evidence evaluation through
story construction, explains perhaps the most important piece of how jurors
make decisions — the way in which they first assemble knowledge into a story
form.
As the following analysis
illustrates, they are thus particularly susceptible to implicit racial biases.
170 Pennington and Hastie propose that three types of knowledge combine
to assemble a juror’s knowledge into story form: case-specific information
about the trial, knowledge about similar crimes, and the juror’s expectations
about what makes a complete story.171 Discussing these components, Penning-
ton and Hastie summarize: “some of these inferences may be suggested by the
attorney and some may be constructed solely by the juror. Whatever their
source, the inferences will serve to fill out the episode structure of the story.”172
The second component, representation of the decision alternatives by
learning verdict category attributes, describes how jurors come to understand
their decision-making options.
As the next subsection will discuss, considering the relationship between these
three types of knowledge and racial stereotypes, it is not difficult to imagine
how implicit biases may function to influence the inferences that jurors make,
therefore altering their story construction in racially biased ways.
173 According to Pennington and Hastie, jurors
learn their options both through the judge’s instructions on the law and through
their “prior ideas regarding the meaning of the verdict categories.”174
166 Id. at 190.
As we will
discuss, if the judge’s instructions unintentionally introduce implicit bias, or if
jurors prior ideas about verdict categories are already biased, or both, this
167 Id. at 192.
168 Id.
169 Reid Hastie, Conscious and Nonconscious Cognitive Processes in Jurors’ Decision, in
BETTER THAN CONSCIOUS? DECISION-MAKING, THE HUMAN MIND, AND IMPLICATIONS FOR
INSTITUTIONS 371, 384 (Christoph Engel & Wolf Singer eds., 2008).
170 Explaining the Evidence, supra note 163, at 190–91.
171 Id. at 190. Although this section does not address jurors’ expectations about what makes a
complete story, it is also quite plausible that jurors’ story expectations are affected by implicit
bias.
172 Id. at 191.
173 Id. at 191. See also Nancy Pennington & Reid Hastie, The Story Model for Juror Decision-
Making, in INSIDE THE JUROR: THE PSYCHOLOGY OF JUROR DECISION-MAKING 192, 200 (Reid
Hastie ed., 1993) [hereinafter The Story Model].
174 Explaining the Evidence, supra note 163, at 191.
342 WEST VIRGINIA LAW REVIEW [Vol. 112
second component of the Story Model illustrates yet another way that implicit
bias can affect decision-making.
The third component of the Story Model describes how jurors reach a
decision through the classification of their constructed story into the best-fitting
verdict category.175 This component thus describes how the first two compo-
nents come together. In addition, this component also incorporates legal prin-
ciples such as the presumption of innocence.176
B. Memory Errors, Biased Evidence, Implicit Associations, and More
If the first two components of
the Story Model have been tainted by implicit bias, the third component will
therefore automatically become infected. Similarly, because during this stage
jurors apply the presumption of innocence in determining how their already
constructed stories fit into the verdict categories, if implicit bias distorts the pre-
sumption of innocence, it can further bias the jurors’ decisions.
This subsection begins an amplification of the Story Model in light of
evidence on implicit bias. It considers the three components of the Story Model,
in the same order as described by Pennington and Hastie.
Biased Evidence Hypothesis and biased memory errors each have the
potential to incorporate implicit bias into the first component of the Story Mod-
el.177
Implicit memory bias demonstrates another way that story construction
can become tainted with implicit bias. In a project previously described,
As the empirical study described in Section IV demonstrated, biased evi-
dence evaluations were found to predict verdicts. Because Biased Evidence
Hypothesis fits so squarely into the first stage of the Story Model, “evidence
evaluation through story construction,” we therefore turn to other evidence that
implicit bias can taint the first component of the Story Model.
178 Le-
vinson claimed that jurors misremember case facts in racially biased ways.
Drawing upon empirical studies to support his argument, Levinson explained
that social cognition “research in memory and decision-making indicates that
memory errors are normal, occur in stereotype-driven ways, and can manifest in
a variety of ways that ultimately affect legal decision-making.”179
175 Id.
After ex-
176 Id. See also The Story Model, supra note 173, at 201.
177 In addition to Biased Evidence Hypothesis, the first component of the Story Model may also
allow for implicit bias introduction in another way. According to this component in the Story
Model, jurors also rely upon knowledge of similar crimes to help them construct their stories.
Thus, if jurors’ knowledge of similar crimes is tainted by racial stereotypes, these stereotypes will
again become incorporated into the Story Model. A study by Skorinko and Spellman found that
people hold clear stereotypes that associate certain ethnic groups with specific crimes. See infra
note 191 and accompanying text.
178 See Levinson, Forgotten Racial Equality, supra note 1. See supra notes 50–55 and accom-
panying text.
179 Id. at 374.
2010] DIFFERENT SHADES OF BIAS 343
amining how these principles apply to the legal decision-making context, Levin-
son then conducted an empirical study that supported his hypothesis.180 The
results of the study showed that study participants were more likely to remem-
ber aggressive story facts when they read about an African American actor
compared to a Caucasian actor.181 For some facts, participants were even more
likely to possess false memories, such that they erroneously believed that the
African American actor had taken aggressive actions when he had not.182
Levinson’s hypotheses and the social cognition studies underlying it fit
directly into the Story Model’s first component, and specifically into the sub-
component describing how jurors heavily rely upon case-specific information
about the trial. Simply stated, if jurors’ memories of case facts are affected by
racial bias, then the stories they construct based on those memories will be bi-
ased. Story Model research by Pennington and Hastie confirms that the Story
Model is “memory-dependent.”
183 In a study of memory’s role in decision-
making, Pennington and Hastie presented mock-jurors with information from a
murder trial.184 The mock jurors were later quizzed in order to test what infor-
mation they remembered.185 The results of the study showed that mock-jurors
were more likely to remember trial evidence when that evidence supported their
verdict.186
The second component of the Story Model is also susceptible to implicit
bias, although there are fewer empirical studies in this domain. We first consid-
er a judge’s instructions to jurors on the law, and next discuss the role of jurors’
“prior ideas regarding the meaning of verdict categories.” In considering an
amplification of this component, there are two main points to consider: first,
judges’ instructions on the law may prime racial stereotypes, and second, jurors
“prior ideas regarding the meaning of verdict categories” may incorporate
crime-specific racial stereotypes into decision-making. Although more targeted
research is needed in this domain, we briefly explain how these points might
each introduce implicit bias into decision-making.
The results thus demonstrated that memory is a critical function in
decision-making. Combining Levinson’s study with the Story Model indicates
that racially biased memories of case facts likely predict biased verdicts.
180 Id. at 390–406.
181 Id. at 398–401.
182 Id. at 400. The results also showed that “susceptibility to misremembering facts based on
race cannot be attributed simply to more overtly racist people — those who were susceptible to
racial misremembering sometimes embraced less explicitly racist attitudes.” Id. at 350.
183 Id. at 193–95. See also Nancy Pennington & Reid Hastie, Explanation-Based Decision
Making: Effects of Memory Structure on Judgment, 14 J. EXPERIMENTAL PSYCHOL.: LEARNING,
MEMORY, AND COGNITION 521 (1988) [hereinafter Memory Structure].
184 Levinson, Forgotten Racial Equality, supra note 1, at 197–98 (citing David A. Schum &
Anne W. Martin, Formal and Empirical Research on Cascaded Inference in Jurisprudence, 17
LAW & SOC’Y REV. 105 (1982)); Memory Structure, supra note 183, at 525–26.
185 Memory Structure, supra note 183, at 525–26.
186 Id. at 526–27.
344 WEST VIRGINIA LAW REVIEW [Vol. 112
First, it is entirely plausible that judges’ instructions on the law might
prime jurors’ racial biases.187 As we have described,188 people are extremely
susceptible to automatic stereotype priming by racial cues. Although research-
ers have yet to examine specifically whether a judge’s instructions can prime
racial bias, Levinson has suggested two ways that legal processes may change
jurors’ cognitive processes in undesirable ways. First, jury duty itself can trig-
ger implicit knowledge structures, perhaps even those including racial stereo-
types that carry forward historical inequality in the legal system.189 And second,
certain supposedly race-neutral legal processes, such as “death qualification” in
a capital trial, may unintentionally trigger juror racial biases.190
Second, it is also possible that jurors’ “prior ideas about verdict catego-
ries” are racially biased, particularly in cases with alternative verdict categories.
If jurors hold stronger implicit associations between members of stereotyped
groups and one particular verdict category (such as intentional murder or drug
dealing) relative to another (such as reckless homicide or drug possession), then
implicit bias has the potential even to affect the way jurors interpret the verdict
categories. Although researchers have not yet looked at this specific issue,
some studies have found that jurors hold strong race-specific stereotypes related
to certain crimes. For example, Jeanine Skorinko and Bobbie Spellman found
that study participants associated intentional murder with African Americans.
It would not be
particularly surprising, then, if certain elements of judges’ instructions to juries
unintentionally triggered racial stereotypes in jurors. Future research should
explore this possibility.
191
Finally, the third component of the Story Model is also susceptible to
implicit bias. As previously described,
If jurors in a homicide case are given instructions for intentional murder and
reckless homicide (or jurors in a narcotics case are given instructions for drug
dealing and drug possession), for example, then previous race-relevant stereo-
types held about these crimes might affect the way jurors make decisions.
192 Levinson, Cai and Young investigated
implicit bias in the presumption of innocence by designing and running a
Guilty/Not Guilty IAT.193
187 Researchers have yet to investigate whether supposedly race-neutral jury instructions might
also prime racial stereotypes, but such a possibility should be investigated. If it were confirmed, it
would fit specifically into the second component of the Story Model.
The results of the study showed that participants as-
188 See supra notes 89, 92–120 and accompanying text.
189 See Levinson, Suppressing Community Values, supra note 50.
190 See Levinson, The Complicitous Mind, supra note 2.
191 Jeanine L. Skorinko & Barbara A. Spellman, Stereotypic Crimes: How Group-Crime Asso-
ciations Affect Memory and (Sometimes) Verdicts and Sentencing Appx. C (June 2006) (unpub-
lished manuscript, on file with the author). Some crimes, including the non-violent crimes of
identity fraud and ecstasy possession, are stereotypic White crimes.
192 See supra notes 73–76 and accompanying text.
193 See generally Levinson et al., supra note 73.
2010] DIFFERENT SHADES OF BIAS 345
sociated Black and Guilty, compared to White and Guilty.194 In light of the
results, the authors questioned whether the presumption of innocence is truly a
race-neutral concept.195
The results of the Guilty/Not Guilty IAT demonstrate another way that
implicit bias affects decision-making. The third component of the Story Model
states that jurors consider the first two components in light of the presumption
of innocence and the burden of proof. Jurors thus consider the presumption of
innocence when determining whether their story fits the decision alternatives. If
jurors hold implicit biases manifested as cognitive associations between Black
and Guilty, a finding that calls into question the racial equality underlying the
presumption of innocence, then this third Story Model component likely allows
the introduction of a new element of implicit bias into decision-making.
Considered together, it is not difficult to see that nearly all of the impor-
tant elements in jury decision-making have the potential to become tainted by
implicit bias. Future research, particularly empirical endeavors, should continue
to examine this possibility and continue to build an implicit bias model of deci-
sion-making.
VI. CONCLUSION: SETTING A RESEARCH AGENDA
Because only a few empirical studies have investigated implicit bias in
the legal context,196 future research must continue to investigate the ways in
which implicit bias leads to racial injustice in the legal system. Some of these
endeavors should examine implicit bias in legal decision-making, an area which,
as we have described, is still tremendously ripe for discovery. Others should
explore implicit bias in non-decision-making contexts. For example, implicit
bias may exist in the administration of a wide range of laws, including property
law, contract law, immigration law, trusts and estates law, and more.197
194 Id. at 16–17. In addition, IAT results predicted the way the participants evaluated ambi-
guous trial evidence (using the evidence measure discussed in this Article). Id. at 18–19.
Pur-
suing such a broad research agenda will help in seeking to understand continu-
ing racial inequality in American society — racial inequality that the legal sys-
tem has yet to resolve. It is our prediction that, as more research is conducted,
these projects will portray implicit bias as an ever-present hidden danger in our
legal system.
195 Courts and many scholars have assumed that the presumption of innocence is a bedrock of
fairness. See Coffin v. United States, 156 U.S. 432, 453 (1895); Rinat Kitai, Presuming Inno-
cence, 55 OKLA. L. REV. 257, 260–61 (2002)); Levinson et al., supra note 73, at 13–14 (citing
Estelle v. Williams, 425 U.S. 501, 503 (1976) (stating that “[t]he presumption of innocence . . . is
a basic component of a fair trial under our system of criminal justice).
196 See Section II.A.3 for a review of these studies.
197 See, e.g., Kang, supra note 1, at 1536–37 (listing a range of areas that should be investi-
gated).
346 WEST VIRGINIA LAW REVIEW [Vol. 112
Future projects should also test ways of lessening the harms of implicit
bias. Although little progress has been made in understanding how to reduce
implicit biases across society, social scientists have found that several types of
interventions have at least temporarily reduced the harmful effects of implicit
bias in limited settings.198 For example, Jennifer Richeson and Richard Nuss-
baum found that multiculturalism training temporarily reduced implicit racial
biases.199 And Nilanjana Dasgupta and Anthony Greenwald found that briefly
exposing participants to photos of “famous and admired” African American
exemplars, such as Denzel Washington, temporarily decreased implicit biases
against African Americans.200 None of these debiasing or bias-lessening stu-
dies, however, have been conducted in the context of law or a courtroom. Fu-
ture projects should undertake this important endeavor. In the case of Biased
Evidence Hypothesis, future research should evaluate whether any interventions
might at least help to dampen the harmful effects of an evidence evaluation bias.
Some possibilities include: training jurors on multiculturalism, confronting
jurors with their biases,201 or drafting new jury instructions designed to reduce
the chances of priming stereotypes.202
Before charging into this research agenda, however, a bit of caution is in
order regarding the bias reduction strategies we have suggested. Considered in
societal context, narrowly focused bias-reduction strategies represent an inade-
quate and only temporary response to a culturally based problem.
203
198 For explanations of some of these studies, see, e.g., Kang, supra note
The
1, at 1494; Kang &
Banaji, supra note 2, at 1065; Levinson, Forgotten Racial Equality, supra note 1, at 411; Page,
supra note 26, at 160.
199 Jennifer A. Richeson & Richard J. Nussbaum, The Impact of Multiculturalism Versus Col-
or-Blindness on Racial Bias, 40 J. EXPERIMENTAL SOC. PSYCHOL. 417, 420 (2004). The research-
ers were comparing training on multiculturalism versus training on “color-blindness.” See also
Laurie A. Rudman et al., Unlearning Automatic Biases: The Malleability of Implicit Prejudice and
Stereotypes, 81 J. PERSONALITY & SOC. PSYCHOL. 856, 865 (2001).
200 Nilanjana Dasgupta & Anthony G. Greenwald, On the Malleability of Automatic Attitudes:
Combating Automatic Prejudice with Images of Admired and Disliked Individuals, 81 J.
PERSONALITY & SOC. PSYCHOL. 800, 803 (2001).
201 Research has indicated that at least some, but not all, ways of confronting implicit biases
have the potential to reduce the power of harmful stereotypes. See Alexander M. Czopp et al.,
Standing up for a Change: Reducing Bias Through Interpersonal Confrontation, 90 J.
PERSONALITY & SOC. PSYCHOL. 784, 799 (2006). For a caution on using such an intervention with
jurors, see Levinson, Forgotten Racial Equality, supra note 1, at 414 n.322.
202 Longer-term efforts might include increasing diversity in the ranks of judges and attorneys.
Studies have shown that exposing people to counterstereotypic exemplars can reduce biases. See
Nilanjana Dasgupta & Shaki Asgari, Seeing Is Believing: Exposure to Counterstereotypic Women
Leaders and Its Effect on the Malleability of Automatic Gender Stereotyping, 40 J. EXPERIMENTAL
SOC. PSYCHOL. 642, 645 (2004). See also Kang & Banaji, supra note 2 (suggesting the hiring of
counterstereotypic exemplars in a variety of fields).
203 See Levinson, Forgotten Racial Equality, supra note 1, at 417–20 (reviewing debiasing
techniques as a second-best alternative to cultural change). See also Ralph R. Banks & Richard T.
Ford, (How) Does Unconscious Bias Matter?: Law, Politics, and Racial Inequality, 58 EMORY
L.J. 1053 (2009) (claiming that the focus should be on the pervasiveness of discrimination, not on
2010] DIFFERENT SHADES OF BIAS 347
second-best strategies that we have proposed, such as the development of de-
biasing instructions for jurors, may temporarily lessen jurors’ biases for a few
crucial hours, but are no long term match for the powerful and pervasive harms
of implicit racial bias. Nor can debiasing techniques dismantle the continuing
racial inequality present in America. Unfortunately, however, for now only bias
reduction strategies exist. Bias elimination strategies do not exist because, short
of cultural change and the elimination of all forms of racial inequality, these
societal-based biases will remain. 204
The investigation of Biased Evidence Hypothesis has revealed yet
another deeply concerning way that implicit bias threatens racial justice and
legal fairness. If jurors are unable to evaluate trial evidence without the intru-
sion of harmful racial stereotypes, racial justice in the legal system and beyond
is surely not at hand.
We therefore propose bias reduction strat-
egies with the understanding that the long-term goal must include taking cultural
responsibility for racial inequality and implicit bias.
whether the bias is conscious, unconscious or covert); Unconscious Racism Revisited, supra note
15 (questioning whether some legal scholarship on implicit biases might unintentionally mask the
“societal disease” underlying continuing racism and “collective responsibility” for eliminating it).
204 See generally Unconscious Racism Revisited, supra note 15.
348 WEST VIRGINIA LAW REVIEW [Vol. 112
APPENDIX A
Items of Evidence Evaluated by Participants
The defendant purchased an untraceable handgun three weeks
before the robbery.
The store owner identified the defendant’s voice in an audio
line-up.
A week after the robbery, the defendant purchased jewelry for
his girlfriend.
The defendant’s brother is in jail for trafficking narcotics.
The defendant recently lost his job.
The defendant used to be addicted to drugs.
The defendant has been served with a notice of eviction from
his apartment.
The defendant was videotaped shopping at the same Mini Mart
two days before the robbery.
The defendant frequently shops at a variety of Mini Mart stores.
The defendant used to work at this particular Mini Mart.
The defendant is left handed.
The defendant was a youth Golden Gloves boxing champ in
2006.
The defendant belongs to a local gun club called Safety Shot:
The Responsible Firing Range.
The defendant had a used movie ticket stub for a show that
started 20 minutes before the crime occurred.
The defendant wore a plaster cast on his broken right arm
around the time of the robbery.
The defendant is a member of an anti-violence organization.
2010] DIFFERENT SHADES OF BIAS 349
The defendant’s fingerprints were not found at the scene of the
crime.
The defendant does not have a driver’s license or car.
The defendant has no prior convictions.
The defendant graduated high school with good grades.
350 WEST VIRGINIA LAW REVIEW [Vol. 112
APPENDIX B
Photo from Security Camera — Lighter Skin Tone
Photo from Security Camera — Darker Skin Tone