VOLUME 122

JANUARY 2009

NUMBER 3

© 2009 by The Harvard Law Review Association

ARTICLE
WHOSE EYES ARE YOU GOING TO BELIEVE? SCOTT V.
HARRIS AND THE PERILS OF COGNITIVE ILLIBERALISM
Dan M. Kahan, David A. Hoffman, and Donald Braman
TABLE OF CONTENTS
INTRODUCTION ............................................................................................................................838
I.

IT’S OBVIOUS! .......................................................................................................................843

II.

TAKING THE SCOTT CHALLENGE....................................................................................848
A. Four Members of the American Venire ..........................................................................849
B. Theoretical Background: Motivated Cognition of Legally Consequential Facts.......851
C. Study Design and Hypotheses.........................................................................................854
1. Sample ..........................................................................................................................854
2. Stimulus ........................................................................................................................854
3. Response Measures .....................................................................................................856
4. Individual Characteristics ..........................................................................................859
(a) Demographic Characteristics ...............................................................................859
(b) Political Ideology and Party Affiliation ..............................................................859
(c) Cultural Worldviews .............................................................................................859
5. Hypotheses ...................................................................................................................860
D. Results ................................................................................................................................864
1. Preliminary Analyses: Main Effects and Group Differences. ...............................864
2. Clarifying the Data: Statistical Simulations ............................................................870
(a) Overview.................................................................................................................870
(b) Ron, Linda, Bernie, and Pat ................................................................................872
E. Summary and Discussion ................................................................................................879

III. EVALUATING SCOTT .............................................................................................................881
A. Dissensus, Deliberation, and Legitimacy.......................................................................881
B. How To Defend the Outcome in Scott: Reasons, Not Perceptions .............................887
C. Cognitive Illiberalism and Judicial Humility ...............................................................894
CONCLUSION .................................................................................................................................903

837

WHOSE EYES ARE YOU GOING TO BELIEVE? SCOTT V.
HARRIS AND THE PERILS OF COGNITIVE ILLIBERALISM
Dan M. Kahan,∗ David A. Hoffman,∗∗ and Donald Braman∗∗∗
This Article accepts the unusual invitation to “see for yourself” issued by the Supreme
Court in Scott v. Harris, 127 S. Ct. 1769 (2007). Scott held that a police officer did not
violate the Fourth Amendment when he deliberately rammed his car into that of a
fleeing motorist who refused to pull over for speeding and instead sought to evade the
police in a high-speed chase. The majority did not attempt to rebut the arguments of the
single Justice who disagreed with its conclusion that “no reasonable juror” could find
that the fleeing driver did not pose a deadly risk to the public. Instead, the Court
uploaded to its website a video of the chase, filmed from inside the pursuing police
cruisers, and invited members of the public to make up their own minds after viewing it.
We showed the video to a diverse sample of 1350 Americans. Overall, a majority agreed
with the Court’s resolution of the key issues, but within the sample there were sharp
differences of opinion along cultural, ideological, and other lines. We attribute these
divisions to the psychological disposition of individuals to resolve disputed facts in a
manner supportive of their group identities. The Article also addresses the normative
significance of these findings. The result in the case, we argue, might be defensible, but
the Court’s reasoning was not. Its insistence that there was only one “reasonable” view
of the facts itself reflected a form of bias — cognitive illiberalism — that consists in the
failure to recognize the connection between perceptions of societal risk and contested
visions of the ideal society. When courts fail to take steps to counteract that bias, they
needlessly invest the law with culturally partisan overtones that detract from the law’s
legitimacy.

INTRODUCTION

F

or Craig Jones, it had to be one of those sinking moments when a
lawyer arguing a case before the Supreme Court realizes, with
stomach-turning certitude, that all is lost. Not only were the Justices
broadcasting in their questions at oral argument that they were disposed to rule against him. They were making clear that their inclinations rested on a foundation that simply does not admit of counterargument: brute sense impressions.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
∗
∗∗
∗∗∗

Elizabeth K. Dollard Professor of Law, Yale Law School.
Associate Professor of Law, Temple University Beasley School of Law.
Associate Professor of Law, The George Washington University Law School.
Research for this paper was generously supported by the National Science Foundation, Grant
SES 0621840; the Oscar M. Ruebhausen Fund at Yale Law School; and the Clifford Scott Green
Research Fund in Law at Temple Law School. Thanks to Bruce Ackerman, Harry Edwards, Ted
Eisenberg, Linda Elliott, Neal Feigenson, Brandon Garrett, Craig Green, Linda Greenhouse,
Christopher Lasch, Dan Markel, Tracey Meares, Andrew Pincus, Richard Posner, Jeff Rachlinski,
Giovanna Shay, Chris Slobogin, and Paul Slovic for comments; and to Kenworthey Bilz for advice on the study design.

838

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

839

Jones was representing Victor Harris, a motorist who had been rendered a quadriplegic when police officer Timothy Scott deliberately
rammed Harris’s vehicle at the end of a high-speed chase, causing it to
flip over an embankment and crash.1 Jones’s task was to defend the
judgment of the Eleventh Circuit Court of Appeals, which had ruled
that Scott was not entitled to summary judgment in Harris’s suit for
violation of his Fourth Amendment rights.2 But within seconds of beginning to explain why there was a “genuine issue” of fact3 on whether
Harris’s flight posed a danger to the public sufficient to justify use of
deadly force against him, Jones was confronted with what Justice after
Justice cited as dispositive evidence to the contrary: a videotape of the
chase filmed from inside police cruisers.
“Mr. Jones,” Justice Alito started, “I looked at the videotape on this.
It seemed to me that [Harris] created a tremendous risk [to] drivers on
that road.”4 “He created the scariest chase I ever saw since ‘The
French Connection,’” Justice Scalia immediately chimed in, provoking
laughter throughout the courtroom.5
Probably even more dispiriting was the exchange that came next
with Justice Breyer, whose vote Jones had likely been counting on for
affirmance. “I was with you when I read . . . the opinion of the court
below,” Justice Breyer related.6 “Then I look at that tape, and I have
to say that when I looked at the tape, my reaction was somewhat similar to Justice Alito’s.”7 As Jones attempted to offer a less damaging interpretation of the tape’s contents, reinforced by an invocation of the
jury’s prerogatives as factfinder, Justice Breyer sharply retorted:
JUSTICE BREYER: . . . What am I supposed to assume? . . . I mean,
I looked at the tape and that tape shows he is weaving on both sides of the
lane, swerving around automobiles that are coming in the opposite direction with their lights on, goes through a red light where there are several
cars that are right there, weaves around them, and there are cars coming
the other way, weaves back, goes down the road.
. . . [A]m I supposed to pretend I haven’t seen that? . . .
MR. JONES: Well, I didn’t see that.
JUSTICE BREYER: You didn’t see that?

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
1
2

See Scott v. Harris, 127 S. Ct. 1769, 1773 (2007).
Harris v. Coweta County, 433 F.3d 807 (11th Cir. 2005), rev’d sub nom. Scott, 127 S. Ct.

1769.
3 See FED. R. CIV. P. 56(c) (summary judgment appropriate when “the pleadings, the discovery and disclosure materials on file, and any affidavits show that there is no genuine issue as to
any material fact and that the movant is entitled to judgment as a matter of law”).
4 Transcript of Oral Argument at 27, Scott, 127 S. Ct. 1769 (No. 05-1631).
5 Id. at 28.
6 Id. at 30–31.
7 Id. at 31.

840

HARVARD LAW REVIEW

[Vol. 122:837

....
MR. JONES: But those are not the facts that were found by the court
below in this —
JUSTICE BREYER: Well that’s, that’s what I wonder. If the court
says that isn’t what happened, and I see with my eyes that is what happened, what am I supposed to do?8

Later, during the rebuttal argument of opposing counsel (at which
point a helpless Jones was reduced to watching mute), Justice Breyer
continued, “I look at the tape and I end up with Chico Marx’s old
question with respect to the Court of Appeals: Who do you believe, me
or your own eyes?”9
When the opinion was handed down some ten weeks later, a majority of Justices had indeed decided to “call it” as they “saw it.” Acknowledging that although normally “courts are required to view the
facts and draw reasonable inferences ‘in the light most favorable to the
party opposing the [summary judgment] motion,’”10 Justice Scalia,
writing for an eight-Justice majority, stated, “[t]here is . . . an added
wrinkle in this case: existence in the record of a videotape capturing
the events in question.”11 “When opposing parties tell two different
stories, one of which is blatantly contradicted by the record, so that no
reasonable jury could believe it, a court should not adopt that version
of the facts for purposes of ruling on a motion for summary judgment,”12 Justice Scalia reasoned. “Respondent’s version of events is so
utterly discredited by the record that no reasonable jury could have
believed him.”13
Somewhat inconveniently for the majority, however, one Justice
who had watched the tape had in fact taken Harris’s view. “[T]he
tape actually confirms, rather than contradicts, the lower courts’ appraisal of the factual questions at issue,”14 Justice Stevens announced
in dissent.
But having grounded its decision in its senses, the majority saw no
need to resort to reasoned elaboration of its position to rebut Stevens’s
apprehension of facts “no reasonable juror” could have seen. Instead,
creating the Court’s first (and so far only) multimedia cyber-opinion, it
supplied a URL for a digital rendering of the tape that had been up–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
8
9
10

Id. at 44–45.
Id. at 54.
Scott, 127 S. Ct. at 1774 (alteration in original) (quoting United States v. Diebold, Inc., 369
U.S. 654, 655 (1962)).
11 Id. at 1775.
12 Id. at 1776.
13 Id.
14 Id. at 1781 (Stevens, J., dissenting).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

841

loaded to the Court’s website. “We are happy,” Scalia wrote, “to allow
the videotape to speak for itself.”15
Well, does the Scott v. Harris16 tape speak for itself? If so, what
does it say?
We decided to conduct an empirical study to answer these questions. We showed the video to a diverse sample of approximately 1350
Americans. We then asked them to tell us what they saw, and give us
their views on the issues that the Court had identified as dispositive.
Our subjects didn’t see things eye to eye. A fairly substantial majority did interpret the facts the way the Court did. But members of
various subcommunities did not. African Americans, low-income
workers, and residents of the Northeast, for example, tended to form
more pro-plaintiff views of the facts than did the Court. So did individuals who characterized themselves as liberals and Democrats.
Individuals with these characteristics tend to share a cultural orientation that prizes egalitarianism and social solidarity. Various highly
salient, “symbolic” political issues — from gun control to affirmative
action, from the death penalty to environmental protection — feature
conflict between persons who share this recognizable cultural profile
and those who hold an opposing one that features hierarchical and individualistic values.17 We found that persons who subscribed to the
former style tended to perceive less danger in Harris’s flight, to attribute more responsibility to the police for creating the risk for the public,
and to find less justification in the use of deadly force to end the chase.
Indeed, these individuals were much more likely to see the police,
rather than Harris, as the source of the danger posed by the flight and
to find the deliberate ramming of Harris’s vehicle unnecessary to avert
risk to the public.
Thus, the question posed by the data is not, as Justice Breyer
asked, whether to believe one’s eyes, but rather whose eyes the law
should believe when identifiable groups of citizens form competing factual perceptions. That’s a normative question, which this Article will
also try to answer.
We will argue that the Court in Scott was wrong to privilege its
own view. Although an admitted minority of American society, citizens disposed to see the facts differently from the Scott majority share
a perspective founded on common experiences and values. By insisting that a case like Scott be decided summarily, the Court not only denied those citizens an opportunity, in the context of jury deliberations,
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
15
16
17

Id. at 1775 n.5 (majority opinion).
127 S. Ct. 1769.
See generally JOSEPH R. GUSFIELD, SYMBOLIC CRUSADE: STATUS POLITICS AND THE
AMERICAN TEMPERANCE MOVEMENT (2d ed. 1986) (describing historical and contemporary
disputes between cultural groups).

842

HARVARD LAW REVIEW

[Vol. 122:837

to inform and possibly change the view of citizens endowed with a different perspective. It also needlessly bound the result in the case to a
process of decisionmaking that deprived the decision of any prospect
of legitimacy in the eyes of that subcommunity whose members saw
the facts differently. Told that the law must be made in a fashion that
rigorously excludes their understanding — which the opinion in Scott
stigmatizes as being one only “unreasonable” people could hold —
those who disagree with the outcome cannot divorce their assent to it
from acceptance of their status as defeated outsiders. We won’t attempt to articulate a comprehensive theory of the function of the jury
in promoting the democratic legitimacy of law. But we will argue that
no normatively credible theory of this sort can be reconciled with the
Scott Court’s inattention to the shared identity of citizens likely to
form a different understanding of the risks depicted in the videotape.
This is not to say that we believe, necessarily, that there is no
credible normative defense of the result in Scott. One might take the
position, for example, that Scott and cases like it should be decided
summarily not because the understanding of social reality associated
with a particular cultural subcommunity is inherently “unreasonable,”
but because allowing the law to credit it could result in inconsistent
jury verdicts across jurisdictions or within jurisdictions over time. Alternatively, one might take the position that courts should not allow
the law to credit a subcommunity’s view of the facts because doing so
could result in bad practical consequences for law enforcement. Or
one might argue that cases like Scott should be dismissed because the
appropriateness of high-speed police chases should not be determined
by courts, but by democratically accountable legislative bodies. We
aren’t fully convinced by these arguments. But we are certain that
any of these rationales would have been superior to the one the Court
offered, because none would have incurred the cost to democratic legitimacy associated with labeling the perspective of persons who share
a particular cultural identity “unreasonable” and hence unworthy of
consideration in the adjudicatory process.
In addition to explaining how the Court got it wrong in Scott, we
will also venture a diagnosis of why it did. The Court’s mistake, we’ll
argue, reflects a type of decisionmaking hubris that has cognitive origins and that has deleterious consequences that extend far beyond the
Court’s decision in Scott. Social psychology teaches us that our perceptions of fact are pervasively shaped by our commitments to shared
but contested views of individual virtue and social justice. It also tells
us that although our ability to perceive this type of value-motivated
cognition in others is quite acute, our power to perceive it in ourselves

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

843

tends to be quite poor.18 We thus simultaneously experience overconfidence in the unassailable correctness of the factual perceptions we
hold in common with our confederates and unwarranted contempt for
the perceptions associated with our opposites. When these dispositions
become integrated into law — as we believe they did in Scott — they
generate needless cultural and political conflict that enervates the
law’s political legitimacy.19 Judges, legislators, and ordinary citizens
should therefore always be alert to the influence of this species of
“cognitive illiberalism” and take the precautions necessary to minimize
it.
This Article will present and elaborate on these various claims over
the course of three Parts. Part I will look more closely at the issue and
opinions in Scott, including the Court’s provocative invitation to readers to view the tape and decide for themselves whether the majority or
dissent got it right. Part II will describe the theoretical underpinnings,
the design, and the results of the empirical study we conducted in response to the Court’s invitation. Part III will then set forth our normative analysis of what the study results say about the correctness of
the Scott decision, about the general impact of cognitive illiberalism in
adjudication, and about the steps judges should take to try to counteract it.
I. IT’S OBVIOUS!
We consider first the story of Scott v. Harris. It begins with a relatively familiar challenge — “catch me if you can” — on the roads of
Georgia and ends with a very unusual one — “see for yourself” — in
the pages of a Supreme Court opinion.
Just before 11:00 p.m. on March 29, 2001, on a two-lane highway
in the Atlanta suburbs, the police detected Victor Harris speeding.20
But when the officers attempted to make a traffic stop, Harris hit the
gas pedal, fleeing at high speed.21 Soon a car driven by Officer Timothy Scott joined the chase. Knowing little of the inciting situation,
Scott had decided on his own initiative to help apprehend Harris.22
Following a slow-speed interlude that included a side swiping in an
empty shopping mall parking lot, the chase returned to the road,
reaching speeds in excess of eighty-five miles per hour.23 The pursuit
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
18 See Robert J. Robinson, Dacher Keltner, Andrew Ward & Lee Ross, Actual Versus Assumed
Differences in Construal: “Naive Realism” in Intergroup Perception and Conflict, 68 J. PERSONALITY & SOC. PSYCHOL. 404, 414–16 (1995).
19 See Dan M. Kahan, The Cognitively Illiberal State, 60 STAN. L. REV. 115 (2007).
20 Harris v. Coweta County, 433 F.3d 807, 810 (11th Cir. 2005).
21 Scott, 127 S. Ct. at 1772; Harris, 433 F.3d at 810.
22 See Harris, 433 F.3d at 810.
23 Scott, 127 S. Ct. at 1772–73; Harris, 433 F.3d at 810.

844

HARVARD LAW REVIEW

[Vol. 122:837

ended some six minutes and nine miles after it began, when Scott decided to strike Harris’s rear bumper with his car, causing Harris, as intended, to spin out of control and crash.24 Scott recognized that this
maneuver involved a significant risk of serious injury or death to Harris,25 who in fact suffered a broken neck that left him a quadriplegic.26
Harris filed a lawsuit under 42 U.S.C. § 1983 alleging that the use
of admittedly deadly force to terminate the chase constituted an unreasonable seizure under the Fourth Amendment.27 After the district
court denied Scott’s claim of qualified immunity,28 Scott took an interlocutory appeal to the Eleventh Circuit Court of Appeals, which affirmed.29 In addition to upholding the district court’s ruling on immunity,30 the court of appeals agreed that Scott was not entitled to
summary judgment on the merits of Harris’s Fourth Amendment
claim:
None of the antecedent conditions for the use of deadly force existed in
this case. Harris’ infraction was speeding (73 mph in a 55 mph zone).
There were no warrants out for his arrest for anything, much less for the
requisite “crime involving the infliction or threatened infliction of serious
physical harm.” Indeed, neither Scott nor [a second officer] had any idea
why Harris was being pursued. The use of deadly force is not “reasonable” in a high-speed chase based only on a speeding violation and traffic
infractions where there was little, if any, actual threat to pedestrians or
other motorists, as the roads were mostly empty and Harris remained in
control of his vehicle, and there is no question that there were alternatives
for a later arrest.31

The court also specifically rejected Scott’s argument that “Harris’
driving must, as a matter of law, be considered sufficiently reckless to
give Scott probable cause to believe that he posed a substantial threat
of imminent physical harm to motorists and pedestrians”:32
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
24 Scott, 127 S. Ct. at 1773; Harris, 433 F.3d at 810. Scott had been authorized by his dispatcher to use a so-called “Precision Intervention Technique” (PIT) maneuver, a collision technique that is designed to cause a pursued vehicle to spin out and stop, but not necessarily to crash.
See Harris, 433 F.3d at 810–11. However, Scott (who was not trained in the PIT maneuver, see
id.) decided that he was going too fast to execute the maneuver and that it was necessary to
“ram[] his cruiser directly into Harris’s vehicle.” Id. at 811.
25 See Harris, 433 F.3d at 814 n.8.
26 See id. at 811; Harris v. Coweta County, No. 3:01CV148, 2003 WL 25419527, at *2 (N.D.
Ga. Sept. 25, 2003).
27 Scott, 127 S. Ct. at 1773.
28 See Harris, 2003 WL 25419527, at *6.
29 See Harris, 433 F.3d at 821. The court did, however, reverse in part, holding that a second
defendant, the officer who authorized the less dangerous PIT maneuver, was entitled to summary
judgment. See id. at 816–17.
30 See id. at 821.
31 Id. at 815 (citation omitted) (quoting Tennessee v. Garner, 471 U.S. 1, 11–12 (1985)).
32 Id.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

845

This is a disputed issue to be resolved by a jury. As noted by the district
court judge, taking the facts from the non-movant’s viewpoint, Harris remained in control of his vehicle, slowed for turns and intersections, and
typically used his indicators for turns. He did not run any motorists off
the road. . . . [B]y the time . . . Scott rammed Harris, the motorway had
been cleared of motorists and pedestrians allegedly because of police
blockades of the nearby intersections.33

The Supreme Court granted certiorari. Framed by the questions
presented for review and by the briefs, the case appeared to hinge on
two issues. One was whether Scott was entitled to immunity from suit
on the ground that any violation of Harris’s Fourth Amendment rights
was not based on law “clearly established” at the time of the chase.34
The other was the relevance of Tennessee v. Garner,35 which held that
police could not use deadly force in the form of shooting a fleeing suspect “unless . . . the officer has probable cause to believe that the suspect poses a significant threat of death or serious physical injury to the
officer or others.”36 The Eleventh Circuit had relied heavily upon
Garner;37 Scott argued for a less restrictive standard in the context of a
high-speed police chase.38
But it was the chase videotape, an exhibit in support of the defendants’ motion for summary judgment, that proved decisive. For the
Court, the facts revealed in the video made it so indisputably clear
that Scott was entitled to summary judgment that it found no need to
resolve the immunity issue.39 Justice Scalia wrote:
[W]e see respondent’s vehicle racing down narrow, two-lane roads in the
dead of night at speeds that are shockingly fast. We see it swerve around
more than a dozen other cars, cross the double-yellow line, and force cars
traveling in both directions to their respective shoulders to avoid being hit.

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
33
34

Id. at 815–16 (footnote omitted) (citations omitted).
See Brief for Petitioner at 24, Scott v. Harris, 127 S. Ct. 1769 (2007) (No. 05-1631), 2006 WL
3693418 (“Scott’s decision to use his push bumper to protect the lives of innocent persons from the
risks created by Harris’s dangerous driving did not violate ‘clearly established’ law. To find otherwise would reduce the requirement of ‘fair and clear warning’ that Scott’s conduct violated the
Fourth Amendment to no warning at all.”).
35 471 U.S. 1.
36 Id. at 3.
37 See Harris, 433 F.3d at 816–17.
38 See Brief for Petitioner, supra note 34, at 14 (“Garner’s framework should be reserved for
cases in which the officer would know with certainty that his use of force will be deadly. Applying Garner to vehicle-contact cases such as this one would provide the police with very little guidance on when they can use direct contact to stop a fleeing vehicle.”).
39 The Court reasoned that “because the constitutional question with which we are presented
is . . . easily decided,” it was not necessary to “address the wisdom” of the disputed precedent,
Saucier v. Katz, 533 U.S. 194 (2001), which generally requires the merits of a constitutional claim
to be considered prior to any immunity claim. Scott, 127 S. Ct. at 1774 n.4. Although he agreed
with the Court’s assessment of the merits, Justice Breyer, in his concurring opinion, took issue
with the Court’s failure to consider Scott’s immunity argument first. See id. at 1780–81 (Breyer,
J., concurring).

846

HARVARD LAW REVIEW

[Vol. 122:837

We see it run multiple red lights and travel for considerable periods of
time in the occasional center left-turn-only lane, chased by numerous police cars forced to engage in the same hazardous maneuvers just to keep
up. Far from being the cautious and controlled driver the lower court depicts, what we see on the video more closely resembles a Hollywood-style
car chase of the most frightening sort, placing police officers and innocent
bystanders alike at great risk of serious injury.40

Referring to the conventional rule that disputed facts should be construed in favor of the nonmoving party in evaluating a motion for
summary judgment, Justice Scalia concluded that “[t]he Court of Appeals should not have relied on such visible fiction; it should have
viewed the facts in the light depicted by the videotape.”41
The Court also found that the tape so manifestly demonstrated the
“reasonableness” of the use of deadly force that there was no need to
puzzle over how to adapt Garner to a car chase.42 The Court evaluated the reasonableness of Scott’s actions by looking at several factors,
starting first with the risks posed to the police, the public, and Harris
by the chase:
Although there is no obvious way to quantify the risks on either side, it is
clear from the videotape that respondent posed an actual and imminent
threat to the lives of any pedestrians who might have been present, to
other civilian motorists, and to the officers involved in the chase. It is
equally clear that Scott’s actions posed a high likelihood of serious injury
or death to respondent — though not the near certainty of death posed by,
say, shooting a fleeing felon in the back of the head . . . .43

The Court then attempted to balance these factors by framing the
issue as one of comparative fault:
So how does a court go about weighing the perhaps lesser probability of
injuring or killing numerous bystanders against the perhaps larger probability of injuring or killing a single person? We think it appropriate in
this process to take into account not only the number of lives at risk, but
also their relative culpability. It was respondent, after all, who intentionally placed himself and the public in danger by unlawfully engaging in the
reckless, high-speed flight that ultimately produced the choice between
two evils that Scott confronted. Multiple police cars, with blue lights
flashing and sirens blaring, had been chasing respondent for nearly 10
miles, but he ignored their warning to stop. By contrast, those who might
have been harmed had Scott not taken the action he did were entirely
innocent.44

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
40
41
42
43
44

Scott, 127 S. Ct. at 1775–76 (majority opinion) (footnotes omitted).
Id. at 1776.
Id. at 1778.
Id.
Id.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

847

The Court apparently viewed the conclusion of this “relative culpability” analysis as likewise so far beyond dispute that no contrary jury determination would be sustainable: “We have little difficulty in concluding it was reasonable for Scott to take the action that he did.”45
As noted, Justice Stevens, alone, dissented. Justice Stevens reported his own impression that “the tape actually confirms, rather than
contradicts, the lower courts’ appraisal of the factual questions at issue.”46 In what must have struck the majority as a strangely flattering
rebuke, Justice Stevens, at eighty-seven the Court’s oldest Justice, attributed his colleagues’ contrary perceptions to their comparative
youth: “Had they learned to drive when most high-speed driving took
place on two-lane roads rather than on superhighways — when splitsecond judgments about the risk of passing a slow-poke in the face of
oncoming traffic were routine — they might well have reacted to the
videotape more dispassionately.”47
Just as strangely, if not as flatteringly, the Court majority did not
counter Justice Stevens’s dissent with argument. As noted, the Court
replied curtly, “[w]e are happy to allow the videotape to speak for itself.”48 Answering Chico Marx’s question, Justice Breyer, in a concurring opinion, seconded the Court’s “see for yourself” rejoinder:
Because watching the video footage of the car chase made a difference to
my own view of the case, I suggest that the interested reader take advantage of the link in the Court’s opinion and watch it. Having done so, I do
not believe a reasonable jury could, in this instance, find that Officer
Timothy Scott (who joined the chase late in the day and did not know the
specific reason why the respondent was being pursued) acted in violation
of the Constitution.49

Indeed, Justice Breyer and Justice Ginsburg, who also wrote separately, were arguably even more emphatic about the impact of the
video. Whereas the majority opinion appeared to endorse a general
rule that all uses of deadly force to end dangerous high-speed chases
should be treated as constitutional,50 these Justices stressed the need to
review such pursuits case by case. “[T]he video,” Justice Breyer wrote,
“makes clear the highly fact-dependent nature of this constitutional de-

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
45
46
47
48
49
50

Id.
Id. at 1781 (Stevens, J., dissenting).
Id. at 1781 n.1.
Id. at 1775 n.5 (majority opinion).
Id. at 1780 (Breyer, J., concurring) (citation omitted).
See id. at 1779 (majority opinion) (“[W]e lay down a . . . sensible rule: A police officer’s attempt to terminate a dangerous high-speed car chase that threatens the lives of innocent bystanders does not violate the Fourth Amendment, even when it places the fleeing motorist at risk of
serious injury or death.”).

848
termination.”51
concurrence:

HARVARD LAW REVIEW

[Vol. 122:837

Justice Ginsburg echoed this point in her separate

I do not read today’s decision as articulating a mechanical, per se rule.
The inquiry described by the Court is situation specific. Among relevant
considerations: Were the lives and well-being of others (motorists, pedestrians, police officers) at risk? Was there a safer way, given the time,
place, and circumstances, to stop the fleeing vehicle? “[A]dmirable” as
“[an] attempt to craft an easy-to-apply legal test in the Fourth Amendment
context [may be],” the Court explains, “in the end we must still slosh our
way through the factbound morass of ‘reasonableness.’”52

For Justices Breyer and Ginsburg, then, the video mandated summary judgment not merely because it foreclosed reasonable disagreement about whether a dangerous chase had occurred, but also because,
for them at least, it foreclosed reasonable disagreement on whether
chasing Harris at all promoted public safety, whether Harris or the police were more culpable for the danger of the chase to the public, and
ultimately whether the use of deadly force was justified in light of the
risk that Harris posed.
II. TAKING THE SCOTT CHALLENGE
For those familiar with the Court’s commitments both to reasoned
justification and to safeguarding its exclusive power to interpret the
Constitution, the invitation to members of the public at large to judge
the correctness of the decision for themselves by simply applying their
senses was a conclusion to the case every bit as spectacular as the
metal-contorting crash that ended Harris’s flight from the police.53
There is, however, an obvious problem with the Court’s invitation. In
reporting that he, at least, saw something different, Justice Stevens
was plainly advancing the claim that the tape doesn’t speak for itself
— that different people, with different experiences, can see different
things in it. No individual who watches the tape and comes away
agreeing with the Court will be in a position to rebut Justice Stevens’s
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
51
52

Id. at 1780 (Breyer, J., concurring).
Id. at 1779 (Ginsburg, J., concurring) (alterations in original) (citations omitted) (quoting id.
at 1777–78 (majority opinion)).
53 Not surprisingly, this aspect of the Court’s decision has provoked debate in the legal
academy. Compare Posting of Orin Kerr to The Volokh Conspiracy, http://volokh.com/posts/
1172720514.shtml (Feb. 28, 2007, 22:23) (“The right answer is that Justice Breyer should believe
his own eyes.”), with The Supreme Court, 2006 Term—Leading Cases, 121 HARV. L. REV. 185, 222
(2007) (arguing that the Court’s analytical approach pushed it to make a bad ruling on an incomplete record), Jessica Silber, Justices Taken in by Illusion of Film, Balt. Sun, May 13, 2007, at
21A (“[A]ll film manifests a distinct point of view . . . . Films never speak for themselves . . . .”),
and Posting of Dave Hoffman to Concurring Opinions, http://www.concurringopinions.com/
archives/2007/04/the_death_of_fa.html (Apr. 30, 2007, 13:41) (“[E]ach Justice saw the risk of
speeding through his or her own cultural prism.”). See generally Howard M. Wasserman, Video
Evidence and Summary Judgment: The Procedure of Scott v. Harris, 91 Judicature 108 (2008).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

849

claim, because however clearly that person perceives things, the fact
remains that she is able to see only what she sees and not what anyone
else does. The testing method the Court proposes, in sum, is hopelessly solipsistic.
There is only one way we can think of to accept the Scott majority’s proposal that its conclusion be tested by viewing the tape. It is to
get a large number of people of diverse experiences and characteristics
to watch it. If there are members of this group who do see things differently from the majority, one can then try to determine whether there
is anything that seems to unite the former other than that they are
simply “unreasonable.”
That’s what we decided to do. We conducted a study to see who
sees what in the Scott tape and why. In this Part, we describe that
study. We start essentially with the bottom line — a narrative account
of what people of varying characteristics and backgrounds would
likely see were they to watch the Scott tape. We then identify the social psychological theories that explain why one might expect certain
defining characteristics to affect such individuals’ perceptions of facts.
Then we show how these theories informed the design and hypotheses
of our study. Finally, we offer a more fine-grained reporting of the
study results themselves, before summing up.
One point of clarification, however, must be stressed at the outset.
Although our subjects were instructed to view the video from the perspective of a juror deciding the case, we do not understand the results
of the study to indicate how a case like Scott would come out if it
went to trial. Obviously, jurors in an actual case would be exposed to
more evidence than just the video. They would also hear arguments
from counsel. Most important of all, they would deliberate. They
might conclude, on such a basis, to vote for a verdict contrary to their
initial reaction to the tape, particularly if that reaction were equivocal.
Our study is designed only to help evaluate the assertion that the
tape “speak[s] for itself”54 — and that what it says is so unequivocal
that it leaves no room to “believe a reasonable jury could, in this instance, find that Officer Timothy Scott . . . acted in violation of the
Constitution.”55
A. Four Members of the American Venire
Imagine four Americans (Figure 1). Ron, a white male who lives in
Arizona, overcame his modest upbringing to become a self-made millionaire businessperson. He deeply resents government interference
with markets but is otherwise highly respectful of authority, which he
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
54
55

Scott, 127 S. Ct. at 1775 n.5.
Id. at 1780 (Breyer, J., concurring).

850

HARVARD LAW REVIEW

[Vol. 122:837

believes should be clearly delineated in all spheres of life. Politically,
he identifies himself as a conservative Republican. Bernie, another
white male, is a university professor who makes a modest salary and
lives in Burlington, Vermont. He will go along with the left wing of
the Democratic Party, but thinks of himself as a “social democrat.” He
advocates highly egalitarian conditions in the home, in the workplace,
and in society at large, and strongly supports government social welfare programs and regulations of every stripe. Linda is an African
American woman employed as a social worker in Philadelphia, Pennsylvania. She is a staunch Democrat and unembarrassed to be characterized as a “liberal.” Finally, there is Pat. Pat is the average American in every single respect: Pat earns the average income, has the
average level of education, is average in ideology, is average in party
identification, holds average cultural values, and is average even in
race and gender. If placed on a jury, apprised of the basic issues and
law, and asked to watch the video in Scott v. Harris, what will these
four individuals see on the tape, and how will the tape affect their
views of how the case should come out prior to deliberating?
Figure 1. Four Representative Members
of the American Venire

Ron

Linda

Bernie

Pat

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

851

Here is the answer. Ron will see exactly what the Scott majority
saw: a very dangerous driver, whom the police correctly pursued from
the outset and against whom they appropriately used deadly force.
Pat will tend in the same direction, but with less certainty; in particular, Pat will be ambivalent about whether the risk to the public was
worth the chase, but will conclude, fairly unequivocally, that terminating it with deadly force was warranted given the danger that Harris’s
flight posed.
Linda and Bernie, however, will feel quite differently. They will be
fairly adamant that the police, after detecting that Harris was speeding, made a serious mistake in conducting a high-speed chase when he
refused to pull over. They will tend to agree that the chase posed lethal risks to the public and the police, but in contrast to Ron and Pat,
they will be somewhat equivocal in that judgment. They will not be
equivocal about blame: they will perceive the police to be as culpable
as Harris or even more culpable for the risk the chase created for the
public. And they will be fairly strongly disposed to find that the police
were not justified in using deadly force to terminate the chase in light
of the danger Harris himself posed.
At least, those are the impressions they’ll have before they start to
exchange their views with one another and explain them in deliberations. In what follows, we identify the theoretical grounds from which
to expect this initial array of perceptions and then turn to the details of
the study that support this particular account of what individuals like
these will likely “see” in the Scott tape.
B. Theoretical Background: Motivated Cognition
of Legally Consequential Facts
Why, to start, might we expect people with characteristics as diverse as Ron’s, Linda’s, Pat’s, and Bernie’s not to see “eye to eye” on
the Scott video? There are a number of social psychological mechanisms that explain how the group identities and values associated with
such characteristics influence cognition of facts. They form the theoretical basis for the design of our study and its hypotheses.
One of these mechanisms is the culpable control model of blame.56
That theory starts with the premise — confirmed by a diverse body of
research in psychology, anthropology, and linguistics57 — that people
are generally disposed to “blame” someone for an action only if they
perceive that the putatively blameworthy agent made a voluntary
choice to act in a manner that caused some harm or injury. Neverthe–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
56

See generally Mark D. Alicke, Culpable Control and the Psychology of Blame, 126 PSY-

CHOL. BULL. 556 (2000).
57 See, e.g., id. at 556–57

(discussing previous research on blame).

852

HARVARD LAW REVIEW

[Vol. 122:837

less, experiments demonstrate that people’s perceptions of the elements
of this blaming template are highly sensitive to circumstances extrinsic
to the template itself. In particular, people are much more disposed to
perceive voluntariness, action, causation, and harm (as well as other
conditions of blaming) when the putatively blameworthy agent is behaving in ways that defy social norms (perhaps by engaging in an interracial relationship or by using drugs), including norms contested
across distinct subcommunities.58 In effect, cognition of blamerelevant “facts” (volition, action, causation, harm) is motivated by the
subconscious desire to form blame attributions that accord with moral
evaluations of the agent’s character or lifestyle.
A second theory is identity-protective cognition.59 Individual wellbeing, material and emotional, is bound up with membership in various self-defining groups. Rejecting factual beliefs widespread within
such a group can undermine individual well-being, either by threatening to estrange a person from his peers or by forcing that person to
contemplate the social incompetence of those he identifies with. As a
means of psychological self-defense, then, people tend to process information in a selective fashion that bolsters beliefs dominant within
their self-defining groups.60
Finally, there is the cultural cognition of risk.61 This theory posits
that people tend to conform factual beliefs about risk to their cultural
evaluations of putatively dangerous behavior. As a result of various
cognitive mechanisms, people are motivated to believe that behavior
they find noble is also socially beneficial (or at least benign) and behavior they find base is also socially harmful. For example, people
who hold individualist values discount claims that commerce and industry harm the environment because they value markets and other
forms of private ordering. People who hold egalitarian values, in contrast, readily credit claims of environmental risk, the widespread acceptance of which would justify restricting behavior — such as commerce and industry — that those people associate with inequalities in
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
58 See id. at 564; see also Mark D. Alicke, Culpable Causation, 63 J. PERSONALITY & SOC.
PSYCHOL. 368 (1992).
59 See generally Dan M. Kahan, Donald Braman, John Gastil, Paul Slovic & C.K. Mertz, Culture and Identity-Protective Cognition: Explaining the White-Male Effect in Risk Perception, 4 J.
EMPIRICAL LEGAL STUD. 465 (2007).
60 See generally Joshua Aronson & Claude M. Steele, When Beliefs Yield to Evidence: Reducing Biased Evaluation by Affirming the Self, 26 PERSONALITY & SOC. PSYCHOL. BULL. 1151
(2000); Geoffrey L. Cohen, Roger Giner-Sorolla & Shelly Chaiken, Selective Use of Heuristic and
Systematic Processing Under Defense Motivation, 23 PERSONALITY & SOC. PSYCHOL. BULL.
84 (1997); Geoffrey L. Cohen, David K. Sherman, Anthony Bastardi, Lillian Hsu, Michelle
McGoey & Lee Ross, Bridging the Partisan Divide: Self-Affirmation Reduces Ideological ClosedMindedness and Inflexibility in Negotiation, 93 J. PERSONALITY & SOC. PSYCHOL. 415 (2007).
61 See generally Dan M. Kahan, Paul Slovic, Donald Braman & John Gastil, Fear of Democracy: A Cultural Evaluation of Sunstein on Risk, 119 HARV. L. REV. 1071, 1083–87 (2006).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

853

wealth. Alternatively, people who subscribe to hierarchical values impute risk to homosexuality, drug use, abortion, subversive speech, and
other behaviors that defy the authority of traditional, stratified norms
of behavior, whereas egalitarians and individualists reject these risk
claims as specious.62
Combined, these theories furnish strong grounds to expect individuals of diverse identities and commitments — like Ron, Linda,
Bernie, and Pat — to form different perceptions of the facts revealed
by the Scott tape. The facts highlighted by Justice Scalia’s analysis —
how much care Harris was taking to avoid colliding with other vehicles,63 how feasible it was for other drivers to pull to the side to avoid
exposure to collision,64 the relative harm of chasing Harris or breaking
off the pursuit and letting him escape65 — all relate to moral (and legal) attributions of blame. Perceptions of those facts, the culpable control model suggests, are likely to be motivated by extrinsic moral
evaluations of the putatively blameworthy actors — Harris and the
police.
Beliefs about the extent to which the police in general abuse their
authority (particularly against minorities), and correspondingly the
relative preponderance of licit and illicit reasons for attempting to
avoid police encounters, vary across sociodemographic and political
groups.66 Identity-protective cognition thus suggests that people are
likely to construe the facts depicted in the tape in a way that reinforces
the beliefs that predominate among their peers.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
62 See id. at 1086–87. This theory is amply supported by a growing body of empirical research. See, e.g., Karl Dake, Orienting Dispositions in the Perception of Risk: An Analysis of
Contemporary Worldviews and Cultural Biases, 22 J. CROSS-CULTURAL PSYCHOL. 61 (1991);
Hank C. Jenkins-Smith, Modeling Stigma: An Empirical Analysis of Nuclear Images of Nevada,
in RISK, MEDIA, AND STIGMA: UNDERSTANDING PUBLIC CHALLENGES TO MODERN SCIENCE AND TECHNOLOGY 107 (James Flynn, Paul Slovic & Howard Kunreuther eds., 2001);
Kahan, Braman, Gastil, Slovic & Mertz, supra note 59; Dan M. Kahan, Donald Braman, Paul
Slovic, John Gastil & Geoffrey Cohen, Cultural Cognition of the Risks and Benefits of Nanotechnology, 4 NATURE NANOTECHNOLOGY (forthcoming Feb. 2009); Ellen Peters & Paul Slovic,
The Role of Affect and Worldviews As Orienting Dispositions in the Perception and Acceptance of
Nuclear Power, 26 J. APPLIED SOC. PSYCHOL. 1427, 1445 (1996); Dan M. Kahan, Donald Braman, Paul Slovic, John Gastil & Geoffrey L. Cohen, The Second National Risk and Culture
Study: Making Sense of — and Making Progress in — The American Culture War of Fact (Yale
Law Sch., Public Law Working Paper No. 154; George Washington Univ., Legal Studies Research
Paper No. 370; George Washington Univ., Law School Public Law Research Paper No. 370, 2007)
[hereinafter Kahan et al., Risk and Culture Study], available at http://ssrn.com/abstract=1017189.
63 Scott v. Harris, 127 S. Ct. 1769, 1775–76 (2007).
64 See id. at 1775.
65 Id. at 1778–79.
66 See Daniel Lempert, Belief in a Just World and Perceptions of Fair Treatment by Police (2007), available at http://www.electionstudies.org/resources/papers/Pilot2006/
nes012058.pdf (reporting in Part I results of survey on perceived fairness of police toward criminal suspects). Cf. Illinois v. Wardlow, 528 U.S. 119, 132–34 (2000) (Stevens, J., concurring in part
and dissenting in part) (noting community-specific reactions to avoidance of police).

854

HARVARD LAW REVIEW

[Vol. 122:837

Finally, the facts at issue relate to the risks posed by the parties to
the chase. Consistent with the theory of cultural cognition, we can expect individuals to form risk perceptions that reflect the competing,
culturally grounded affective responses a high-speed police chase is
likely to evoke: from fear of those who defy lawful authority; to resentment of abuses of power by the police; to distrust of authority generally; to anger at apparent indifference to the well-being of innocent
bystanders.67
C. Study Design and Hypotheses
These understandings of the potential sources of motivated cognition suggest a relatively straightforward way to determine whether the
Scott tape admits of competing interpretations. By showing the tape
to a sufficiently large and diverse group of persons and soliciting their
reactions to it, one can determine whether differences of belief on key
issues vary across persons in patterns predicted by cultural cognition,
identity-protective cognition, and the culpable control theory. As we
explain in greater detail, this is how we proceeded.
1. Sample. — The study sample consisted of approximately 1350
individuals. The subjects were drawn randomly from a demographically diverse panel of some 40,000 online adult Americans assembled
by Knowledge Networks for participation in scholarly public opinion
analysis.68 The sample was 51% male and 49% female; and 9% Hispanic, 6% African American, and 6% other minority. The mean age
was 47, mean annual household income $40,000 to $49,000, and the
mean education level was “some college.” The subjects participated in
the study over a one-month period between late September 2007 and
late October 2007.
2. Stimulus. — Subjects were exposed to a stimulus consisting of
two parts. The first was a textual introduction. The introduction advised subjects that the point of the survey was to determine how they
“would decide a real-life lawsuit if [they] were on the jury.” In addition to setting forth the nature of the suit — one brought by Harris al–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
67 See generally Dan M. Kahan, Two Conceptions of Emotion in Risk Regulation, 156 U. PA.
L. REV. 741 (2008) (describing evidence of the role that affective states play in connecting risk
perceptions to cultural values).
68 Knowledge Networks is a leading firm in online opinion studies. Numerous studies have
shown that the online samples and testing methods of Knowledge Networks yield results equivalent in reliability to conventional random-digit-dial surveys, and studies based on those samples
and methods are routinely published in academic journals. See KNOWLEDGE NETWORKS,
KNOWLEDGE NETWORKS BIBLIOGRAPHY: ARTICLES AND PRESENTATIONS BASED ON KN
COLLECTED PANEL DATA, ANALYSIS, OR METHODOLOGY, available at http://www.
knowledgenetworks.com/ganp/docs/KN%20Bibliog%205-29-2007%20External.pdf; J.M. Dennis,
Methodological Information About Knowledge Networks for Reviewers, http://www.
knowledgenetworks.com/ganp/reviewer-info.html (last visited Dec. 6, 2008).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

855

leging that he had suffered injuries when Scott used “unreasonable
force” to end a high-speed chase — the introduction set forth a list of
facts on which “the parties agree.” The list consisted of facts not disputed by the parties when the case was before the Supreme Court:69
The police clocked Harris driving 73 miles per hour on a highway in a 55
mile-per-hour zone at around 11 pm.
The police decided to pursue Harris when Harris ignored the police car’s
flashing lights and kept driving rather than pulling over.
The chase lasted around seven minutes and covered eight to nine miles.
The police determined from the license plate number that the vehicle had
not been reported stolen.
Officer Scott joined the chase after it started. He did not know why the
other officers had originally tried to stop Harris.
Scott knew that other police officers had blocked intersections leading to
the highway but did not know if all of the intersections were blocked.
Officer Scott deliberately used his police cruiser’s front bumper to hit the
rear of Harris’s car[,] hoping to cause Harris’s car to spin out and come to
a stop.
Officer Scott knew there was a high risk that ramming the car in this
manner could seriously injure or kill Harris.
Harris lost control, crashed, and suffered severe injuries, including permanent paralysis from the neck down.70

The second part of the stimulus consisted of the chase video. The
video was identified as having been filmed from inside the pursuing
police cruisers. It was also described as “key evidence” relating to
facts on which the “[p]arties disagree.” Subjects were requested to
“closely watch the video . . . just as [members of the jury in the case]
would.”71
The video is approximately six minutes in length. It starts when
the police activate their sirens and terminates with a scene of Harris’s
flipped-over vehicle engulfed in thick smoke. The study video displayed an image comparable in size to the video uploaded to the Supreme Court website. Like the Supreme Court video, the study video
also included sound, which consisted of radio communication between
the pursuing cars and the police dispatcher. Shortly before Scott rams
Harris’s vehicle, the dispatcher is heard issuing the instruction to “take
him out.”
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
69 Scott, 127 S. Ct. at 1772–73; Brief for Petitioner, supra note 34, at 2–5; Brief for Respondent
at 12, Scott, 127 S. Ct. 1769 (No. 05-1631), 2007 WL 118977.
70 The survey instrument is on file with the Harvard Law School Library.
71 The study video can be accessed at http://www.youtube.com/watch?v=DBY2y2YsmN0.

856

HARVARD LAW REVIEW

[Vol. 122:837

The video shown to the subjects was actually a composite of two
from the trial record, both of which were uploaded to the Supreme
Court’s website.72 The two tapes were recorded by the two pursuing
police cars, which, at around the midpoint of the chase, swap positions
relative to the fleeing Harris. The study video consists of those portions of each tape recorded when the filming vehicle was the lead car
in the chase and omits those portions of each recorded when the car
filming was in the trailing position. It was necessary to combine the
tapes in this fashion to keep the total running time of the study short
enough to avert high dropout rates among study subjects. Because
only the footage shot from inside the lead vehicle permits observation
of Harris, the study video nevertheless contained all portions of both
tapes that bear on the factual disagreements between the Scott majority and dissent.
Like the tapes in the Supreme Court record, but unlike the file displayed on the Supreme Court website, the study tape was in color. As
a result, it permitted subjects to observe the color of overhead traffic
lights and also clearly to discern the tail lights (including the braking
lights) of Harris’s car and of vehicles passed during the pursuit. The
black and white file displayed on the Supreme Court website, in addition to masking these pertinent details, contains myriad indistinct
patches of bright or flashing light; in the color video, these flashes are
revealed to be objects such as illuminated roadside street signs, lights
in roadside structures, and the headlights of cars that are either pulled
over, approaching the chase from an intersecting street, or traveling in
the opposite direction. Beyond furnishing a more vivid (and potentially more chilling) depiction of the chase than the file uploaded to the
Court’s website, the study tape in this respect also more closely resembled the footage observed by the Justices themselves.
3. Response Measures. — On exactly which facts the Court understood the video to be dispositive in Scott is itself open to debate. Although Justice Scalia considered a range of issues — the degree of risk
Harris’s driving posed;73 how to “weigh[] the perhaps lesser probability
of injuring or killing numerous bystanders against the perhaps larger
probability of injuring or killing a single person”; 74 the “relative culpability” of Harris and the police for creating risk to the public75 — his
opinion could be read to indicate that the constitutionality of deadly
force turns on only one fact: whether the police and the fleeing driver
were involved in a “high-speed car chase that threatens the lives of in–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
72 We were kindly supplied with the record versions of the tapes by Andrew Clarke, counsel
for Harris.
73 See Scott, 127 S. Ct. at 1775–76.
74 Id. at 1778.
75 Id.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

857

nocent bystanders.”76 Justices Breyer and Ginsburg, however, indicated that they did not understand the Court’s decision “as articulating
a mechanical, per se rule.”77 For them, the entire “inquiry described
by the Court”78 involved “highly fact-dependent” questions, the answers to which were “ma[de] clear” by a viewing of the video.79 We
decided to solicit our subjects’ reactions to the entire range of determinations that figured in the Court’s reasoning, both because the opinion
was arguably ambiguous about which of them were matters of fact for
the jury to decide and because the possibility that people might disagree about them after watching the video struck us as relevant to
evaluating which of those facts ought to be ones for a jury in a case
like Scott.
One of the key issues was the degree of risk that Harris’s driving
posed. Justice Scalia used various formulations to characterize the degree of lethal danger that a jury would be required to find in a case
like Scott.80 “Great risk of death” struck us as the one most likely to
be understood clearly and uniformly among the laypersons in our subject pool.81 Accordingly, to assess our subjects’ perceptions of that
risk, we asked them to indicate, on a six-point scale, their level of disagreement or agreement with these two statements:
During the pursuit, Harris drove in a manner that put members of the
public at great risk of death.
During the pursuit, Harris drove in a manner that put the police at serious
risk of death.

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
76 Id. at 1779 (“[W]e lay down a more sensible rule: A police officer’s attempt to terminate a
dangerous high-speed car chase that threatens the lives of innocent bystanders does not violate
the Fourth Amendment, even when it places the fleeing motorist at risk of serious injury or
death.”).
77 Id. (Ginsburg, J., concurring); see also infra p. 848.
78 Scott, 127 S. Ct. at 1779 (Ginsburg, J., concurring).
79 Id. at 1780 (Breyer, J., concurring).
80 See, e.g., id. at 1775–76 (majority opinion) (“[T]he video . . . closely resembles a Hollywoodstyle car chase of the most frightening sort, placing police officers and innocent bystanders alike
at great risk of serious injury.” (emphasis added)); id. at 1778 (“[I]t is clear from the videotape that
respondent posed an actual and imminent threat to the lives of any pedestrians . . . , to other civilian motorists, and to the officers . . . .” (emphasis added)); id. at 1779 (“The car chase that respondent initiated in this case posed a substantial and immediate risk of serious physical injury to others; no reasonable jury could conclude otherwise.” (emphasis added)).
81 Because the Court emphasized that the tape left no ground for reasonable disagreement on
“the factual issue whether respondent was driving in such fashion as to endanger human life,” id.
at 1776 (emphasis added), we interpret the Court’s references to the “risk of serious injury,” id. at
1772, 1776, and a “substantial and immediate risk of serious physical injury,” id. at 1779, as contemplating a degree of injury that would be life-threatening. To avoid the possibility that subjects
would construe the term “physical injury” as extending to non-life-threatening harm, we chose not
to use this phrase and instead to refer simply to “great risk of death,” which we judged would uniformly be understood to embrace the risk of life-threatening physical injury.

858

HARVARD LAW REVIEW

[Vol. 122:837

The majority opinion in Scott also evaluated the degree of risk the
police created and its magnitude in relation to the risk posed by Harris. To assess our subjects’ perceptions of that issue, we asked them to
indicate their level of disagreement or agreement (again on a six-point
scale) with this statement:
It just wasn’t worth the danger to the public for the police to engage in a
high-speed chase of Harris when he refused to pull over for speeding. Instead, they should have tried to find and arrest him later.

They were also asked to indicate their assessment of the “relative
culpability”82 of the police and Harris:
Please indicate how much you think the parties were at fault for the risk
posed to the public by the chase: (1) the police were much more at fault
than Harris; (2) the police were slightly more at fault than Harris; (3) the
police and Harris were equally at fault; (4) Harris was slightly more at
fault than the police; (5) Harris was much more at fault than the police.

Finally, we asked subjects to indicate their level of disagreement or
agreement with a statement relating to the outcome of the case:
The danger that Harris’s driving posed to the police and the public justified Officer Scott’s decision to end the chase in a way that put Harris’s
own life in danger.

Responses to these five items, when combined, formed a highly reliable scale (α = 0.81).83 To facilitate analysis, the scale was coded to
reflect agreement with the Scott majority and labeled “Agree with
Court.”84
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
82
83

Id. at 1778.
Cronbach’s alpha (α) is a statistic for measuring the internal validity of attitudinal scales.
In effect, it measures the degree of intercorrelation that exists among various items within a scale;
a high score suggests that the items can be treated as a valid measure of a latent, or unmeasured,
attitude or trait. Generally, α ≥ 0.70 suggests scale validity. See generally Jose M. Cortina, What
Is Coefficient Alpha? An Examination of Theory and Applications, 78 J. APPLIED PSYCHOL. 98
(1993). Because a high Cronbach’s alpha confirms that responses to items of opposing valence
(e.g., that “[d]uring the pursuit, Harris drove in a manner that put members of the public at great
risk of death” versus “[i]t just wasn’t worth the danger to the public for the police to engage in a
high-speed chase of Harris when he refused to pull over for speeding”) are indeed negatively correlated, it helps to dispel the concern that subjects’ responses reflected acquiescence bias, the tendency “to ‘agree’ with a statement just to avoid seeming disagreeable,” RUSSELL K. SCHUTT,
INVESTIGATING THE SOCIAL WORLD 265 (6th ed. 2008).
84 Because subjects responded after watching a car crash that they were told resulted in serious injury, one might reasonably wonder whether their responses were affected by “hindsight
bias,” which consists in overestimating the ex ante likelihood of an event based on ex post knowledge of its occurrence. See generally Jeffrey J. Rachlinski, A Positive Psychological Theory of
Judging in Hindsight, 65 U. CHI. L. REV. 571 (1998). However, the Scott majority based its ruling not on its anxiety that jurors might overreact to the tape (an alternative basis for decision we
discuss below, see infra p. 891), but on its professed confidence that jurors would rule for the defendant on the basis of the tape. The possibility of hindsight bias — whatever it might imply
about what jurors in a case like Scott should be allowed to see — does not pose any difficulty for
a study aimed at testing disputed claims about how jurors would react if allowed to view it.
Moreover, because it is an influence to which all subjects are exposed, the possible impact of

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

859

4. Individual Characteristics. — (a) Demographic Characteristics.
— We collected data relating to the individual characteristics of the
subjects. These included conventional sociodemographic characteristics, such as gender, race, age, household income, education, community type (urban or nonurban) of residence, and region of residence.
(b) Political Ideology and Party Affiliation. — Subjects indicated their party affiliation — either Republican, Democrat, or neither.
They also indicated their political ideology on a seven-point
scale that ran from “extremely liberal” to extremely conservative
(“conservativism”).
(c) Cultural Worldviews. — We also collected data on our subjects’
cultural orientations and worldviews. The orientations reflected a
scheme developed by the late anthropologist Mary Douglas, who characterized worldviews, or preferences for how society should be organized, along two cross-cutting dimensions — “group” and “grid.”85 A
“high group” worldview generates a preference for a communitarian
ordering in which the interests of the individual are subordinated to
the needs of the collective, which is in turn responsible for securing the
conditions of individual flourishing. A “low group” worldview, in contrast, coheres with a preference for an individualist ordering in which
individuals are expected to secure the conditions of their own flourishing without interference or assistance from the collective. A “high
grid” worldview corresponds to a preference for a relatively hierarchical ordering, in which entitlements, obligations, opportunities, and offices are all assigned on the basis of conspicuous and largely fixed attributes, such as gender, race, lineage, class, and the like. A “low grid”
worldview, in contrast, generates a preference for an egalitarian ordering that emphatically rejects the proposition that such distinctions
should figure in this way in societal conditions.86
The two dimensions of worldview contemplated by “group-grid”
were measured (in pre-screening surveys conducted several weeks in
advance of the study) with two scales, “Hierarchy-Egalitarianism” (or
simply, “Hierarchy”) and “Individualism-Communitarianism” (“Individualism”), used in previous studies of the cultural cognition of risk.87
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
hindsight bias also does not pose any obstacle to testing hypotheses about individual differences in
perceptions based on cultural worldviews and other pertinent characteristics.
85 See generally MARY DOUGLAS, NATURAL SYMBOLS: EXPLORATIONS IN COSMOLOGY
(1996).
86 See generally Steve Rayner, Cultural Theory and Risk Analysis, in SOCIAL THEORIES OF
RISK 83 (Sheldon Krimsky & Dominic Golding eds., 1992).
87 The scales consist of multiple “agree-disagree” items that express attitudes associated with
one or another worldview (e.g., “[a] lot of problems in our society today come from the decline in
the traditional family, where the man works and the woman stays home,” for hierarchy; “[w]e
need to dramatically reduce inequalities between the rich and the poor, whites and people of color,
and men and women,” for egalitarianism; “[t]oo many people today expect society to do things for

860

HARVARD LAW REVIEW

[Vol. 122:837

As in previous studies, the scales were highly reliable measures of the
latent disposition of subjects toward those respective sets of worldviews (Individualism, α = 0.86; Hierarchy, α = 0.85). To facilitate
comparisons of subjects identified by their worldviews, we assigned
subjects to cultural groups. Based on the relationship of their scores to
the median on each scale, we classified subjects as either “Hierarchs”
or “Egalitarians” and as either “Individualists” or “Communitarians.”
5. Hypotheses. — We formulated two major hypotheses. One was
that even if a majority of subjects agreed with the interpretation of the
Supreme Court majority in Scott, reactions to the study tape would
display significant variation across persons of different characteristics.
The second major hypothesis was that these characteristics would be
suggestive of the impact of culpable control, identity-protective, and
cultural cognition.
What these characteristics would be and how they would matter
support a number of subhypotheses. Some of the relevant characteristics, we surmised, would be sociodemographic. We thus anticipated
differences along dimensions such as age and community type (urban
or nonurban). Such differences could be a consequence of the mundane contribution that experience in general makes to understanding
events: witnessed events are, in effect, the equivalent of minor premises in a cognitive syllogism that generates conclusions only when
combined with major premises drawn from experience. Justice Stevens clearly had this point in mind when he argued that older people
who learned to pass slow-moving automobiles on poorly lit, windy,
two-lane country roads are likely to see Harris’s driving as creating
less risk than those who grew up in an era of straight, brightly illuminated, four-lane highways. One might expect differences in various
other experientially relevant characteristics — such as education — to
have some effect too.
But other demographic characteristics, we expected, would also
contribute to differences in perception through mechanisms of motivated cognition. Race and gender, for example, are variously depicted
both as sources of group identity and as proxies for identities that consist in shared values.88 African Americans, we predicted, would be inclined to view the facts in a manner different from the Scott majority
because they are more likely to have had negative experiences with po–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
them that they should be doing for themselves,” for individualism; and “[i]t’s society’s responsibility to make sure everyone’s basic needs are met,” for communitarianism). Kahan, Braman, Gastil,
Slovic & Mertz, supra note 59, at 504–05.
88 See, e.g., W.E.B. DU BOIS, THE SOULS OF BLACK FOLK (Alfred A. Knopf, Inc. 1993)
(1903) (offering classic statement of African American identity based on shared values); CAROL
GILLIGAN, IN A DIFFERENT VOICE: PSYCHOLOGICAL THEORY AND WOMEN’S DEVELOPMENT (1993) (developing theory of gender identities focused on contrasting values).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

861

lice, to know other persons who have, and (as a result) to have friends
and family members who believe the police are disposed to abuse their
authority.89 Perceiving the facts in the video in a manner that suggests
the police response was out of proportion to the risk created by Harris
is congenial to this evaluation, and thus predicted by both the culpable-control and identity-protective cognition theories.
We also predicted that gender would influence perceptions of the
facts in the Scott tape. Women, we surmised, were likely to view the
facts in a manner relatively more favorable to Harris than were men.
The basis for this conjecture is that gender correlates with values —
including cultural egalitarianism90 and ideological liberalism91 — that
we have reason to think (as described below) would motivate cognition
hostile to the behavior of the police in this case.
Other characteristics we expected to matter are explicitly connected
to group commitments that figure in theories of motivated cognition.
These include political ideology and political party affiliation. Understandings predominant among politically defined groups are among
those to which individuals, for identity-protective reasons, tend to conform their own beliefs about policy-related and legally consequential
facts.92 Cultural worldviews also matter from the standpoint of motivated cognition. As noted, previous work suggests that cultural
worldviews more powerfully explain differences of risk perception93
and legally consequential facts94 than do other characteristics. Gender,
race, income, and region of residence are likely proxies for the affinities constituted by such values.95 Moreover, they also furnish greater

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
89 See Sandra Lee Browning et al., Race and Getting Hassled by the Police: A Research Note,
POLICE STUD.: INT’L REV. OF POLICE DEV., NO. 1, at 1, 1–11 (1994); Yolander G. Hurst,
James Frank & Sandra Lee Browning, The Attitudes of Juveniles Toward the Police: A Comparison of Black and White Youth, 23 POLICING: INT’L J. POLICE STRATEGIES & MGMT. 37, 44–
50 (2000).
90 See Melissa L. Finucane et al., Gender, Race, and Perceived Risk: The ‘White Male’ Effect,
2 HEALTH, RISK & SOC’Y 159, 166–67 (2000).
91 See Richard A. Brody & Jennifer L. Lawless, Political Ideology in the United States: Conservatism and Liberalism in the 1980s and 1990s, in CONSERVATIVE PARTIES AND RIGHTWING POLITICS IN NORTH AMERICA: REAPING THE BENEFITS OF AN IDEOLOGICAL VICTORY? 53 (Rainer-Olaf Schultze, Roland Sturm & Dagmar Eberle eds., 2003).
92 See Geoffrey L. Cohen, Party over Policy: The Dominating Impact of Group Influence on
Political Beliefs, 85 J. PERSONALITY & SOC. PSYCHOL. 808 (2003).
93 See generally Kahan, Braman, Gastil, Slovic & Mertz, supra note 59.
94 See Dan M. Kahan & Donald Braman, The Self-Defensive Cognition of Self-Defense, 45
AM. CRIM. L. REV. 1, 54–56 (2008).
95 See RICHARD E. NISBETT & DOV COHEN, CULTURE OF HONOR: THE PSYCHOLOGY
OF VIOLENCE IN THE SOUTH (1996) (using region as a proxy for shared values); Kahan, Braman, Gastil, Slovic & Mertz, supra note 59, at 471 (suggesting that race and gender correlate with
shared values).

862

HARVARD LAW REVIEW

[Vol. 122:837

heuristic guidance than do political ideologies for persons who (like
most Americans) have at best only modest interest in politics.96
To focus our hypotheses reflecting these influences, we predicted
that reactions to the Scott tape would vary across subjects divided into
two prominent and recognizable cultural styles, which we designated
“aleph” and “bet.” Alephs hold conspicuously hierarchical and individualistic cultural worldviews and (on most questions, at least) highly
conservative political leanings. Demographically, they are more likely
to be affluent white males.97 Alephs also are more likely to live in the
South or far West, where hierarchical and individualistic values predominate.98 Bets, in contrast, hold disproportionately egalitarian and
communitarian views. Their politics are more liberal and Democratic.
Relative to alephs, at least, bets will also be disproportionately female
and African American. And they are more likely to live in the Northeast, where egalitarian and communitarian cultural worldviews are
relatively predominant.99
We understand these groups to be recognizable contemporary examples of what Gusfield refers to as “status collectivities.”100 “Unlike
groups such as religious and ethnic communities[,] they have no
church, no political unit, and no associational units which explicitly
defend their interests,” but are nevertheless affiliated, in their own selfunderstandings and in the views of others, by largely convergent
worldviews and by common commitments to salient political agendas.101 “They possess subcultures” that play a conspicuous role in political conflict motivated by symbolic status competition.102 The issues
that pit our aleph and bet styles against each other range from affirmative action to gun control, from nuclear power to abortion, from the
death penalty to gay rights.103
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
96 See Aaron Wildavsky, Choosing Preferences By Constructing Institutions: A Cultural Theory of Preference Formation, 81 AM. POL. SCI. REV. 3, 15–18 (1987); see also Kahan et al., Risk
and Culture Study, supra note 62, at 14–15.
97 Indeed, a discrete but sizable subset of white males who hold extremely hierarchical and
individualistic values drive the so-called “white male effect,” which refers to the disposition of
white men to attribute less risk to various putatively dangerous activities than do women and minorities. See generally Kahan, Braman, Gastil, Slovic & Mertz, supra note 59.
98 See generally RAYMOND D. GASTIL, CULTURAL REGIONS OF THE UNITED STATES
(1975) (providing basic cultural descriptions of geographic regions); NISBETT & COHEN, supra
note 95 (contrasting the South with the North, West, and Midwest).
99 See generally Dan M. Kahan, The Secret Ambition of Deterrence, 113 HARV. L. REV. 413,
439–42, 452–53, 462–67 (1999) (describing culturally grounded status conflict between these types
on issues such as the death penalty, gun control, and hate crimes).
100 GUSFIELD, supra note 17, at 21 (emphasis omitted).
101 Id.
102 Id.
103 See, e.g., id. at 140–41, 167–71; KRISTIN LUKER, ABORTION AND THE POLITICS OF
MOTHERHOOD 158–91 (1984); J.M. Balkin, The Constitution of Status, 106 YALE L.J. 2313
(1997) (gay rights and affirmative action); Kahan, supra note 99 (gun control and death penalty).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

863

As the Court’s analysis in Scott illustrates, someone called upon to
evaluate a high-speed car chase must individuate and compare competing risks. Which course of action by the police creates greater risk
for the public — pursuing a driver who refuses to stop or letting him
(and others in his situation) go? What sorts of risk are worse — the
continuous but indeterminately lethal ones to the public and the police
associated with maintaining pursuit of a fleeing driver, or the sudden
and more obviously lethal one to the driver associated with deliberately inducing his car to crash to a halt? Aleph and bet subjects, we
surmised, would form competing perceptions and evaluations relating
to these matters.
Aleph subjects, we hypothesized, would assess the risks in a manner consistent with that of the Scott majority. Alephs morally disapprove of challenges to lawful authority and defiance of dominant
norms. These moral sensibilities make it congenial for them to believe,
among other things, that illicit drug trafficking causes social harm and
that excessive gun control renders law-abiding citizens vulnerable to
predation.104 The same cultural and political sensibilities would likely
trigger negative emotions, including fear of and resentment toward
Harris as a symbol of deviance and law-breaking. People who hold
individualistic and hierarchic worldviews and associated political commitments also tend to approve of highly punitive responses to lawbreaking, and as a result, to believe that such measures (including
capital punishment) reduce the incidence of dangerous or harmful behavior.105 This shared disposition, we surmised, would move subjects
to approve of both the decision to pursue Harris at the outset and the
decision to use deadly force to terminate the chase as risk-reducing on
balance.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Like any scheme that purports to treat “culture” as an explanatory variable — particularly one
amenable to measurement — ours necessarily reflects a highly simplified account of how shared
values, experiences, and patterns of living connect people. Indeed, in referring to aleph and bet as
cultural styles, we mean to emphasize the relatively modest depth of the affinities that connect
individuals who adhere to them. As is clearly implied in Gusfield’s account of status collectivities, aleph and bet comprise subgroups whose members, by virtue of more intimate ties, obviously
possess distinct “cultures” in an even deeper sense. We recognize, too, that there will be some issues that divide the subcommunities that tend to share either the aleph or bet styles. See, e.g.,
Kahan, Braman, Gastil, Slovic & Mertz, supra note 59, at 494–95 (discussing variation among
white and African American egalitarians on issues involving sexual mores). Because the aleph
and bet styles are salient and do figure in a familiar range of political conflicts, we see utility, both
methodological and practical, in incorporating them into our analysis. But we most certainly do
not mean, in adopting this scheme, to make the patently absurd claim that American society consists of just two “cultures” or “cultural types.”
104 See Dan M. Kahan & Donald Braman, Cultural Cognition and Public Policy, 24 YALE L. &
POL’Y REV. 149, 156–57 (2006) (describing theory of cultural cognition); see also Kahan, supra
note 19, at 131–36.
105 Cf. Kahan, supra note 19, at 122–23.

864

HARVARD LAW REVIEW

[Vol. 122:837

We hypothesized that bet subjects would form the opposite set of
reactions. Their egalitarian worldviews and left-leaning political sensibilities can be expected to incline bets to condemn authority figures
for abuses of power much more readily than they condemn putative
deviants for defying authority.106 Reinforced by their communitarian
orientation, these same dispositions tend to make bet persons supportive of social welfare programs.107 Adopting factual beliefs congenial
to these dispositions, such individuals tend to think that capital punishment and other punitive measures are not effective deterrents, and
that permitting individuals to arm themselves with guns for selfprotection, rather than making society safer, increases the risk of crime
and accidents.108 In the same vein, we predicted that individuals who
adhere to the bet style, when they viewed the Scott tape, would be angrier at the police, as symbols of overreaching authority figures indifferent to the danger their use of coercion posed to the well-being of bystanders, not to mention Harris. As a result, they would form the
judgment that the decisions to chase Harris and to use deadly force to
halt his flight did not reduce the net risk to society.
D. Results
We present the results of our study in two steps. First, we offer
preliminary analyses of the reactions of our subjects, overall and
across different groups, to the Scott tape. Second, we use statistical
simulations to explore more systematically how individuals bearing
combinations of characteristics that endow them with identities consistent with the aleph and bet cultural styles would view the facts the
tape reveals.
1. Preliminary Analyses: Main Effects and Group Differences. — A
preliminary examination of the data reveals two conclusions. The first
is that a relatively large majority formed perceptions of the Scott tape
consistent with that of the Supreme Court majority. The second is
that there are nevertheless marked differences in perceptions across
identifiable subgroups.109
As reflected in Figure 2 and Figure 3, overall perceptions of the key
risk issues are highly consistent with those of the Court in Scott. Solid
majorities agreed either “strongly” or “moderately” that Harris’s driv–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
106
107
108
109

See sources cited supra note 66.
See Kahan et al., Risk and Culture Study, supra note 62, at 15.
See Kahan, supra note 19, at 134–36.
To correct for underrepresentation of African Americans in the sample — and thus make
the results more reflective of the likely views of the American public at large — the sample-wide
summary statistics reported in Figure 2 through Figure 4 use population-weighted data. The
group summary statistics reported in Table 1 use unweighted data, as do the multivariate regression analyses reflected in Table 2 and the statistical simulations presented infra section II.D.2.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

865

ing posed a deadly risk to the public, and additional subjects “slightly”
agreed with those statements. The subjects were more equivocal
about the decision of the police to pursue Harris; 45% agreed (28% either “strongly” or “moderately”) that the chase was not worth the risk
it posed to the public, and another 13% only “slightly” agreed that the
chase was worth the risk. Still, 74% of the subjects judged Harris to
be either “much more” or “slightly more” at fault for the risk to the
public.110
Figure 2. Overall Perceptions of Risk

100%

14%

Strongly Agree
Moderately Agree

75%

38%

14%
52%

Slightly Agree
Slightly Disagree
Moderately Disagree

17%

Strongly Disagree

50%

23%

13%
21%
17%

19%

25%

15%
26%
6%

0%
Chase Not
Worth Risk

4%

9%

4%

6%
5%

Deadly Risk to Deadly Risk to
Public
Police

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
110 All percentages reported in the Figures and Tables and discussed in the text are rounded to
the nearest percent.

866

HARVARD LAW REVIEW

[Vol. 122:837

Figure 3. Overall Perceptions of Fault
75%
60%

50%

25%

19%
14%
5%

3%

0%
Harris Much
Harris
About Equal
Police
Police Much
More
Slightly More
Slightly More
More

On the ultimate question (Figure 4), approximately 75% agreed and
26% disagreed that the use of deadly force was warranted. Reflecting
some equivocation, however, nearly a quarter agreed or disagreed only
slightly.
Figure 4. Deadly Force Was Justified
in Response to Risk Harris Posed
40%

37%

30%
23%

20%
15%
9%

10%

8%

9%

0%
Strongly Moderately
Agree
Agree

Slightly
Agree

Slightly Moderately Strongly
Disagree
Disagree Disagree

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

867

Table 1 illustrates the mass of individual variation tucked into the
main effects illustrated in Figures 2 through 4. That table reports
group means on the individual response items and on the composite
“Agree with Court” scale, which reflects the average of the response to
all of the items.111 As predicted, African Americans took a significantly more pro-plaintiff stance across all items. So did Democrats
relative to Republicans, liberals relative to conservatives, and Egalitarians relative to Hierarchs. Communitarians were significantly more
pro-plaintiff than Individualists for every item except risk to the public. Women were also generally more pro-plaintiff, although statistically the difference between the sexes was only marginally significant
(p = 0.07 for “Agree with Court”; p = 0.09 for “Deadly Force Justified”).
Table 1 also reveals some additional sources of variation. Lowerincome subjects were consistently more pro-plaintiff than were higher
income ones. Nevertheless, less educated subjects were overall more
pro-defendant than were more educated subjects. So, unexpectedly,
were married subjects.
Older subjects were more inclined than younger ones to view the
chase as not worth the risk it imposed on the public and the use of
deadly force as not justified by the risk Harris posed. However, contrary to Justice Stevens’s hypothesis, elderly subjects did not perceive
Harris’s driving to be less risky to either the public or the police.
One might have surmised that urban dwellers would react differently from non-urban ones. But our results detected no such effect.
The impact of various individual characteristics relative to each
other is reflected in the ordered logistic regression analyses reported in
Table 2. A multivariate regression model shows how much variance in
any explanatory or “independent” variable (here race, gender, cultural
worldview, and so forth) affects a quantity of interest (in our case, subjects’ answers to our response measures) when the impact of every
other independent variable in the model is held constant.112 The models in Table 2 demonstrate that being African American (as opposed to
white) exerts the largest effect across the various response measures.
How hierarchical or egalitarian subjects’ worldviews are also exerts a
relatively large (and statistically significant) independent effect across
all measures except the perceived risk of Harris’s driving to the public.
Being from the Northeast likewise had a relatively large (and statistically significant) effect across all but two of the measures. Income had
a significant but relatively small effect on three of the five measures as
well.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
111
112

See supra p. 858.
See JACOB COHEN ET AL., APPLIED MULTIPLE REGRESSION/CORRELATION ANALYSIS FOR THE BEHAVIORAL SCIENCES 6–7 (3d ed. 2003).

868

HARVARD LAW REVIEW

[Vol. 122:837

Table 1. Mean Responses
Agree
with
Court
Scale

1–6

Relative

Chase Not

Culpability

Worth Risk

1 (police) to

Harris Posed Lethal Risk
to Public

to Police

Deadly
Force
Justified

Å 1 (strongly disagree) to 6 (strongly agree) Æ

5 (Harris)
Female
Male
Difference
Black
White
Difference
< $35,000/yr
≥ $50,000/yr
Difference
≤ HS degree
≥ BA degree
Difference
18-36 yrs old
> 53 yrs old
Difference
Unmarried
Married
Difference
Urban
Nonurban
Difference
Northeast
South or West
Difference
Democrats
Republicans
Difference
Liberals
Conservatives
Difference
Egalitarians
Hierarchs
Difference
Communitarians
Individualists
Difference

4.37
4.48
0.11
3.70
4.50
0.80
4.23
4.52
0.29
4.50
4.24
0.26
4.44
4.28
0.16
4.24
4.58
0.34
4.44
4.41
0.03
4.23
4.51
0.28
4.16
4.72
0.56
4.13
4.67
0.54
4.21
4.62
0.41
4.28
4.56
0.28

4.18
4.28
0.10
3.57
4.32
0.75
4.05
4.30
0.25
4.30
4.07
0.23
4.19
4.17
0.02
4.10
4.34
0.24
4.23
4.23
0.00
4.13
4.28
0.15
4.00
4.51
0.51
4.01
4.44
0.43
4.05
4.40
0.35
4.09
4.37
0.28

3.25
3.09
0.16
3.87
3.04
0.83
3.35
3.09
0.26
3.17
3.33
0.16
2.96
3.46
0.50
3.28
3.08
0.20
3.02
3.20
0.18
3.37
3.04
0.33
3.55
2.70
0.85
3.45
2.87
0.58
3.52
2.84
0.68
3.38
2.96
0.42

4.92
5.04
0.12
4.42
5.03
0.61
4.66
5.15
0.49
5.02
4.89
0.13
4.89
4.84
0.05
4.72
5.20
0.48
4.92
4.99
0.07
4.78
5.02
0.24
4.85
5.14
0.29
4.74
5.15
0.41
4.90
5.05
0.15
4.94
5.02
0.08

4.61
4.64
0.03
3.83
4.69
0.86
4.53
4.69
0.16
4.75
4.44
0.31
4.57
4.58
0.01
4.47
4.77
0.30
4.61
4.63
0.02
4.40
4.72
0.32
4.44
4.84
0.40
4.35
4.85
0.50
4.51
4.74
0.23
4.52
4.74
0.22

Bolded text indicates difference in means of paired groups significant at p ≤ 0.05.

4.37
4.52
0.15
3.51
4.55
1.04
4.25
4.54
0.29
4.61
4.10
0.51
4.50
4.28
0.22
4.22
4.64
0.42
4.48
4.44
0.04
4.20
4.58
0.38
4.07
4.82
0.75
4.01
4.80
0.79
4.14
4.73
0.59
4.23
4.65
0.42

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

869

Table 2. Ordered Logistic Regression Models
for Response Measures

Female
Black (v. White)
Other Minority (v. White)
Age
Household Income
Education
South (v. West)
Midwest (v. West)
Northeast (v. West)
Urban (v. Nonurban)
Married (v. Unmarried)
Parent (v. No children)
Republican (v. Democrat)
Independent (v. Democrat)
Conservative
Hierarchy
Individualism

Chase Not

Deadly

Deadly

Harris

Worth

Risk to

Risk to

More at

Force

Risk

Public

Police

Fault

Justified

0.03

-0.01

0.09

-0.08

-0.15

(0.10)

(0.11)

(0.10)

(0.12)

(0.10)

0.46

-0.45

-0.60

-0.92

-0.66

(0.22)

(0.22)

(0.22)

(0.22)

(0.22)

-0.03

0.09

0.06

-0.32

0.09

(0.16)

(0.17)

(0.16)

(0.17)

(0.16)

0.01

0.00

0.00

-0.01

0.068

(0.00)

(0.00)

(0.00)

(0.00)

(0.00)

-0.03

0.06

0.02

0.04

0.03

(0.01)

(0.02)

(0.01)

(0.02)

(0.01)

0.03

-0.05

-0.06

-0.02

-0.08

(0.03)

(0.03)

(0.03)

(0.04)

(0.03)

0.08

0.08

-0.10

-0.21

-0.02

(0.13)

(0.14)

(0.14)

(0.16)

(0.14)

0.27

0.06

-0.05

-0.20

-0.21

(0.14)

(0.15)

(0.15)

(0.17)

(0.15)

0.30

-0.17

-0.25

-0.48

-0.33

(0.15)

(0.16)

(0.15)

(0.17)

(0.15)

0.04

0.18

0.13

0.14

0.15

(0.14)

(0.15)

(0.14)

(0.16)

(0.14)

-0.22

0.32

0.16

0.31

0.27

(0.11)

(0.12)

(0.11)

(0.13)

(0.11)

-0.07

0.04

0.15

0.17

-0.01

(0.12)

(0.13)

(0.12)

(0.14)

(0.12)

-0.31

-0.03

-0.04

0.29

-0.01

(0.13)

(0.14)

(0.13)

(0.16)

(0.13)

-0.03

0.00

0.01

0.15

-0.03

(0.31)

(0.33)

(0.33)

(0.38)

(0.33)

0.03

0.09

0.05

0.08

0.05

(0.04)

(0.05)

(0.04)

(0.05)

(0.05)

-0.39

0.10

0.16

0.39

0.46

(0.08)

(0.08)

(0.08)

(0.09)

(0.08)

-0.08

0.04

0.08

0.07

0.07

(0.08)

(0.09)

(0.08)

(0.10)

(0.09)

0.09

0.06

0.04

0.14

0.11

log likelihood

-2296.53

-1731.62

-2049.71

-1393.15

-2060.64

Prob > Chi2

0.00

0.00

0.00

0.00

0.00

2

R (McKelvey/Zavoina)

N = 1,347. Dependent variables are responses to the indicated survey items. Coefficients are
ordered log-odds (ordered logits). Bolded coefficients are significant at p ≤ 0.05. Parentheticals indicate standard errors.

870

HARVARD LAW REVIEW

[Vol. 122:837

Some characteristics of interest had no significant effect in the regression models. These included gender (p = 0.15, for “Force Justified,” for example) and the relative individualistic or communitarian
worldview of subjects (p = 0.45, for “Force Justified”), although those
variables did have the expected signs. This finding, of course, does not
necessarily mean that these characteristics in fact have no influence on
perceptions of the Scott tape. They mean only that these characteristics do not have an influence considered apart from whatever effect
they might be exerting jointly with other characteristics.113
This picture is largely consistent with our hypotheses. But it remains blurry. Differences in means show that characteristics like race,
gender, income, party affiliation, ideology, region of residence, and cultural orientation all tend to matter. But in the real world, people’s
identities are usually not formed with reference only to one or another
of these characteristics but with reference to packages of them that cohere with and reinforce one another in meaningful ways. Far from
eliminating this problem, multivariate regression analysis exaggerates
it by partialing out the joint effects of these characteristics and showing us only what each independently contributes, thus obscuring the
effects they will exert in tandem if they covary.114 Our aleph and bet
cultural styles form packages of related, and typically jointly occurring, characteristics and dispositions. Clarifying the picture of what
individuals holding these recognizable combinations of these characteristics are seeing in the Scott tape, then, requires a more powerful form
of statistical analysis.
2. Clarifying the Data: Statistical Simulations. — (a) Overview.
— Clarify, a statistical application designed by Harvard political scientist Gary King, makes such analysis possible. In conventional regression analysis, the influence of some set of explanatory variables on a
dependent variable is expressed in a mathematical equation, the elements of which (regression coefficients, standard errors, p-values, and
so forth) are reported in a table (such as our Table 2).115 Clarify is intended to generate data analyses that simultaneously extract more information from a multivariate regression and present it more intelligibly.116 Using Clarify, an analyst specifies values for the independent
variables that form a regression model. The application then generates a predicted value for the dependent variable through a statistical
simulation that takes account of the model’s key parameters (including
the standard errors for the regression coefficients). It then repeats that
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
113
114
115
116

See id. at 79.
See id. at 72, 78–79.
See generally id.
See Gary King, Michael Tomz & Jason Wittenberg, Making the Most of Statistical Analyses:
Improving Interpretation and Presentation, 44 AM. J. POL. SCI. 347 (2000).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

871

process. Then it repeats it again. Then it repeats it again and again
and again — as many times as directed by the analyst (typically 1000
times, or enough to approximate the entire probability distribution for
the dependent variable). The resulting array of values for the dependent variable can then be analyzed with techniques that are statistically
equivalent to those used in survey sampling to determine an average
predicted value, plus a precisely calculated margin of error.117
This form of analysis has various advantages over conventional regression analysis. Conventional regression outputs (such as those reflected in Table 2) report the impact and statistical significance of each
predictor or independent variable while controlling for the impact of
every other one.118 As such, these outputs obscure the impact of theoretically relevant combinations of predictors (in our study, the various
characteristics associated with the aleph and bet cultural styles). That
impact can easily be modeled through Clarify simulations, which permit one to compare how competing sets of predictors jointly affect a
quantity of interest (such as responses to our various risk-perception
measures).119 In addition, consistent with the growing movement in
the social sciences to make statistical analysis more accessible,120 the
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
117
118

See id. at 349–51.
The concept of statistical significance is used to characterize a degree of confidence, as reflected in a p-value, that a predictor or independent variable (e.g., length of imprisonment) exerts
an effect on a quantity of interest or dependent variable (e.g., the crime rate). The p-value represents the probability that the difference between any observed effect and zero (or no effect) could
have occurred by chance. Conventionally, an effect is considered statistically significant when p ≤
.05, meaning that the likelihood that one might have observed the measured effect when the predictor in fact has no effect is less than 5%. See generally Jacob Cohen, The Earth Is Round (P <
.05), 49 AM. PSYCHOLOGIST 997 (1994). Accordingly, the “nonsignficant” 0.15 p-value for Female in “Force Justified” indicates that there is a 15% chance that the observed effect of being
female (controlling for all other influences) in reducing support for a pro-defendant outcome occurred by chance.
119 Independent-variable regression coefficients that are not statistically significant considered
in isolation can still have a significant joint effect on the dependent variable when added together.
To capture that effect, nonsignificant predictors can be included, and set to whatever value is desired by the analyst, in a Clarify simulation. See Michael Tomz, Jason Wittenberg & Gary King,
Clarify: Software for Interpreting and Presenting Statistical Results, J. STAT. SOFTWARE, Jan.
2003, at 1, 19 (2003). “This is not problematic because the true quantities of interest are usually
the predicted values, . . . not the coefficients themselves.” Id. The size of the standard error associated with any predictor (i.e., independent variable) included in the simulation will, of course,
affect the precision of the predicted value of the quantity of interest (i.e., the dependent variable).
But because “even coefficients that are not statistically significant can provide important information,” it is more sensible “to focus on the confidence intervals Clarify reports for each quantity it
computes than the standard errors of coefficients.” Id.; see also ANDREW GELMAN & JENNIFER HILL, DATA ANALYSIS USING REGRESSION AND MULTILEVEL/HIERARCHICAL MODELS 42, 69 (2007) (rejecting the position that statistically nonsignificant independent variables
should be excluded from predictive models and identifying circumstances in which their inclusion
“make[s] sense,” id. at 42).
120 See, e.g., Andrew Gelman, Cristian Pasarica & Rahul Dodhia, Let’s Practice What We
Preach: Turning Tables into Graphs, 56 AM. STATISTICIAN 121 (2002). In the legal academy, Lee

872

HARVARD LAW REVIEW

[Vol. 122:837

results of Clarify simulations — predicted values for quantities of interest, subject to specified margins of error — can be graphically displayed in a manner that more clearly illustrates their real-world significance to real-world people than do regression coefficients, which
lack intuitive meaning.121
This form of analysis furnishes a perfect fit for our purposes. We
are interested in the perceptions of subcommunities of people holding
combinations of characteristics associated with the cultural styles
that we have designated aleph and bet. We can form reasonable statistical predictions of those perceptions by setting pertinent characteristics — gender, race, region of residence, political ideology, party affiliation, and cultural worldviews — to appropriate values in Clarify
simulations.
(b) Ron, Linda, Bernie, and Pat. — We used Clarify to simulate
the responses of our four hypothetical jurors, Ron, Linda, Bernie, and
Pat. Ron was endowed with characteristics associated with the aleph
cultural style. He is a white male. He is significantly more hierarchic
and more individualistic (in the top quartile of the population for both
characteristics) than the average American. He is a Republican and
also extremely conservative. He is fairly well-educated (a professional
degree), and his household income is relatively high (over $175,000 per
year). He is from Arizona, or some other “Far West” state. He is 46
years old (the population mean, excluding those under 18). Having no
reason to believe that alephs differ from bets in their family status, we
treated him as being as likely to be married and have children as the
average American.122
We set Linda’s characteristics to those of a recognizable bet. She is
an African-American female. She is a liberal Democrat. She lives in
Pennsylvania, or another Northeastern state. She is significantly more
egalitarian and more communitarian in her cultural outlooks than the
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Epstein and her coauthors have forcefully espoused the use of simulations and other forms of
analysis that make the practical impact of multivariate regressions more accessible. See Lee Epstein, Andrew D. Martin & Matthew M. Schneider, On the Effective Communication of the Results of Empirical Studies, Part I, 59 VAND. L. REV. 1811 (2006); Lee Epstein, Andrew D. Martin
& Christina L. Boyd, On the Effective Communication of the Results of Empirical Studies, Part
II, 60 VAND. L. REV. 801 (2007).
121 See King, Tomz & Wittenberg, supra note 116, at 347 (explaining that Monte Carlo statistical simulations used in Clarify “extract . . . currently overlooked information” that “(1) convey[s]
numerically precise estimates of the quantities of greatest substantive interest, (2) include[s] reasonable assessments of uncertainty about those estimates, and (3) require[s] little specialized
knowledge to understand”); id. at 360.
122 As can be seen in Table 2, being married and having children tend to predict greater agreement with the Court majority. Simulations suggest that, all else being equal, being married with
children, as opposed to unmarried and childless, predicts about a 5% (± 1%) increase in the likelihood of agreement with the Court’s disposition on the “ultimate issue” — whether deadly force
was justified in light of the risk that Harris posed.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

873

population mean (in the top quartile for both). She has an associate’s
degree (or more precisely, some college education short of a bachelor’s
degree) and makes a modest salary (around $25,000 per year) as an occupational therapist’s assistant. We made no assumptions about her
marital or family status.
Bernie also fits the profile for a bet. As a white male, his inclusion
in the analysis helps to show how that particular cultural style is not
confined to African Americans or women. Because white bets tend to
be more egalitarian in their attitudes toward sexual mores than do African American ones,123 we assigned Bernie a more egalitarian worldview (in the top decile of the population) than Linda. He is coded as a
liberal Democrat. As one would expect of a university professor, he is
extremely well educated (he holds a doctorate degree), but makes a
more modest salary (that of the population mean). Like Linda, he
lives in the Northeast. In other respects (including marital status and
parenthood), his characteristics were set to the population mean.
Pat, of course, is an “average American.” All of his/her values (including gender) have been set to the American population mean. Neither aleph nor bet (or perhaps a bit of both), he/she is included as a
heuristic benchmark for assessing the views of the other representative
members of our hypothetical venire.
Using Clarify, we simulated the responses of Ron, Bernie, Linda,
and Pat to the statements used to assess subjects’ perceptions of the
Scott tape. The results appear in Figures 5 to 9 and Tables 3 to 7.124
In the cases of Ron, Linda, and Bernie, the percentages can be interpreted as reflecting either the likelihood that a person with his or her
characteristics would respond to an item in the indicated manner or
the percentage of people with his or her characteristics who would respond that way. For Pat, the percentages can be understood to reflect
either the likelihood that any person picked from the American population would respond as indicated or the percentage of persons in the
general population who would so respond.
Figure 5 and Table 3 furnish a vivid image of the deep dissensus
that exists over whether the police should have engaged in a highspeed chase to apprehend Scott in the first place. For Ron, this is a
no-brainer: approximately three-quarters (76%, ± 2%) of the persons
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
123 See Ted G. Jelen & Clyde Wilcox, Causes and Consequences of Public Attitudes Toward
Abortion: A Review and Research Agenda, 56 POL. RES. Q. 489, 492 (2003) (finding that African
Americans are more opposed to abortion than are whites); Gregory B. Lewis, Black-White Differences in Attitudes Toward Homosexuality and Gay Rights, 67 PUB. OPINION Q. 59, 63 (2003)
(reviewing studies finding African Americans more opposed to homosexuality than whites controlling for other influences); see also Kahan, Braman, Gastil, Slovic & Mertz, supra note 59, at 494–
95 (finding that egalitarianism influences abortion risk perceptions less powerfully among African
Americans than among whites).
124 The margins of error for all estimates reflect a 95% level of confidence.

874

HARVARD LAW REVIEW

[Vol. 122:837

who share his defining characteristics disagree — about two-thirds
(66%, ± 3%) either moderately or strongly — with the proposition that
the chase “wasn’t worth the danger to the public.” Bernie and Linda,
in contrast, generally agree with that same statement: 59% (± 3%) of
the persons who share Linda’s characteristics either strongly or moderately agree the chase wasn’t worth the risk, and another considerable slice (18%, ± 4%) “slightly agree”; 73% (± 3%) of the persons who
share Bernie’s characteristics agree (about half moderately or strongly)
that the chase wasn’t worth it. Pat leans toward Ron but is equivocal:
55% (± 2%) of the members of the general population (according to the
simulation) reject the claim that the chase wasn’t worth the risk to the
public, but the median citizen is only “slightly” inclined toward that
position.
Figure 5. Chase Not Worth the Risk

100%

75%

5%
7%

11%

12%

14%

Strongly Agree
31%

36%

10%

23%

12%

24%

0%
Pat

Strongly Disagree

20%
18%
9%

45%

Ron

Moderately Disagree

22%

21%

19%

25%

Slightly Agree
Slightly Disagree

19%

50%

Moderately Agree

8%

10%

8%

9%

7%

Bernie

Linda

Table 3. Chase Not Worth the Risk
Ron

Pat

Bernie

Linda

Strongly Agree

5% ( ± 2%)

11% ( ± 2%)

31% ( ± 9%)

36% ( ± 12%)

Moderately Agree

7% ( ± 3%)

14% ( ± 2%)

22% ( ± 4%)

23% ( ± 3%)

Slightly Agree

12% ( ± 3%)

19% ( ± 2%)

20% ( ± 3%)

18% ( ± 4%)

Slightly Disagree

10% ( ± 2%)

12% ( ± 2%)

9% ( ± 2%)

8% ( ± 3%)

Moderately Disagree

21% ( ± 3%)

19% ( ± 2%)

10% ( ± 3%)

8% ( ± 4%)

Strongly Disagree

45% (± 10%)

24% ( ± 2%)

9% ( ± 4%)

7% ( ± 4%)

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

875

Figure 6 and Figure 7, along with Table 4 and Table 5, address the
risk that Harris’s driving posed. Much like the majority in Scott,
Americans on average (represented by Pat) are strongly disposed to see
Harris as a lethal menace: over 85% believe (more than 70% either
moderately or strongly) that he posed a risk to the public, and about
80% believe (over 60% either moderately or strongly) that he posed a
deadly risk to the police. People who see the world the way Ron does
hold these beliefs even more decidedly.
Bernie and Linda believe Harris’s driving posed deadly risks too,
but their views are somewhat more equivocal. Fewer than one-half
(44%, ± 4%) of the persons who share Linda’s defining characteristics
believe either moderately or strongly that Harris posed a lethal danger
to the public, and another quarter (23%, ± 3%) of those persons agree
only “slightly” with that proposition. Even fewer (37%, ± 5%) of the
Lindas in American society hold a firm conviction (“moderately” or
“strongly agree”) that Harris posed a lethal danger to the police.
Fewer than half of the persons with Bernie’s characteristics (39%, ±
4%) would have a firm conviction (“moderately” or “strongly agree”)
that Harris put the police at deadly risk. There is a high likelihood
(75%, ± 4%) that someone with his characteristics will agree that Harris’s escapade posed a lethal threat to the public, although only about
half (54%, ± 3%) would agree either “moderately” or “strongly.”
Figure 6. Harris a Lethal Danger to the Public

100%
Strongly Agree
30%

23%

Moderately Agree
Slightly Agree

75%

53%

67%

21%

Slightly Disagree
Moderately Disagree

24%

Strongly Disagree

50%
23%
22%

21%
9%

25%

17%

7%

13%
9%

0%

2% 5%
2%

3%

Ron

Pat

14%

4%

10%

3%

7%

10%

Bernie

Linda

876

HARVARD LAW REVIEW

[Vol. 122:837

Table 4. Harris a Lethal Danger to the Public
Ron

Pat

Bernie

Linda

Strongly Agree

67% ( ± 11%)

53% ( ± 3%)

30% ( ± 10%)

23% ( ± 10%)

Moderately Agree

17% ( ± 5%)

22% ( ± 2%)

24% ( ± 3%)

21% ( ± 4%)

Slightly Agree

9% ( ± 3%)

13% ( ± 2%)

21% ( ± 4%)

23% ( ± 3%)

Slightly Disagree

2% ( ± 1%)

4% ( ± 1%)

7% ( ± 3%)

9% ( ± 3%)

Moderately Disagree

3% ( ± 1%)

5% ( ± 1%)

10% ( ± 4%)

14% ( ± 5%)

Strongly Disagree

2% ( ± 1%)

3% ( ± 1%)

7% ( ± 4%)

10% ( ± 6%)

Figure 7. Harris a Lethal Danger to Police

100%
18%

17%

Strongly Agree
Moderately Agree

37%

46%

75%

Slightly Agree
21%

20%

Slightly Disagree
Moderately Disagree

50%

26%

23%

23%

15%

15%

13%

14%

11%

12%

Bernie

Linda

Strongly Disagree

25%
18%

25%

15%
8%
6%

6%
5%

3%

0%
Ron

4%

Pat

Table 5. Harris a Lethal Danger to Police
Ron

Pat

Bernie

Linda

Strongly Agree

46% ( ± 11%)

37% ( ± 3%)

18% ( ± 7%)

17% ( ± 7%)

Moderately Agree

25% ( ± 3%)

26% ( ± 2%)

21% ( ± 4%)

20% ( ± 5%)

Slightly Agree

15% ( ± 4%)

18% ( ± 2%)

23% ( ± 3%)

23% ( ± 3%)

Slightly Disagree

6% ( ± 2%)

8% ( ± 1%)

15% ( ± 3%)

15% ( ± 4%)

Moderately Disagree

5% ( ± 2%)

6% ( ± 1%)

13% ( ± 5%)

14% ( ± 5%)

Strongly Disagree

3% ( ± 1%)

4% ( ± 1%)

11% ( ± 5%)

12% ( ± 6%)

The simulations also furnish a clear picture of the divisions hidden
within the societal consensus over the “relative culpability” of Harris
and the police (Figure 8 and Table 6). The overwhelming majority of

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

877

the persons who hold Ron’s defining attributes feel Harris was either
“much more” (88%, ± 6%) or “slightly more” (6%, ± 2%) at fault. Pat,
representing the average American, is almost as emphatic in his/her
condemnation: there is a 79% (± 2%) chance, the simulation suggests,
that a member of the general public will see Harris as more at fault,
and only a 5% (± 1 %) chance that he/she will regard the police as
more culpable.
Figure 8. Relative Culpability

100%
Harris Much More at
Fault
Harris Slightly More
at Fault
Equally at Fault

16%
29%
13%

75%
64%
18%
88%

Police Slightly More
at Fault
Police Much More at
Fault

35%

50%
32%

6%

0%

12%

15%

25%

1%

8%

15%
5% 2%
1%

Ron

24%
13%

3%

Pat

Bernie

Linda

Table 6. Relative Culpability
Ron

Pat

Bernie

Linda

Harris Much More at Fault

88% ( ± 6%)

64% ( ± 3%)

29% ( ± 11%)

16% ( ± 8%)

Harris Slightly More at Fault

6% ( ± 2%)

15% ( ± 2%)

18% ( ± 3%)

13% ( ± 4%)

Equally at Fault

5% ( ± 2%)

15% ( ± 2%)

32% ( ± 5%)

35% ( ± 5%)

Police Slightly More at Fault

1% ( ± < 1%)

2% ( ± 1%)

8% ( ± 4%)

12% ( ± 4%)

Police Much More at Fault

1% ( ± < 1%)

3% ( ± 1%)

13% ( ± 6%)

24% ( ± 11%)

Bernie and Linda, though, see matters differently. About one-half
(53%, ± 5%) of the people who share Bernie’s defining characteristics
will say either that the parties were equally at fault or the police more
at fault. At most a third (29%, ± 4%) of the persons who share
Linda’s characteristics will say that Harris was more at fault; a person
who shares her characteristics is at least five times more likely (36%, ±
4%) to say that the police were more at fault than is an average member of the general population (5%, ± 1%).
Bernie and Linda also don’t agree with the Scott majority on the
ultimate issue (Figure 9 and Table 7). Over three-fifths (65%, ± 2%) of

878

HARVARD LAW REVIEW

[Vol. 122:837

the persons who share Linda’s characteristics disagree — about onehalf either strongly or moderately — with the statement that “[t]he
danger that Harris’s driving posed to the police and the public justified Officer Scott’s decision to end the chase in a way that put Harris’s
own life in danger.” Nearly three-fifths (58%, ± 2%) of the persons
who hold Bernie’s characteristics are also likely to believe that deadly
force was unreasonable.
Figure 9. Deadly Force Termination Justified

100%
11%
37%

75%

16%

55%

9%

Strongly Agree

13%

Moderately Agree
Slightly Agree

14%
15%
11%

50%

27%

Slightly Disagree
Moderately Disagree
Strongly Disagree

12%
21%
20%

23%

14%

25%

7%

9%
4%

0%
Ron

4%

8%

4%

7%

Pat

26%

Bernie

33%

Linda

Table 7. Deadly Force Termination Justified
Bernie

Linda

Strongly Agree

55% ( ± 11%)

Ron

37% ( ± 3%)

Pat

11% ( ± 5%)

9% ( ± 4%)

Moderately Agree

23% ( ± 4%)

27% ( ± 2%)

16% ( ± 5%)

13% ( ± 5%)

Slightly Agree

9% ( ± 3%)

14% ( ± 2%)

15% ( ± 3%)

14% ( ± 3%)

Slightly Disagree

4% ( ± 2%)

7% ( ± 1%)

12% ( ± 2%)

11% ( ± 2%)

Moderately Disagree

4% ( ± 2%)

8% ( ± 1%)

20% ( ± 4%)

21% ( ± 4%)

Strongly Disagree

4% ( ± 1%)

7% ( ± 1%)

26% ( ± 9%)

33% ( ± 11%)

Pat does agree with the Scott majority, although not without a bit
of equivocation. There is a 64% (± 2%) chance that a person drawn
randomly from the population would either moderately or strongly
agree that the police were justified in using deadly force. There is,
however, a 14% (± 2%) chance that he/she would be only “slightly” inclined to agree, and over a 20% chance that he/she would conclude
upon watching the tape that use of deadly force was unreasonable.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

879

Ron again is emphatic. Over 80% of the individuals who share his
characteristics would find the police acted reasonably.
E. Summary and Discussion
The results of our study strongly confirmed our hypotheses. With
the exception of whether the police should have initiated a high-speed
chase — a matter on which subjects sharply disagreed — reactions to
the Scott tape reflect constrained dissensus. A very sizable majority of
our diverse, nationally representative sample agreed with the Scott
majority that Harris’s driving exposed the public and the police to lethal risks, that Harris was more at fault than the police for putting the
public in danger, and that deadly force ultimately was reasonable to
terminate the chase. However, dissent from these perceptions and
evaluations was not random; the minority of subjects who disagreed
about the appropriateness of deadly force were connected by a core of
identity-defining characteristics. Indeed, so too were members of a
minority who formed a view of the facts most unequivocally in line
with those of the Scott majority.
The relationship between these perceptions, on the one hand, and
the relevant group identities, on the other, fits our hypothesis that reactions to the Scott tape would be shaped by various sources of valuemotivated cognition. As we predicted, there were sharp differences of
perception among persons bearing characteristics and commitments
typical of two recognizable cultural styles. Individuals (particularly
white males) who hold hierarchical and individualist cultural worldviews, who are politically conservative, who are affluent, and who reside in the West were likely to form significantly more pro-defendant
risk perceptions. Individuals who hold egalitarian and communitarian
views, whose politics are liberal, who are well educated but likely less
affluent, and whose ranks include disproportionately more African
Americans and women, in contrast, were significantly more likely to
form pro-plaintiff views and to reject the conclusion that the police
acted reasonably in using deadly force to terminate the chase. The
conspicuous competition between these recognizable cultural styles (or
“status collectivities”125) on issues ranging from gun control to climate
change, from abortion to the death penalty, attests to the power of the
images reflected in the Scott tape to provoke perceptions protective of
observers’ identities.126
The results of the study also make clear that this form of identityprotective cognition operates unevenly across the types of risk percep–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
125
126

GUSFIELD, supra note 17, at 21 (emphasis omitted).
See supra p. 862 & note 103 (describing relationship between cultural styles and deeper
forms of cultural affinity).

880

HARVARD LAW REVIEW

[Vol. 122:837

tions and evaluations that the Court’s own analysis in Scott reflects.
Persons subscribing to the bet cultural style disagreed with those subscribing to the aleph style about the risk posed by Harris’s driving, but
they even more strongly disagreed about the apportionment of fault
between Harris and the police for creating that risk. The latter assessment, likely combined with a similar discrepancy between these
groups on whether the chase was worth the risk to the public to begin
with, is apparently what explains the disagreement over whether
deadly force was justified. The stake that those adhering to these
styles have in protecting their respective identities, then, impels them
into disagreement most strongly on whether the behavior of the police
was risk-reducing or risk-enhancing on net, likely because that “fact”
has the closest connection to whether we should view those in authority with trust or suspicion.
As we have discussed, the Court’s opinion admits of some ambiguity on exactly which issues the video was deemed dispositive.127 Were
a case like Scott to be submitted to a jury, of course, the jury would be
called upon to decide (in the form of a general verdict) all the issues
that the Court identified as decisive to its analysis. That is, in considering whether the use of deadly force to terminate the chase was reasonable, the jury would be required not only to gauge the degree of
risk that the fleeing driver’s behavior imposed on the public and the
police, but also to assess the “relative culpability” of the fleeing driver
and the pursuing police officer for creating that risk.128
As a result of the Court’s decision in Scott, though, in no case will
a jury be permitted to decide any of those issues. The Court’s decision
effectively determined that, regardless of whatever other evidence
might be presented in the case and whatever might transpire in the
course of jury deliberations, there could be no room for “reasonable”
disagreement on either the magnitude of the risks involved in the case
or the role of the police in reducing or exacerbating those risks.
Our analysis suggests that this conclusion cannot be based on the
ground that in fact no identifiable group of people would disagree with
the Court on these matters after watching the Scott tape. The Court’s
decision can be justified only if the members of that group — those
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
127
128

See supra pp. 856–57.
See, e.g., COMM. ON PATTERN JURY INSTRUCTIONS, DIST. JUDGES ASS’ N, ELEVENTH
CIRCUIT, PATTERN JURY INSTRUCTIONS (CIVIL CASES) 192–93 (2005) (“The third aspect of
the Plaintiff’s [§ 1983, Fourth Amendment] claim is that excessive force was used by the Defendants in effecting the Plaintiff’s arrest. . . . Whether a specific use of force is excessive or unreasonable turns on factors such as the severity of the crime, whether the suspect poses an immediate
violent threat to others, and whether the suspect is resisting or fleeing. You must decide whether
the force used in making an arrest was excessive or unreasonable on the basis of that degree of
force that a reasonable and prudent law enforcement officer would have applied in making the
arrest under the same circumstances disclosed in this case.”).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

881

who do see something different from the Court majority — are necessarily “unreasonable.” We consider next whether such a conclusion
can possibly be an appropriate one for the law to make.
III. EVALUATING SCOTT
Just as the Scott tape doesn’t “speak for itself” in a way that generates a single, indisputably correct answer to the factual issues posed by
the case, so our study results don’t “speak for themselves” in a way
that generates a single and indisputably correct answer to whether
Scott was correctly decided. Our results show that a substantial majority of the American public would likely see the key facts in the
manner the Supreme Court majority did. One could argue that this
finding supports the Court’s conclusion that summary judgment was
warranted. We would not, because we think the shared values and
other defining characteristics of the citizens who would see things differently make it inappropriate to dismiss their minority perspective as
“unreasonable.” But we realize full well that this position demands a
reasoned defense, which we endeavor to supply in this Part. We also
identify alternative grounds on which the Court could have reached
the same result in Scott without insisting that the videotape supported
only one “reasonable” view of the facts.
That insistence, we argue, is the only thing that is manifestly wrong
about the decision. The Court’s failure to recognize the culturally partial view of social reality that its conclusion embodies is symptomatic
of a kind of cognitive bias that is endemic to legal and political decisionmaking and that needlessly magnifies cultural conflict over and
discontent with the law.
A. Dissensus, Deliberation, and Legitimacy
Our study suggests that a fairly sizable majority of Americans
would agree after viewing the Scott tape that Harris’s driving created
lethal risks to the public that warranted the police using deadly force
to terminate the chase. Such a finding is a necessary condition for
concluding that the Court was correct to hold that the case should be
decided summarily, but is it sufficient to justify that outcome?
To answer the question requires a theoretical understanding of the
properties of consensus that justify dispensing with civil jury decisionmaking.129 How large, how intense, and how uniform across
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
129 We mean to address this issue not just as it relates to summary judgment but also to judgment for directed verdict and judgment notwithstanding the verdict as well. Although variously
worded under the Federal Rules, compare FED. R. CIV. P. 56(c) (standard for summary judgment),
with FED. R. CIV. P. 50(a)(1) (standard for judgment as a matter of law), “the inquiry [for all these
dispositions] is the same: whether the evidence presents a sufficient disagreement to require sub-

882

HARVARD LAW REVIEW

[Vol. 122:837

groups must such agreement be for a court to say there is no point in
obliging the large majority that sees things one way to deliberate with
an identifiable minority that sees them differently? We will offer a
modest theoretical framework for answering that question — modest
not necessarily because we have only a modest view of our own ability
as theorists, but rather because we think nothing more than midlevel
theorizing is needed to show that the necessary properties of consensus
were lacking in Scott.130
Presumably, a judge in a case like Scott should not base her decision to decide summarily on whether she thinks deliberation would increase the likelihood of a correct verdict enough to justify the cost of a
jury trial. The reason is that verdict accuracy at a reasonable cost is
not the point of jury decisionmaking. If accuracy were the goal, it
seems unlikely the law would use ordinary citizens to decide facts in a
one-off fashion; it would almost certainly rely instead on professionals,
whose expertise in “getting it right” would reflect both prior training
and experience from repeated decisionmaking. Or in other words, the
law would presumably use judges as factfinders.131 This is not to say
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
mission to a jury or whether it is so one-sided that one party must prevail as a matter of law.”
Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 251–52 (1986); see also 9B CHARLES WRIGHT &
ARTHUR MILLER, FEDERAL PRACTICE AND PROCEDURE § 2532 (3d ed. 2008). Nothing in
our discussion, however, bears on summary disposition under Federal Rule 12(b)(6) or its statelaw equivalents, which require a case to be dismissed for “failure to state a claim,” a disposition
that goes to the adequacy of a plaintiff’s legal theory, not the strength of the plaintiff’s factual
evidence.
130 Our theory is modest, too, in the sense that we do not take on the task of justifying the jury
system. There is a rich literature on this subject, many contributions to which we draw on in the
discussion that follows. The Court in Scott, of course, did not premise its decision on a radical
repudiation of the jury and its conventional justifications. Nevertheless, we can easily imagine
that the Court was motivated to overstate the conclusiveness of the video on the key facts in the
case by an unstated resistance to jury trials generally. See Patricia M. Wald, Summary Judgment
at Sixty, 76 TEX. L. REV. 1897, 1907–14 (1998) (linking evolving Supreme Court summary judgment standards to growing acceptance of limiting access to jury trials); see also John Bronsteen,
Against Summary Judgment, 75 GEO. WASH. L. REV. 522 (2007) (noting and criticizing the same
trend). But if so, the Court could have achieved the same result, without duplicity and without
the cost to legitimacy we attribute to its reasoning, if it had based its decision on the grounds we
discuss in section III.B.
131 The case for professionalization of factfinding has a long pedigree and is well developed.
See, e.g., ERWIN GRISWOLD, 1962–1963 HARVARD LAW SCHOOL DEAN’S REPORT 5–6 (“Why
should anyone think that 12 persons brought in from the street, selected in various ways, for their
lack of general ability, should have any special capacity for deciding controversies between persons?”), quoted in HARRY KALVEN, JR. & HANS ZEISEL, THE AMERICAN JURY 5 (1966). See
generally JOHN H. LANGBEIN, THE ORIGINS OF ADVERSARY CRIMINAL TRIAL 318–31
(2003) (documenting the historical antagonism between truth-seeking and the growing centrality
of the jury in an adversary system of adjudication). Although the comparison presents methodological challenges, the weight of the evidence seems to suggest that judges generally render more
accurate verdicts than juries. See, e.g., Bruce D. Spencer, Estimating the Accuracy of Jury Verdicts, 4 J. EMPIRICAL LEGAL STUD. 305 (2007) (surveying past studies and using innovative statistical methods to determine relative error rates of judges and juries based on rates of disagree-

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

883

that a theory of when jury decisionmaking is worth the bother should
be indifferent to accuracy, but it is to say such a theory must also be
sensitive to other properties that a jury is uniquely or specially fitted to
add to a (reasonably) accurate verdict.
Those qualities are familiar and are reflected in standard accounts
of the benefits of jury decisionmaking. One is the simple ordinariness
of jurors’ perspectives.132 Facts “speak for themselves” only against
the background of preexisting understandings of social reality that invest those facts with meaning. Those understandings come from experiences and social influences that vary across groups of persons in
systematic ways. In particular, we can expect ordinary citizens to form
different understandings of social reality from judges or other professional factfinders precisely because the process of legal and judicial
professionalization involves experiences and social influences alien to
the rest of society.133 Maybe those special experiences and influences
make the judges “smarter” and better equipped to give facts their
proper meaning. But since the judgments of fact on which the law
bases its commands are ones that govern the lives of ordinary citizens,
the law would face a fairly obvious difficulty if its view of the “facts”
didn’t take ordinary citizens’ understandings of reality into account.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
ment). Some dispute this conclusion. See, e.g., Robert J. MacCoun, Epistemological Dilemmas in
the Assessment of Legal Decisionmaking, 23 LAW & HUM. BEHAV. 723, 726 (1999). However, the
most common reply is to demur; the exclusive focus on accuracy fails to account for the myriad
other political benefits associated with giving ordinary citizens a conspicuous and (arguably)
meaningful voice in the administration of justice. See, e.g., VALERIE P. HANS & NEIL VIDMAR,
JUDGING THE JURY 249 (1986) (arguing that the “political functions of the jury” deserve just as
much consideration as its “fact-finding functions . . . in judging the jury’s role in society”); Arie M.
Rubenstein, Verdicts of Conscience: Nullification and the Modern Jury Trial, 106 COLUM. L.
REV. 959, 983 (2006) (“[T]he jury is not at its core a mechanism for seeking truth; it is a tool for
injecting democracy into the judicial process . . . .”); see also Robert J. MacCoun & Tom R. Tyler,
The Basis of Citizens’ Perceptions of the Criminal Jury: Procedural Fairness, Accuracy, and Efficiency, 12 LAW & HUM. BEHAV. 333 (1988) (presenting experimental evidence that ordinary citizens prefer the jury on grounds of fairness to other modes of adjudication that appear to surpass
it in accuracy).
132 See, e.g., JEFFREY ABRAMSON, WE, THE JURY: THE JURY SYSTEM AND THE IDEAL
OF DEMOCRACY 18 (Harvard Univ. Press 2000) (1994) (“[L]ocal knowledge . . . qualifies the juror[s] to understand the facts of the case and to pass judgment in ways that a stranger . . . could
not. . . . [T]hey know the conscience of the community and can apply the law in ways that resonate with the community’s moral values and common sense.”).
133 See, e.g., Maher v. People, 10 Mich. 212, 222 (1862) (“[J]urors, from the mode of their selection, coming from the various classes and occupations of society, and conversant with the practical affairs of life, are, in my opinion, much better qualified to judge of the sufficiency and tendency of a given provocation, and much more likely to fix, with some degree of accuracy, the
standard of what constitutes the average of ordinary human nature, than the Judge whose habits
and course of life give him much less experience of the workings of passion in the actual conflicts
of life.”); HANS & VIDMAR, supra note 131, at 248 (“[T]he overwhelming majority of judges are
still white males who come from a privileged sector of our society. Often their views of the world
reflect their backgrounds. Some rather rigidly adhere to a narrow perspective of justice and fairness that is not consistent with that of the general community.”).

884

HARVARD LAW REVIEW

[Vol. 122:837

That difficulty would be of various forms, all of which relate, essentially, to legitimacy. “Legitimacy” in a descriptive sense refers to
the political acceptability of law — its power to command voluntary
compliance.134 Citizens would be unlikely to assent to legal determinations that seem to reflect inaccurate judgments of fact — inaccurate
because of their lack of correspondence to ordinary citizens’, as opposed to specialized professionals’, understandings of how the world
works.135
“Legitimacy” in its normative sense refers to qualities that make the
law morally worthy of assent.136 Here too the jury plays a critical role.
Broadly speaking, law has democratic legitimacy when the process of
its formation is sufficiently connected to (determined by, solicitous of,
respectful toward) the will of those who are governed by it that we can
impute the law’s commands to them.137 Jury factfinding is such a
process: the understanding of the facts reflected in the law can be morally imputed to those governed by the law when the law uses a factfinding process that is informed by their view of social reality.
Understandings of social reality vary, of course, not only between
judges and citizens, but also across citizens of diverse experiences and
social identities. The diversity of citizens magnifies the contribution
that jury factfinding makes to legitimacy in all these various senses.
Just as citizens would be unlikely to assent to verdicts rendered by
professionals (say, judges) whose understandings of social reality were
alien to theirs, so diverse citizens would be unlikely to assent to verdicts rendered only by other citizens whose understandings of social
reality were alien to theirs.138 By affording a factfinding role to citizens from diverse subcommunities, whose understandings of reality re–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
134 This is the positive conception of legitimacy associated with the social sciences. See generally C.K. Ansell, Legitimacy: Political, in INTERNATIONAL ENCYCLOPEDIA OF THE SOCIAL
& BEHAVIORAL SCIENCES 8704, 8704 (Neil J. Smelser & Paul B. Baltes eds., 2001).
135 See HANS & VIDMAR, supra note 131, at 248–49 (arguing that the public will accept verdicts more readily when rendered by jury than by judge, particularly in a controversial matter);
cf. Laurence H. Tribe, Trial By Mathematics: Precision and Ritual in the Legal Process, 84 HARV.
L. REV. 1329, 1376 (1971) (modes of adjudication that “fail[] to penetrate or convince the untutored contemporary intuition threaten to make the legal system seem even more alien and inhuman than it already does to distressingly many”).
136 See Ansell, supra note 134, at 8704 (describing the normative conception of “legitimacy” of
concern to political philosophers).
137 This is the unifying theme of a diverse collection of “social contract” theories of legitimacy.
See generally IAN SHAPIRO, THE MORAL FOUNDATIONS OF POLITICS 109–15 (2003).
138 See, e.g., Batson v. Kentucky, 476 U.S. 79, 87 (1986) (“The harm from discriminatory jury
selection extends beyond that inflicted on the defendant and the excluded juror to touch the entire
community. Selection procedures that purposefully exclude black persons from juries undermine
public confidence in the fairness of our system of justice.”); Kim Taylor-Thompson, Empty
Votes in Jury Deliberations, 113 HARV. L. REV. 1261, 1316–17 (2000) (stressing the role of representative juries in dissipating resistance to verdicts likely to be controversial among discrete
subcommunities).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

885

flect experiences and social influences peculiar to those subcommunities, the jury contributes to the law’s legitimacy in the descriptive
sense.139 To take an obvious example, the Jim Crow exclusion of African Americans from juries140 led to a cynical and dispirited view of legal institutions in minority communities. Having no voice in that
mode of the law’s production, those citizens naturally distrusted the
rules that came from the jury.141
Likewise, by involving diverse citizens in factfinding, the jury contributes to the law’s democratic legitimacy in the moral sense. The
experience of interacting with others whose understandings of social
reality differ from theirs — and thus learning that their own understandings, and hence their views of the facts, are partial — might
cause jurors of diverse identities to converge on a common view of the
facts, particularly where one side’s initial view is less intensely held
than the other’s.142 Such convergence would furnish assurance to the
diverse citizens whom such jurors represent that the law embodies a
view of the facts consistent with their shared experiences and defining
commitments. This assurance would in turn make the expectation of
generalized obedience morally compelling.
But probably even more important, jury deliberation can invest
law with democratic legitimacy even when factual understandings
born of diverse experiences and social influence persist. Necessarily in
that circumstance, subcommunities whose views of the facts are rejected by the law will not be able to see the law as reflecting their understanding of reality in substance. Accordingly, they can be expected
to see the law as theirs in the sense that morally warrants an expectation of assent only if the law arises from a process that shows due respect for their understanding of reality and hence for their identities.143
Jury factfinding is a procedural strategy of that sort. It assures that
those who win in a contest between competing understandings were
obliged to listen — under circumstances geared toward maximizing
the prospects of changes in, and convergences of, perspectives — to
those who have lost. And in so doing, it enables the latter to assent
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
139 See, e.g., Glasser v. United States, 315 U.S. 60, 86 (1942) (“[T]he proper functioning of the
jury system, and, indeed, our democracy itself, requires that the jury be a ‘body truly representative of the community,’ and not the organ of any special group or class.” (quoting Smith v. Texas,
311 U.S. 128, 130 (1940))).
140 See ABRAMSON, supra note 132, at 108–12.
141 See ROBERT A. CARO, THE YEARS OF LYNDON JOHNSON: MASTER OF THE SENATE
946–47 (2002) (describing civil rights leaders’ resistance to jury trial enforcement of the 1957 Civil
Rights Act because of distrust of unrepresentative juries).
142 See ABRAMSON, supra note 132, at 100–01.
143 Cf. HANS & VIDMAR, supra note 131, at 51, 248–49 (explaining that subcommunities aggrieved by controversial verdicts are more likely to see verdicts as fair if the jury contains their
members).

886

HARVARD LAW REVIEW

[Vol. 122:837

without the experience of subjugation and domination that estranges
them from the law.144
This account brings into sharper focus the features that factual
consensus must have before it can justify dispensing with jury deliberation. Precisely because juries can lend legitimacy to law by assuring minorities that their perspective is being respected, it surely isn’t
enough that the facts in a particular case “speak for themselves” for a
large majority. If the minority’s view of the facts reflects the minority’s view of social reality, summary adjudication will deny the minority a basis to accept, or for the majority to demand that it accept, the
law’s view of the facts as its own. Before summary adjudication can
be justified, then, the consensus that attends a particular set of factual
findings must be more than (or simply different from) “large.” It must
also be devoid of any partial understanding of social reality the endorsement of which by the law would alienate or stigmatize an identifiable subcommunity, whose perspective has been excluded from consideration. Or, in a word, it must be mundane.
Most — probably the overwhelming majority — of the cases that
strike judges as presenting “no genuine issue as to any material fact”145
will pass this test. Almost every time a judge declares no such issue to
exist, whether in a complex commercial dispute or a routine slip-andfall suit, she can be confident that some small fraction of potential jurors might well perceive the facts differently; statistical outliers are inevitable. But if these individuals are mere outliers — if they don’t
share experiences and an identity that endow them with a distinctive
view of reality, if the factual perceptions in question don’t arise from
their defining group commitments — summary judgment will not convey the message of exclusion that delegitimizes the law in the eyes of
identifiable subcommunities.
Scott, however, was not a case of that sort. Our data suggest that
the minority who would see things differently from the Court after
watching the tape are not idiosyncratic statistical outliers; they are
members of groups who share a distinctive understanding of social reality that informs their view of the facts. Those facts, moreover, involve a type of police-citizen encounter fraught with competing connotations in our society: a civil liberties case. Perhaps the disclosure of
the experiences and social influences on which the minority’s understanding rests could change the majority’s view of social reality, and
hence its view of the facts.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
144 See generally CHRISTOPHER J. ANDERSON, ANDRÉ BLAIS, SHAUN BOWLER, TODD
DONOVAN & OLA LISTHAUG, LOSERS’ CONSENT: ELECTIONS AND DEMOCRATIC LEGITIMACY (2005).
145 FED. R. CIV. P. 56(c).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

887

But even if we assume — as we think is likely — that such exposure would not change the perceptions of the “vast majority,” it still
does not follow that the Court was right to order that Scott be decided
summarily. For it is exactly when the law is certain to endorse a factual position that aligns it with one contested view of how the world
works rather than another that the process of jury deliberation performs its greatest function in conserving democratic legitimacy in a diverse society. If the law has not only rejected their view of social reality, but has refused even to permit the articulation of it in the process
of the law’s determination of the facts, those who disagree lack any resources for understanding the law as theirs. Indeed, if the law has
adopted procedures designed rigorously to insulate judicial determinations from the minority’s view of reality because a court deems that
view to be one no “reasonable” citizen could possibly hold, members of
that minority cannot understand (or be expected to understand) assent
as anything other than acquiescence in their status as defeated and
subjugated outsiders.
Scott put identifiable subgroups of citizens in exactly that position.
Even though constrained, the nature of the dissensus surrounding the
facts revealed in the tape shows that Americans interpret those facts
against the background of competing subcommunity understandings of
social reality. Under these circumstances, ordering that the case be decided summarily based on the video was wrong precisely because doing so denied a dissenting group of citizens the respect they were owed,
and hence denied the law the legitimacy it needs, when the law adopts
a view of the facts that divides citizens on social, cultural, and political
lines. In so doing, the Scott majority transformed an inevitably partial
view of social reality reflected in law into a needlessly partisan one.
B. How To Defend the Outcome in Scott: Reasons, Not Perceptions
We have suggested that the Court in Scott was wrong to order
summary judgment on the ground that it was entitled to “believe its
own eyes” after watching the tape. Its decision to privilege its view of
a set of facts on which even a minority of persons who share a set of
defining commitments would disagree stigmatizes those citizens as outsiders and in so doing delegitimizes the law.
But nothing we have said in that regard implies, necessarily, that
the result in Scott was incorrect. We can think of at least three plausible alternative grounds for the decision. We don’t know whether in
the end any of them is persuasive (in part because we disagree among
ourselves about the merits of at least one of them). But we do believe
that they all avoid the sort of criticism we have developed of the
Court’s reasoning — or lack thereof — in the case.

888

HARVARD LAW REVIEW

[Vol. 122:837

One such ground would have emphasized the unique approach that
courts often take to factfinding in constitutional settings.146 So-called
“constitutional facts,” unlike those decided in cases presenting no constitutional issue, are often reviewed de novo in the Supreme Court.147
One reason for independent factfinding is to assure adequate enforcement of constitutional guarantees toward which there is majority antagonism that could seep into jury factfinding.148 Another is the importance of uniformity and predictability in constitutional rules.149
This concern is particularly compelling where a court can perceive
that enforcement of a constitutional norm will turn on a type of factual
perception that a discrete subcommunity does not share, for in that
case summary adjudication is necessary to avoid inconsistent verdicts
across jurisdictions and within particular jurisdictions over time.150
These concerns have had a conspicuous influence in Fourth
Amendment jurisprudence. The Supreme Court has warned against
“standards requiring sensitive, case-by-case determinations of government need, lest every discretionary judgment in the field be converted
into an occasion for constitutional review.”151 To avoid that, Fourth
Amendment doctrine is replete with rule-like presumptions of reasonableness for generically defined fact patterns (for example, the arrest
and jailing of suspected misdemeanants without regard to their per–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
146 See generally Henry P. Monaghan, Constitutional Fact Review, 85 COLUM. L. REV. 229,
247–54 (1985) (history of constitutional fact review).
147 Id. at 261–62.
148 See, e.g., Monitor Patriot Co. v. Roy, 401 U.S. 265, 276–77 (1971) (justifying judicial scrutiny
of jury findings in cases involving the First Amendment because the jury itself is “unlikely to be
neutral with respect to the content of speech and holds a real danger of becoming an instrument
for the suppression of those ‘vehement, caustic, and sometimes unpleasantly sharp attacks’ which
must be protected if the guarantees of the First and Fourteenth Amendments are to prevail,” id.
at 277 (citation omitted) (quoting New York Times Co. v. Sullivan, 376 U.S. 254, 270 (1964))).
149 See, e.g., Ornelas v. United States, 517 U.S. 690, 697–98 (1996) (“[D]e novo review [of facts
that bear on Fourth Amendment determinations] tends to unify precedent and will come closer to
providing law enforcement officers with a defined ‘set of rules which, in most instances, makes it
possible to reach a correct determination beforehand as to whether an invasion of privacy is justified in the interest of law enforcement.’” (quoting New York v. Belton, 453 U.S. 454, 458 (1981))).
150 These, in any case, are the conventional accounts of a certain type of aggressive judicial
factfinding in constitutional cases. For an alternative account, see Suzanne B. Goldberg, Constitutional Tipping Points: Civil Rights, Social Change, and Fact-Based Adjudication, 106 COLUM.
L. REV. 1955 (2006). Goldberg argues that courts tend to disguise contestable normative judgments (sometimes ones that support repression of marginalized groups but also sometimes ones
that resist such repression) as unassailable “facts” in order to minimize political resistance to their
decisions. See id. at 1961; see also Alafair S. Burke, Equality, Objectivity, and Neutrality, 103
MICH. L. REV. 1043, 1048–51 (2005) (book review) (describing tension between empirical and
normative strands of Fourth Amendment search jurisprudence); David L. Faigman, “Normative
Constitutional Fact-Finding”: Exploring the Empirical Component of Constitutional Interpretation, 139 U. PA. L. REV. 541, 544–45 (1991) (identifying normative commitments in factfinding).
151 Atwater v. City of Lago Vista, 532 U.S. 318, 347 (2001).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

889

ceived dangerousness;152 “suspicionless” interviews of travelers at bus
terminals;153 the stopping of a vehicle where probable cause exists to
believe a traffic violation has taken place, without regard to officers’
subjective motives for the stop154) that spare police the uncertainty
that would attend minute, case-specific factual inquiries, whether in
the context of pretrial motions to exclude evidence from criminal
prosecutions or of civil actions for damages.155
The Court in Scott could easily have reversed on the basis of a presumption of reasonableness defended in this way. Indeed, its decision
ended by announcing a “rule” — “[a] police officer’s attempt to terminate a dangerous high-speed car chase that threatens the lives of innocent bystanders does not violate the Fourth Amendment, even when it
places the fleeing motorist at risk of serious injury or death”156 — that
going forward could well shield from judicial scrutiny any police chase
of a fleeing motorist.157
To be sure, any decision on this basis could have been criticized.
Every bit as often as the Court has insisted on the importance of
“bright line” rules, it has also recognized the need for flexible standards
that can accommodate the fact sensitivity of Fourth Amendment reasonableness determinations.158 Because such standards do enlarge the
role for juries in civil damages cases, this branch of the doctrine is said
to accord with “[t]he genius of the framers’ [understanding] . . . that
juries of ordinary Americans can sometimes decide which intrusions

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
152
153
154
155

See id. at 354.
See Florida v. Bostick, 501 U.S. 429 (1991).
See Whren v. United States, 517 U.S. 806 (1996).
Cf. Mitchell N. Berman, Constitutional Decision Rules, 90 VA. L. REV. 1, 110–11 (2004) (describing Atwater as a constitutional decision rule).
156 Scott v. Harris, 127 S. Ct. 1769, 1779 (2007).
157 But see id. (Ginsburg, J., concurring) (“I do not read today’s decision as articulating a mechanical, per se rule. The inquiry described by the Court is situation specific.” (citations omitted));
id. at 1780 (Breyer, J., concurring) (“[T]he video makes clear the highly fact-dependent nature of
this constitutional determination.”). See generally pp. 856–57 (describing the Scott opinion’s ambiguity on the rule to be applied in car chase cases and on the identity of the issues subject to determination by the factfinder).
158 See Kathryn R. Urbonya, Rhetorically Reasonable Police Practices: Viewing the Supreme
Court’s Multiple Discourse Paths, 40 AM. CRIM. L. REV. 1387, 1417 (2003) (“In addition to the
Court’s construction of broad, shifting reasonableness propositions, it has also inconsistently characterized the Fourth Amendment as requiring bright-line rules or case-by-case adjudication.”); see
also Robbins v. California, 453 U.S. 420, 443 (1981) (Rehnquist, J., dissenting) (attacking one
Fourth Amendment “bright line” rule against warrantless search of container during lawful search
of interior of car: “Our entire profession is trained to attack ‘bright lines’ the way hounds attack
foxes. Acceptance by the courts of arguments that one thing is the ‘functional equivalent’ of the
other, for example, soon breaks down what might have been a bright line into a blurry impressionistic pattern.”), overruled by United States v. Ross, 456 U.S. 798, 824 (1982) (adopting factsensitive standard for reviewing searches of containers in vehicles subject to search).

890

HARVARD LAW REVIEW

[Vol. 122:837

are so unreasonable” as to violate the Fourth Amendment.159 The
concern that jury factfinding might be corrupted by insufficient community support for the protections the Fourth Amendment guarantees
is, at least sometimes, counteracted by the stake that ordinary citizens
clearly have in “strik[ing] a sensible balance between liberty and order[:] If they unreasonably handcuff the cops, their community will
suffer; and if they allow the cops to handcuff citizens unreasonably,
they are likewise putting themselves at risk.”160 Finally, the prospect
that jury decisionmaking might result in nonuniform verdicts, far from
being decried as a vice, might be thought by some to be a virtue that
perfects democratic rule by giving a persistent dissenting group “temporally or spatially restricted power to express their views.”161
We take no position on the merits of judicial factfinding animated
by concerns for uniform application of constitutional principles. We
note only that had the Court decided Scott on this ground, as it easily
could have, it would have avoided stigmatizing an identifiable subcommunity’s view of social reality as too “unreasonable” to be given
consideration in the administration of justice. No group of citizens’
views of the facts having been treated as privileged by the law, the
members of subcommunities that had the minority view would have
had no reason to see the law as less legitimately binding on them than
the members of the majority, who likewise are expected to assent to
judicial determinations that reflect the special competence of courts as
expositors of the Constitution.162
The same would have been true had the Court in Scott squarely
based reversal on a second basis: the institutional advantage that
courts have in determining the systemic consequences of particular legal rulings. If the law is trying to figure out whether one man
“slapped [another] . . . on the elbow” or “merely touched him,”163 “jurors, from the mode of their selection, coming from the various classes
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
159 Akhil Reed Amar, An Unreasonable View of the 4th Amendment, L.A. TIMES, Apr. 29,
2001, at M1.
160 Id.
161 Heather K. Gerken, Dissenting by Deciding, 57 STAN. L. REV. 1745, 1756 (2005).
162 This is not to say that the minority — like the majority — will not be irritated by courts’
assertion of power. Even today, the (large) minority of the population who disagree with Roe and
Casey protest its rule that courts, rather than criminal juries, weigh the social consequences and
contested facts involved in the law of abortion. That protest, in turn, excites political action. See
Robert Post & Reva Siegel, Roe Rage: Democratic Constitutionalism and Backlash, 42 HARV.
C.R.-C.L. L. REV. 373, 399 (2007) (arguing that dissensus is “frequently expressed in legislation,
which offers countless opportunities for judicial critics to interpose practical obstacles to the realization of constitutional norms advanced by a challenged decision”). Here we invoke only the
conventional argument that when courts exercise their constitutional interpretive power in the
manner conventionally assumed to be appropriate, their decisions are legitimate in the moral
sense.
163 Duncan v. Louisiana, 391 U.S. 145, 147 (1968).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

891

and occupations of society, and conversant with the practical affairs of
life” have an advantage in determining all manner of historical fact
over judges, “whose habits and course of life give [them] much less experience of the workings of passion in the actual conflicts of life.”164
But by the same token, precisely because of their distinctive “habits
and course of life,” not to mention the special “mode of their selection,”
judges are likely to gain certain insights into the workings of the practical affairs of legal institutions. Whereas most citizens, for example,
have only fleeting contact with law enforcement, judges have recurring
occasion to observe police officers and prosecutors at work. They are
thus, arguably, much more cognizant than are individual jurors of how
damages verdicts — even intermittent ones — might change law enforcement behavior. Indeed, far from taking these more abstract and
largely unobserved costs into effect, juries might be riveted by the
vivid consequences of policing gone bad in individual cases, and thus
overestimate the likelihood of such misfortunes generally.165 Even if
one does not believe that these sorts of influences should disqualify
ordinary citizens from central participation in the administration
of justice generally, one might still conclude that they warrant constraining ordinary citizens’ role in deciding certain matters, which for
that reason should be characterized as “matters of law” for courts to
determine.166
Elements of such reasoning do appear in the Court’s opinion in
Scott. Rejecting the claim that the police should have discontinued the
chase as a less drastic alternative to deadly force, the Court stated, “[i]t
is obvious the perverse incentives such a rule would create: Every fleeing motorist would know that escape is within his grasp, if only he accelerates to 90 miles per hour, crosses the double-yellow line a few
times, and runs a few red lights.”167 Because such perverse incentives
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
164
165

Maher v. People, 10 Mich. 212, 222 (1862).
This feature of jury decisionmaking reflects the unique vulnerability of juries to distorting
influences such as vividness, representativeness bias, hindsight bias, and the like. See, e.g., Dorothy K. Kagehiro, Ralph B. Taylor, William S. Laufer & Alan T. Harland, Hindsight Bias and
Third-Party Consentors to Warrantless Police Searches, 15 LAW & HUM. BEHAV. 305 (1991) (analyzing effects of hindsight bias on lay perceptions of third-party consent to warrantless searches);
Michael J. Saks & Robert F. Kidd, Human Information Processing and Adjudication: Trial by
Heuristics, 15 LAW & SOC’Y REV. 123 (1981) (analyzing cognitive biases affecting lay juror interpretation of quantitative evidence); cf. Catherine T. Struve, Constitutional Decision Rules for Juries, 37 COLUM. HUM. RTS. L. REV. 659, 680–81 (2006) (arguing that issues on which Fourth
Amendment excessive force claims turn should be allocated between judge and jury with such
considerations in mind).
166 Cf. Chris Guthrie, Jeffrey J. Rachlinski & Andrew J. Wistrich, Blinking on the Bench: How
Judges Decide Cases, 93 CORNELL L. REV. 1 (2007) (noting that in performing certain tasks
structured to involve legal reasoning, judges appear to display relatively greater resistance to biases than lay persons).
167 Scott v. Harris, 127 S. Ct. 1769, 1779 (2007).

892

HARVARD LAW REVIEW

[Vol. 122:837

could be generated by any jury verdict in favor of the plaintiff in a
case involving a high-speed chase, the Court could have treated this
reasoning as dispositive of the case as a matter of law wholly apart
from whatever facts any “reasonable juror” might have seen in the
tape.
Again, we are not necessarily arguing that this would have been a
persuasive basis for reversing. Indeed, one could argue that judges are
the ones who lack a realistic understanding of the costs and benefits of
various policing techniques,168 particularly in communities whose
members have considerable experience with the coercive incidences of
them.169 Moreover, judges may be subject to capture by the authorities they are supposed to regulate, especially in state systems, where
electoral endorsement by the police and the district attorney tend to be
necessary in judicial elections. It is also a bit odd for members of the
Supreme Court — who encounter the workings of criminal law enforcement from a remote, law-bookish perspective only — to think
they have a better grasp of the consequences of Fourth Amendment
rulings than do individual district court judges, whose dockets are
likely to be dominated by criminal matters. Nevertheless, as a ground
for decision firmly established in familiar understandings of the institutional competence of courts in general, had the Court relied entirely
on the bad consequences of permitting any jury to award damages in a
high-speed police chase, it would have avoided gratuitously insulting
any class of citizens in particular.
Finally, the Court in Scott could have reversed on the ground that
democratic political checks adequately assure the “reasonableness” of
high-speed police chases. Like many other bodies of constitutional
doctrine,170 the Court’s Fourth Amendment jurisprudence is highly
sensitive to the generality of coercive state behavior. In a case like
Tennessee v. Garner — in which the police shot a criminal suspect who
was fleeing them on foot — coercion is concentrated on a single individual whose well-being is likely to be a matter of indifference (at
best)171 to the general public. Popularly accountable officials thus
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
168 See Margaret Raymond, Down on the Corner, Out in the Street: Considering the Character
of the Neighborhood in Evaluating Reasonable Suspicion, 60 OHIO ST. L.J. 99, 137 n.114 (1999)
(suggesting that “availability” heuristic distorts judicial perception of likelihood of accuracy of
police investigatory judgments because courts see “primarily . . . criminal cases in which police
intuition proved accurate”).
169 See generally Dan M. Kahan & Tracey L. Meares, The Coming Crisis of Criminal Procedure, 86 GEO. L.J. 1153 (1998) (defending deference to inner-city communities’ balance of order
and liberty struck by policing techniques that affect them).
170 See generally JOHN HART ELY, DEMOCRACY AND DISTRUST: A THEORY OF JUDICIAL
REVIEW (1980).
171 Although the Supreme Court famously avoided referring to this aspect of the case, the defendant in Garner was an African-American youth, and the challenge to the state common law
authorization of the use of deadly force against nondeadly fleeing felons was based primarily on

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

893

have little incentive to police the police in this context to assure the toll
police behavior exacts on individual liberty is compensated for by its
contribution to public order — the test for “reasonableness” under the
Fourth Amendment.172 Accordingly, courts regulate the exercise of police power in this and like circumstances through strict enforcement of
judicially defined reasonableness tests.
Many other types of law enforcement authority, however, do impose a burden on citizens generally, either directly or indirectly
through their impact on parties whose interests citizens share. Examples include sobriety173 and other types of vehicle checkpoints,174
which burden ordinary citizens as drivers; random drug tests of high
school students,175 which burden ordinary citizens as parents; and
warrantless “administrative” searches of businesses,176 which burden
ordinary citizens as consumers. Precisely because the coercive effects
of such exercises of law enforcement authority are meaningfully visited
on the citizenry at large, their approval by politically accountable actors furnishes compelling evidence that these policies strike a reasonable balance between liberty and order.177 Judicially enforced rules
are thus not necessary in these settings to test whether such exercises
of power are “reasonable” under the Fourth Amendment.178
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
the racially disparate impact of this law enforcement technique. See Tracey Maclin, Race and the
Fourth Amendment, 51 VAND. L. REV. 333, 339–40 (1998).
172 See, e.g., New Jersey v. T.L.O., 469 U.S. 325, 337 (1985) (stating that “the standard of reasonableness” under the Fourth Amendment turns on a “balanc[ing]” of the affected “individual’s
legitimate expectations of privacy and personal security” and the government’s interest in securing “public order”).
173 See Mich. Dep’t of State Police v. Sitz, 496 U.S. 444 (1990).
174 See Illinois v. Lidster, 540 U.S. 419 (2004) (upholding checkpoints aimed at apprehending
hit-and-run drivers).
175 See Bd. of Educ. of Indep. Sch. Dist. No. 92 v. Earls, 536 U.S. 822 (2002) (upholding drug
testing program for students involved in extracurricular activities generally); Vernonia Sch. Dist.
47J v. Acton, 515 U.S. 646, 650 (1995) (upholding drug testing program for high school athletes).
176 See New York v. Burger, 482 U.S. 691 (1987).
177 See Kahan & Meares, supra note 169, at 1173–74.
178 See, e.g., Lidster, 540 U.S. at 421, 426 (explaining no “rule is needed” to assure roadside
checkpoints are “reasonable, [and] hence, constitutional” because “practical considerations —
namely, limited police resources and community hostility to related traffic tieups” — can be expected to constrain the unwarranted use of this technique); Earls, 536 U.S. at 841 (Breyer, J., concurring) (citing as evidence of reasonableness of drug-testing policy that a “democratic, participatory process” in which parents were involved “revealed little, if any, objection”); Acton, 515 U.S. at
665 (citing “unanimous approval” of parents at public meeting as evidence that drug testing does
not unreasonably burden privacy of students).

894

HARVARD LAW REVIEW

[Vol. 122:837

This reasoning could well have been adopted in Scott.179 When
police decide to initiate a high-speed chase — and to persist in it until
they manage to force the suspect to lose control of his vehicle and
crash — they create immense risk for members of the public generally.
Indeed, innocent bystanders are often injured when struck either by a
fleeing motorist or police pursuers.180 Accordingly, even if she is indifferent (or hostile) to the interests of the fleeing suspect, the ordinary
citizen as driver or pedestrian clearly has a stake in the police not resorting to this potentially deadly seizure technique unless doing so generates commensurate benefits for public order. Because the ordinary
citizen has such a stake, politically accountable officials have an incentive to police their own police to make sure they don’t engage in highspeed chases without good reason. Indeed, hundreds of municipalities
— responsive to the opinions of citizens like Bernie and Linda — have
already adopted regulations that place severe restrictions on the use of
high-speed chases and prohibit them from being used to apprehend
drivers who resist being pulled over for minor traffic infractions.181
Once more, we are not necessarily arguing that the Court should
have identified the adequacy of political checks as grounds for summarily deciding Scott. But had it done so, far from sending a message of
exclusion to citizens like Bernie and Linda, it would have been conveying in the most emphatic terms available to a court that the decision
was to be made through a process — the enactment of law by democratically accountable representatives — in which the voices of all must
theoretically be heard.182
C. Cognitive Illiberalism and Judicial Humility
We have come not to bury Scott but rather to identify the lessons
that can be gleaned from it. We have argued that members of the
Court majority were mistaken merely to “trust their own eyes” in
viewing the Scott tape. But the upshot can’t be that judges should
never trust their own perceptions of the facts when determining
whether to resolve cases summarily — instead sending every case, no
matter how flimsy or outright specious the factual disputes are, to ju–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
179 It was in fact a disposition urged on the Court by an amicus group consisting of various
municipal and state governments and officials. See Brief of the National Ass’n of Counties et al.
as Amici Curiae Supporting Petitioner at 9–17, Scott v. Harris, 127 S. Ct. 1769 (2007) (No. 051631), 2006 WL 3693465. This group was represented by one of the authors (Kahan) in his capacity as an instructor in the Yale Law School Supreme Court Clinic.
180 See id. at 12 & n.9.
181 See id. at 15.
182 See generally JEREMY WALDRON, THE DIGNITY OF LEGISLATION (1999) [hereinafter
WALDRON, DIGNITY] (developing theory of how legislation recognizes dignity of the minority at
the same time it recognizes authority of the majority); JEREMY WALDRON, LAW AND DISAGREEMENT 108–13 (1999) (same).

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

895

ries. Scott, we’ve argued, presents certain special features that made it
inappropriate for the Court to treat it as warranting summary decision
even though the case would have presented a “genuine issue [of] material fact”183 only for an admitted minority of citizens. The task, then,
is to try to figure out what sorts of cues judges can look for when they
are trying to distinguish the relatively few cases that possess these features from the vast run of cases, including ones that rely on video and
like forms of demonstrative evidence, that are otherwise fit for summary adjudication. Performing this task, we suggest, requires identifying and taking effective steps to neutralize a set of interlocking socialpsychological dynamics that are likely to distort judicial decisionmaking on factual issues that divide competing cultural and social groups.
The foundation of such distortion is naïve realism.184 Social psychologists use this term to refer to an asymmetry in the ability of most
people to identify the effects of value-motivated cognition. We are
good at detecting when those who disagree with us about matters of
fact are influenced by the congeniality of their beliefs to their defining
group commitments.185 That’s the realism part. The naïve part is
that we are correspondingly poor at identifying how the motivation to
form beliefs congenial to our own group commitments operates in
us.186
In the realm of legal and political life, this tendency has pernicious
consequences. Many important policies turn on issues of disputed fact,
or predictions based on fact: Does the death penalty deter murder?
Does global warming exist, is it caused by humans, and does it pose
significant economic and environmental threats? Does the minimum
wage make working-class people better off or worse off? Will vaccinating school-age girls against HPV cause them to engage in promiscuous, unprotected sex and thus increase the incidence of teen pregnancy and HIV infection? Because they are not generally aware of
their own disposition to form factual beliefs that cohere with their cultural commitments, legislators, policy analysts, and ordinary citizens
manifest little uncertainty about their answers to these questions. But
much worse, because they can see full well the influence that cultural
predispositions have on those who disagree with them, participants in
policy debates often adopt a dismissive and even contemptuous posture toward their opponents’ beliefs. This attitude in turn provokes
resentment on the part of their opponents, who, as a result of naïve realism, bridle at the suggestion that they are conforming their factual
beliefs to their values yet see exactly that sort of process going on in
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
183
184
185
186

FED. R. CIV. P. 56(c).
See Robinson, Keltner, Ward & Ross, supra note 18, at 404–05.
Id. at 405.
Id.

896

HARVARD LAW REVIEW

[Vol. 122:837

the minds of their (annoyingly smug, it seems) antagonists. They naturally proclaim as much — initiating an escalating cycle of recrimination and distrust.187
The result is a state of cognitive illiberalism. The vast majority of
citizens in our society do not desire to impose their values on others.
They accept the basic liberal premise that law and policy should be
confined to attainment of secular goods — security, health, prosperity
— that are fully accessible to persons of all cultural outlooks.188 But
because the factual beliefs they form about the sorts of behavior that
threaten those goods are (subconsciously) motivated by their cultural
appraisals of those activities, such citizens naturally divide into opposing cultural factions on the policies the law should pursue to achieve
their common welfare. Locked into a state of cyclical recrimination,
moreover, members of these factions become perfectly (painfully)
aware of this alignment between competing factual beliefs and opposing cultural identities. In such a climate, challenges to groupdominant beliefs become indistinguishable from indictments of the integrity and competence of the groups’ members. Opposing groups find
themselves engaged in relentless, symbolic status competition — not
over whose partisan view of the good the law will endorse, but over
whose culturally partisan view of the facts it will credit.189
We argue that the decision in Scott reflects and reinforces these dynamics. The Justices in the majority couldn’t literally have perceived
that no one could see the facts on the tape differently from how they
saw them. The evidence that some citizens might was staring them,
literally, in the face: Justice Stevens, who presumably indicated even in
conference that he was not of the view that the case was fit for summary disposition. Even apart from Justice Stevens’s interpretation of
the tape, though, the case was replete with cues that a decision on the
grounds the Court settled on would provoke at least some generalized
societal dissent. It involved a coercive, near-deadly encounter between
police and a citizen, always a potentially divisive matter in our society;
numerous public interest groups had filed briefs in support of the respondent; coverage of the case in the media, too, suggested the decision
would be controversial. Consistent with naïve realism, the Court
might well have concluded, hardly without reason, that any dissatisfaction various social groups would have with reversal would reflect
the motivating impact of these citizens’ group commitments on their
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
187 See Kahan, supra note 19, at 130–42 (tracking this process in political debates concerning
sodomy, drugs, guns, smoking, and nuclear energy and global warming).
188 See generally MORRIS P. FIORINA WITH SAMUEL J. ABRAMS & JEREMY C. POPE, CULTURE WAR? THE MYTH OF A POLARIZED AMERICA (2006) (canvassing evidence that the vast
majority of the public cares more about material welfare issues than symbolic moral ones).
189 See generally Kahan, supra note 19.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

897

perceptions of the facts as well as their interpretation of the relevant
legal precedents. But what likely did not occur to the Justices in the
majority was the degree to which their own perceptions (not to mention the perceptions of those who would agree with them upon watching the tape) would be just as bound up with cultural, ideological, and
other commitments that disposed them to see the facts in a particular
way.190
The basis on which the Court justified its decision exhibits the tendency of naïve realism toward culturally grounded recrimination and
distrust. By declaring, in particular, that “no reasonable juror” could
have formed beliefs contrary to the Court’s own, the Court inevitably
called into question the integrity, intelligence, and competence of identifiable subcommunities whose members in fact held those dissenting
beliefs. Those individuals, the Court should have foreseen, would in
turn react with resentment and (not insupportable) suspicion that
members of the Court majority (and any who agreed with them) were
motivated by their values to declare their perceptions alone to be “reasonable.” As in other settings, then, in which the law picks sides in
factual disputes that arise from culturally conflicting worldviews, the
decision itself could well have been expected to deepen illiberal status
competition.191
We believe this sort of outcome is avoidable. Judges can but
needn’t inevitably compound the dynamics of cognitive illiberalism.
Indeed, we believe they are uniquely equipped to help counteract those
dynamics. The remedy is a form of judicial humility.
In a recent article, Professor Cass Sunstein argues that an appropriate posture of humility counsels courts to be sensitive to community
outrage.192 Judges, he notes, are boundedly rational, just like the rest
of us, and as a result are prone to err both about the legal correctness
of their decisions and about the practical consequences of them. One
remedy, Sunstein argues, is a sensitivity to anticipated community out–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
190 Cf. RICHARD A. POSNER, HOW JUDGES THINK 116 (2008) (observing that when judges
are confronted with ambiguous facts that touch on charged issues they, like everyone else, “fall
back on their intuitions” and display “[t]he kind of telescoped reasoning . . . called . . . ‘cultural
cognition’”).
191 See generally Goldberg, supra note 150 (arguing that judicial factfinding responds to and
reinforces cultural conflict); Kahan & Braman, supra note 94 (making this point about factual determinations in controversial self-defense cases). It is precisely because this aspect of Scott generalizes that the decision furnishes a constructive target for criticism on this basis. Our point is not
that Scott, considered by itself, deprives the law generally of legitimacy; obviously, the degree to
which any particular display of cultural partisanship undermines civic identification with the law
will be negligible. Our point is that the case displays a form of bias that is in fact pervasive in our
legal and political system and that as a whole is responsible for illiberal conflict. See generally
Kahan, supra note 19.
192 See Cass R. Sunstein, If People Would Be Outraged by Their Rulings, Should Judges Care?,
60 STAN. L. REV. 155 (2007).

898

HARVARD LAW REVIEW

[Vol. 122:837

rage conventionally thought to be antithetical to the judicial mindset.193 Humility — born of mindfulness of the limits of her own reasoning power — counsels the judge to treat the foreseeability of such
outrage as a cue that maybe she is in fact wrong; it gives her, at a
minimum, reason to rethink, and might, in some cases, furnish her a
reason to decide a case in a manner contrary to her own inclinations.194 Sunstein suggests that outrage normally serves as a corrective
heuristic of this sort only when it is experienced by a large majority,
unless the judge perceives that an outraged minority has some special
expertise or might be specially situated to resist the decision in a way
that has bad consequences for society overall.
But our argument shows how humility might enlarge the circumstances in which judges should attend to the potential for minority
outrage. Judges, like the rest of us, lack full insight into how the
mechanisms of value-motivated cognition shape their and others’ perceptions of particular facts. But just like the rest of us, they are perfectly capable of understanding that these dynamics exist and can
adversely affect the quality of their decisionmaking. One way to compensate for the partiality, and the incipient partisanship, of their own
factual perceptions is to attend to cues that a cultural subcommunity
will react with outrage should judges privilege their own factual perceptions. For in that situation, the anticipated reaction will furnish
judges with evidence that committing the law to a particular fact
risks creating the delegitimizing forms of cultural conflict that we have
described.
More concretely, we recommend that a judge engage in a sort of
mental double check when ruling on a motion that would result in
summary adjudication. Again, almost any time a judge does conclude
that there is no genuine dispute about some set of material facts, she
will be able to anticipate that some small percentage of actual jurors
would nevertheless dispute them. Before concluding, then, that no
reasonable juror could find such facts, the judge should try to imagine
who those potential jurors might be. If, as will usually be true, she
cannot identify them, or can conjure only the random faces of imaginary statistical outliers, she should proceed to decide the case summarily. But if instead she can form a concrete picture of the dissenting
jurors, and they are people who bear recognizable identity-defining
characteristics — demographic, cultural, political, or otherwise — she
should stop and think hard. Due humility obliges her to consider
whether privileging her own view of the facts risks conveying a deni–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
193
194

See id. at 164.
See id. at 175–78.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

899

grating and exclusionary message to members of such subcommunities.
If it does, she should choose a different path.
There is nothing radical in our suggestion that a judge temper her
decisionmaking with a prudential sensibility of this sort. On the contrary, that judges can and should exercise this form of discernment is
the premise of Professor Alexander Bickel’s celebrated defense of the
“passive virtues,” which counsels a rich array of avoidance techniques
to steer the law clear of legitimacy-enervating gestures of partisanship.195 All we are doing is calling attention to another legitimacydepleting gesture — the endorsement of culturally partisan views of
facts — that courts should use this sensibility to avoid.
But will judges inevitably succumb to the subconscious influence of
their cultural predispositions even as they exercise the particular corrective we have urged to avoid cognitively illiberal judicial factfinding? Maybe.196 Research is growing, however, on the power of the
judicial role to impart habits of mind that counteract certain types of
biases,197 including ones that distort moral reasoning.198 There is certainly no reason, then, to dismiss out of hand the possibility that the
device we are recommending — that judges pause to consider whether
what strikes them as an “obvious” matter of fact might in fact be
viewed otherwise by a discrete and identifiable subcommunity199 — is
one that would function as an effective debiasing strategy for cognitive
illiberalism. Indeed, the very gesture of attempting to do so in good
faith would go a long way to counteracting the message of exclusion
associated with a decision like Scott.200
Connecting our proposal to Bickel’s prudential minimalism makes
more concrete the limited scope of our caution about summary adjudication. Courts are rarely impelled to adopt the self-abnegating style
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
195 See ALEXANDER M. BICKEL, THE LEAST DANGEROUS BRANCH: THE SUPREME
COURT AT THE BAR OF POLITICS 111–98 (Vail-Ballor Press 2d ed. 1986) (1962); see also
Sunstein, supra note 192, at 168–75 (grounding his defense of “humility” in the theory of passive
virtues).
196 Cf. Richard A. Posner, The Supreme Court, 2004 Term—Foreword: A Political Court, 119
HARV. L. REV. 32, 81–82 (2005) (“Reading Bickel’s [The Supreme Court, 1960 Term—Foreword:
The Passive Virtues, 75 HARV. L. REV. 40 (1961)], one realizes that he had definite ideas about
where the public policy of the United States should be moving and that these ideas were his
‘principles.’”).
197 See Guthrie, Rachlinski & Wistrich, supra note 166.
198 See Jeffrey J. Rachlinski, Sheri Lynn Johnson, Andrew J. Wistrich & Chris Guthrie, A
Thumb on the Scale: Investigating Implicit Bias on the Bench, 84 NOTRE DAME L. REV. (forthcoming Mar. 2009).
199 Cf. Linda Greenhouse, Justices Indicate They May Uphold Voter ID Rules, N.Y. TIMES,
Jan. 10, 2008, at A1 (reporting Justice Roberts’s skepticism toward the claim that obtaining official identification at the county seat as a prerequisite to voting is burdensome and reply of counsel
that “[i]f you’re indigent, [the 17-mile bus trip from urban Gary to the county seat is] a significant
burden” (internal quotation marks omitted)).
200 We are grateful to Judge Richard Posner for focusing our attention on this issue.

900

HARVARD LAW REVIEW

[Vol. 122:837

associated with the passive virtues. As we have emphasized, the Court
certainly could have foreseen, because of the contentious symbolism of
the case, that its view of the facts in Scott would be disputed by persons of a particular cultural outlook. We can certainly think of other
kinds of cases too — ones involving gun control,201 say, or environmental regulation202 — that courts could readily anticipate would result in culturally polarized understandings of fact. But those cases are
far and away the exception and not the rule in ordinary litigation.
Consequently, the prudential brake we are urging on summary adjudication is one courts should rarely feel constrained to apply in cases
that otherwise seem fit for such disposition.
It should also be clear that we are not advancing any sweeping indictment of judicial consideration of visual or other demonstrative evidence. It is not unprecedented for the Supreme Court to attach photographs, maps, pictures, and exhibits to its opinions as support and to
refer readers to them.203 In the aftermath, too, of Scott, judges might
well feel emboldened to give more decisive weight to the factual inferences they themselves are inclined to draw from videos or photographs
in deciding summary judgment motions.204 There might well be compelling grounds for objecting to these and like practices,205 but the particular criticism we are making of Scott doesn’t go to the propriety of
what might be called a “sensory jurisprudence” generally.
Our concern with the Court’s reliance on the Scott tape is much
more focused. At least within the terms of our argument, there is
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
201 Cf. Dan M. Kahan, Donald Braman & John Gastil, A Cultural Critique of Gun Litigation,
in SUING THE GUN INDUSTRY: A BATTLE AT THE CROSSROADS OF GUN CONTROL AND
MASS TORTS 105 (Timothy D. Lytton ed., 2005) (arguing that use of litigation to regulate guns
exacerbates cultural conflict).
202 Cf. David Hunter & James Salzman, Negligence in the Air: The Duty of Care in Climate
Change Litigation, 155 U. PA. L. REV. 1741, 1744 (2007) (describing resort to tort cases in response
to global warming).
203 See, e.g., Carnival Cruise Lines, Inc. v. Shute, 499 U.S. 585, 597 (1991) (Stevens, J., dissenting) (“[O]nly the most meticulous passenger is likely to become aware of the forum-selection provision [in the agreement printed on each cruise-vacation ticket]. I have . . . appended to this opinion a facsimile of the relevant text, using the type size that actually appears in the ticket itself. A
careful reader will find the forum-selection clause in the 8th of the 25 numbered paragraphs.”).
See generally Hampton Dellinger, Words are Enough: The Troublesome Use of Photographs, Maps,
and Other Images in Supreme Court Opinions, 110 HARV. L. REV. 1704, 1707 (1997) (stating that
“photographs, maps, and other attachments operate as communicative vehicles of economy, all
offering the possibility of an impact more powerful than words”).
204 See, e.g., Green v. N.J. State Police, No. 06-4111, 2007 WL 2453580, at *2 n.1 (3d Cir. Aug.
29, 2007) (using video evidence to establish facts but affirming denial of summary judgment);
Sharp v. Fisher, No. 406CV020, 2007 WL 2177123 (S.D. Ga. Jul. 26, 2007) (granting summary
judgment on the basis of the facts as depicted in a police video, and posting the video to the
court’s website).
205 See Dellinger, supra note 203; Jessica Silbey, Cross-Examining Film, 8 U. MD. L.J. RACE,
RELIGION, GENDER & CLASS (forthcoming 2009), available at http://papers.ssrn.com/sol3/
papers.cfm?abstract_id=1202043; Wasserman, supra note 53.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

901

nothing problematic about a court deciding summarily based on its
sensory impressions when the factual inference it is drawing isn’t one
that is likely to divide potential jurors along cultural lines (in a personal injury case, say, where the object the plaintiff is seen to be tripping on in a video is his own untied shoelace rather than a raised
board in the floor of the defendant’s store). In addition, where a factual inference would likely provoke cultural dissensus, our argument
would counsel the judge not to draw it even if the basis for the inference is nothing other than her mere sensory impression.
What should a judge do in cases in which she does perceive
that summary adjudication would create the sort of cultural divisions
we are describing? As we have suggested, there are a number of
possibilities.
One of them, of course, is to permit the case to be decided by the
jury notwithstanding the judge’s justifiable belief that a large majority
of jurors would see the facts as she does. Even if doing so did not result in a verdict in line with the factual views of the minority groups’
members, mandating that the case be decided in a manner that assures
their perspective is given a respectful hearing makes it possible for
them to assent to the outcome as one consistent with recognition of
their status and competence.
Alternatively, the court can decide the case summarily on some announced basis that doesn’t stigmatize the potentially aggrieved subcommunity’s view of reality as flawed. The Court could have done
that in Scott, we pointed out, by emphasizing either the need for verdict uniformity in this area, the adverse policy consequences that
would attend a verdict for the plaintiff, or the appropriateness of political rather than judicial regulation of high-speed police pursuits.
That is, appropriate humility does not forbid judges to select an
outcome that is likely to be more congenial to one cultural style or another, but only to justify that outcome in terms that avoid cultural partisanship. Had Scott been decided on one of these alternative grounds,
the outcome would likely not have been any less contentious in the
eyes of persons who subscribe to what we have characterized as the
bet cultural style. But it would have been less demeaning to them.
The law explicitly takes sides all the time on issues that pit conflicting cultural values against one another. In the ordinary tort case, for
example, communitarians might object to using a cost-benefit calculus
to determine the scope of the duty to use care.206 In an action for
breach of contract, individualists might believe that contracting parties
should be bound to all promises, including those that courts routinely
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
206 See, e.g., Stephen G. Gilles, The Invisible Hand Formula, 80 VA. L. REV. 1015, 1020–21
(1994) (describing and criticizing this position in academic commentary).

902

HARVARD LAW REVIEW

[Vol. 122:837

dismiss as “puffery.”207 In antitrust cases, the dominant consumerwelfare philosophy will likely be much less congenial to egalitarians,
who favor a more redistributional approach, and hierarchs, who might
prefer a more corporatist one, than to individualists.208
But when judges enforce the law in a manner that forbids dissenting jurors to override the values reflected in such doctrines, they do so
within a decisionmaking environment that (normally, at least) evinces
respect for losers. The ability of defeated parties to identify with such
decisions, notwithstanding their disagreement with them, is preserved,
in part, through the law’s genesis, and continued amenability to revision, in democratic politics.209 But just as important, the dignity of
dissenters is protected by idioms of justification, including formalism,
that disavow the law’s endorsement of a cultural orthodoxy. Indeed,
the array of techniques associated with judicial minimalism is animated by a recognition on the part of the judiciary that promoting liberal pluralism in law requires judges to attend carefully to the language they use to justify their decisions.210
Far from promoting that end, proclaiming that there is only one
“reasonable” view of the facts in a culturally contentious case needlessly burdens the law with partisanship, detracting from its legitimacy. That is the simple, but important, lesson of Scott.

–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
207 See generally David A. Hoffman, The Best Puffery Article Ever, 91 IOWA L. REV. 1395
(2006) (describing multiple areas of law in which puffing claims are taken from juries).
208 See generally Herbert Hovenkamp, Antitrust Policy After Chicago, 84 MICH. L. REV. 213
(1985) (refuting neoclassical efficiency model of antitrust as insufficient in addressing strategic
behavior); Herbert Hovenkamp, The Rationalization of Antitrust, 116 HARV. L. REV. 917, 920
(2003) (reviewing RICHARD A. POSNER, ANTITRUST LAW (2001)) (describing connection between ideological antitrust and competing antitrust theories).
209 See WALDRON, DIGNITY, supra note 182.
210 See David A. Strauss, Legal Argument and the Overlapping Consensus 20–21 (July 12,
1998) (unpublished manuscript, on file with the Harvard Law School Library); see also CASS R.
SUNSTEIN, LEGAL REASONING AND POLITICAL CONFLICT 132 (1996). The contribution
that idioms of justification make to liberal culture independent of the content of the policies being
justified is well-developed in the study of democratic politics generally. See, e.g., AMY GUTMANN & DENNIS THOMPSON, DEMOCRACY AND DISAGREEMENT 14–16 (1996) (defending
idioms of justification that evidence “reciprocity” and seek to uncover common ground among
persons of diverse values); STEPHEN HOLMES, PASSIONS AND CONSTRAINT: ON THE THEORY OF LIBERAL DEMOCRACY 234 (1995) (defending democratic “gag rules” that constrain appeal to divisive issues of value in order to enable “citizens who differ greatly in outlook on life [to]
work together to solve common problems”). How idioms of justification in judicial decisions
should be structured to promote liberal accommodation warrants a commensurate degree of attention and empirical study. For an existing treatment that ought to serve as a model for such
investigation, see MARY ANN GLENDON, ABORTION AND DIVORCE IN WESTERN LAW 25–
39 (1987), discussing the West German Supreme Court’s attention to the possibility of injecting
plurality of social meanings into abortion law.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

903

CONCLUSION
To our knowledge, Scott v. Harris is the only case in which the Supreme Court has invoked brute sense impressions to justify its decision.211 Its remarkable invitation to members of the public to
download the Scott video and decide for themselves what to make of
Justice Stevens’s dissent reflects the premise that perceptions of “facts”
don’t stand in need (or even admit) of the sort of reasoned defense we
expect when courts make potentially contentious normative judgments. Our goal in this Article was to motivate a critical appraisal of
the Court’s premise by determining empirically who would and
wouldn’t agree with the majority after viewing the tape.
Our empirical study found that when we “allow the videotape to
speak for itself,”212 what it says depends on to whom it is speaking. To
be sure, a substantial majority of American society is inclined to see in
Harris’s flight from the police the sort of lethal threat to public safety
that in turn warrants a potentially deadly response on the part of the
police. But this view is not uniform across subcommunities. Whites
and African Americans, high-wage earners and low-wage earners,
Northeasterners and Southerners and Westerners, liberals and conservatives, Republicans and Democrats — all varied significantly in their
perceptions of the risk that Harris posed, of the risk the police created
by deciding to pursue him, and of the need to use deadly force against
Harris in the interest of reducing public risk.
These patterns suggest the influence of value-motivated cognition.
Comprising several discrete mechanisms, value-motivated cognition
refers to the tendency of people to resolve factual ambiguities in a
manner that generates conclusions congenial to self-defining values.
This conclusion was compellingly borne out by the relationship of
cultural outlooks to views of the tape in Scott. Individuals disposed to
form extreme views of the facts in Scott tend to be united by adherence to one of two competing sets of preferences about how society
should be organized — one that is egalitarian and communitarian, and
another that is hierarchical and individualistic. We found that after
viewing the Scott tape, individuals of the latter outlook formed views
emphatically in line with those of the Court majority. Those of the
former outlook, in contrast, were more likely to see the police, not Harris, as the source of the risk to the public and to conclude that use of
deadly force was not a justifiable response given the risk that Harris
posed. This division brings the conclusions of our study into line with
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
211 The closest analogue is probably Justice Stewart’s famous and frequently mocked statement, “I know [obscenity] when I see it, and the motion picture involved in this case is not that.”
Jacobellis v. Ohio, 378 U.S. 184, 197 (1964) (Stewart, J., concurring).
212 Scott v. Harris, 127 S. Ct. 1769, 1775 n.5 (2007).

904

HARVARD LAW REVIEW

[Vol. 122:837

those on cultural cognition generally, which show that competing cultural outlooks of these varieties dispose people to disagree about the
facts of all manner of putative dangers — from climate change to gun
control to HPV vaccinations for school-age girls.213
What should be made of these findings? We have argued that they
are the basis for strongly objecting to the Court’s reasoning in Scott.
Our legal system conspicuously holds forth jury decisionmaking as a
means of making the law responsive to, and hence legitimately binding
on, individuals of diverse backgrounds. The basis of the Court’s decision in Scott cannot be reconciled with this understanding. It’s true
that a majority of Americans would indeed see the facts the way the
Court did after watching the Scott tape. But the minority of persons
in our society who would perceive the facts differently are not bare
statistical outliers; they are a cohesive and recognizable subcommunity
whose members share an identity based on common values and experiences. By asserting that the view of the facts these people came
away with was one no “reasonable juror” could have formed, the Scott
majority not only denied jurors of this identity a chance to persuade
those of another identity to see things a different way. It also assured
that the law established by the case would be seen by members of that
subcommunity as deriving from a process calculated to exclude their
voices from even being heard.
There were multiple avenues available to the Court for reversing in
Scott, we have suggested. But the justification it chose was the one
that maximized the experience of exclusion for a recognizable segment
of the American citizenry, needlessly infusing the decision with culturally partisan overtones that detracted from the law’s legitimacy.
We do not mean, however, to attribute bad faith to the Court. On
the contrary, we have suggested that it’s likely that the Court in Scott
unwittingly fell prey to cognitive illiberalism — and dragged the rest
of us along with it. Sincerely perceiving the facts to be unambiguous,
the Court concluded, with considerable foundation, that only those in
the grip of a partisan set of cultural commitments could see things otherwise. But in saying that, the Court necessarily made itself into a culturally partisan decisionmaker in the eyes of those who saw, with just
as much foundation, that the Court’s own view of the facts was culturally motivated. This is how honest disagreements of fact in our society mutate into recriminatory sources of cultural conflict among persons who in truth have no desire to enforce a moral orthodoxy through
law.
The incompatibility of liberal principles with the use of law to impose a partisan vision of the good is well understood. Our legal and
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
213

See Kahan, supra note 19, at 123, 134–36, 139–42.

2009]

WHOSE EYES ARE YOU GOING TO BELIEVE?

905

political practices are replete with devices that not only forbid fullblooded sectarianism of this type but also effectively stifle even latent
forms of it in their incipiency, lest rhetorical misadventure impel citizens into illiberal conflict against their will. In political life, these include the norm of public reason, which requires legislators and ordinary citizens to justify policies on grounds accessible to persons of
diverse moral and cultural persuasions instead of in terms that reflect
a partisan conception of the good.214 In the legal arena, they include
devices like formalism and minimalism, which enable judges to defend
outcomes and elaborate the law without resort to more contentious
moral claims.215
What’s not nearly so well understood, however, is the threat that
competing understandings of fact pose to a liberal society. Indeed,
forms of advocacy that feature seemingly neutral factual claims about
how to promote societal welfare (“optimal deterrence,” “cost-benefit
analysis,” “contingent valuation” and the like) are thought to be among
the practices that dissipate illiberal conflict by avoiding reference to
more contentious judgments of value.216 It might seem natural to see
judicial idioms that focus on “facts” as conflict-avoiding for the same
reason.217 But because we inevitably recur to our cultural values to
evaluate empirical claims about what conditions threaten our welfare
and what policies promote it, styles of argumentation that feature facts
can polarize us every bit as much as one that deals with differences of
value in a transparent way.218
We have proposed a form of judicial humility as one technique for
ameliorating such conflict. But such a strategy, we concede, will on its
own make only a modest contribution to this end. Much more attention to solving the problem of cognitive illiberalism is needed, in both
the legal and the political domains.
Accepting the Court’s peculiar invitation in Scott hasn’t shown
that every “reasonable” person would see the facts in that case the way
the majority did. But it has, we think, helped make it possible for
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
214 See, e.g., JOHN RAWLS, POLITICAL LIBERALISM 175, 217–18 (1993) (articulating a norm
of “public reason” that prohibits political actors in most contexts from invoking “comprehensive”
doctrines that “include[] conceptions of what is of value in human life, as well as ideals of personal virtue and character,” id. at 175, and instead requires them to “explain . . . how the principles and policies they advocate and vote for can be supported by” considerations, id. at 217, consistent with “a diversity of reasonable religious and philosophical doctrines,” id. at 218).
215 See generally SUNSTEIN, supra note 210; Strauss, supra note 210.
216 See Martin Rein & Christopher Winship, The Dangers of “Strong” Causal Reasoning in Social Policy, SOCIETY, July–Aug. 1999, at 38, 39 (identifying the appeal of empirical modes of policy analysis in their provision of “objective procedures and criteria” that seem “decidedly divorced
from statements about morality”).
217 See, e.g., Goldberg, supra note 150, at 1984 (imputing this approach to courts in constitutional law contexts).
218 See Kahan, supra note 19, at 144.

906

HARVARD LAW REVIEW

[Vol. 122:837

every reasonable citizen to see something much more important: that
just as critical to liberalism as developing strategies for dissipating cultural conflict over values is the development of strategies for dissipating cultural conflict over facts.