Journal of Criminal Law and Criminology
Volume 103 | Issue 1

Article 1

Winter 2013

The Innocent Defendant's Dilemma: An
Innovative Empirical Study of Plea Bargaining's
Innocence Problem
Lucian E. Dervan
Vanessa A. Edkins Ph.D.

Follow this and additional works at: http://scholarlycommons.law.northwestern.edu/jclc
Part of the Criminal Law Commons
Recommended Citation
Lucian E. Dervan and Vanessa A. Edkins Ph.D., The Innocent Defendant's Dilemma: An Innovative Empirical Study of Plea Bargaining's
Innocence Problem, 103 J. Crim. L. & Criminology 1 (2013).
http://scholarlycommons.law.northwestern.edu/jclc/vol103/iss1/1

This Criminal Law is brought to you for free and open access by Northwestern University School of Law Scholarly Commons. It has been accepted for
inclusion in Journal of Criminal Law and Criminology by an authorized administrator of Northwestern University School of Law Scholarly Commons.

0091-4169/13/10301-0001
THE JOURNAL OF CRIMINAL LAW & CRIMINOLOGY
Copyright © 2013 by Northwestern University School of Law

Vol. 103, No. 1
Printed in U.S.A.

CRIMINAL LAW
THE INNOCENT DEFENDANT’S DILEMMA:
AN INNOVATIVE EMPIRICAL STUDY OF
PLEA BARGAINING’S INNOCENCE
PROBLEM
LUCIAN E. DERVAN*
AND
VANESSA A. EDKINS, Ph.D.**
In this Article, Professors Dervan and Edkins discuss a recent
psychological study they completed regarding plea bargaining and
innocence. The study, involving dozens of college students and taking place
over several months, revealed that more than half of the innocent
participants were willing to falsely admit guilt in return for a benefit.
These research findings bring significant new insights to the long-standing
Special thanks to Professors Robert Ahdieh, Peter Alexander, Albert Alschuler, Shima
Baradaran, Christopher Behan, Doug Berman, William Berry, III, Katrice Copeland, Russell
Covey, Deborah Dinner, George Fisher, Oren Gazal-Ayal, Michael Heise, Richard Hertling,
Christopher Hines, Virginia Harper Ho, Rebecca Hollander-Blumoff, Jeniffer Horan, John
Inazu, Lea Johnston, Hon. Sterline Johnson, Jr., Sam Jordan, Marc Miller, Matthew Miner,
Karen Petroski, Ellen Podgor, Laurent Sacharoff, Nadia Sawicki, William Schroeder,
Michael Seigel, Hon. William Sessions, III, Christopher Slobogin, Paul Van de Graaf,
Bobby Vassar, and Verity Winship for the comments and conversations and to the following
research assistants: Brian Lee, Alexandra Novak, Elisabeth Beasley, Matthew Martin,
Geraldine Castillo, Joseph Guccione, Alexa Weinberg, and Alison Koenig. Thanks also to
Washington University School of Law for the opportunity to present this piece as part of its
workshop series. This study and its results were also the subject of testimony before the
U.S. House Judiciary Committee in 2012.
*
Visiting Assistant Professor of Law at the University of Georgia School of Law and
Assistant Professor of Law at the Southern Illinois University School of Law. Former
member of the King & Spalding LLP Special Matters and Government Investigations
practice group.
**
Florida Institute of Technology, Department of Psychology.

1

2

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

debate regarding the extent of plea bargaining’s innocence problem. The
Article also discusses the history of bargained justice and examines the
constitutional implications of the study’s results on plea bargaining, an
institution the Supreme Court reluctantly approved of in 1970 in return for
an assurance that it would not be used to induce innocent defendants to
falsely admit guilt.
TABLE OF CONTENTS
I. INTRODUCTION........................................................................................... 2
II. THE HISTORICAL RISE OF PLEA BARGAINING AND ITS INNOCENCE
PROBLEM ....................................................................................................... 5
A. The Rise of Plea Bargaining ................................................................. 7
B. Plea Bargaining’s Innocence Debate .................................................. 15
III. LABORATORY EVIDENCE OF PLEA BARGAINING’S INNOCENCE
PROBLEM ..................................................................................................... 24
A. Study Methodology—Confronting a Devil’s Bargain ........................ 28
B. Study Results—The Innocent Defendant’s Dilemma Exposed .......... 33
1. Pleading Rates for Guilty and Innocent Students ............................ 34
2. The Impact of Sentencing Differentials ........................................... 38
IV. THE CONSTITUTIONALITY OF THE INNOCENT DEFENDANT’S
DILEMMA .................................................................................................... 43

I. INTRODUCTION
In 1989, Ada JoAnn Taylor sat quietly in a nondescript chair
contemplating her choices.1 On a cold February evening four years earlier,
a sixty-eight-year-old woman was brutally victimized in Beatrice,
Nebraska.2 Police were now convinced that Taylor and five others were
responsible for the woman’s death.3 The options for Taylor were stark.4 If
she pleaded guilty and cooperated with prosecutors, she would be rewarded
1

See Know the Cases: Ada JoAnn Taylor, THE INNOCENCE PROJECT,
www.innocenceproject.org/Content/Ada_JoAnn_Taylor.php (last visited Oct. 26, 2012)
[hereinafter Taylor, THE INNOCENCE PROJECT].
2
See id. (“Sometime during the night of February 5, 1985, 68-year-old Helen Wilson
was sexually assaulted and killed in the Beatrice, Nebraska, apartment where she lived
alone.”).
3
But see id. (“An FBI analysis of the Wilson murder and the three other [related] crimes
concluded that ‘we can say with almost total certainty that this crime was committed by one
individual acting alone.’”).
4
See id.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

3

with a sentence of ten to forty years in prison.5 If, however, she proceeded
to trial and was convicted, she would likely spend the rest of her life behind
bars.6
Over a thousand miles away in Florida, and more than twenty years
later, a college student sat nervously in a classroom chair contemplating her
options.7 Just moments before, a graduate student had accused her of
cheating on a logic test being administered as part of a psychological study.
The young student was offered two choices. If she admitted her offense and
saved the university the time and expense of proceeding with a trial before
the Academic Review Board, she would simply lose her right to
compensation for participating in the study. If, however, she proceeded to
the review board and lost, she would lose her compensation, her faculty
advisor would be informed, and she would be forced to enroll in an ethics
course.
In Beatrice, Nebraska, the choice for Taylor was difficult, but the
incentives to admit guilt were enticing.8 A sentence of ten to forty years in
prison meant she would return home one day and salvage at least a portion
of her life.9 The alternative, a lifetime behind bars, was grim by
comparison.10 After contemplating the options, Taylor pleaded guilty to
aiding and abetting second-degree murder.11 Twenty years later, the college
student made a similar calculation.12 While the loss of compensation for

5

See id. (“Ada JoAnn Taylor agreed with prosecutors to plead guilty and testify at the
trial of co-defendant Joseph White regarding her alleged role in the murder. In exchange for
her testimony, she was sentenced to 10 to 40 years in prison.”).
6
See id.
7
See infra Part III (discussing the plea-bargaining study).
8
See Taylor, THE INNOCENCE PROJECT, supra note 1.
9
See id.
10
See id.; see also Wayne A. Logan, Proportionality and Punishment: Imposing Life
Without Parole on Juveniles, 33 WAKE FOREST L. REV. 681, 712 (1998) (discussing the
severity of life in prison and noting that some death row inmates “waive their appeals out of
fear that they will perhaps succeed and be faced with a mandatory LWOP sentence”). As
noted by one philosopher:
What comparison can there really be, in point of severity between consigning a man to the short
pang of a rapid death, and immuring him in a living tomb, there to linger out what may be a long
life in the hardest and most monotonous toil, without any of its alleviation or rewards—debarred
from all pleasant sights and sounds, and cut off from all earthly hope, except a slight mitigation
of bodily restraint, or a small improvement of diet?

Id. (quoting LEON SHASKOLSKY SHELEFF, ULTIMATE PENALTIES: CAPITAL PUNISHMENT, LIFE
IMPRISONMENT, PHYSICAL TORTURE 60 (1987) (quoting John Stuart Mill, Parliamentary
Debate on Capital Punishment Within Prisons Bill (Apr. 21, 1868))).
11
Taylor, THE INNOCENCE PROJECT, supra note 1.
12
See infra Part III (discussing the plea-bargaining study).

4

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

participating in the study was a significant punishment, it was certainly
better than being forced to enroll in a time-consuming ethics course.13 Just
as Taylor had decided to control her destiny and accept the certainty of the
lighter alternative, the college student admitted that she had knowingly
cheated on the test.14
That Taylor and the college student both pleaded guilty is not the only
similarity between the cases. Both were also innocent of the offenses of
which they had been accused.15 After serving nineteen years in prison,
Taylor was exonerated after DNA testing proved that neither she nor any of
the other five defendants in her case were involved in the murder.16 As for
the college student, her innocence is assured by the fact that, unbeknownst
to her, she was actually part of an innovative new study into plea bargaining
and innocence.17 The study, conducted by the authors, involving dozens of
college students and taking place over several months, not only recreated
the innocent defendant’s dilemma experienced by Taylor, but also revealed
that plea bargaining’s innocence problem is not isolated to an obscure and
rare set of cases.18 Strikingly, the study demonstrated that more than half of
the innocent participants were willing to falsely admit guilt in return for a
perceived benefit.19 This finding brings new insights to the long-standing
debate regarding the possible extent of plea bargaining’s innocence problem
and ignites a fundamental constitutional question regarding an institution
the Supreme Court reluctantly approved of in 1970 in return for an
assurance that it would not be used to induce innocent defendants to falsely
admit guilt.20
This Article begins in Part II by examining the history of plea
13

See infra Part III.
See infra Part III.
15
See Taylor, THE INNOCENCE PROJECT, supra note 1.
16
See id. It should also be noted that five of the six defendants in the Wilson murder
case pleaded guilty. As described above, DNA evidence showed that all six defendants were
innocent and played no role in the sexual assault or murder of Wilson. See id.; see also
Know the Cases: Debra Shelden, THE INNOCENCE PROJECT, www.innocenceproject.org/
Content/Debra_Shelden.php (last visited Jan. 1, 2012) (“Debra Shelden agreed with
prosecutors to plead guilty and testify falsely to her alleged role in the crime at the trial of
co-defendant Joseph White in exchange for a lighter sentence.”); Know the Cases: James
Dean, THE INNOCENCE PROJECT, www.innocenceproject.org/Content/James_Dean.php (last
visited Jan. 1, 2012) (“Joseph White was the only defendant in this case to go to trial, and
three of his five co-defendants testified against him in exchange for shorter sentences than
those they may have received had their own cases gone to trial.”).
17
See infra Part III (discussing the plea-bargaining study).
18
See infra Part III.
19
See infra Part III.
20
See infra Part III.
14

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

5

bargaining in the United States, including an examination of the current
debate regarding the prevalence of plea bargaining’s innocence problem.21
In Part III, this Article discusses the psychological study of plea bargaining
conducted by the authors.22 This Part reviews the methodology and results
of the study.23 Finally, Part III analyzes the constitutional limits placed on
plea bargaining by the Supreme Court in its landmark 1970 decision, Brady
v. United States.24 In this decision, the Supreme Court stated that plea
bargaining was a tool for use only when the evidence of guilt was
overwhelming and the defendant might benefit from the opportunity to
bargain.25 According to the Court, if it became evident that plea bargaining
was being used more broadly to create incentives for questionably guilty
defendants to “falsely condemn themselves,” the entire institution of plea
bargaining and its constitutionality would require reexamination.26 Perhaps,
as a result of this new study, a time for such reevaluation has arrived.
II. THE HISTORICAL RISE OF PLEA BARGAINING AND ITS
INNOCENCE PROBLEM
On December 23, 1990, a twenty-one-year-old woman was robbed and
sexually assaulted by an unknown assailant in New Jersey.27 Three days
after the attack, and again a month later, the victim identified John Dixon as
the perpetrator from a photo array.28 Dixon was arrested on January 18,
1991, and ventured down a road familiar to criminal defendants in the
United States.29 Threatened by prosecutors with a higher prison sentence if
he failed to cooperate and confess to his alleged crimes, Dixon pleaded
guilty to sexual assault, kidnapping, robbery, and unlawful possession of a
weapon.30 He received a sentence of forty-five years in prison.31 Ten years
21
See infra Part II (discussing the historical rise of plea bargaining and its innocence
problem).
22
See infra Part III (discussing the plea-bargaining study).
23
See infra Part III.
24
See Brady v. United States, 397 U.S. 742 (1970).
25
Id. at 752.
26
Id. at 757–58; see also Lucian E. Dervan, Bargained Justice: Plea-Bargaining’s
Innocence Problem and the Brady Safety-Valve, 2012 UTAH L. REV. 51.
27
Know
the
Cases:
John
Dixon,
THE
INNOCENCE
PROJECT,
http://www.innocenceproject.org/Content/John_Dixon.php (last visited Jan. 23, 2012)
[hereinafter Dixon, THE INNOCENCE PROJECT] (describing the story of John Dixon, who
pleaded guilty to rape charges for fear that he would receive a harsher sentence if he
proceeded to trial but who was later exonerated by DNA evidence).
28
See id.
29
See id.
30
See id.; see also Richard Klein, Due Process Denied: Judicial Coercion in the Plea

6

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

later, however, Dixon was released from prison after DNA evidence
established that he could not have been the perpetrator of the crime.32
While the story of an innocent man pleading guilty and serving a decade in
prison before exoneration is a tragedy, perhaps it should not be surprising
given the prominence and power of plea bargaining in today’s criminal
justice system.33
Plea bargaining, however, was not always such a dominant force in the
United States.34 In fact, when appellate courts first began to see an influx of
such bargains around the time of the American Civil War, most struck
down the deals as unconstitutional.35 Despite these early judicial rebukes,
plea bargaining continued to linger in the shadows as a tool of corruption. 36
Bargaining Process, 32 HOFSTRA L. REV. 1349, 1398 (2004).
By the time of the plea allocution it is clear that the defendant has decided to take the plea
bargain and knows or has been instructed by counsel to tell the court that he did indeed do the
crime. Predictably, the National Institute of Justice survey found that judges rejected guilty pleas
in only two percent of cases. Since efficiency and speed is the name of the game, it is not
unexpected that meaningful questioning of the defendant does not occur and it is not surprising
that the Institute concluded that the plea allocution procedure is “close to being a new kind of
‘pious fraud.’”

Id. (citations omitted); see also Ronald F. Wright, Trial Distortion and the End of Innocence
in Federal Criminal Justice, 154 U. PA. L. REV. 79, 93 (2005) (“But when it comes to the
defendant’s ‘voluntariness’—the second half of the formula—courts have walked away. The
proper knowledge, together with a pro forma statement from the defendant that her guilty
plea was not coerced, normally suffices.”).
31
See Dixon, THE INNOCENCE PROJECT, supra note 27.
32
See id.
33
See U.S. SENTENCING COMM’N, 2010 SOURCEBOOK OF FEDERAL SENTENCING
STATISTICS,
fig.C
[hereinafter
2010
SOURCEBOOK,
fig.C],
available
at
http://www.ussc.gov/Data_and_Statistics/Annual_Reports_and_Sourcebooks/2010/FigureC.
pdf (documenting that almost 97% of convicted defendants in the federal criminal justice
system plead guilty).
34
See Dervan, supra note 26, at 58; Lucian E. Dervan, Plea Bargaining’s Survival:
Financial Crimes Plea Bargaining, A Continued Triumph in a Post-Enron World, 60 OKLA.
L. REV. 451, 478 (2007); Mark H. Haller, Plea Bargaining: The Nineteenth Century Context,
13 LAW & SOC’Y REV. 273, 273 (1979) (“[Alschuler and Friedman] agree that plea
bargaining was probably nonexistent before 1800, began to appear during the early or midnineteenth century, and became institutionalized as a standard feature of American urban
criminal courts in the last third of the nineteenth century.”). For further discussion regarding
the early history of plea bargaining, see John Baldwin & Michael McConville, Plea
Bargaining and Plea Negotiation in England, 13 LAW & SOC’Y REV. 287 (1979); John H.
Langbein, Understanding the Short History of Plea Bargaining, 13 LAW & SOC’Y REV. 261
(1979); Lynn M. Mather, Comments on the History of Plea Bargaining, 13 LAW & SOC’Y
REV. 281 (1979).
35
See Dervan, supra note 26, at 58–59.
36
See Albert W. Alschuler, Plea Bargaining and Its History, 79 COLUM. L. REV. 1, 19–
24 (1979).

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

7

Then, in response to growing pressures on American courts due to
overcriminalization in the early twentieth century, plea bargaining began a
spectacular rise to power.37 That today almost 97% of convictions in the
federal system result from pleas of guilt, such as John Dixon’s in New
Jersey in 1991, is both a testament to the institution’s resilience and a
caveat about its power of persuasion.38
A. THE RISE OF PLEA BARGAINING
While most discussions regarding the rise of plea bargaining begin in
the late nineteenth century, the full history of plea bargaining dates back
hundreds of years to the advent of confession law.39 As Professor Albert
Alschuler noted, “[T]he legal phenomenon that we call a guilty plea has
existed for more than eight centuries . . . [as] a ‘confession.’”40
Interestingly, early legal precedent regarding confessions prohibited the
offering of any inducement to prompt the admission.41 As an example, in
the 1783 case of Rex v. Warickshall, an English court stated, “[A]
confession forced from the mind by the flattery of hope, or by the torture of
fear, comes in so questionable a shape . . . that no credit ought to be given
to it.”42 While plea bargaining as it exists today relies upon the use of
incentives, common law prohibitions on such inducements persisted until
well into the twentieth century.43
37

George Fisher, Plea Bargaining’s Triumph, 109 YALE L.J. 857, 859 (2000)
[hereinafter Fisher, Plea Bargaining’s Triumph (Yale)] (“There is no glory in plea
bargaining. In place of a noble clash for truth, plea bargaining gives us a skulking truce . . . .
But though its victory merits no fanfare, plea bargaining has triumphed . . . . The battle has
been lost for some time.”); see also GEORGE FISHER, PLEA BARGAINING’S TRIUMPH: A
HISTORY OF P LEA BARGAINING IN AMERICA (2003) [hereinafter FISHER, PLEA BARGAINING’S
TRIUMPH].
38
See 2010 SOURCEBOOK, fig.C, supra note 33.
39
See Alschuler, supra note 36, at 12.
40
See id. at 13.
41
See id. at 12.
42
See id. (“It soon became clear that any confession ‘obtained by [a] direct or implied
promise[], however slight’ could not be received in evidence. Even the offer of a glass of
gin was a ‘promise of leniency’ capable of coercing a confession.” (footnotes omitted)).
43
See Dervan, supra note 26, at 65–66 (discussing the evolution of the doctrine that
guilty pleas must be voluntary); see also Albert W. Alschuler, The Changing Plea
Bargaining Debate, 69 CALIF. L. REV. 652, 657 (1981) (“Plea negotiation works . . . only
because defendants have been led to believe that their bargains are in fact bargains. If this
belief is erroneous, it seems likely that the defendants have been deluded into sacrificing
their constitutional rights for nothing.”); Russell D. Covey, Signaling and Plea Bargaining’s
Innocence Problem, 66 WASH. & LEE L. REV. 73, 77–78 (2009) (“Assuming that prosecutors
seek to maximize and defendants seek to minimize sentences, the price of any plea should be
the product of the anticipated trial sentence and the likelihood of conviction, discounted by

8

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

The first influx of plea-bargaining cases at the appellate level in the
United States occurred shortly after the Civil War.44 Relying on past
confession precedent prohibiting the offering of incentives in return for
admissions of guilt, various courts summarily rejected these bargains and
permitted the defendants to withdraw their statements.45 These early
American appellate decisions, however, did not prevent plea bargaining
from continuing to operate in the shadows.46 Plea bargains continued to be
used during this period, despite strong precedential condemnation, at least
in part as a tool of corruption.47 As an example, and as Professor Alschuler
has previously noted, there are documented accounts that by 1914 a defense
attorney in New York would “stand out on the street in front of the Night
Court and dicker away sentences in this form: $300 for ten days, $200 for
twenty days, $150 for thirty days.”48 Such bargains were not limited to
New York.49 One commentator in 1928 discussed the use of “fixers,” who
negotiated bargains between the government and the defense in Chicago,
Illinois:
some factor to reflect the resources saved by not having to try the case.”).
44
See Alschuler, supra note 36, at 19–21.
45
See id. Alschuler provides several examples of statements made by the appellate
courts examining plea bargains in the late nineteenth century.
The least surprise or influence causing [the defendant] to plead guilty when he had any
defense at all should be sufficient cause to permit a change of the plea from guilty to not guilty.
...
No sort of pressure can be permitted to bring the party to forego any right or advantage
however slight. The law will not suffer the least weight to be put in the scale against him.
[W]hen there is reason to believe that the plea has been entered through inadvertence . . . and
mainly from the hope that the punishment to which the accused would otherwise be exposed may
thereby be mitigated, the Court should be indulgent in permitting the plea to be withdrawn.

Id. at 20 (citations omitted). A legal annotation from the period stated:
We would conclude, from an examination of all the cases upon the subject, that where there is an
inducement of any kind held out to the prisoner, by reason of which he enters the plea of guilty,
it will . . . better comport with a sound judicial discretion to allow the plea to be withdrawn . . .,
and especially so when counsel and friends represent to the accused that it has been the custom
and common practice of the court to assess a punishment less than the maximum upon such a
plea . . . .

Id. at 24 (quoting M.W. Hopkins, Withdrawal of Plea of Guilty, 11 CRIM. L. MAG. 479, 484
(1889)).
46
See Alschuler, supra note 36, at 22.
47
See id. at 24 (“The gap between these judicial denunciations of plea bargaining [in the
late nineteenth century] and the practices of many urban courts at the turn of the century and
thereafter was apparently extreme. In these courts, striking political corruption apparently
contributed to a flourishing practice of plea bargaining.”).
48
Id. (citations omitted).
49
See id. at 24–25.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

9

This sort of person is an abomination and it is a serious indictment against our system
of criminal administration that such a leech not only can exist but thrive. The “fixer”
is just what the word indicates. As to qualifications, he has none, except that he may
50
be a person of some small political influence.

The use of plea bargaining by such “fixers” ensured that the practice would
survive despite judicial repudiation, though a later phenomenon ultimately
brought it out of the shadows.51
While corruption kept plea bargaining alive during the late nineteenth
and early twentieth centuries, overcriminalization necessitated plea
bargaining’s emergence into mainstream criminal procedure and its rise to
dominance.52 According to one analysis of individuals arrested in Chicago
in 1912, “more than one half were held for violation of legal precepts which
did not exist twenty-five years before.”53 As the number of criminal
statutes—and, as a result, criminal defendants—swelled, court systems
became overwhelmed.54 In searching for a solution, prosecutors turned to
bargained justice, the previous bastion of corruption, as a mechanism by
which official and “legitimate” offers of leniency might ensure defendants
waived their rights to trial and cleared cases from the dockets.55 The
50

Id. This quotation is attributed to Albert J. Harno, Dean, University of Illinois Law
School. See id.
51
See Dervan, supra note 26, at 59 (“While corruption introduced plea bargaining to the
broader legal community, it was the rise in criminal cases before and during Prohibition that
spurred its growth and made it a legal necessity.”).
52
See id.; see also Donald A. Dripps, Overcriminalization, Discretion, Waiver: A Survey
of Possible Exit Strategies, 109 PENN. ST. L. REV. 1155, 1156–61 (2005) (discussing the
relationship between broadening legal rules and plea bargaining); William J. Stuntz, The
Pathological Politics of Criminal Law, 100 MICH. L. REV. 505, 519–20 (2001) (discussing
the influence of broader laws on the rate of plea bargaining). For a definition of
“overcriminalization,” see Lucian E. Dervan, Overcriminalization 2.0: The Symbiotic
Relationship Between Plea Bargaining and Overcriminalization, 7 J.L. ECON. & POL’Y 645,
645–46 (2011). Similarly, consider the significant ramifications that would follow should
there no longer be overcriminalization:
The law would be refined and clear regarding conduct for which criminal liability may attach.
Individual benefits, political pressure, and notoriety would not incentivize the invention of novel
legal theories upon which to base liability where none otherwise exists, despite the already
expansive size of the United States criminal code. Further, novel legal theories and overly-broad
statutes would not be used to create staggering sentencing differentials that coerce defendants,
even innocent ones, to falsely confess in return for leniency.

Id. at 645–46.
53
See Alschuler, supra note 36, at 32.
54
See Dervan, supra note 52, at 650 (“In return for agreeing not to challenge the
government’s legal assertions and for assisting in lessening the strain created by
overcriminalization, defendants were permitted to plead guilty to reduced charges and in
return for lighter sentences.”).
55
See id.

10

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

reliance on bargains during this period is evidenced by the observed rise in
guilty plea rates.56 Between 1908 and 1916, the number of federal
convictions resulting from pleas of guilty rose from 50% to 72%.57
The passage of the Eighteenth Amendment and advent of the
Prohibition era in 1919 only exacerbated the overcriminalization problem
and required further reliance on plea bargaining to ensure the continued
functionality of the justice system.58 As George Fisher noted in his seminal
work on plea bargaining, prosecutors had little option other than to continue
attempting to create incentives for defendants to avoid trial.59 By 1925,
almost 90% of criminal convictions were the result of guilty pleas.60 By the
end of the Prohibition era, plea bargaining had successfully emerged from
the shadows of the American criminal justice system to take its current
place as an indispensable solution for an overwhelmed structure.61
Though plea-bargaining rates rose significantly in the early twentieth
century, appellate courts were still reluctant to approve such deals when
appealed.62 For example, in 1936, Jack Walker was charged with armed
robbery.63 In a scene common in today’s criminal justice system,
prosecutors threatened to seek a harsh sentence if Walker failed to
cooperate, but offered a lenient alternative in return for a guilty plea.64
Facing a sentence twice as long if he lost at trial, Walker pleaded guilty. 65
The United States Supreme Court found the bargain constitutionally
impermissible, noting that the threats and inducements had made Walker’s
plea involuntary.66
56

See Alschuler, supra note 36, at 33.
See id. at 27.
58
See Scott Schaeffer, The Legislative Rise and Populist Fall of the Eighteenth
Amendment: Chicago and the Failure of Prohibition, 26 J.L. & POL. 385, 391–98 (2011)
(discussing the history of the passage of the Eighteenth Amendment).
59
See FISHER, PLEA BARGAINING’S TRIUMPH, supra note 37, at 210; see also Alschuler,
supra note 36, at 28 (“The rewards associated with pleas of guilty were manifested not only
in the lesser offenses of which guilty-plea defendants were convicted but also in the lighter
sentences that they received.”).
60
Alschuler, supra note 36, at 27.
61
See Dervan, supra note 26, at 60 (“As Prohibition was extinguished, the United States
continued its drive to create new criminal laws, a phenomenon that only added to the courts’
growing case loads and the pressure to continue to use bargaining to move cases through the
system.”).
62
See, e.g., Walker v. Johnston, 312 U.S. 275, 279–80 (1941).
63
See id.
64
See id. at 280.
65
Id. at 281.
66
See id. at 279–86; see also Hallinger v. Davis, 146 U.S. 314, 324 (1892) (requiring
that defendant voluntarily avail himself of the option to plead guilty).
57

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

11

[Walker] was deceived and coerced into pleading guilty when his real desire was to
plead not guilty or at least to be advised by counsel as to his course. If he did not
voluntarily waive his right to counsel, or if he was deceived or coerced by the
67
prosecutor into entering a guilty plea, he was deprived of a constitutional right.

Once again, despite plea bargaining’s continued presence in the court
system, the Supreme Court was reluctant to embrace the notion of
bargained justice and coerced confessions.68
By 1967, despite a continued rejection of plea bargaining by appellate
courts, even the American Bar Association (ABA) was beginning to see the
benefits of the practice.69 In a report regarding the criminal justice system,
the ABA noted that the use of plea bargaining allowed for the resolution of
many cases without a trial, which was necessary given the system’s lack of
resources.70 In particular, the report noted that “the limited use of the trial
process for those cases in which the defendant has grounds for contesting
the matter of guilt aids in preserving the meaningfulness of the presumption
of innocence.”71
67

Walker, 312 U.S. at 286; see also ALISA SMITH & SEAN MADDAN, NAT’L ASS’N OF
CRIM. DEF. LAWYERS, THREE-MINUTE JUSTICE: HASTE AND WASTE IN FLORIDA’S
MISDEMEANOR COURTS 15 (2011) (noting that a study of misdemeanor cases in Florida
courts found that 66% of defendants appeared at arraignment without counsel and almost
70% of defendants pleaded guilty or no contest at arraignment). According to the NACDL
report, “[t]rial judges failed to advise the unrepresented defendants of their right to counsel
in open court . . . only 27% of the time.” Id. In less than 50% of the cases, the judges asked
the defendants if they wanted an attorney. See id. Finally, the report stated, “only about
one-third of the time did the trial judge discuss the importance and benefits of counsel or
disadvantages of proceeding without counsel.” Id.
68
During the period between 1941 and 1970, several additional appellate cases
challenged the constitutionality of plea bargaining. See, e.g., United States v. Jackson, 390
U.S. 570, 571–72 (1968) (striking down a statute that allowed for the death penalty only
when a defendant failed to plead guilty and moved forward with a jury trial as an
“impermissible burden upon the exercise of a constitutional right”); Machibroda v. United
States, 368 U.S. 487, 491–93 (1962) (finding a prosecutor’s offer of leniency and threats of
additional charges an improper inducement that stripped the voluntariness of defendant’s
guilty plea); Shelton v. United States, 242 F.2d 101, 113 (5th Cir. 1957), judgment set aside,
246 F.2d 571 (5th Cir. 1957) (en banc), rev’d per curiam, 356 U.S. 26 (1958) (involving a
defendant the court determined was induced to plead guilty by the promise of a light
sentence and the dismissal of other pending charges). In Shelton, the court stated, “[j]ustice
and liberty are not the subjects of bargaining and barter.” 242 F.2d at 113.
69
See AM. BAR ASS’N, PROJECT ON MINIMUM STANDARDS FOR CRIMINAL JUSTICE,
STANDARDS RELATING TO P LEAS OF GUILTY 2 (Tentative Draft 1967) [hereinafter ABA
PROJECT].
70
See id.
71
Id.
[A] high proportion of pleas of guilty and nolo contendere does benefit the system. Such pleas
tend to limit the trial process to deciding real disputes and, consequently, to reduce the need for
funds and personnel. If the number of judges, courtrooms, court personnel and counsel for

12

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

Three years after the ABA embraced plea bargaining as a necessary
tool in an overburdened system, the United States Supreme Court finally
directly addressed the constitutionality of modern plea bargaining in the
case of Brady v. United States.72 The case involved a defendant charged
with kidnapping in violation of federal law.73 The charged statute permitted
the death penalty, but only where recommended by a jury.74 This meant
that a defendant could avoid capital punishment by pleading guilty.75
Realizing his chances of success at trial were minimal given that his
codefendant had agreed to testify against him, Brady pleaded guilty and
was sentenced to fifty years in prison.76 He later changed his mind,
however, and sought to have his plea withdrawn, arguing that his act was
induced by his fear of the death penalty.77
Prior precedent regarding plea bargaining suggested that the Supreme
Court would look with disfavor upon the defendant’s decision to plead
guilty in return for the more lenient sentence, but plea bargaining’s rise
during the previous century and its unique role by 1970 protected the
practice from absolute condemnation.78 Instead of finding plea bargaining
unconstitutional, the Court acknowledged the necessity of the institution to
protect crowded court systems from collapse.79 The Court then went on to
prosecution and defense were to be increased substantially, the funds necessary for such
increases might be diverted from elsewhere in the criminal justice process. Moreover, the
limited use of the trial process for those cases in which the defendant has grounds for contesting
the matter of guilt aids in preserving the meaningfulness of the presumption of innocence.

Id.
72

See Brady v. United States, 397 U.S. 742, 743 (1970).
See id. Interestingly, the defendant in Brady was charged under the same federal
statute at issue in the 1968 case of United States v. Jackson. See Jackson, 390 U.S. at 583;
see also Dervan, supra note 26, at 75–76 (“With regard to the federal kidnapping statute,
[the Jackson court stated that] the threat of death only for those who refuse to confess their
guilt is an example of a coercive incentive that makes any resulting guilty plea invalid.”).
74
The law, 18 U.S.C. § 1201(a), read as follows:
73

Whoever knowingly transports in interstate . . . commerce, any person who has been
unlawfully . . . kidnap[p]ed . . . and held for ransom . . . or otherwise . . . shall be punished (1) by
death if the kidnap[p]ed person has not been liberated unharmed, and if the verdict of the jury
shall so recommend, or (2) by imprisonment for any term of years or for life, if the death penalty
is not imposed.

Jackson, 390 U.S. at 570–71.
75
See Brady, 397 U.S. at 743.
76
See id. at 743–44.
77
See id. at 744.
78
See supra notes 44–68 and accompanying text.
79
See Brady, 397 U.S. at 752–58; see also Miranda v. Arizona, 384 U.S. 436 (1966)
(describing the protection against self-incrimination); Gideon v. Wainwright, 372 U.S. 335
(1963) (describing the right to counsel); Mapp v. Ohio, 367 U.S. 643 (1961) (describing the

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

13

describe the type of bargains that would be acceptable80:
Of course, the agents of the State may not produce a plea by actual or threatened
physical harm or by mental coercion overbearing the will of the defendant. But
nothing of the sort is claimed in this case; nor is there evidence that Brady was so
gripped by fear of the death penalty or hope of leniency that he did not or could not,
with the help of counsel, rationally weigh the advantages of going to trial against the
81
advantages of pleading guilty.

The Court continued:
[A] plea of guilty entered by one fully aware of the direct consequences, including the
actual value of any commitments made to him by the court, prosecutor, or his own
counsel, must stand unless induced by threats (or promises to discontinue improper
harassment), misrepresentation (including unfulfilled or unfulfillable promises), or
perhaps by promises that are by their nature improper as having no proper relationship
82
to the prosecutor’s business (e.g. bribes).

After Brady, plea bargaining was permitted and could fully emerge into the
mainstream of the American criminal justice system.83 As long as the plea
was “voluntary,” which meant that it was not induced “by actual or
threatened physical harm or by mental coercion overbearing the will of the
defendant,” the bargain would be permitted.84
Plea bargaining continued its rise over the next four decades and,
today, over 96% of convictions in the federal system result from pleas of
guilt rather than decisions by juries.85 While plea bargaining was a
exclusionary rule); Dervan, supra note 26, at 81 (“[T]he Supreme Court imposed the
‘exclusionary rule’ for violations of the Fourth Amendment, granted the right to counsel, and
imposed the obligation that suspects be informed of their rights prior to being
interrogated.”).
80
See Brady, 397 U.S. at 750–51.
81
Id.
82
Id. at 755 (quoting Shelton v. United States, 246 F.2d 571, 572 n.2 (5th Cir. 1957) (en
banc), rev’d per curiam, 356 U.S. 26 (1958)). Interestingly, the language used by the
Supreme Court in Brady is the same as language proposed by the United States Court of
Appeals for the Fifth Circuit several years earlier to address “voluntariness.” See Shelton v.
United States, 242 F.2d 101, 115, judgment set aside, 246 F.2d 571 (5th Cir. 1957) (en
banc), rev’d per curiam, 356 U.S. 26 (1958). The Shelton case almost rose to the United
States Supreme Court for review of the constitutionality of plea bargaining in 1958, but was
surreptitiously withdrawn prior to argument after the government admitted that the guilty
plea may have been improperly obtained. See Dervan, supra note 26, at 73 (“According to
Professor Albert Alschuler, evidence indicates that the government likely confessed its error
for fear that the Supreme Court would finally make a direct ruling that all manner of plea
bargaining was wholly unconstitutional.”).
83
See Brady, 397 U.S. at 750–55.
84
Id. at 750.
85
See U.S. SENTENCING COMM’N, 2011 SOURCEBOOK OF FEDERAL SENTENCING
STATISTICS, fig.C, available at http://www.ussc.gov/Data_and_Statistics/Annual_

14

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

powerful force in 1970, the ability of prosecutors to create significant
incentives for defendants to accept plea offers grew exponentially after
Brady with the implementation of sentencing guidelines throughout much
of the country.86 As one commentator explained, “By assigning a fixed and
narrow penalty range to almost every definable offense, sentencing
guidelines often empower prosecutors to dictate a defendant’s sentence by
manipulating the charges.”87 Through charge selection and influence over
sentencing ranges, prosecutors today possess striking powers to create
significant sentencing differentials, a term used to describe the difference
between the sentence a defendant faces if he or she pleads guilty versus the
sentence risked if he or she proceeds to trial and is convicted.88 Many have
Reports_and_Sourcebooks/2011/FigureC.pdf.
86
See FISHER, PLEA BARGAINING’S TRIUMPH, supra note 37, at 210 (“[Sentencing
Guidelines] invest prosecutors with the power, moderated only by the risk of loss at trial, to
dictate many sentences simply by choosing one set of charges over another.”); see also Mary
Patrice Brown & Stevan E. Bunnell, Negotiating Justice: Prosecutorial Perspectives on
Federal Plea Bargaining in the District of Columbia, 43 AM. CRIM. L. REV. 1063, 1066–67
(2006) (“Like most plea agreements in federal or state courts, the standard D.C. federal plea
agreement starts by identifying the charges to which the defendant will plead guilty and the
charges or potential charges that the government in exchange agrees not to prosecute.”);
Geraldine Szott Moohr, Prosecutorial Power in an Adversarial System: Lessons from
Current White Collar Cases and the Inquisitorial Model, 8 BUFF. CRIM. L. REV. 165, 177
(2004) (“The power of the prosecutor to charge is two-fold; the power to indict or not . . .
and the power to decide what offenses to charge.”); Joy A. Boyd, Comment, Power, Policy,
and Practice: The Department of Justice’s Plea Bargaining Policy as Applied to the Federal
Prosecutor’s Power Under the United States Sentencing Guidelines, 56 ALA. L. REV. 591,
592 (2004) (“Not only may a prosecutor choose whether to pursue any given case, but she
also decides which charges to file.”); Jon J. Lambiras, Comment, White-Collar Crime: Why
the Sentencing Disparity Despite Uniform Guidelines?, 30 PEPP. L. REV. 459, 512 (2003)
(“Charging decisions are a critical sentencing matter and are left solely to the discretion of
the prosecutor. When determining which charges to bring, prosecutors may often choose
from more than one statutory offense.”).
87
FISHER, P LEA BARGAINING’S TRIUMPH, supra note 37, at 17; see also Marc L. Miller,
Domination & Dissatisfaction: Prosecutors as Sentencers, 56 STAN. L. REV. 1211, 1252
(2004) (“The overwhelming and dominant fact of the federal sentencing system, beyond the
Commission and the guidelines and mandatory penalties, is the virtually absolute power the
system has given prosecutors over federal prosecution and sentencing.”); Boyd, supra note
86, at 591–92 (“While the main focus of the Sentencing Guidelines appeared to be
narrowing judicial discretion in sentencing, some critics argued that the Sentencing
Guidelines merely shifted the federal judges’ discretionary power to federal prosecutors.”).
88
See Alschuler, supra note 43, at 652–53. Professor Alschuler stated, “Criminal
defendants today plead guilty in overwhelming numbers primarily because they perceive that
this action is likely to lead to more lenient treatment than would follow conviction at trial. A
number of studies suggest that this perception is justified.” Id. at 652–53. Among the
studies cited by Professor Alschuler in support of his statement are the following: MARVIN
ZALMAN ET AL., SENTENCING IN MICHIGAN: REPORT OF THE MICHIGAN FELONY SENTENCING
PROJECT 268 (1979) (noting that proceeding to trial tended to increase the probability of
serving prison time); H. Joo Shin, Do Lesser Pleas Pay?: Accommodations in the Sentencing

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

15

surmised that the larger the sentencing differential, the greater the
likelihood a defendant will forego his or her right to trial and accept the
deal.89
B. PLEA BARGAINING’S INNOCENCE DEBATE
In 2004, Lea Fastow, wife of former Enron Chief Financial Officer
and Parole Processes, 1 J. CRIM. JUST. 27, 31 (1973) (noting that defendants charged with
robbery and felonious assault who proceeded to trial received sentences almost twice as long
as those who pleaded guilty); Franklin E. Zimring et al., Punishing Homicide in
Philadelphia: Perspectives on the Death Penalty, 43 U. CHI. L. REV. 227, 236 (1976) (noting
that no homicide defendants who pleaded guilty received a sentence of life or death, as
compared to 29% of those convicted at trial); Patrick R. Oster & Roger Simon, Jury Trial a
Sure Way to Increase the Rap, CHI. SUN TIMES, Sept. 17, 1973, at 4 (noting a disparity
between sentences of murder defendants who pleaded guilty and those who proceeded to
trial); see also Alschuler, supra note 43, at 653 n.2; Stephanos Bibas, Bringing Moral Values
into a Flawed Plea-Bargaining System, 88 CORNELL L. REV. 1425, 1425 (2003) (“The
criminal justice system uses large sentence discounts to induce guilty pleas. Of course these
discounts exert pressure on defendants to plead guilty.”); Dervan, supra note 26, at 64
(“[P]lea bargaining’s rise to dominance during the nineteenth and twentieth centuries
resulted from prosecutors gaining increased power over the criminal justice system and,
through such power, the ability to offer increasingly significant incentives to those willing to
confess their guilt in court.”); Lucian E. Dervan, The Surprising Lessons from Plea
Bargaining in the Shadow of Terror, 27 GA. ST. U. L. REV. 239, 245 (2011) (“Key to the
success of prosecutors’ use of increasing powers to create incentives that attracted
defendants was their ability to structure plea agreements that included significant differences
between the sentence one received in return for pleading guilty and the sentence one risked if
he or she lost at trial.”).
89
One study analyzed robbery and burglary defendants in three California jurisdictions
and found that defendants who went to trial received significantly higher sentences. See
David Brereton & Jonathan D. Casper, Does It Pay to Plead Guilty? Differential Sentencing
and the Functioning of Criminal Courts, 16 LAW & SOC’Y REV. 45, 55–59 (1981–1982);
Daniel Givelber, Punishing Protestations of Innocence: Denying Responsibility and Its
Consequences, 37 AM. CRIM. L. REV. 1363, 1382 (2000) (“The differential in sentencing
between those who plead and those convicted after trial reflects the judgment that defendants
who insist upon a trial are doing something blameworthy.”); Shin, supra note 88, at 27
(finding that charge reduction directly results in reduction of the maximum sentence
available and indirectly results in lesser actual time served); Tung Yin, Comment, Not a
Rotten Carrot: Using Charges Dismissed Pursuant to a Plea Agreement in Sentencing
Under the Federal Guidelines, 83 CALIF. L. REV. 419, 443 (1995) (“Curiously, the arena of
plea bargaining pits the concepts of duress and consideration against each other: a large
sentencing differential makes it more likely that a defendant is coerced into pleading guilty,
and yet it also increases the benefit offered in exchange for the guilty plea.”). The Brereton
and Casper study stated:
The point of the preceding discussion is simple enough: when guilty plea rates are high, expect
to find differential sentencing. We believe that recent arguments to the effect that differentials
are largely illusory do not withstand serious scrutiny, even though this revisionist challenge has
been valuable in forcing us to examine more closely what is too often taken to be self-evidently
true.

Brereton & Casper, supra, at 89.

16

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

Andrew Fastow, was accused of engaging in six counts of criminal conduct
related to the collapse of the Texas energy giant.90 Though conviction at
trial under the original indictment carried a prison sentence of ten years
under the Federal Sentencing Guidelines, the government offered Fastow a
plea bargain.91 In return for assisting in their prosecution, she would be
eligible for a mere five months in prison.92 With small children to consider
and a husband who would certainly receive a lengthy prison sentence,
Fastow accepted the offer.93 The question that remained, however, was
whether Fastow had pleaded guilty because she had committed the alleged
offenses, or whether the plea bargaining machine had become so powerful
that even innocent or questionably guilty defendants were now becoming
mired in its powerful grips.94
90
See Indictment, United States v. Fastow, Cr.No. H-03- (S.D. Tex. Apr. 30, 2003),
available at http://fl1.findlaw.com/news.findlaw.com/hdocs/docs/enron/usleafstw43003ind.
pdf; see also Michelle S. Jacobs, Loyalty’s Reward—A Felony Conviction: Recent
Prosecutions of High-Status Female Offenders, 33 FORDHAM URB. L.J. 843 (2006); Mary
Flood, Lea Fastow in Plea-Bargain Talks, HOUS. CHRON., Nov. 7, 2003, at 1A.
91
See Bruce Zucker, Settling Federal Criminal Cases in the Post-Enron Era: The Role of
the Court and Probation Office in Plea Bargaining Federal White Collar Cases, 6 FLA.
COASTAL L. REV. 1, 3–5 (2004). The ten-year sentence is calculated using the 2002
sentencing guidelines for fraud and the allegations contained in Fastow’s indictment. Given
an alleged loss amount of $17 million and more than fifty victims, Fastow, who had no prior
criminal record, faced a sentencing range of 97–121 months. See U.S. SENTENCING
GUIDELINES MANUAL § 2B1.1 & ch. 5, pt. A (2002).
92
See Zucker, supra note 91, at 3. In Fastow’s eventual plea agreement, the prosecutors
used a federal misdemeanor charge as a mechanism by which to ensure the judge could not
sentence Fastow beyond the terms of the arrangement. See Mary Flood, Fastows to Plead
Guilty Today, HOUS. CHRON., Jan. 14, 2004, at 1A.
93
See Greg Farrell & Jayne O’Donnell, Plea Deals Appear Close for Fastows, USA
TODAY, Jan. 8, 2004, § B, at 1 (“One of the reasons that Lea Fastow wants to limit her jail
time to five months is that she and her husband have two young children, and they’re trying
to structure their pleas so they’re not both in jail at the same time.”); see also Flood, supra
note 92, at A1 (“The plea bargains for the Fastows, who said they wanted to be sure their
two children are not left parentless, have been in limbo for more than a week.”).
Interestingly, the judge in the case later rejected the government’s attempts to utilize a
binding plea agreement containing the five-month offer. See Farrell & O’Donnell, supra,
§ B, at 1 (“U.S. District Judge David Hittner told Lea Fastow Wednesday that he refused to
be locked in to the five-month prison sentence that her lawyers had negotiated with
prosecutors.”). In response, the government withdrew the original charges and allowed Lea
Fastow to plead guilty to a single misdemeanor tax charge. See New Plea Bargain for Lea
Fastow in Enron Case, N.Y. TIMES, Apr. 30, 2004, at C13. The judge then sentenced her to
one year in prison. See Lea Fastow Enters Prison, CNNMONEY (July 12, 2004, 12:52 PM),
http://money.cnn.com/2004/07/12/news/newsmakers/lea_fastow/index.htm.
94
See Dervan, supra note 26, at 56 (“Today, the incentives to bargain are powerful
enough to force even an innocent defendant to falsely confess guilt in hopes of leniency and
in fear of reprisal.”); see also Larry E. Ribstein, Agents Prosecuting Agents, 7 J.L. ECON. &
POL’Y 617, 628 (2011) (“[P]rosecutors can avoid having to test their theories at trial by using

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

17

It is unclear how many of the more than 96% of defendants who are
convicted through pleas of guilt each year are actually innocent of the
charged offenses, but it is clear that plea bargaining has an innocence
problem.95 As Professor Russell D. Covey has stated, “When the deal is
good enough, it is rational to refuse to roll the dice, regardless of whether
one believes the evidence establishes guilt beyond a reasonable doubt, and
regardless of whether one is factually innocent.”96 While almost all
commentators agree with Covey’s statement that some innocent defendants
will be induced to plead guilty, much debate exists regarding the extent of
this phenomenon.97
Some argue that plea bargaining’s innocence problem is significant
significant leverage to virtually force even innocent, or at least questionably guilty,
defendants to plead guilty.”).
95
See Michael O. Finkelstein, A Statistical Analysis of Guilty Plea Practices in the
Federal Courts, 89 HARV. L. REV. 293, 295 (1975) (“[T]he pressure on defendants to plead
guilty in the federal courts has induced a high rate of conviction by ‘consent’ in cases in
which no conviction would have been obtained if there had been a contest.”); Robert E. Scott
& William J. Stuntz, Plea Bargaining as Contract, 101 YALE L.J. 1909, 1950–51 (1992)
(discussing plea bargaining’s innocence problem); David L. Shapiro, Should a Guilty Plea
Have Preclusive Effect?, 70 IOWA L. REV. 27, 27 (1984); see also Covey, supra note 43, at
74 (“Plea bargaining has an innocence problem.”); Oren Gazal-Ayal, Partial Ban on Plea
Bargains, 27 CARDOZO L. REV. 2295, 2295–96 (2006) (arguing for a partial ban on plea
bargaining to reduce the likelihood innocent defendants will plead guilty); Andrew D.
Leipold, How the Pretrial Process Contributes to Wrongful Convictions, 42 AM. CRIM. L.
REV. 1123, 1154 (2005).
96
Russell D. Covey, Longitudinal Guilt: Repeat Offenders, Plea Bargaining, and the
Variable Standard of Proof, 63 FLA. L. REV. 431, 450 (2011) (“The risk of inaccurate results
in the plea bargaining system thus seems substantial.”); see also Gregory M. Gilchrist, Plea
Bargains, Convictions and Legitimacy, 48 AM. CRIM. L. REV. 143, 148 (2011).
That plea bargaining represents something of an affront to the rule against coerced confessions
has been oft-noted and more often ignored. The objections that have been leveled against plea
bargaining are numerous and diverse, but most stem from a common problem: plea bargaining
reduces the ability of the criminal justice system to avoid convicting the innocent.

Gilchrist, supra, at 148; see also Gazal-Ayal, supra note 95, at 2306 (“In all these cases, an
innocent defendant might accept the offer in order to avoid the risk of a much harsher result
if he is convicted at trial, and thereby plea bargaining could very well lead to the conviction
of factually innocent defendants.”); Leipold, supra note 95, at 1154 (“Yet we know that
sometimes innocent people plead guilty, and we know some of the reasons why . . .
[S]ometimes the prosecutor offers such a generous discount for admitting guilt that the
defendant feels he simply can’t take the chance of going to trial.”).
97
It is worth mentioning that even Joan of Arc and Galileo Galilei fell victim to the
persuasions of plea bargaining. See Alschuler, supra note 36, at 41 (“[Joan of Arc]
demonstrated that even saints are sometimes unable to resist the pressures of plea
negotiation.”); Kathy Swedlow, Pleading Guilty v. Being Guilty: A Case for Broader Access
to Post-Conviction DNA Testing, 41 CRIM. L. BULL. 575, 575 (2005) (describing Galileo’s
decision to admit his belief in the theory that the earth was the center of the universe in
return for a lighter sentence).

18

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

and brings into question the legitimacy of the entire criminal justice
system.98 Professor Ellen S. Podgor wrote recently of plea bargaining,
“[O]ur existing legal system places the risk of going to trial, and in some
cases even being charged with a crime, so high, that innocence and guilt no
longer become the real considerations.”99 But even for those who believe
that plea bargaining leads to large numbers of innocent defendants pleading
guilty, an uncertainty persists regarding exactly how susceptible innocent
defendants are to bargained justice.100 This is troubling because it prevents
an accurate assessment of what must be done in response to this potential
injustice.101
Others argue, however, that plea bargaining’s innocence problem is
“exaggerated” and the likelihood of persuading an innocent defendant to
falsely confess is minimal.102 This argument rests, in part, on a perception
98

See Dervan, supra note 26, at 97 (“That plea-bargaining today has a significant
innocence problem indicates that the Brady safety-valve has failed and, as a result, the
constitutionality of modern day plea bargaining is in great doubt.”); Gilchrist, supra note 96,
at 147 (“By failing to generate results correlated with the likely outcome at trial, plea
bargaining undermines the legitimacy of the criminal justice system.”); F. Andrew Hessick
III & Reshma M. Saujani, Plea Bargaining and Convicting the Innocent: The Role of the
Prosecutor, the Defense Counsel, and the Judge, 16 BYU J. PUB. L. 189, 197 (2002) (“While
the concept of convicting an innocent person is a terrible imperfection of our justice system,
an innocent person pleading guilty is inexcusable.”).
99
Ellen S. Podgor, White Collar Innocence: Irrelevant in the High Stakes Risk Game, 85
CHI.-KENT L. REV. 77, 77–78 (2010); see also Covey, supra note 43, at 80 (“[A]s long as the
prosecutor is willing and able to discount plea prices to reflect resource savings, regardless
of guilt or innocence, pleading guilty is the defendant’s dominant strategy. As a result, nonfrivolous accusation—not proof beyond a reasonable doubt—is all that is necessary to
establish legal guilt.”).
100
See Dervan, supra note 26, at 96–97 (discussing plea bargaining’s innocence
problem, but acknowledging that the exact impact of bargained justice on innocent
defendants is, as of yet, unknown); see also Scott W. Howe, The Value of Plea Bargaining,
58 OKLA. L. REV. 599, 631 (2005) (“The number of innocent defendants who accept
bargained guilty pleas is uncertain.”).
101
See Ric Simmons, Private Plea Bargains, 89 N.C. L. REV. 1125, 1173 (2011) (“If the
plea bargaining process is indeed a reasonable replacement for a trial, then plea bargaining
should be encouraged . . . On the other hand, if the results are dependent on factors
unrelated to what would occur at trial, then society should work to reform, limit, or abolish
the practice.”).
102
See Shapiro, supra note 95, at 40 (“[Plea bargaining’s] defenders deny that the
chances of convicting the innocent are substantial . . . .”); Avishalom Tor et al., Fairness and
the Willingness to Accept Plea Bargain Offers, 7 J. EMPIRICAL LEGAL STUD. 97, 114 (2010)
(“[I]f innocents tend to reject offers that guilty defendants accept, the concern over the
innocence problem may be exaggerated.”); Oren Gazal-Ayal & Limor Riza, PleaBargaining and Prosecution 13 (European Ass’n of Law & Econ., Working Paper No. 0132009, 2009) (“Since trials are designed to reveal the truth, an innocent defendant would
correctly estimate that his chances at trial are better than the prosecutor’s offer suggests. As
a result, innocent defendants tend to reject offers while guilty defendants tend to accept

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

19

that innocent defendants will reject prosecutors’ plea offers and instead will
proceed to trial backed by the belief that their factual innocence will protect
them from conviction.103 One commentator noted that supporters of the
plea-bargaining system believe “[p]lea agreements are not forced on
defendants . . . they are only an option. Innocent defendants are likely to
reject this option because they expect an acquittal at trial.”104
Such skeptics are in good company. Even the Supreme Court in its
landmark Brady decision permitting bargained justice rejected concerns that
innocent defendants would falsely confess to crimes they did not commit.105
The Court stated:
We would have serious doubts about this case if the encouragement of guilty pleas by
offers of leniency substantially increased the likelihood that defendants, advised by
competent counsel, would falsely condemn themselves. But our view is to the
contrary and is based on our expectations that courts will satisfy themselves that pleas
of guilty are voluntarily and intelligently made by competent defendants with
adequate advice of counsel and that there is nothing to question the accuracy and
reliability of the defendants’ admissions that they committed the crimes with which
106
they are charged.

This sentiment was expressed by the Court again eight years later in
Bordenkircher v. Hayes.107 In Bordenkircher, the Court stated that as long
as the defendant is free to accept or reject a plea bargain, it is unlikely an
innocent defendant will be “driven to false self-condemnation.”108 Even
those who argue that plea bargaining’s innocence problem is exaggerated,
however, rely mainly on speculation regarding how innocent defendants
will respond in such situations.109
them.”); see also Josh Bowers, Punishing the Innocent, 156 U. PA. L. REV. 1117, 1165
(2008).
When an innocent defendant rationally chooses to plead guilty, the system should want to protect
access. It should recognize that at least for the innocent defendant it is not bad that some deals
are more than just sensible—they would be improvident to reject. Particularly where process
costs are high and the consequences of conviction low, a bargained-for conviction of an innocent
accused is no evil; it is the constructive minimization thereof—an unpleasant medicine softening
the symptoms of separate affliction.

Bowers, supra, at 1165.
103
See Gazal-Ayal, supra note 95, at 2298.
104
See id.
105
See Brady v. United States, 397 U.S. 742, 757–58 (1970).
106
Id. at 758.
107
Bordenkircher v. Hayes, 434 U.S. 357 (1978).
108
Id. at 363 (“Indeed, acceptance of the basic legitimacy of plea bargaining necessarily
implies rejection of any notion that a guilty plea is involuntary in a constitutional sense
simply because it is the end result of the bargaining process.”).
109
See supra notes 102–104 and infra notes 111–123 and accompanying text.

20

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

The need by both sides of the innocence debate to gather more data
regarding the extent to which innocent defendants might be vulnerable to
the persuasive power of plea bargaining has led to numerous studies.110
Several legal scholars have conducted examinations of exoneration statistics
in an effort to identify examples where innocent defendants were convicted
by guilty pleas.111 Professor Samuel Gross conducted one of the most
comprehensive studies in 2005.112 While Professor Gross’s research
explored exonerations in the United States broadly, he also specifically
discussed plea bargaining’s innocence problem.113 His study stated that
twenty of 340 exonerees had pleaded guilty.114 Although Professor Gross
found a relatively low number among those exonerated who falsely pleaded
guilty, there are significant limitations to using this study to disprove the
innocence problem surrounding guilty pleas.115 Upon closer examination of
this and other exoneration studies, one realizes that while exoneration data
is vital to our understanding of wrongful convictions generally, it cannot
accurately or definitively explain how likely innocent defendants are to
110

See infra note 111.
See Baldwin & McConville, supra note 34, at 296–98 (discussing plea bargaining’s
innocence problem in England); Brandon L. Garrett, Judging Innocence, 108 COLUM. L.
REV. 55, 74 (2008) (noting that nine of the first two hundred individuals exonerated by the
Innocence Project had pleaded guilty); Samuel R. Gross et al., Exonerations in the United
States 1989 Through 2003, 95 J. CRIM. L. & CRIMINOLOGY 523, 524, 536 (2005) (examining
the number of persons exonerated who pleaded guilty); D. Michael Risinger, Innocents
Convicted: An Empirically Justified Factual Wrongful Conviction Rate, 97 J. CRIM. L. &
CRIMINOLOGY 761, 778–79 (2007) (examining DNA exonerations for capital rape–murder
convictions); George C. Thomas III, Two Windows into Innocence, 7 OHIO ST. J. CRIM. L.
575, 577–78 (2010) (“McConville and Baldwin concluded that two percent of the guilty
pleas were of doubtful validity. As there were roughly two million felony cases filed in
2006, if two percent result in conviction of an innocent defendant, 40,000 wrongful felony
convictions occur per year.”).
112
See Gross et al., supra note 111, at 523.
113
See id. at 524, 536.
114
Id. (observing that of this number, fifteen were murder defendants, four were rape
defendants, and one was a gun-possession defendant facing life in prison as a habitual
offender). Professor Gross goes on to note that in two cases of mass exoneration involving
police misconduct, a subset of cases not included in his study, a significant number of the
defendants pleaded guilty. See id. (“By contrast, thirty-one of the thirty-nine Tulia
defendants pled guilty to drug offenses they did not commit, as did the majority of the 100 or
more exonerated defendants in the Rampart scandal in Los Angeles.”).
115
See Howe, supra note 100, at 631 (“Particularly if many innocent defendants who go
to trial are acquitted, [Professor Gross’s] figure does not support claims that innocent
defendants are generally more risk averse regarding trials than factually guilty defendants or
that prosecutors frequently persuade innocent defendants with irresistibly low plea offers.”).
Howe goes on, however, to caution those who might rely on this study in such a manner
because of the difficulty in gaining an exoneration following a guilty plea as opposed to
following a conviction by trial. See id.
111

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

21

plead guilty.116
As noted by other scholars in the field, three problems exist with
exoneration data when applied to plea-bargaining research.117 First,
exoneration data predominantly focuses on serious felony cases such as
murder or rape where there is available DNA evidence and where the
defendants’ sentences are lengthy enough for the exoneration process to
work its way through the system.118 This means that exoneration data does
not examine the role of innocence and plea bargaining in the vast majority
of criminal cases, those not involving murder or rape, including
misdemeanor cases.119 Second, because many individuals who plead guilty
do so in return for a reduced sentence, it is highly likely that innocent
defendants who plead guilty have little incentive or insufficient time to
pursue exoneration.120 Finally, even if some innocent defendants who
pleaded guilty had the desire and time to move for exoneration, many
would be prohibited from challenging their convictions by the mere fact
that they had pleaded guilty.121 As such, innocent defendants who plead
guilty are not accurately captured by the exoneration data sets and,

116
See Howe, supra note 100, at 631; Russell Covey, Mass Exoneration Data and the
Causes of Wrongful Convictions 1 (Aug. 22, 2011) (unpublished manuscript), available at
ssrn.com/abstract=1881767.
117
See Howe, supra note 100, at 631; Covey, supra note 116, at 1.
118
See Covey, supra note 116, at 1 (“[The post-conviction testing of DNA] dataset has
significant limitations, chief of which is that it is largely limited to the kinds of cases in
which DNA evidence is available for post-conviction testing.”).
119
The Federal Bureau of Investigation crime statistics indicate that in 2010 there were
1,246,248 violent crimes and 9,082,887 property crimes in the United States. See U.S.
DEP’T OF JUSTICE, F.B.I., CRIME IN THE UNITED STATES, at tbl.1 (2010), available at
http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2010/crime-in-the-u.s.2010/tables/10tb
l01.xls. Of this number, murder accounted for 1.2% and forcible rape accounted for 6.8% of
the violent crimes. See id. Further, in 2011, the National Association of Criminal Defense
Attorneys released a report regarding misdemeanor cases in Florida. See SMITH & MADDAN,
supra note 67. The report noted that nearly a half-million misdemeanor cases are filed in
Florida each year, and over 70% of those cases are resolved with a guilty plea at
arraignment. See id. at 10.
120
See Jon B. Gould & Richard A. Leo, One Hundred Years Later: Wrongful
Convictions After a Century of Research, 100 J. CRIM. L. & CRIMINOLOGY 825, 834–35
(2010).
121
See JH Dingfelder Stone, Facing the Uncomfortable Truth: The Illogic of PostConviction DNA Testing for Individuals Who Pleaded Guilty, 45 U.S.F. L. REV. 47, 50–52
(2010) (discussing restrictions on the ability of defendants who pleaded guilty to utilize
postconviction DNA testing); see also Howe, supra note 100, at 631 (“Those relying on
[Professor Gross’s] study, however, should do so cautiously. The proportion of false
convictions due to guilty pleas probably exceeds the exoneration figure from the study,
because pleading guilty, as opposed to being convicted after trial, likely makes subsequent
exoneration more difficult.”).

22

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

therefore, it is highly likely that the true extent of plea bargaining’s
innocence problem is significantly underestimated by these studies.122
Consequently, one must look elsewhere to determine the true likelihood that
an innocent defendant might falsely condemn himself in return for an offer
of leniency in the form of a plea bargain.123
One such source of information are psychological studies regarding
plea bargaining and the decisionmaking processes of defendants in the
criminal justice system.124 Unfortunately, these studies are also problematic
and fail to resolve definitively plea bargaining’s innocence debate because
the majority merely employ vignettes in which participants are asked to
imagine themselves as guilty or innocent and faced with a hypothetical
decision regarding whether to accept or reject a plea offer.125 As a result of
the utilization of such imaginary and hypothetical scenarios, these studies
are unable to capture either the full impact of a defendant’s knowledge that
she is factually innocent or the true gravity of the choices she must make
when standing before the criminal justice system accused of a crime she did

122

Even Professor Gross acknowledges that his study fails to capture many innocent
defendants who plead guilty. In concluding his discussion regarding the Tulia and Rampart
mass exoneration cases, he notes that these cases received attention because they involved
large-scale police corruption. He goes on to state, “If these same defendants had been
falsely convicted of the same crimes by mistake—or even because of unsystematic acts of
deliberate dishonesty—we would never have known.” Gross et al., supra note 111, at 537;
see also Allison D. Redlich & Asil Ali Özdoğru, Alford Pleas in the Age of Innocence, 27
BEHAV. SCI. & L. 467, 468 (2009) (“Determining the prevalence of innocents is
methodologically challenging, if not impossible. There is no litmus test to definitively
determine who is innocent and who is guilty. Exonerations are long, costly, and arduous
processes; efforts towards them are often unsuccessful for reasons having little to do with
guilt or innocence.”).
123
See infra notes 124–140 (discussing psychological studies of plea bargaining).
124
The majority of psychological studies to date have only looked at the phenomenon
from the perspective of the attorney and his or her decisionmaking process. See Vanessa A.
Edkins, Defense Attorney Plea Recommendations and Client Race: Does Zealous
Representation Apply Equally to All?, 35 LAW & HUM. BEHAV. 413, 413 (2011); see also
Greg M. Kramer et al., Plea Bargaining Recommendations by Criminal Defense Attorneys:
Evidence Strength, Potential Sentence, and Defendant Preference, 25 BEHAV. SCI. & L. 573,
573 (2007); Hunter A. McAllister & Norman J. Bregman, Plea Bargaining by Prosecutors
and Defense Attorneys: A Decision Theory Approach, 71 J. APPLIED PSYCHOL. 686, 686
(1986).
125
See Kenneth S. Bordens, The Effects of Likelihood of Conviction, Threatened
Punishment, and Assumed Role on Mock Plea Bargaining Decisions, 5 BASIC & APPLIED
SOC. PSYCHOL. 59, 63–65 (1984) (discussing the methodology of the study); W. Larry
Gregory et al., Social Psychology and Plea Bargaining: Applications, Methodology, and
Theory, 36 J. PERSONALITY & SOC. PSYCHOL. 1521, 1522–28 (1978) (discussing the
methodology of the study); Tor et al., supra note 102, at 103–09 (discussing the
methodology of the study).

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

23

not commit.126 Nevertheless, these studies do offer some preliminary
insights into the world of the innocent defendant’s dilemma.
One of the first psychological studies attempting to understand a
defendant’s plea-bargaining decisionmaking process through the use of
vignettes was conducted by Professors Larry Gregory, John Mowen, and
Darwyn Linder in 1984 (Gregory study).127 In the Gregory study, students
were asked to “imagine that they were innocent or guilty of having
committed an armed robbery.”128 The students were then presented with
the evidence against them and asked to make a decision regarding whether
they would plead guilty or proceed to trial.129 As might be expected, the
study revealed that students imagining themselves to be guilty were
significantly more likely to plead guilty than those who were imagining
themselves to be innocent.130 In the experiment, 18% of the “innocent”
students and 83% of the “guilty” students pleaded guilty.131 While these
results might lend support to the argument that few innocent defendants in
the criminal justice system falsely condemn themselves—if you can
consider 18% to be an insignificant number—the study suffered from its
utilization of hypotheticals.132 As has been shown in social psychological
studies for decades, what people say they will do in a hypothetical situation
126

See supra note 125.
See Gregory et al., supra note 125.
128
Id. at 1522. The Gregory study involved 143 students. Interestingly, the study only
utilized male participants. The study stated, “Since most armed robberies are committed by
men, only male students were used.” Id. The methodological explanation went on to
describe the particulars of the study:
127

After listening to a tape recording of their defense attorney’s summary of the evidence that
would be presented for and against them at their trial, students opened an experimental booklet
that contained information about the charges against them (four versus one), the punishment they
would face if convicted (ten to fifteen years in prison versus one to two years in prison), and the
details of the plea bargain that was offered them. Students then indicated whether they accepted
or rejected the plea bargain, responded to manipulation checks, indicated their perceived
probability of conviction, and indicated how sure their defense attorney and the judge were of
their innocence or guilt.

Id.
129

Id. The study also discussed the results of different students facing differing
punishments and numbers of charges. The study found that the severity of punishment and
the number of charges only affected the guilty condition, not the innocent condition. Those
in the guilty condition behaved as would be expected: most likely to accept a plea with a
large number of charges and a severe penalty attached (100%), and least likely with a few
number of charges and a low penalty attached (63%). The innocent defendants had a low
rate of plea bargaining regardless of condition (11%–33%). Id. at 1524, tbl.1.
130
See id. at 1524–26.
131
See id.
132
See supra notes 125–126 and accompanying text.

24

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

and what they would do in reality are two very different things.133
Perhaps acknowledging the unreliable nature of a study relying merely
on vignettes to explore such an important issue, Gregory attempted to create
a more realistic innocent defendant’s dilemma in a subsequent
experiment.134 In the study, students were administered a “difficult exam
after being given prior information by a confederate that most of the
answers were ‘B’ (guilty condition) or after being given no information
(innocent condition).”135 After the test, the students were accused of the
“crime” of having prior knowledge of the answers and told they would have
to appear before an ethics committee.136 The participants were then offered
a plea bargain that required their immediate admission of guilt in return for
a less severe punishment.137 Unfortunately, the second study was only
successfully administered to sixteen students, too few to draw any
significant conclusions.138 Nevertheless, Gregory was finally on the right
path to answering the lingering question pervading plea bargaining’s
innocence debate. How likely is it that an innocent defendant might falsely
plead guilty to a crime he or she did not commit?139
III. LABORATORY EVIDENCE OF PLEA BARGAINING’S INNOCENCE
PROBLEM
In 2006, a wave of new accounting scandals pervaded the American
corporate landscape.140 According to federal prosecutors, numerous
companies were backdating stock options for senior executives to increase
compensation without disclosing such expenses to the public as required by
133

See Richard E. Nisbett & Timothy D. Wilson, Telling More Than We Can Know:
Verbal Reports on Mental Processes, 84 PSYCHOL. REV. 231, 246 (1977).
134
See Gregory et al., supra note 125, at 1526–27.
135
Id. at 1526.
136
See id.
137
See id.
138
See id. at 1528. The results of the second study by Gregory and colleagues were that
six of eight guilty students accepted the deal and zero of eight innocent students accepted the
deal. See id. These findings led to further research regarding the effect of an innocent
defendant’s belief that he or she would succeed at trial. In their work regarding fairness and
plea negotiations, Tor, Gazal-Ayal, and Garcia showed that “guilty” participants were more
likely to accept a plea than the “innocent” participants. See Tor et al., supra note 102, at
113–14.
139
See infra Part IV (discussing the results of the authors’ plea-bargaining study).
140
Companies including Broadcom, Brocade Communications, McAfee, and Comverse
Technologies were targeted by the government during the stock options backdating
investigations. See Peter J. Henning, How the Broadcom Backdating Case Went Awry, N.Y.
TIMES DEALBOOK (Dec. 15, 2009, 1:37 PM), http://dealbook.nytimes.com/2009/12/
14/how-the-broadcom-backdating-case-has-gone-awry/.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

25

Securities and Exchange Commission regulations.141 Prosecutors alleged
that one such company was Broadcom, a large semiconductor manufacturer
in California.142 After Broadcom restated $2.2 billion in charges because of
backdating in January 2007, the government indicted Dr. Henry Samueli,
cofounder and former Chief Technical Officer of the company.143 Dr.
Samueli pleaded guilty and, as part of his deal, agreed to testify for the
prosecution against Henry T. Nicholas III, Broadcom’s other cofounder,
and William J. Ruehle, the company’s Chief Financial Officer.144 After Dr.
Samueli offered his testimony at trial, however, U.S. District Judge Cormac
J. Carney voided Dr. Samueli’s guilty plea, dismissed the charges against
all the defendants, and called the prosecutors’ actions a “shameful”
campaign of intimidation.145 The judge stated in open court that “there was
no evidence at trial to suggest that Dr. Samueli did anything wrong, let
alone criminal. Yet, the government embarked on a campaign of
intimidation and other misconduct to embarrass him and bring him down.”
The judge went on to state, “One must conclude that the government
engaged in this misconduct to pressure Dr. Samueli to falsely admit guilt
and incriminate [the other defendants] or, if he was unwilling to make such
a false admission and incrimination, to destroy Dr. Samueli’s credibility as
a witness for [the other defendants].”146 With this unusual public rebuke of
141
See Events in the Broadcom Backdating Case, L.A. TIMES (Dec. 16, 2009),
http://articles.latimes.com/2009/dec/16/business/la-fi-broadcom-timeline16-2009dec16
(“Stock options, typically used as incentive pay, allow employees to buy stock in the future
at current prices. Broadcom Corp. and other companies also backdated the options to a
previously lower price to give employees a little extra when they cashed in the options.”).
142
See Mike Koehler, The Façade of FCPA Enforcement, 41 GEO. J. INT’L L. 907, 940–
41 (2010) (discussing the Broadcom case); Ribstein, supra note 94, at 630 (discussing the
Broadcom case).
143
See Press Release, Dep’t of Justice, Broadcom Co-Founder Pleads Guilty to Making
False Statement to the SEC in Backdating Investigation (June 23, 2008), available at
http://www.justice.gov/usao/cac/Pressroom/pr2008/086.html.
144
See Stuart Pfeifer & E. Scott Reckard, Judge Throws Out Stock Fraud Charges
Against Broadcom Co-Founder, Ex-CFO, L.A. TIMES, Dec. 16, 2009, at A16; see also
Indictment, United States v. Nicholas, SA CR 08-00139 (C.D. Cal. June 4, 2008), available
at http://lawprofessors.typepad.com/whitecollarcrime_blog/files/broadcom_nicholasruehlein
dictment.pdf.
145
See Reporter’s Transcript of Proceedings at 5195, United States v. Ruehle, No. SACR
08-00139-CJC (C.D. Cal. Dec. 15, 2009) [hereinafter Transcript of Proceedings, Ruehle]
(“Based on the complete record now before me, I find that the Government has intimidated
and improperly influenced the three witnesses critical to Mr. Ruehle’s defense. The
cumulative effect of that misconduct has distorted the truth-finding process and
compromised the integrity of the trial.”).
146
Id. at 5197–99 (“Needless to say, the government’s treatment of Dr. Samueli was
shameful and contrary to American values of decency and justice.”); see also Michael
Hilzik, Judicial System Takes a Hit in Broadcom Case, L.A. TIMES, July 18, 2010, at B3

26

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

the prosecutorial tactics that forced an innocent defendant into a plea
bargain, the judge in the Broadcom case demonstrated once again the
existence of the innocent defendant’s dilemma.147
While the Gregory study attempted to capture the likelihood that an
innocent defendant such as Dr. Samueli might falsely plead guilty, the
study’s utilization of hypotheticals prevented it from offering an accurate
glimpse inside the mind of the accused.148 Shortly before the Broadcom
prosecution, however, a study regarding police interrogation tactics
utilizing an experimental design similar to Gregory’s second study offered a
path forward for plea bargaining’s innocence inquiry.149 In 2005,
Professors Melissa Russano, Christian Meissner, Fadia Narchet, and Saul
Kassin initiated a study (Russano study) in which students were accused by
a research assistant of working together after being instructed this was
(noting that in an attempt to pressure defendant Nicholas, the government had “threatened to
force Nicholas’ [thirteen]-year-old son to testify about his father and drugs”). Judge Carney
listed some of the prosecution’s misconduct as the following:
Among other wrongful acts, the Government, one, unreasonably demanded that Dr. Samueli
submit to as many as 30 grueling interrogations by the lead prosecutor.
Two, falsely stated and improperly leaked to the media that Dr. Samueli was not cooperating
in the Government’s investigation.
Three, improperly pressured Broadcom to terminate Dr. Samueli’s employment and remove
him from the board.
Four, misled Dr. Samueli into believing that the lead prosecutor would be replaced because of
misconduct.
Five, obtained an inflammatory indictment that referred to Dr. Samueli 72 times and accused
him of being an unindicted coconspirator when the government knew, or should have known,
that he did nothing wrong.
And six, crafted an unconscionable plea agreement pursuant to which Dr. Samueli would
plead guilty to a crime he did not commit and pay a ridiculous sum of $12 million to the United
States Treasury.

Transcript of Proceedings, Ruehle, supra note 145, at 5198.
147
See Koehler, supra note 142, at 941 (“In pleading guilty, Samueli did what a
‘disturbing number of other people have done: pleaded guilty to a crime they didn’t commit
or at least believed they didn’t commit’ for fear of exercising their constitutional right to a
jury trial, losing, and ‘getting stuck with a long prison sentence.’” (citation omitted));
Ribstein, supra note 94, at 630 (“In the Broadcom backdating case, particularly egregious
prosecutorial conduct caused defendants to plead guilty to crimes they knew they had not
committed . . . .”); Ashby Jones, Are Too Many Defendants Pressured into Pleading Guilty?,
WALL ST. J.L. BLOG (Dec. 21, 2009, 8:50 AM), http://blogs.wsj.com/law/2009/12/21/aretoo-many-defendants-pressured-into-pleading-guilty/ (“Samueli did what lawyers and legal
scholars fear a disturbing number of other people have done: pleaded guilty to a crime either
they didn’t commit or at least believed they didn’t commit.”).
148
See supra notes 127 and 133 and accompanying text.
149
See Melissa B. Russano et al., Investigating True and False Confessions with a Novel
Experimental Paradigm, 16 PSYCHOL. SCI. 481 (2005).

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

27

prohibited.150 Some of the students accused of this form of “cheating”
were, in fact, guilty of the charge, while others were not.151 Russano
wanted to test the effect of two types of police interrogation on the rates of
guilty and innocent suspects confessing to the alleged crime.152 The first
interrogation tactic utilized to exact admissions from the students was
minimization.153 Minimization is the process by which interrogators
minimize the seriousness and anticipated consequences of the suspect’s
conduct.154 The second interrogation tactic utilized to exact admissions
from the students involved offering the students a “deal.”155 Students were
told that if they confessed, the matter would be resolved quickly and they
would merely be required to return to retake the test at a later date.156 If the
students rejected the offer, the consequences were unknown and would be
decided later by the course’s professor.157 Russano found that utilizing
these tactics together, 43% of students falsely confessed and 87% of
students truthfully confessed.158 When only the “deal” was offered,
however, only 14% of the students in Russano’s study falsely confessed.159
150

See id. at 481.
See id. at 482 (“In the current paradigm, participants were accused of breaking an
experimental rule, an act that was later characterized as ‘cheating.’”).
152
See id. at 481 (“In the first demonstration of this paradigm, we explored the influence
of two common police interrogation tactics: minimization and an explicit offer of leniency,
or a ‘deal.’”).
153
See id. at 482.
154
See id.
151

Researchers have categorized the interrogation methods promoted by interrogation manuals into
two general types, namely, maximization and minimization. Maximization involves so-called
scare tactics designed to intimidate suspects: confronting them with accusations of guilt, refusing
to accept their denials and claims of innocence, and exaggerating the seriousness of the situation.
This approach may also include presenting fabricated evidence to support the accusation of guilt
(e.g., leading suspects to think that their fingerprints were lifted from the murder weapon). In
contrast, minimization encompasses strategies such as minimizing the seriousness of the offense
and the perceived consequences of confession, and gaining the suspect’s trust by offering
sympathy, understanding, and face-saving excuses.

Id. (citations omitted) (internal quotation marks omitted).
155
See id.
156
See id. at 483.
157
See id. (“They were also told that if they did not agree to sign the statement, the
experimenter would have to call the professor into the laboratory, and the professor would
handle the situation as he saw fit, with the strong implication being that the consequences
would likely be worse if the professor became further involved.”).
158
See id. at 484.
159
See id.
Condition

True Confessions

False Confessions

No Tactic

46%

6%

28

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

In 2011, utilizing the Russano study as a guide, we constructed a new
investigatory paradigm that would better reflect the mechanics of the
criminal justice system and more precisely focus the inquiry on the innocent
defendant’s dilemma.160 The new study was administered to eighty-two
students from a small, southeastern, private technical university.161 The
results of the study were significant and established what Gregory and
Russano had hinted at in their earlier forays into the plea-bargaining
machine.162
A. STUDY METHODOLOGY—CONFRONTING A DEVIL’S
BARGAIN
Participants in the study were all college students at a small technical
university in the southeastern United States.163 The study participants had
each signed up for what they believed was a psychological inquiry into
individual versus group problem-solving performance. When a study
participant arrived for the problem-solving experiment, he or she was met
by another student pretending to be participating in the exercise also.
Unbeknownst to the study participant, however, the second student was
actually a confederate working with the authors.164 At this point, a research
assistant, also working with the authors, led the two students into a private
room and explained the testing procedures.165 The research assistant

Deal

72%

14%

Minimization

81%

18%

Minimization + Deal

87%

43%

Id. at tbl.1.
160
See infra Part III.B (discussing the results of the authors’ plea-bargaining study).
161
See id.
162
See id.
163
See Vanessa A. Edkins & Lucian E. Dervan, Pleading Innocents: Laboratory
Evidence of Plea Bargaining’s Innocence Problem 9 (2012) (unpublished short research
report) (on file with authors). The study was administered to eighty-two students. Id. Six
students were removed from the study because of their suspicion as to the study’s actual
focus, an inability to complete the study, or a refusal to assist the confederate when asked to
render assistance in answering the questions. Id. Thus, seventy-six participants remained.
Id. Of this number, thirty-one indicated they were female and forty-five indicated they were
male. Id. Of the study population, 52.6% identified as Caucasian, 21.1% identified as
African-American, 13.2% identified as Hispanic, 5.3% identified as Asian, and 7.9%
identified as “Other.” Id. at 10. Forty-eight students identified themselves as U.S. citizens,
while twenty-eight students identified themselves as non-U.S. citizens. Id.
164
See id. Two female students served as confederates in the study. One was twenty
years of age and the other was twenty-one years of age.
165
See id. Two research assistants were used in this experiment. One research assistant
was a twenty-seven-year-old male. The other was a twenty-four-year-old female.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

29

informed the students that they would be participating in an experiment
about performance on logic problems. According to the research assistant,
the two students would be left alone to complete three logic problems
together as a team.166 The research assistant then informed them that after
the first problems were completed, the students would receive three
additional logic problems that must be completed individually. When these
problems were distributed, the research assistant’s script required the
following statement, “Now I will hand out the individual problems,
remember that you are to work alone. I will give you 15 minutes to
complete these.”
While the study participant and the confederate were solving the
individual logic problems, one of two conditions would occur. In half of
the cases, the confederate asked the study participant for assistance in
answering the questions, a clear violation of the research assistant’s explicit
instructions. First, the confederate asked the study participant, “What did
you get for number 2?” If the study participant did not respond with the
answer, the confederate followed up by saying, “I think it is ‘D’ because
[some scripted reasoning based on the specifics of the problem].” Finally,
if necessary, the confederate would ask, “Did you get ‘E’ for # 3?”167 It is
worth noting that all but two study participants asked by the confederate to
offer assistance violated the requirement that each student work alone.168
Those study participants offering assistance were placed in the “guilty
condition,” because they had “cheated” by violating the research assistant’s
instructions. In the other half of the cases, the confederate sat quietly and
did not ask the study participant for assistance.169 The study participants in
166

See Application by Vanessa A. Edkins & Lucian E. Dervan to the Florida Institute
of Technology Institutional Review Board, The Function of Sentence Disparity on Plea
Negotations 16 (Nov. 3, 2009) (on file with authors). The research script required the
research assistants to make the following statement during the introduction:
We are studying the performance of individuals versus groups on logic problems. You will be
given three logic problems to work through together and then three problems to work through on
your own. It is very important that you work on the individual problems alone. You have 15
minutes for each set of problems. Even if you run out of time, you must circle an answer for
each question. First, you’ll be working on the group problems. I will leave the room and be
back in 15 minutes. If you finish before that time, one of you can duck your head out the door
and let me know.

Id.
167
See id. at 20. The study protocols also instructed the confederate that “[i]f they [the
study participant] refuse after this prodding, stop asking and record (on the demographic
sheet, at the end of the study) that the individual was in the cheat condition but refused to
cheat. Give specific points explaining what you tried to do to instigate the cheating.” Id.
168
See Edkins & Dervan, supra note 163, at 10. The two students who refused to offer
assistance were removed from the study.
169
See Edkins & Dervan, supra note 166, at 20. The study protocol stated:

30

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

this scenario were placed in the “innocent condition,” because they had not
“cheated” by violating the research assistant’s instructions.
After completing the second set of logic problems, the research
assistant, who did not know whether cheating had occurred, collected the
logic problems and asked that the students remain in the room for a few
minutes while the problems were graded.170 Approximately five minutes
later, the research assistant reentered the room and said, “We have a
problem. I’m going to need to speak with each of you individually.” The
research assistant looked at the sign-in sheet and read off the confederate’s
name and the two then left the room together. Five minutes later, the
research assistant reentered the room, sat down near the student, and made
the following statement.
You and the other student had the same wrong answer on the second and third
individual questions. The chances of you both getting the exact same wrong answer
are really small—in fact they are like less than 4%—because of this, when this occurs,
we are required to report it to the professor in charge and she may consider this a form
of academic dishonesty.171

In early trials of the study design, it was determined that study
participants did not understand how getting the same wrong answer on
questions two and three indicated they may have cheated. As a result, there
was a perception that no actual evidence of guilt existed. Because actual
criminal trials involve evidence of guilt, even trials where the individual is
actually innocent, it was determined that the study would more accurately
capture the criminal process if one piece of evidence leading to the
accusation was explained. Therefore, as described above, the subject was
informed that statistically, given that there were five available choices for
each question, there was only a 4% chance that the students provided the
same incorrect answers by coincidence. This explanation of the logic
behind the research assistant’s accusation certainly did not mean the subject
was guilty. To the contrary, the research assistant actually noted that there

Do not speak to the participant and do not respond if they ask for assistance.
Be sure that the participant cannot see what answers you are choosing—he/she needs to believe
that you both answered two questions the same way and if they see your paper they may know
that this was not the case. We need to make sure that no matter what, cheating does NOT occur
in this condition.

Id.
170
See Edkins & Dervan, supra note 163, at 10–11. The research assistants were not
informed of whether cheating had occurred to ensure that their approach to each study
participant—during the plea-bargaining component of the study—was consistent and not
influenced by omnipotent knowledge of guilt or innocence that would not be available to a
prosecutor or investigator in the actual criminal justice system.
171
Id. at 11.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

31

was a 4% chance there was no cheating. As with all studies of this nature,
difficult decisions must be made in an effort to create as realistic an
environment as possible. While some might argue that mentioning the
statistical evidence leading to the accusation might lead to a perception of
an overly strong case against the study participant, it was decided that the
benefits of explaining the reasoning for the charge outweighed any potential
influence this data might have on the study results.172
To ensure the study participant was unable to argue that he had
answered questions two and three correctly, the second set of logic
questions were designed to have no correct answer. The research assistant
then informed the student that this had occurred before and she had been
given authority to offer two alternatives.173 The first alternative the
research assistant offered was a “plea” in which the study participant would
be required to admit he or she cheated and, as punishment, would lose all
compensation promised for participating in the experiment.174 This
particular offer was made to all study participants and was constructed to be
akin to an offer of probation or time served in the actual criminal justice
system.175 The research assistant then offered each study participant one of

172
This conclusion was reached for several reasons. First, an actual criminal case should
not reach the trial stage without at least one piece of significant evidence or a multitude of
smaller pieces of evidence. As such, in designing the study, we did not believe offering this
single piece of evidence would unduly influence the subject’s decisionmaking or
unreasonably influence the study’s results. Second, it is difficult in a short study to build the
same, often complex, foundation that is inherent in a criminal case. To rectify this inherent
design limitation, we devised one simple piece of evidence to explain the basis for the
accusation. The offered explanation, however, did leave room for the possibility that the
individual was innocent, thus allowing the subject an argument upon which to rely in
professing their innocence during the plea-negotiation process or during a trial before the
ARB. Third, even though many innocent defendants may not be confronted with as strong
an indicator of guilt, it does not change the fact that any innocent defendant, no matter the
evidence, necessarily falls within the margins of a case where there is evidence pointing to
guilt, but the defendant is, in fact, innocent. Even if our margin is smaller than most, the
argument could be made that it does not change the fact that the person is innocent and,
according to many commentators, should be motivated to maintain that innocence and
proceed to trial.
173
See id. The research assistants also informed the study participants that this situation
had arisen before and that the described protocol must be followed or the research assistants
might lose their research positions.
174
See id. at 12. The compensation offered for participating in the study was research
participation credit—something required for students to successfully complete their
Introduction to Psychology course.
175
See id.; see also Bowers, supra note 102, at 1136–37.

The trial course is long; even if convicted, the defendant often has already served any
postconviction sentence, and then some. In this way, conviction may counterintuitively
inaugurate freedom. Moreover, the costs of conviction are minimal; an additional misdemeanor

32

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

two alternative options if the plea offer was rejected.
In roughly half of the cases (referred to as the “harsh sentencing
condition”), the research assistant informed the student that if the “deal”
was not accepted, the professor leading the experiment would bring the
matter before the Academic Review Board (ARB). The research assistant
explained that the ARB was a group of ten to twelve faculty and staff
members that ruled on such matters. To make the ARB sound similar to a
jury in an actual criminal trial, the research assistant described it as being a
forum in which the student had the option of telling his or her version of
events, presenting evidence, and arguing for his or her position. Again, to
better reflect the actual mechanics of the criminal justice system, the
research assistant also informed the student that “the majority of students,
like 80–90%, are usually found guilty” before the ARB. This percentage
was selected and communicated because it is consistent with the actual
current conviction rate of defendants proceeding to trial in the United
States.176 While it is impossible to predict how common it is for defense
counsel to relate such statistics to their clients, we believed that this
information would, at a minimum, be considered by counsel during their
own assessment of the case and in preparing to advise their clients of the
risks and rewards of each option. As such, we felt it important to offer this
information to the participants in this study to utilize during their personal
assessment processes. The research assistant then informed the student that
if he or she were “convicted” by the ARB, she would lose her study
compensation, her faculty advisor would be notified, and she would have to
enroll in an ethics course that met for three hours each week during the
semester. The course was described as a pass/fail class that would be
offered free of charge, but it would require mandatory weekly attendance
and the completion of a paper and a final examination.
In roughly the other half of the cases (referred to as the “lenient
sentencing condition”), the research assistant provided the same
information to the student regarding the ARB process, but informed the
student that if he was “convicted” by the ARB, he would lose his study

conviction does little to further mar an already-soiled record because the recidivist defendant has
already suffered most of the corollary consequences that typically stem from convictions. If the
defendant can get a plea to a misdemeanor and time served, then the process constitutes the
whole punishment. Any plea that frees this defendant may be more than advisable—it may be
salvation. No matter how certain of acquittal, she is better off pleading guilty. She is the
defendant who benefits most from plea bargaining, and she is the very defendant who most
frequently is innocent in fact.

Bowers, supra note 102, at 1136–37 (footnotes omitted).
176
See Edkins & Dervan, supra note 163, at 12; see also Gregory et al., supra note 125,
at 1529.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

33

compensation, his faculty advisor would be notified, and he would undergo
nine hours of ethics training in the form of three three-hour seminars. The
seminars were described as free of charge but requiring mandatory
attendance and the completion of a final examination. Half the students
were offered the harsh sentencing condition and the other half were offered
the lenient sentencing condition to test the impact of “sentencing
differentials” on the rate of innocent and guilty students accepting the plea
offer rather than proceeding to trial before the ARB.
Once the study participants were presented with their options of
pleading guilty or proceeding to the ARB, the research assistant presented
them each with a piece of paper. The paper outlined their options and asked
that they circle their selection.177 To ensure study participants did not
become distraught under the pressure of the scenario, the research assistant
was instructed to terminate the experiment and debrief the student regarding
the true nature of the study if he or she took too long to select an option,
seemed overly stressed, or tried to leave the room.178
B. STUDY RESULTS—THE INNOCENT DEFENDANT’S DILEMMA
EXPOSED
While academic discipline is not precisely equivalent to traditional
criminal penalties, the anxiety experienced by students anticipating
punishment is similar in form, if not intensity, to the anxiety experienced by
an individual charged with a criminal offense. As such, this study sought to
recreate the innocent defendant’s dilemma in as real a manner as possible
by presenting two difficult and discernible choices to students and asking
them to make a decision. This is the same mentally anguishing decision
defendants in the criminal justice system must make every day. 179 While it
177

See Edkins & Dervan, supra note 166, at 17–18. The research assistants had scripted
answers to common questions that might be asked while the students deliberated on their
choices. For example, answers were prepared for questions such as “I didn’t do it,” “What
did the other person say?” “How can I be in trouble if this isn’t a class?” etc. This was done
to ensure the research assistants’ interactions with the study participants were uniform and
consistent. See Edkins & Dervan, supra note 163, at 12.
178
See id. After making their selection, the study participants were probed for suspicion
and eventually debriefed regarding the true nature of the experiment. During this debriefing
process, the students were informed that helping other students outside the classroom setting
was a very kind action and that they were, in fact, in no trouble. The research assistants
ensured that prior to leaving the room, the study participants understood that the nature of
the study needed to remain confidential.
179
See id. One important distinction between the experimental methodology used in the
authors’ study and previous studies is that the former included a definitive top end to the
sentencing differential. This better reflects the reality of modern sentencing, particularly in
jurisdictions utilizing sentencing guidelines, and thus better captures the decisionmaking

34

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

was anticipated that this plea-bargaining study would reveal that innocent
students, just like innocent defendants, sometimes plead guilty to an offense
they did not commit in return for promises of leniency, the rate at which
such false pleas occurred exceeded our estimations and should lead to a
reevaluation of the role and method of plea bargaining today.
1. Pleading Rates for Guilty and Innocent Students
As had been anticipated, both guilty and innocent students accepted
the plea bargain and confessed to the alleged conduct.180 In total, almost
nine out of ten guilty study participants accepted the deal, while slightly
fewer than six out of ten innocent study participants took the same path.181
Figure 1
Number and Percentage of Students by Condition (Guilty or Innocent)
Rejecting and Accepting the Plea Offer
Condition
Guilty
Innocent

Rejected Plea Offer
No.
%
4
10.8
17
43.6

Accepted Plea Offer
No.
%
33
89.2
22
56.4

processes of criminal defendants faced with plea-bargaining decisions. See Russano et al.,
supra note 149, at 483 (discussing the lack of a definitive sentence for those who failed to
accept the deal).
180
See Edkins & Dervan, supra note 163, at 12–14. We first tested our sample to see if
there were any demographic differences with regards to the decision to accept a plea.
Participants did not differ in their choices based on gender, 2(1, N = 76) = 0.24, p = 0.63
(continuity correction applied), ethnicity 2(4, N = 76) = 0.51, p = 0.97, citizenship status
2(1, N = 76) = 0.16, p = 0.90 (continuity correction applied), or whether or not English was
the participant’s first language 2(1, N = 76) = 0.34, p = 0.56 (continuity correction applied).
We also ensured that the decision of the participants did not differ by the experimenter 2(1,
N = 76) = 0.83, p = 0.36. Reported results, therefore, are collapsed across all of the
previously mentioned groups.
181
See id. at 13. We conducted a three-way loglinear analysis to test the effects of guilt
(guilt vs. innocence) and type of sanction (lenient vs. harsh) on the participant’s decision to
accept the plea bargain. The highest order interaction (guilt x sanction x plea) was not
significant, 2(1, N = 76) = 0.26, p = 0.61. What was significant was the interaction between
guilt and plea, 2(1, N = 76) = 10.95, p < 0.01. To break down this effect, a separate chisquare test was performed looking at guilt and plea, collapsed across type of sanction.
Applying the continuity correction for a 2 x 2 contingency table, there was a significant
effect of guilt, 2(1, N = 76) = 8.63, p < 0.01, with the odds ratio indicating that those who
were guilty were 6.38 times more likely to accept a plea than those who were innocent.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

35

Two important conclusions stem from these results.182 First, as had
been predicted by others, guilty defendants are more likely to plead guilty
than innocent defendants.183 In our study, guilty defendants were 6.39
times more likely to accept a plea than innocent defendants given the same
sentencing options.184

Percent of Individuals Tested

Figure 2
Percentage of Students by Condition (Guilty or Innocent)
Accepting the Plea Offer
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%

10.8%
43.6%

Rejected
Plea Offer
Accepted
Plea Offer

89.2%
56.4%

Guilty (N = 37)

Innocent (N = 39)

Interestingly, these results are consistent with predictions made by other
scholars relying on case studies to predict the impact of innocence on pleabargaining decisions.185

182

See id. at 13–14.
See id.; see also Tor et al., supra note 102, at 113 (arguing that innocent defendants
tend to reject plea offers more than guilty defendants); Covey, supra note 116, at 34.
184
See Edkins & Dervan, supra note 163, at 13.
185
See Covey, supra note 116, at 1.
183

36

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

In his recent article entitled Mass Exoneration Data and the Causes of
Wrongful Convictions, Professor Covey examined two mass-exoneration
cases and predicted, based on the choices of defendants in those cases, that
innocence mattered.186
While Professor Covey concedes that his
examination of case studies only permits “some tentative comparisons,” it is
fascinating to observe that the actions of the defendants in these two massexoneration cases mirror the actions of our study participants.187

Figure 3
Percentage of Individuals by Condition (Guilty or Innocent)
Accepting the Plea Offer in the Study and in Professor Covey’s Studies on
Mass Exonerations
Condition Dervan/Edkins Study
%
Guilty
89.2
Innocent
56.4

Covey Mass Exonerations Studies
%
89.0
77.0

As the numbers reflect, guilty defendants in Professor Covey’s mass
exoneration cases acted almost exactly as did guilty students in our
experiment.188 In both cases, nine out of ten guilty individuals accepted the
deal.189 While not as precise, in both the mass-exoneration cases and the
plea-bargaining study, well over half of innocent individuals also selected
the bargain over proceeding to trial.190 These similarities not only lend
credibility to the results of our new study, but once again support the
arguments of those who previously predicted that plea bargaining’s
186
See id. (examining the mass exonerations in the Rampart case in California and the
Tulia case in Texas).
187
See id. at 34.

Although the numbers are small, they are large enough to permit some tentative comparison.
With respect to plea rates, the data show that innocence does appear to make some
difference . . . . Actually innocent exonerees thus plead guilty at a rate of 77%. In comparison,
22 of those who were not actually innocent pled guilty while 3 were convicted at trial. In other
words, 88% of those who were not innocent pled guilty. Finally, of the remaining group of “may
be innocents,” 17 pled guilty while two were convicted at trial, providing an 89% guilty plea
rate.

Id.
188
189
190

See id.
See id.; Edkins & Dervan, supra note 163, at 13.
See Covey, supra note 116, at 34; Edkins & Dervan, supra note 163, at 13.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

37

innocence problem affected more than just an isolated few.191
The second and, perhaps, more important conclusion stemming from
the study is that well over half of the innocent study participants, regardless
of whether the lenient or harsh sentencing condition was employed, were
willing to falsely admit guilt in return for a reduced punishment.192
Previous research has argued that plea bargaining’s innocence problem is
minimal because defendants are risk prone and willing to defend
themselves before a tribunal.193 Our research, however, demonstrates that
when study participants are placed in real, rather than hypothetical,
bargaining situations and are presented with accurate information regarding
their statistical probability of success, just as they might be so informed by
their attorneys or the government during criminal plea negotiations,
innocent individuals are actually highly risk averse.194
Based on examination of the detailed notes compiled during the
debriefing of each study participant, two common concerns drove the
participants’ risk-averse behavior. First, study participants sought to avoid
the ARB process and move directly to punishment.195 Second, study
191

See Bowers, supra note 102, at 1136–37.
See Edkins & Dervan, supra note 163, at 5. While design constraints prevented the
incorporation of counsel into our study, we believe that this omission does not lessen the
significance of these findings. First, while the presence of counsel may have resulted in a
slight shift in outcomes, it is unlikely such representation would have dramatically altered
the study results because the underlying decisionmaking factors presented to the participants
would remain the same. Second, it is important to note that many individuals in the U.S.
criminal justice system proceed without counsel. See SMITH & MADDAN, supra note 67, at 9.
Finally, the results of this study are relevant for other institutions employing models based
on the criminal justice system, many of which do not utilize an equivalent to counsel. That
students will acquiesce in such a manner should not only bring the criminal justice system’s
use of plea bargaining into question, but also all other similar forms of adjudication
throughout society. For example, this would include reevaluation of student conduct
procedures that contain offers of leniency in return for admissions of guilt.
193
See Tor et al., supra note 102, at 106 (arguing based on a study utilizing an email
questionnaire that innocent defendants are risk prone and on average were willing to proceed
to trial rather than accept a plea); see also Stephanos Bibas, Plea Bargaining Outside the
Shadow of Trial, 117 HARV. L. REV. 2464, 2507 (2004) (“Defendants’ attitudes toward risk
and loss will powerfully shape their willingness to roll the dice at trial.”).
194
See Edkins & Dervan, supra note 163, at 6; see also Bibas, supra note 193, at 2509
(discussing risk aversion and loss aversion). Professor Bibas notes that “most people are
inclined to gamble to avoid sure losses and inclined to avoid risking the loss of sure gains;
they are risk averse, but they are even more loss averse. When these gains and losses are
uncertain probabilities rather than certain, determinate amounts, the phenomenon is
reversed.” Id.
195
See Edkins & Dervan, supra note 163, at 6; see also Bowers, supra note 102, at
1136–37.
192

Likewise, over fifty percent of all misdemeanor charges that ended in conviction resulted in

38

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

participants sought a punishment that would not require the deprivation of
direct future liberty interests.196 Further research is necessary in this area to
fully understand these motivations, but one key trend is worth noting at this
juncture. The study participants’ actions appear to be directly mimicking a
phenomenon that has drawn much debate and concern in recent years197:
the students appear to have been selecting “probation” and immediate
release rather than risking further “incarceration” through forced
participation in a trial and, if found guilty, “confinement” in an ethics
course or seminar.198 In essence, the study participants simply wanted to go
home.199 This study suggests, therefore, that one needs to be concerned not
only that significant sentencing differentials might lead felony defendants to
falsely condemn themselves through plea bargaining, but also that
misdemeanor defendants might be pleading guilty based on factors wholly
distinct from their actual factual guilt.200
2. The Impact of Sentencing Differentials
One goal of the study was to offer two distinct punishments as a result
of conviction by the ARB to determine if the percentage of guilty and
innocent study participants accepting the plea offer rose as the sanction they
risked if they lost at trial increased.201
As discussed previously,
approximately half of the study participants were informed of the harsh
sentencing condition and the other half were informed of the lenient
sentencing condition.202
nonjail dispositions. Of the so-called jail sentences, fifty-seven percent were sentences of time
served. Even for defendants with combined felony and misdemeanor records, the rate of timeserved sentences dropped only to near fifty percent. Further, the percentage of express timeserved sentences significantly underestimates the number of sentences that were in fact
equivalent to time served, because most defendants with designated time sentences actually had
completed those sentences at disposition.

Bowers, supra note 102, at 1144.
196
See Edkins & Dervan, supra note 163, at 16.
197
See Bibas, supra note 193, at 2492–93 (noting that pretrial detention can exceed the
eventual prison sentence after trial); SMITH & MADDAN, supra note 67, at 7 (“But even where
no jail time is imposed, and the court and the prosecutor keep their promises and allow a
defendant to pay his fine and return to his home and job the same day, there are real
punishments attendant to a misdemeanor conviction that have not yet begun.”).
198
See Bowers, supra note 102, at 1136–37.
199
See id.
200
See SMITH & MADDAN, supra note 67, at 7 (discussing concerns regarding
uncounseled defendants pleading guilty in quick arraignments and returning home the same
day without understanding the collateral consequences of their decisions).
201
See Edkins & Dervan, supra note 163, at 3.
202
See id.

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

39

Figure 4
Percentage of Students by Condition (Guilty or Innocent) and
Sentencing Condition (Harsh or Lenient) Accepting the Plea Offer
Condition

Guilty
Innocent
Diagnosticity

Rejected Plea Offer
Harsh
Lenient
%
%
5.9
15.0
38.9
47.6

Accepted Plea Offer
Harsh
Lenient
%
%
94.1
85.0
61.1
52.4
1.54
1.62

As the table above demonstrates, the subjects facing the harsh
sentencing condition, regardless of guilt or innocence, accepted the plea
offer at a rate almost 10% higher than the subjects facing the lenient
sentencing condition.203 Unfortunately, this shift is not statistically
significant due to the limited size of the study population, but the data does
demonstrate that perhaps the study was on the right track; more research
with a larger pool of participants and a greater “sentencing differential” is
needed to examine this phenomenon further.204 Significant questions
remain regarding how large a sentencing differential can become before the
rate at which innocent and guilty defendants plead guilty becomes the same
and regarding how sentencing differentials that include probation, as
opposed to a prison sentence, influence a defendant’s decisionmaking.
Such questions, however, must be reserved for future study.
Just as interesting as the above shift in the percentage of study
participants pleading guilty, perhaps, is the diagnosticity data collected
during this portion of the study.205 Diagnosticity, as used in this study, is a
calculation that ascertains whether one action or decision (e.g., the decision
to accept a plea bargain) is indicative of some truth (e.g., guilt); in other
words, acceptance of a plea bargain would be diagnostic of guilt if it was
significantly more likely to occur with guilty defendants than with innocent
defendants.206 Akin to an odds ratio, diagnosticity levels can be quite high,
but commonly numbers hover around the single digits or low double digits.
For example, a similar test was applied in the Russano study of

203

See id.
See id.
205
See id.
206
See id.; see also Russano et al., supra note 149, at 484 (noting that diagnosticity in
that study illustrated the “ratio of true confessions to false confessions”).
204

40

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

interrogation tactics.207 When Russano’s interrogators did not use any
tactics to elicit a confession, the diagnosticity of the interrogation process
was 7.67.208 By comparison, when Russano’s interrogators applied two
interrogation tactics, the number of false confessions jumped to almost 50%
and the diagnosticity of the process dropped to 2.02.209 This drop in
diagnosticity meant that as Russano applied various interrogation tactics,
the ability of the interrogation procedure to identify only guilty subjects
diminished.210 Taken to the extreme, if one were to torture a suspect during
interrogation, one would anticipate a diagnosticity of 1.0, which would
indicate that the process was just as likely to capture innocent as guilty
defendants.211
In our study, the diagnosticity of the plea-bargaining process utilized
was extremely low, a mere 1.54.212 That the diagnosticity of our pleabargaining process was considerably lower than the diagnosticity of
Russano’s combined interrogation tactics is significant.213 First, it is
important to note that plea bargaining’s diagnosticity in this study was
strikingly low, despite the fact that our process did not threaten actual
prison time or deprivations of significant liberty interests as happens every
day in the actual criminal justice system.214 Further, this diagnosticity result
indicates that innocent defendants may be more vulnerable to coercion in
the plea-bargaining phase of their proceedings than even during a police
interrogation. While much focus has been given to increasing constitutional
protections during police interrogations over the last half-century, perhaps
the Supreme Court should begin focusing more attention on creating
protections within the plea-bargaining process.215
207

See Russano et al., supra note 149, at 484.
See id. (7.67 diagnosticity was the result of only 6% of test subjects falsely
confessing). The Russano study stated, “[D]iagnosticity was highest when neither of the
techniques was used and lowest when both were used. More specifically, diagnosticity was
reduced by nearly 40% with the use of a single interrogation technique . . . and by 74% when
both techniques were used in combination.” Id.
209
See id.
210
See id.
211
See id.
212
See Edkins & Dervan, supra note 163, at 14.
213
Russano et al., supra note 149, at 484; Edkins & Dervan, supra note 163, at 14.
214
John H. Langbein, Torture and Plea Bargaining, 46 U. CHI. L. REV. 3, 12–13 (1978)
(arguing that plea bargaining’s sentencing differential means “[p]lea bargaining, like torture,
is coercive”).
215
See Richard A. Leo & Richard J. Ofshe, The Consequences of False Confessions:
Deprivations of Liberty and Miscarriages of Justice in the Age of Psychological
Interrogation, 88 J. CRIM. L. & CRIMINOLOGY 429, 495–96 (1998) (“When police are trained
to seek both independent evidence of a suspect’s guilt and internal corroboration for every
208

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

41

The other important aspect of our study’s diagnosticity data is that the
diagnosticities of the harsh and lenient sentencing conditions were very
similar.216 This was surprising, because it had been anticipated that the
efficiency of the process would suffer greatly as we increased the
punishment risked at trial.217 That the diagnosticity did not drop in this way
when the harsh sentencing condition was applied means further research is
necessary to better understand the true impact of sentencing differentials.
Though further research is warranted, we suggest two hypotheses that
might offer an explanation of the diagnosticity element of this study. First,
perhaps future studies will demonstrate that diagnosticity here did not drop
significantly because it had little place left to go.218 The diagnosticity for
the lenient sentencing condition was already at 1.62, which, as discussed
above, is exceptionally low. That it did not drop meaningfully below this
threshold when the sentencing differential was increased, therefore, may not
be surprising, particularly given that a diagnosticity of 1.0 would mean that
sentence severity had no ability to predict truthful plea deals.219 Second,
perhaps future studies will reveal that the diagnosticity of our pleabargaining process began so low and failed to drop significantly when a
harsher sentencing condition was applied because sentencing differentials
operate in a manner other than previously predicted.220 Until now, many
observers have predicted that sentencing differentials operate in a linear
fashion (Figure 5), which means there is a direct relationship between the
size of the sentencing differential and the likelihood a defendant will accept
the bargain.221
confession before making an arrest . . . the damage wrought and the lives ruined by the
misuse of psychological interrogation methods will be significantly reduced.”); Russano et
al., supra note 149, at 485 (“[W]e encourage police investigators to carefully consider the
use of interrogation techniques that imply or directly promise leniency, as they appear to
reduce the diagnostic value of an elicited confession.”); see also Missouri v. Frye, 132 S. Ct.
1399, 1407 (2012) (“Because ours ‘is for the most part a system of pleas, not a system of
trials,’ it is insufficient simply to point to the guarantee of a fair trial as a backstop that
inoculates any errors in the pretrial process.”) (citation omitted).
216
See Edkins & Dervan, supra note 163, at 3, 5.
217
See id.
218
See Dervan, supra note 34, at 475 (discussing a similar phenomenon with regard to
plea-bargaining rates, which are now in excess of 96% at the federal level).
219
See Langbein, supra note 214, at 12–13.
220
See Dervan, supra note 88, at 282 (“[I]n a simplistic plea bargaining system the
outcome differential and the sentencing differential track closely.”); Yin, supra note 89, at
443 (“Curiously, the arena of plea bargaining pits the concepts of duress and consideration
against each other: a large sentencing differential makes it more likely that a defendant is
coerced into pleading guilty, and yet it also increases the benefit offered in exchange for the
guilty plea.”).
221
See Dervan, supra note 88, at 282–83; Yin, supra note 89, at 443.

42

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

Figure 5
Predicted Linear Relationship
Between Plea-Bargaining Rates and Sentencing Differentials

Likelihood a
Defendant Will
Plead Guilty

Size of the Sentencing Differential

It may be the case, however, that plea bargaining actually operates as a
“cliff.” This means that a particularly small sentencing differential may
have little to no likelihood of inducing a defendant to plead guilty (Figure
6). However, once the sentencing differential reaches a critical size, its
ability to immediately and markedly influence the decisionmaking process
of a defendant, whether guilty or innocent, becomes almost
overwhelming.222 Such a cliff effect would result in similar diagnosticities
for both the harsh and lenient sentencing conditions because, once the
critical size is reached, there is little additional impact that can be gained
from further increasing the size of the differential.
Figure 6
Possible “Cliff” Relationship
Between Plea-Bargaining Rates and Sentencing Differentials

Likelihood a
Defendant will
Plead Guilty

Size of the Sentencing Differential
222

There are many factors that might shift when this cliff is reached for a particular
defendant. See Bibas, supra note 193 (discussing factors that influence a particular
defendant’s decision to plead guilty).

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

43

If future research indicates that this cliff effect is occurring, then these
findings will be significant for at least three reasons. First, this might mean
that while research suggesting that the answer to plea bargaining’s
innocence problem is better control of sentencing differentials is on the
right track, such proposals will have to account for the cliff effect in
selecting precisely how significant a differential to permit.223 Without such
consideration, it is possible that a proposed limitation on sentencing
differentials that permitted incentives beyond the cliff would have little
positive impact on the coercive nature of subsequent plea offers. Second, if
such cliffs exist and are reached relatively quickly, as was the case in this
study, consideration must be given to limiting the size of sentencing
differentials more drastically then previously proposed.224 Finally, future
research regarding such cliffs might reveal precise mechanisms through
which to increase the efficiency of the plea-bargaining system. For
example, if it were revealed that guilty defendants required a smaller
sentencing differential to reach their cliff, limiting sentencing differentials
to such a size would simultaneously create a significant enough incentive
for most guilty defendants to plead and not so great an incentive as to
capture innocent ones. While further research is necessary to understand
this possible phenomenon better, consideration must now be given to the
implications of a possible finding that small sentencing differentials are
more powerful than previously predicted and operate in a very different
way than previously assumed.
IV. THE CONSTITUTIONALITY OF THE INNOCENT DEFENDANT’S DILEMMA
In 1970, the same year the Supreme Court ruled that plea bargaining
was a permissible form of justice in the Brady decision, the Court also
accepted the case of North Carolina v. Alford.225 In Alford, the defendant

223
See Russell D. Covey, Fixed Justice: Reforming Plea Bargaining with Plea-Based
Ceilings, 82 TUL. L. REV. 1237, 1245 (2008) (discussing the benefits of fixed-plea discounts,
including that such fixed discounts “prevent prosecutors from offering discounts so large that
innocent defendants are essentially coerced to plead guilty to avoid the risk of a dramatically
harsher sentence”); see also Donald G. Gifford, Meaningful Reform of Plea Bargaining: The
Control of Prosecutorial Discretion, 1983 U. ILL. L. REV. 37, 81 (“Dean Vorenberg suggests
that a sentence discount of ten or twenty percent should encourage the requisite number of
desired pleas. This figure appears to be a reasonable one with which to begin . . . .
Excessive sentence discounts should be constitutionally suspect because they place a burden
on the defendant’s exercise of constitutional rights and negate the voluntary nature of his
plea.”).
224
Gifford, supra note 223, at 81.
225
North Carolina v. Alford, 400 U.S. 25 (1970).

44

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

was indicted for first-degree murder.226 After Alford’s attorney questioned
witnesses in the case and determined that there was a strong indication of
guilt, he recommended Alford plead guilty to the prosecution’s offer of
second-degree murder.227 Alford agreed but, during the plea hearing,
continued to declare his innocence and stated that he was pleading guilty
only to avoid the possibility of the death penalty.228 Despite the
proclamations from Alford, the trial judge accepted the plea and sentenced
the defendant to thirty years in prison.229 In approving of the trial court’s
actions, the Supreme Court stated that it was permissible for a defendant to
plead guilty even while maintaining his or her innocence.230 The Court
stated, however, that there must be a “record before the judge contain[ing]
strong evidence of actual guilt” to ensure the rights of the truly innocent are
protected and guilty pleas are the result of “free and intelligent choice.”231
Forty years later, three men serving sentences ranging from life in prison to
death would use this form of bargained justice to walk free after almost two
decades in prison for a crime they may never have committed.232
In May 1993, the mutilated bodies of three eight-year-old boys were
discovered in a drainage canal in Arkansas.233 Spurred by growing concern
regarding satanic cults, police desperately searched for the killer or
killers.234 As part of their investigation, police focused on a seventeen-yearold named Jessie Lloyd Misskelley Jr. Subjected to a twelve-hour
interrogation, Misskelley eventually confessed to committing the killings

226

See id. at 26–27.
See id. at 27.
228
See id. at 28.
229
See id. at 29.
230
Id. at 37; see also Leipold, supra note 95, at 1156 (“An Alford plea, where the
defendant pleads guilty but simultaneously denies having committed the crime, clearly puts
the court on notice that this guilty plea is problematic . . . .”).
231
Alford, 400 U.S. at 37, 38 n.10. Currently, the federal system, the District of
Columbia, and forty-seven states permit Alford pleas. See Stephanos Bibas, Harmonizing
Substantive-Criminal-Law Values and Criminal Procedure: The Case of Alford and Nolo
Contendere Pleas, 88 CORNELL L. REV. 1361, 1372–73 n.52 (2003).
232
See Campbell Robertson, Rare Deal Frees 3 in ’93 Arkansas Child Killings, N.Y.
TIMES, Aug. 20, 2011, at A1; see also Paul G. Cassell, The Guilty and the ‘Innocent’: An
Examination of Alleged Cases of Wrongful Conviction from False Confessions, 22 HARV.
J.L. & PUB. POL’Y 523, 557–60 (1999) (discussing facts of the case); Leo & Ofshe, supra
note 215, at 461–62 (discussing the Misskelley confession); Mara Leveritt, Are ‘Voices For
Justice’ Heard? A Star-Studded Rally on Behalf of the West Memphis Three Prompts the
Delicate Question, 33 U. ARK. LITTLE ROCK L. REV. 137, 150–53 (2011) (discussing
publicity surrounding the case).
233
See Robertson, supra note 232, at A1, A12.
234
See id. at A12.
227

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

45

along with two others teenagers, Damien Echols and Jason Baldwin, though
his confession was “inconsistent with the facts of the case, was not
supported by any evidence, and demonstrated that he lacked personal
knowledge of the crime.”235 Though Misskelley later recanted his
statement, all three teenagers were convicted at trial and became known as
the “West Memphis Three.”236 Misskelley and Baldwin received life
sentences, while Echols received the death penalty.237
Following their convictions, the three young men continued to
maintain their innocence and gradually, publicity regarding the case began
to grow.238 Though many had argued for years that the West Memphis
Three were innocent of the alleged offense, concern regarding the case
reached a crescendo in 2007 after DNA testing conducted on items from the
crime scene failed to match any of the three.239 Significantly, however, the
DNA testing did find a match.240 Hair from the ligatures used to bind one
of the victims matched Terry Hobbs, one of the victims’ stepfathers.241
Though Hobbs had claimed not to have seen the murdered boys at all on the
day of their disappearance, several witnesses came forward after the DNA
test results were released to say they had seen him with the boys shortly

235

See Leo & Ofshe, supra note 215, at 461.
See Robertson, supra note 232, at A12.
237
See id.
238
See id.
239
See Leveritt, supra note 232, at 151–52. In considering the significance of plea
bargaining’s innocence problem, one must also consider how likely it is that police
inadvertently target the wrong suspect in a particular case—something that might eventually
lead to an innocent suspect being offered a plea bargain in return for a false confession. See
Thomas, supra note 111, at 576.
236

Despite Risinger’s wisdom about not attempting a global estimate of how many innocents are
convicted, I continue to try to at least surround the problem. We do know some things for
certain. An Institute of Justice monograph published in 1999 contained a study of roughly
21,000 cases in which laboratories compared DNA of the suspect with DNA from the crime
scene. Remarkably, the DNA tests exonerated the prime suspect in 23% of the cases. In another
16%, the results were inconclusive. Because the inconclusive results must be removed from the
sample, the police were wrong in one case in four. The prime suspect was innocent in one case
out of four!

Id.
240

See Leveritt, supra note 232, at 151.
See id. (discussing the release of this DNA evidence by singer Natalie Maines during
a rally for the West Memphis Three). Further evidence in the case came to light as a result
of a defamation lawsuit filed by Hobbs against Maines. Id. at 151–52. During a deposition
in the defamation case, Hobbs stated that he had not seen the victims on the day of the
murders. Id. When this information was released to the public, several witnesses came
forward to state that they had seen Hobbs with the victims shortly before their
disappearance. Id.
241

46

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

before their murder.242
By 2011, the newly discovered evidence in the case was deemed
sufficient to call a hearing to determine if there should be a new trial.243 For
the prosecution, however, the prospect of retrying the defendants given the
weak evidence offered at the original trial and the new evidence indicating
the three might be innocent was unappealing.244 According to the lead
prosecutor, there was no longer sufficient evidence to convict the three at
trial.245 Despite the strong language in Alford indicating that it was
appropriate only in cases where the evidence of guilt was overwhelming
and conviction at trial was almost ensured, the government offered the West
Memphis Three a deal.246 They could continue to maintain their innocence,
but would be required to enter an Alford plea of guilty to the 1993 murders
of the three boys.247 In return, they would be released immediately.248
While Baldwin was reluctant to accept the offer, he agreed to ensure Echols
would be released from death row.249 Baldwin stated, “[T]his was not
justice. However, they’re trying to kill Damien.”250 On August 19, 2011,
the West Memphis Three walked out of an Arkansas courtroom free men,
though they will live with the stigma and collateral consequences of their
guilty pleas for the rest of their lives.251 Whether they were guilty of the
charged offenses may never be truly known, but it is clear that despite
insufficient evidence to convict them at trial and strong indications that they
were innocent, the three were enticed by the power of the plea-bargaining
machine.252
While the Supreme Court acknowledged the need for plea bargaining
in Brady and approved bargained justice as a form of adjudication in the
American criminal justice system, the Court also offered a cautionary note
regarding the role of innocence.253 At the same time the Court made clear
242

See id.
See Robertson, supra note 232, at A12.
244
See id.
245
See id.
246
See id.
247
See id.
248
See id. (“Under the seemingly contradictory deal, Judge David Laser vacated the
previous convictions, including the capital murder convictions for Mr. Echols and Mr.
Baldwin. After doing so, he ordered a new trial, something the prosecutors agreed to if the
men would enter so-called Alford guilty pleas.”).
249
See id.
250
Id.
251
See id.
252
See id.
253
See Brady v. United States, 397 U.S. 742, 752–58 (1970).
243

2013]

PLEA BARGAINING'S INNOCENCE PROBLEM

47

its belief that innocent defendants were not vulnerable to the powers of
bargained justice, the Court reserved the ability to reexamine the entire
institution should it become evident it was mistaken.254 The Court stated:
For a defendant who sees slight possibility of acquittal, the advantages of pleading
guilty and limiting the probable penalty are obvious—his exposure is reduced, the
correctional processes can begin immediately, and the practical burdens of a trial are
eliminated. For the State there are also advantages—the more promptly imposed
punishment after an admission of guilt may more effectively attain the objectives of
punishment; and with the avoidance of trial, scarce judicial and prosecutorial
resources are conserved for those cases in which there is a substantial issue of the
defendant’s guilt or in which there is substantial doubt that the State can sustain its
255
burden of proof.

Continuing to focus more directly on the possibility of an innocence issue,
the Court stated:
This is not to say that guilty plea convictions hold no hazards for the innocent or that
the methods of taking guilty pleas presently employed in this country are necessarily
valid in all respects. This mode of conviction is no more foolproof than full trials to
the court or to the jury. Accordingly, we take great precautions against unsound
results, and we should continue to do so, whether conviction is by plea or by trial. We
would have serious doubts about this case if the encouragement of guilty pleas by
offers of leniency substantially increased the likelihood that defendants, advised by
256
competent counsel, would falsely condemn themselves.

This caveat about the power of plea bargaining has been termed the Brady
safety valve, because it allows the Supreme Court to reevaluate the
constitutionality of bargained justice if the persuasiveness of plea offers
becomes coercive and surpasses a point at which it begins to ensnarl an
unacceptable number of innocent defendants.257
Interestingly, Brady is not the only Supreme Court plea-bargaining
case to include mention of the innocence issue and the safety valve.258 In

254
255
256
257

See id. at 757–58; see also Dervan, supra note 26, at 87–88.
Brady, 397 U.S. at 752 (emphasis added).
Id. at 757–58 (emphasis added).
See Dervan, supra note 26, at 88.

Safety-valves are intended to relieve pressure when forces within a machine become too great
and, thereby, preserve the integrity of the machine. The Brady safety-valve serves just such a
purpose by placing a limit on the amount of pressure that can constitutionally be placed on
defendants to plead guilty. According to the Court, however, should plea bargaining become so
common that prosecutors offer deals to all defendants, including those whose guilt is in question,
and the incentives to bargain become so overpowering that even innocent defendants acquiesce,
then the Brady safety-valve will have failed and the plea bargaining machine will have ventured
into the realm of unconstitutionality.

Id.
258

See id. at 88–89.

48

LUCIAN E. DERVAN & VANESSA A. EDKINS

[Vol. 103

Alford, for instance, the Court made clear that this form of bargained justice
was reserved only for cases where the evidence against the defendant was
overwhelming and sufficient to overcome easily the defendant’s continued
claims of innocence.259 Where any uncertainty remained, the Supreme
Court expected the case to proceed to trial to ensure that “guilty pleas are a
product of free and intelligent choice,” rather than overwhelming force
from the prosecution.260 The same language requiring that plea bargaining
be utilized in a manner that permits defendants to exercise their free will
was contained in the 1978 case of Bordenkircher v. Hayes.261 In
Bordenkircher, the Court stated that the accused must be “free to accept or
reject the prosecution’s offer.”262 Just as the Court had stated in Brady and
Alford, it concluded its discussion in Bordenkircher by assuring itself that
as long as such free choice existed and the pressure to plead guilty was not
overwhelming, it would be unlikely that an innocent defendant might be
“driven to false self-condemnation.”263
As is now evident from the study described herein, the Supreme Court
was wrong to place such confidence in the ability of individuals to assert
their right to trial in the face of grave choices.264 In our research, more than
half of the study participants were willing to forgo an opportunity to argue
their innocence in court and instead falsely condemned themselves in return
for a perceived benefit.265 That the plea-bargaining system may operate in a
manner vastly different from that presumed by the Supreme Court in 1970
and has the potential to capture far more innocent defendants than predicted
means that the Brady safety valve has failed. Perhaps, therefore, it is time
for the Court to reevaluate the constitutionality of the institution with an eye
towards the true power and resilience of the plea-bargaining machine.

259

North Carolina v. Alford, 400 U.S. 25, 37 (1970); see also ABA PROJECT, supra note
69, at 2 (“Moreover, the limited use of the trial process for those cases in which the
defendant has grounds for contesting the matter of guilt aids in preserving the
meaningfulness of the presumption of innocence.”).
260
Alford, 400 U.S. at 38 n.10.
261
Bordenkircher v. Hayes, 434 U.S. 357 (1978).
262
Id. at 363.
263
Id.
264
See supra Part III (discussing the plea-bargaining study).
265
See Edkins & Dervan, supra note 163, at 13.