Exonerations in The United States 1989 Through 2003: Journal of Criminal Law and Criminology
Exonerations in The United States 1989 Through 2003: Journal of Criminal Law and Criminology
Exonerations in The United States 1989 Through 2003: Journal of Criminal Law and Criminology
Volume 95
Article 5
Issue 2 Winter
Winter 2005
Kristen Jacoby
Daniel J. Matheson
Nicholas Montgomery
Recommended Citation
Samuel R. Gross, Kristen Jacoby, Daniel J. Matheson, Nicholas Montgomery, Exonerations in the United States 1989 through 2003, 95
J. Crim. L. & Criminology 523 (2004-2005)
This Symposium 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 editor of Northwestern University School of Law Scholarly Commons.
0091-4169/05/9502-0523
THEJOURNAL OFCRIMINAL LAw & CRIMINOLOGy Vol. 95, No. 2
Copyright 0 2005 by Northwestern University, School of Law Printed in US.A.
* Thomas & Mabel Long Professor of Law, University of Michigan Law School
([email protected]). The authors thank Phoebe C. Ellsworth and James S. Liebman, for
valuable comments on earlier drafts, and Joanne L.Werdel for excellent research assistance.
We are also grateful to the many people-too many to name-who drew our attention to ex-
onerations we would otherwise have missed. Our work depended on their assistance. The
research for this study was supported by a grant from The Gideon Project of the Open Soci-
ety Institute.
University of Michigan Law School, J.D. candidate, May 2005; Wellesley College,
B.A., 1997.
*..Associate, King & Spalding LLP, Washington, D.C.; University of
Michigan Law
School, J.D., May 2004.
1 University of Michigan Department of Economics and Ford School of Public Policy,
Ph.D. candidate, 2007.
tt Research Biostatistician, Department of Biostatistics and Epidemiology, The Chil-
dren's Hospital of Philadelphia; University of Michigan School of Public Health, MPH,
1998, Ph.D., 2004.
1 Rob Warden, Center on Wrongful Convictions, The Rape That Wasn t: The FirstDNA
Exoneration in Illinois, availableat http://www.law.northwestem.edu/depts/clinic/wrongful/
exonerations/Dotson.htm (last visited Jan. 11, 2005).
GROSS ETAL. [Vol. 95
2 Because men make up over 96% of the total, we generally refer to exonerated defen-
dants using male pronouns.
3 This is a conservative estimate of the direct consequences of these wrongful convic-
tions. We have not counted time spent in custody before conviction. Nor have we included
time spent on probation or parole, or time on bail or other forms of supervised release pend-
ing trial, retrial, or dismissal, even though all of these statuses involve restrictions on lib-
erty-some mild, some onerous.
4 We have excluded any case in which a dismissal or an acquittal appears to have
been
based on a decision that while the defendant was not guilty of the charges in the original
conviction, he did play a role in the crime and may be guilty of some lesser crime that is
based on the same conduct. For our purposes, a defendant who is acquitted of murder on
retrial, but convicted of involuntary manslaughter, has not been exonerated. We have also
excluded any case in which a dismissal was entered in the absence of strong evidence of fac-
tual innocence, or in which--despite such evidence-there was unexplained physical evi-
dence of the defendant's guilt.
5 See Sydney P. Freedberg, He Didn'tDo It, ST. PETERSBURG TIMES, Jan. 7,
2001, at 1A
(discussing Frank Lee Smith); J.M. Lawrence, Ex-ProsecutorApologizes to Salvati, Limone,
BOSTON HERALD, May 12, 2002, at 6; Ralph Ranalli, CongressionalProbe; FBI Used Hit
Man as Informant Transcripts Reveal Bureau Recruited Killer Despite Past, BOSTON GLOBE,
Dec. 4, 2002, at A30 (discussing Louis Greco and Henry Tameleo); Jon Yates & Kevin
Lynch, Confession Leads to 2 Arrests in '75 Killing; Man Convicted in Indiana Case Died in
Prison,CHI. TRIB., Aug. 29, 2002, at 1 (discussing John Jeffers).
2005] EXONERA TIONS IN THE UNITED STA TES
6 Most of the exonerations we include in this database are listed on one or more of the
websites that are maintained by three organizations: The Death Penalty Information Center,
at http://www.deathpenaltyinfo.org; the Innocence Project at Cardozo Law School, at
http://www.innocenceproject.org; and the Center on Wrongful Convictions at Northwestern
University Law School, at http://www.law.northwestern.edu/depts/clinic/wrongful. We have
gathered additional information on most of the cases from these three lists, reviewed them
carefully, and excluded some cases that do not meet our own criteria for an exoneration.
7 An earlier version of this paper was released in April 2004, listing a total of 328 exon-
erations. See Samuel R. Gross, Exonerations in the United States, 1989 Through 2003 (April
9, 2004) (early unpublished manuscript, at http://www.law.umich.edu/newsandinfo/ exon-
erations-in-us.pdf). After that report was released we learned of about fifteen additional ex-
onerations between 1989 and 2003, mostly by way of e-mails from individuals who con-
tacted us about cases we had missed. We have also excluded three cases we listed in the
initial report because additional information revealed that the defendants had not been "ex-
onerated" as we define the term: Edward Ryder in Pennsylvania in 1996, and Dennis Hal-
stead and John Restivo in New York in 2003. Halstead and Restivo were removed from the
list because it remains theoretically possible that charges will be retried. See Chan Lam,
1984 Teen Homicide; HairMay Play a Role in Case, NEWSDAY, Dec. 5, 2004, at A53. More
likely they will be added to the list of exonerees in 2005 or 2006 rather than 2003. See infra
note 32 and accompanying text for a discussion of the delays that often slow down the proc-
ess of exoneration.
GROSS ETAL. [Vol. 95
8 Raymond Bonner, Death Row Inmate is Freed After DNA Test Clears Him, N.Y.
TIMES, Aug. 24, 2001, at A11. This is hardly the only example of prosecutors and police
officers refusing, against all logic, to believe that a defendant they once charged and prose-
cuted could possibly be innocent. In 1993, in Baltimore County, Maryland, Kirk Bloods-
worth became the first defendant in the United States who had been sentenced to death to be
exonerated by DNA evidence. Nine years later, the chief prosecutor of the county said that
the police "still believe [Bloodsworth] did it" and that she herself was "not sure." Lori
Montgomery, EliminatingQuestions ofLife or Death; Prosecutor'sPolicy Raises Questions
in Md., WASH. POST, May 20, 2002, at Bl. More than a decade after Bloodsworth was re-
leased, the police finally, after inexplicable delays, used the DNA evidence at their disposal
to identify the real killer, a Maryland prisoner serving a forty-five-year sentence for bur-
glary, attempted rape and assault with intent to murder. Susan Levine, Ex-Death Row In-
mate Hears Hoped-for Words: We Found Killer, WASH. POST, Sep. 6, 2003, at Al. In 2000,
Virginia Governor Jim Gilmore pardoned Earl Washington after DNA tests cleared Wash-
ington of a rape murder for which he had been sentenced to death and implicated a convicted
serial rapist, Kenneth Tinsley. The Governor ordered a new investigation; four years later,
nothing had happened in that investigation and the law enforcement officers involved con-
tinued to consider Washington a suspect. By then, new DNA tests, commissioned by Wash-
ington's attorneys over the state's objections, conclusively confirmed Tinsley's guilt and re-
confirmed Washington's innocence. Maria Glod, Lawyers Say DNA ClearsEx-Va. Death
Row Inmate; State Defends Testing by Forensic Lab, WASH. POST, Apr. 6, 2004, at B 1;
Frank Green, Justice Undone in 1982 Killing; Victim's HusbandBlasts Lack of Progress
After Washington's Pardon,RICHMOND TIMES-DISPATCH, Mar. 31, 2004, at B 1.
9 Jeffrey Bils & Ted Gregory, "I Just Want To Go Home "; A Nightmare Ends, One
Con-
tinues in Nicarico Case, CHI. TRIB., Dec. 9, 1995, at 1.
20051 EXONERA TIONS IN THE UNITED STA TES
something about the causes of false convictions, and about the operation of
our criminal justice system in general. It is possible that a few of the hun-
dreds of exonerated defendants we have studied were involved in the crimes
for which they were convicted, despite our efforts to exclude such cases.
On the other hand, it is certain-this is the clearest implication of our
study-that many defendants who are not on this list, no doubt thousands,
have been falsely convicted of serious crimes but have not been exonerated.
The rate of exonerations has increased sharply over the fifteen-year pe-
riod of this study, from an average of twelve a year from 1989 through
1994, to an average of forty-two a year since 2000. The highest yearly total
was forty-four, in 2002 and again in 2003. See Figure 1.10
The number of DNA exonerations has increased across this period,
from one or two a year in 1989 to 1991, to an average of six a year from
1992 through 1995, to an average of twenty a year since 2000. Non-DNA
exonerations were less rare initially, and remained relatively stable through
the 1990s, averaging about ten a year. Their numbers have increased rap-
idly in the last several years, averaging twenty-three a year since 2000.
Basis 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Total
DNA 2 1 2 6 4 7 7 13 9 4 10 17 24 21 17 144
Other 9 8 14 8 8 5 13 9 10 9 13 21 19 23 27 195
TotalIt 9 16 14 12 12 20 22 19 13 23 38 43 44 44 340
GROSS ETAL. [Vol. 95
40O
40
.o
x 20 No-O
S20
LUJ
DNA
10
0 1!
1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Year
I These figures were calculated from a list available at The Innocence Project, Other
Projects by State, at http://www.innocenceproject.org/about/other_projects.php (last visited
Jan. 11, 2005).
12 This number includes three defendants who were convicted of manslaughter.
2005] EXONERA TIONS IN THE UNITED STA TES
BASIS
NUMBER OF
CRIME EXONERATIONS DNA Other
Murder 205 (60%) 39 166
Death Sentences 74 (22%) 13 61
Other Murder Cases 131 (39%) 26 105
Rape 121 (36%) 105 16
Other Crimes of Violence 11 (3%) 0 11
Drug and Property Crime 3 (1%) 0 3
TOTAL 340 (101%) 144 196
This highly skewed distribution tells us a great deal about the relation-
ship between exonerations-those erroneous convictions that are discov-
ered and remedied, at least in part-and the larger group of all false convic-
tions, the vast majority of which are never discovered. We consider that
relationship by examining the two major categories of crimes for which ex-
onerations are comparatively common.
13 We coded cases with multiple charges by the most serious crime for which the defen-
dant was convicted, on the following descending scale: murder, rape, other violent crimes,
non-violent crimes. For example, if the exonerated defendant was convicted of murder and
rape, we classified the exoneration as a murder; if he was convicted of robbery and rape, we
classified it as a rape.
GROSS ET AL. [Vol. 95
public order offenses. 14 Why are 90% of the exonerations for non-
homicidal crimes concentrated among the rape cases?
The comparison between rape and robbery is particularly telling.
Robbery and rape are both crimes of violence in which the perpetrator is of-
ten a stranger to the victim. As a result, robberies and rapes alike are sus-
ceptible to the well-known dangers of eyewitness misidentification. In fact,
there is every reason to believe that misidentifications in robberies outnum-
ber those in rapes, by a lot:
(1) Robberies are more numerous than rapes. In 2002, for exam-
ple, the FBI estimates that 95,136 forcible rapes and 420,637 rob-
beries were reported to police departments in the United States,
leading to 20,126 arrests for rape and 77,342 arrests for robbery.' 5
(2) Eyewitness misidentification is almost entirely restricted to
crimes committed by strangers, which includes 16
about three quar-
ters of robberies, but only a third of rapes.
(3) The nature of the crime of rape is such that the victim usually
spends a considerable amount of time in close physical proximity
to the criminal; robberies are frequently quick, and may involve
less immediate physical contact.
14 PAIGE M. HARRISON & ALLEN J. BECK, U.S. DEP'T OF JUSTICE, PRISONERS IN 2002, at
10 tbl.15 (2003).
15 FED. BUREAU OF INVESTIGATION, CRIME IN THE UNITED STATES 2002: UNIFORM CRIME
REPORTS 66 tbl.1, 244 tbl.38 (2002). The arrest figures are limited to the agencies participat-
ing in the FBI's Uniform Crime Reporting Program.
16 CATHY MASTON & PATSY KLAUS, U.S. DEP'T OF JUSTICE, CRIMINAL VICTIMIZATION IN
THE UNITED STATES, 2002 STATISTICAL TABLES 42 tbl.29 (2003).
17 Samuel R. Gross, Loss of Innocence: Eyewitness Identification and Proofof Guilt, 16
J. LEGAL STUD. 395, 413 (1987). The misidentifications in that study did not all involve
convictions. They were unevenly spread across the twentieth century, with the largest num-
ber-thirty-six-in the decade of 1970-79.
2005] EXONERA TIONS IN THE UNITED STATES
ous: DNA. In 1987, the first DNA exoneration in the country was two
years in the future. Since 1989, however, 87% of exonerated rape defen-
dants were cleared by DNA evidence. Only 19% of murder exonerations
included DNA evidence (and none of the other non-rape exonerations), and
all but a couple of those murders also included rape as well.
The implication is clear. If we had a technique for detecting false con-
victions in robberies that was comparable to DNA identification for rapes,
robbery exonerations would greatly outnumber rape exonerations, and the
total number of falsely convicted defendants who were exonerated would be
several times what we report. And even among rape cases, DNA is only
useful if testable samples of biological evidence were preserved and can be
found, which is not always true.
In short, the clearest and most important lesson from the recent spike
in rape exonerations is that the false convictions that come to light are the
tip of an iceberg. Beneath the surface there are other undetected miscar-
riages of justice in rape cases without testable DNA, and a much larger
group of undetected false convictions in robberies and other serious crimes
of violence for which DNA identification is useless.
ever the causes, this is a terrible prospect: that we are most likely
to convict innocent defendants in those cases in which their very
lives are at stake.
Considering the huge discrepancies between the exoneration rates for
death sentences, for other murder convictions, and for criminal convictions
generally, the truth is probably a combination of these two appalling possi-
bilities: We are both much more likely to convict innocent defendants of
murder-and especially capital murder-than of other crimes, and a large
number of false convictions in non-capital cases are never discovered be-
cause nobody ever seriously investigates the possibility of error.
average, get more resources for their defense at trial than other defendants. But even under
the best of circumstances, good defense work will catch some but not all of the errors made
in police investigations, and while capital defendants may receive better defense than other
defendants, on average, "better" does not necessarily mean "good," and many capital defen-
dants get abysmal defense attorneys. The net effect appears to be that many more errors are
generated in the initial investigations of capital cases, and that good defense work catches
some but by no means all of these excess errors. See id. at 148-49. See generally James S.
Liebman, The Overproduction of Death, 100 COLUM. L. REv. 2030 (2000) (procedural in-
centives lead to high number of death sentences and high proportion of legal error among
them).
GROSS ETAL. [Vol. 95
part CRASH officers had routinely lied in arrest reports, shot and
killed or wounded unarmed suspects and innocent bystanders,
planted guns on suspects after shooting them, fabricated evidence,
and framed innocent defendants. In the aftermath of this scandal,
at least 100 criminal defendants who had been framed by Rampart
CRASH officers-and possibly as many as 150-had their con-
victions vacated and dismissed by Los Angeles County judges in
late 1999 and 2000. The great majority were young Hispanic men
who had pled guilty to false felony gun or drug charges.23
In 1999 and 2000, thirty-nine defendants were convicted of drug
offenses in Tulia, Texas, on the uncorroborated word of a single
dishonest undercover narcotics agent. In 2003, thirty-five of them
were pardoned when it was shown that the undercover officer had
systematically lied about these cases, and charged the defendants
with drug sales that had never occurred. (The remaining four
Tulia defendants were not eligible for pardons because their con-
victions had been dismissed,24
or because they were also impris-
oned on unrelated charges.)
The Rampart and Tulia scandals are not the only mass exonerations in
the United States in the past fifteen years. After we released our initial re-
port, in April of 2004, we heard about the Dallas Sheetrock Scandal that
23 For an in-depth look at the Rampart scandal, including links to official reports and re-
views and a summary of the scandal's aftermath, see Frontline:L.A.P.D. Blues (PBS televi-
sion broadcast, May 15, 2001) (transcript available at http://www.pbs.org/wgbh/pages/
frontline/shows/lapd/bare.html). See also Lou Cannon, One Bad Cop, N.Y. TIMES, Oct. 1,
2000, § 6 (Magazine), at 32; Scott Glover & Matt Lait, Lack of Funds Stalls RampartProbe;
The LAPD Seeks PrivateDonationsSo That an Independent Panel Can Begin Investigating
the Department'sHandling of the Scandal, L.A. TIMES, Nov. 6, 2003, at B 1 ($42 million in
settlements paid out in civil settlements); Anna Gorman, For Some, It's Too Late to Over-
turn Convictions: Judges Are Refusing to Review Cases Involving Tainted Officers iflnmate
Is No Longer in Custody, L.A. TIMES, May 19, 2002, at BI (nearly 150 convictions over-
turned); David P. Leonard, Policing the Criminal Justice System: Different Worlds, Different
Realities, 34 LoY. L.A. L. REV. 863, 872-79 (2001) (Rampart residents overwhelmingly La-
tino); Stephen Yagman, Bada Bing, L.A. City Hall Has a Rico Ring, L.A. TIMES, Apr. 25,
2001, at B9 (more than 110 convictions overturned); RAMPART INDEPENDENT REVIEW PANEL,
REPORT OF THE RAMPART INDEPENDENT REVIEW PANEL (2000), available at
http://www.ci.la.ca.us/oig/rirprpt.pdf.
24 See Polly Ross Hughes, Perry Pardons 35 in Tulia Sting, Hous. CHRON., Aug. 23,
2003, at Al; Adam Liptak, $5 Million Settlement Ends Case of Tainted Texas Sting, N.Y.
TIMES, Mar. 11, 2004, at A14; Laura Parker, Texas Scandal Throws Doubt on Anti-Drug
Task Forces, USA TODAY, Mar. 31, 2004, at 3A. Details on the numbers of the Tulia defen-
dants and their outcomes of their cases were provided by Ms. Vanita Gupta, a lawyer for the
NAACP Legal Defense and Educational Fund, who represented them in the proceeding that
led to their exoneration.
2005] EXONERA TIONS IN THE UNITED STA TES
25 See Mark Donald, Dirty or Duped?: Who's to Blamefor the Fake-DrugScandal Rock-
ing Dallas Police? Virtually Everyone, DALLAS OBSERVER, May 2, 2002, available at
www.dallasobserver.com/issues/2002-05-02/news/feature.html.; Paul Duggan, "Sheetrock
Scandal" Hits Dallas Police,WASH. POST, Jan. 18, 2002, at A12.
26 In one of the murder exonerations and four of the rape cases the defendants were ex-
onerated before they were sentenced; one additional rape defendant was sentenced to proba-
tion only.
27 About half of the sentences-other than death or life imprisonment-included a
maximum and a minimum, e.g., "10 to 25 years." In those cases, we have reported the mini-
mums.
GROSS ETAL. [Vol. 95
28 The case of Robert Farnsworth, Jr., who was exonerated in Michigan in 2000, is an
exception that illustrates this rule. Farnsworth was arrested for grand larceny and confessed
under police pressure after a cash deposit bag belonging to his employer went missing. He
immediately recanted his confession and claimed that he had placed the bag in a night de-
posit box at the company's bank, but a jury convicted him and he was sentenced to a six-
month suspended sentence and three years probation. Ordinarily, that would have been the
end of the story. By a fluke, another cash bag deposited in the same night drop box was lost
almost a year after the first, and the owner of the business in question knew the bank presi-
dent and asked him to have the drop box opened and inspected-and both cash bags were
found stuck in the mechanism (plus a third that hadn't been missed). See Bank Finds Lost
Cash Stuck in Vault, Ex-Wendy's Manager Convicted of Stealing is Vindicated, DETROIT
FREE PRESS, Mar. 13, 2000.
29 See, e.g., Barbara Taylor, Trapped on Rikers Island,N.Y. TIMES, Sept. 7, 1996, at 21.
30 In 1998, for example, 90% of those convicted of violent felonies in large urban coun-
ties pled guilty. For all felonies, 96% of those convicted pled guilty. BRIAN A. REAVES,
BUREAU OF JUSTICE STATISTICS, FELONY DEFENDANTS IN LARGE URBAN COUNTIES, 1998, at
24 tbl.23 (2001) (figures recalculated).
31 See supranotes 23 and 24.
2005] EXONERA TIONS IN THE UNITED STA TES
32 Mary Ann Fergus, Josiah Sutton Still Waits for Legal Exoneration, Hous. CHRON.,
Mar. 7, 2004, available at http://www.chron.com/cs/CDA/ssistory.mpl/pagel/2435555.
33 Sutton was finally pardoned on May 14, 2004. Roma Khanna, PerrySigns Pardonfor
Sutton; Man Convicted on Faulty DNA May Be Entitled to $100,000, Hous. CHRON., May
14, 2004, at Al; see also supra note 7 (discussing the cases of Dennis Halstead and John
Restivo).
GROSS ET AL. [Vol. 95
leased.3 4 We have not included McGhee in our data, nor any other defen-
dant who pled guilty in order to be released, regardless of the evidence of
the defendant's innocence. We are examining exonerations, and the final
official act in such a case is not an exoneration but a conviction, however
nominal or misleading. 35 (We have included McGhee's co-defendant, Terry
Harrington, who refused to take a similar deal, and got a dismissal after the
state's star witness at the original trial recanted once more.)
34 Mark Siebert, Case Dismissed Against Man Who Served 25 Years, DES MOINES REG.,
Oct. 25, 2003, at IA, available at http://www.truthinjustice.org/Harrington.htm.
35 For these purposes, a plea of "no contest" (or "nolo contendere") is equivalent to a
guilty plea. McGhee himself entered an Alford plea, which means that he was allowed to
deny participation in the crime and state that he was pleading guilty, despite that denial, to
avoid the risk of trial. All the same, that plea-like any other plea of guilty or no contest-
was the basis of a judgment of conviction, which is not an official exoneration.
We have included three cases in which the exonerated defendant did plead guilty or no
contest, but to a charge that is factually distinct from the crime for which he was originally
convicted. Medell Banks, for example, was charged with capital murder and pled guilty to
manslaughter in Alabama. He was released in 2003, after homicide charges were dropped
because of incontrovertible evidence that the supposed victim-a newborn infant who had
never been seen, alive or dead, by any trustworthy witness-could not have existed: the os-
tensible mother, Mr. Banks's wife Victoria, had a tubal ligation that made pregnancy impos-
sible. See infra text accompanying note 36. In the process, Mr. Banks agreed to plead guilty
to a misdemeanor, tampering with evidence. Carla Crowder, Accused in Killing of Newborn
Who Likely Never Existed, Choctaw County Man Makes Plea Deal, BIRMINGHAM NEWS, Jan.
11, 2003, at 1A. Since that charge (whether true or false) involved conduct totally distinct
from the original homicide charge, we count his case as an exoneration.
36 See id.
20051 EXONERA TIONS IN THE UNITED STA TES
37 For a comprehensive analysis of several of the major childcare and satanic ritual abuse
cases and the phenomenon of allegations of ritual abuse generally, see DEBBIE NATHAN &
MICHAEL SNEDEKER, SATAN'S SILENCE: RITUAL ABUSE AND THE MAKING OF A MODERN
AMERICAN WITCH HUNT (1995).
38 See Frontline: Innocence Lost (PBS television broadcast, May 27, 1997) (tapes, tran-
scripts and additional information, including subsequent developments, available at
http://www.pbs.org/wgbh/pages /frontline/shows/innocence/).
39 See NATHAN & SNEDEKER, supra note 37, at 95-96; see also People v. Pitts, 273 Cal.
Rptr. 757, 774-86 (Cal. Ct. App. 1990); John Johnson, "Kids Don't Lie": Faith in This As-
sumption Led to Dozens of Unjust Molestation Convictions in Bakersfield; Today One Man
Remains in Prison Even After Four of His Original Accusers Said He Never Touched Them,
L.A. TIMES, Aug. 10, 2003, § 9 (Magazine), at 16.
GROSS ET AL. [Vol. 95
from 1984 to 1995, and at least seventy-two were convicted. It is clear that
the great majority were totally innocent; almost all were eventually released
by one means or another before they completed their terms. 40 It is possible,
however, that some of these defendants did commit some acts of sexual mo-
lestation, incidents that later grew into implausible and impossible allega-
tions as the children were interviewed repeatedly by prosecutors and thera-
pists. We have included only one of these cases in our database, a case in
which we know that all of the supposed victims now say that they were
never molested in the first place-that the crime never occurred. Other-
wise, none of the wrongfully convicted victims of this terrible episode in
American legal history are included on this list because they have not been
officially exonerated.
40 In the largest set of ritual sex abuse cases, the Kern County sex abuse rings prosecu-
tions, at least eighteen of the twenty-six convicted defendants have had their convictions re-
versed and the charges dismissed-in at least seven cases, for gross prosecutorial miscon-
duct. Prosecutors declined to retry the cases in which convictions were reversed. The Kern
County defendants, who had been sentenced to as many as 100 years in prison, served be-
tween three and fifteen years. John Stoll is one of the very last defendants to remain in
prison, and by all accounts, the person who has served the longest time of all those convicted
in ritual sex abuse cases across the country; he received a hearing in February 2004, to de-
termine whether he should receive a new trial. Four of the six children, now adults, who tes-
tified against him, now say the abuse never happened. A fifth has no memory either way.
See John Johnson, New HearingIs Granted in ChildAbuse Conviction, L.A. TIMEs, Dec. 2 1,
2003, at B6; Kern County Justice; A Tragic Chapter in Our Justice System Closes with the
Release of Four Peoplefrom Prison 14 Years After Being Convicted of Child Molestation,
FRESNO BEE, Aug. 16, 1996, at B4; Tom Kertscher, Molestation Hysteria Left Sad Legacy;
Painful Lessons Learned in Overzealous Kern County Prosecutions,FRESNO BEE, Sept. 10,
1995 (Telegraph), at Al.
In fact, nearly all of those initially convicted in the childcare and ritual sex abuse cases
have since been reversed, and prosecutors have declined to retry defendants in almost every
case. See Frontline: Innocence Lost, supra note 38 (all charges against Bob Kelly, the last
remaining defendant in the Little Rascals case, were dropped in 1999) (supplemental infor-
mation available at http://www.pbs.org/wgbh/pages/frontline/shows/innocence/etc/ chronol-
ogy.html); id. (conviction of Kelly Michaels, a twenty-three year old day care worker,
charged with 115 counts of child sexual abuse overturned in 1993) (supplemental informa-
tion available at http://www.pbs.org/wgbh/pages/frontline/shows/innocence/etco-
ther.html#3); Update on the Wenatchee, Washington, Child Abuse Case (National Public
Radio broadcast, Aug. 2, 2001) (of the eighteen individuals convicted, all have had their
convictions overturned); When Children Accuse, Who to Believe? (ABC News Broadcast,
Jan. 28, 1999) (across the country, at least 140 people-about three-quarters of the accused
in wave of ritual sex abuse cases-have been acquitted, had their convictions overturned or
charges against them dropped).
2005] EXONERA TIONS IN THE UNITED STA TES
Table 2
Exonerationsby State
Number of
Rank State Exonerations
I Illinois 54
2 New York 35
3 Texas 28
4 California 27
5 Louisiana 17
6 Massachusetts 16
7 Florida 15
8 Pennsylvania 13
9 Oklahoma 11
10 Missouri 10
desperate for leads.4 5 The police themselves may be tempted to cut comers
and falsify evidence to convict a person they believe committed a terrible
murder.4 6
In 71% of the rape exonerations the victims and all other eyewitnesses
who testified were strangers to the falsely convicted defendant. By con-
trast, 85% of exonerated murder defendants knew the victim, or at least one
supposed eyewitness, before the crime. The central problem in most rape
investigations that go wrong is the mistaken identification of a defendant
who is otherwise unknown to those involved. The common problem in the
investigations of the murder cases we studied is deliberately false evidence
implicating an innocent defendant with a known relationship to the victim
or the lying witness.
An eyewitness misidentification by a stranger is easy to spot, once you
know that the person identified is innocent. Detecting a deliberate lie is
harder; there may be no simple way to tell if a statement was false, and if so
whether the falsehood was intentional. As a result, our information on per-
jury understates the extent of the problem. Even so, known perjury is a
surprisingly common feature of the trials that led to the convictions of these
exonerated defendants.
In at least sixty of the 340 exonerations, the defendant was deliberately
falsely accused at trial by someone who claimed to have witnessed the
crime: a supposed victim, participant, or eyewitness. About a quarter of
these false accusations (14/60) occurred in rape cases; in each, the false ac-
cuser was a complaining witness who lied about the occurrence of the
crime. Almost three-quarters of the exonerated defendants who were
falsely accused were convicted of murder (44/60). In two cases the false
accusers were surviving victims; most of the rest were (or claimed to be)
participants in the crimes. In other words, deliberate false accusations were
a major cause of misidentification in murder exonerations. In 43% of the
murder exonerations in which the defendant was misidentified by one or
more eyewitnesses (44/102), we also have information that at least one of
those witnesses misidentified the defendant deliberately.
In five of the exonerations that we have studied there are reports of
perjury by police officers. In an additional twenty-four we have similar in-
formation on perjury by forensic scientists testifying for the government. In
at least seventeen exoneration cases the real criminal lied under oath to get
the defendant convicted; in at least ninety-seven cases a civilian witness
who did not claim to be directly involved in the crime committed perjury-
usually a jailhouse snitch or another witness who stood to gain from the
false testimony.
Overall, in 43% of all exonerations (146/340) at least one sort of per-
jury is reported-including 56% of murder exonerations (114/205), and
25% of rape exonerations (30/121). See Table 3.
Table 3
Causes of False Convictionsfor Exonerations in Murder andRape Cases*
Murder Rape
(205) (121)
Eyewitness Mis-
identification 50% 88%
Reported Perjury 56% 25%
False Confession 20% 7%
The columns add up to more than 100% because some false convictions had multiple
causes.
47 In over half the false confessions (28/51) coercion is apparent from the record we
have; in about 10% (5/51) it appears that the false confession was volunteered; and in about
a third (18/51) we have too little information to say.
2005] EXONERA TIONS IN THE UNITED STA TES
dure is generally reserved for the most serious cases where there is no other
evidence sufficient to convict-which usually means a murder with no sur-
viving eyewitnesses.
False confessions are heavily concentrated among the most vulnerable
groups of innocent defendants. 48 Thirty-three of the exonerated defendants
were under eighteen at the time of the crimes for which they were con-
victed, and fourteen of these innocent juveniles falsely confessed-42%,
compared to 13% of older exonerees. Among the youngest of these juve-
nile exonerees-those aged twelve to fifteen-69% (9/13) confessed to
homicides (and one rape) that they did not commit.
False confessions are even more prevalent among exonerees with men-
tal disabilities.49 Our data indicate that sixteen of the 340 exonerees were
mentally retarded; 69% of them-over two thirds-falsely confessed. An-
other ten exonerees appear to have been suffering from mental illnesses;
seven of them falsely confessed. Among all other exonerees (some of who
may also have suffered from mental disabilities of which we are unaware)
the false confession rate was 11% (33/313). Overall, 55% of all the false
confessions we found were from defendants who were under eighteen, or
mentally disabled, or both. Among adult exonerees without known mental
disabilities, the false confession rate was 8% (23/272). See Table 4.
Table 4
False Confessions by Age and Mental Disability
American charged with killing a white retired police captain, did not con-
fess to murder in Iowa in 1978-but his sixteen-year-old friend Kevin
Hughes did, and that confession, which was later repeatedly retracted, led to
false murder convictions for Harrington and his co-defendant Curtis
McGhee. Hughes was never prosecuted.50 Similarly, in 1978 Paula Gray, a
seventeen-year-old borderline retarded girl, falsely confessed to participat-
ing in a double murder and rape in Chicago, and implicated four innocent
men. After she recanted, she was prosecuted for rape, murder and perjury,
and sentenced to fifty years in prison. The four men she named were also
all convicted, and two were sentenced to death. All five were exonerated
after DNA testing cleared the men of the rape; the real killers have since
been identified, linked to the rape by DNA, and confessed.5 And in 1988
in Austin, Texas, Christopher Ochoa falsely confessed to rape and murder
in order to avoid the threat of the death penalty-and along the way falsely
implicated his friend, Richard Danzinger; both were sentenced to life in
prison, and both were exonerated by DNA in 2001, three years after the real
52
criminal sent a letter to Governor Bush confessing to these same crimes.
VI. RACE
Table 5
Race of ExoneratedDefendants, by Crime
Murder Rape All Cases
(193) (107) (311)
White 34% 28% 32%
Black 50% 64% 55%
Hispanic 16% 7% 13%
Other 1% 0% 1%
TOTAL* 101% 99% 101%
54 Id. The proportion of whites is slightly higher among those arrested for rape-63% in
2001. BUREAU OF JUSTICE STATISTICS, U.S. DEP'T OF JUSTICE, SOURCEBOOK OF CRIMINAL
JUSTICE STATISTICS 356 tbl.4.10 (2002), available at http://www.albany.edu/sourcebook/
pdf/section4.pdf (statistics recalculated).
55 Black offenders accounted for an average of approximately 10% of all rapes and sex-
ual assaults of white victims between 1996 and 2002. BUREAU OF JUSTICE STATISTICS, U.S.
DEP'T OF JUSTICE, CRIMINAL VICTIMIZATION IN THE UNITED STATES, 1996-2002, availableat
http://www.ojp.usdoj.gov/bjs/abstract/cvusst.htm (based on the National Criminal Victimiza-
tion Survey; the statistic fluctuates from year to year because for each year it is extrapolated
from a sample of ten or fewer survey responses). Another Bureau of Justice Statistics
study-based on the National Incident-Based Reporting System, reports that in 88% of rapes
the victim and the offender are of the same race, and that the victims of rape are approxi-
GROSS ETAL. [Vol. 95
the race of both parties, almost exactly half (39/80) involve a black man
who was falsely convicted of raping a white woman.
There are many possible explanations for this disturbing pattern. Of
all the problems that plague the American system of criminal justice, few
are as incendiary as the relationship between race and rape. Nobody would
be surprised to find that bias and discrimination continue to play a role in
rape prosecutions. Still, the most obvious explanation for this racial dispar-
ity is probably also the most powerful: the perils of cross-racial identifica-
tion. Virtually all of the inter-racial rape convictions in our data were
based, at least in part, on eyewitness misidentifications, 56 and one of the
strongest findings of systematic studies of eyewitness evidence is that white
Americans are much more likely to mistake one black57
person for another
than to do the same for members of their own race.
mately evenly divided between whites and blacks. LAWRENCE A GREENFIELD, BUREAU OF
JUSTICE STATISTICS, U.S. DEP'T OF JUSTICE, AN ANALYSIS OF DATA ON RAPE AND SEXUAL
ASSAULT: SEX OFFENSES AND OFFENDERS 11 (1997), available at http://www.rainn.org/
Linked%20files/soo.pdf. It follows that the proportion of all rapes that have white victims
and black offenders is about 5-6%.
56 Consider the case of Ronald Cotton, a black man, who was convicted of raping Jenni-
fer Thompson, a white woman in 1985, in Burlington, North Carolina. Thompson was the
only eyewitness at the trial, and by all accounts she was very effective. She was absolutely
confident of her identification, in part because she spent a considerable amount of time with
the rapist and was determined to observe him closely so that she would be able to identify
him later on. She was equally confident when Cotton was retried 1987, convicted again, and
sentenced a second time to life in prison. Even so, she was wrong. Cotton was pardoned in
1995 after DNA tests proved that he was innocent, and that the real rapist was a different
black man, who was in prison on other charges. What makes the case most remarkable is
what happened after the exoneration. Ms. Thompson went to great lengths to make amends
to Mr. Cotton, and to speak out and publicize the case and the terrible mistake she had made,
so others could learn from it-which is why she is identified here by name. See Helen
O'Neill, The Perfect Witness, WASH. POST, Mar. 4, 2001, at F1. See generally EDWARD
CONNORS ET AL., CONVICTED BY JURIES, EXONERATED BY SCIENCE: CASE STUDIES IN THE USE
OF DNA EVIDENCE TO ESTABLISH INNOCENCE AFTER TRIAL 43-44 (1996), available at
http://www.ncjrs.org/pdffiles/dnaevid.pdf.
57 Christian A. Meissner & John C. Brigham, Thirty Years of Investigatingthe Own-Race
Bias in Memory for Faces: A Meta-Analytic Review, 7 PSYCHOL., PUB. POL. & L. 3-35
(2001).
20051 EXONERA TIONS IN THE UNITED STA TES
Table 6
Juvenile Exonerations by Race and Crime
found that 41% of defendants in juvenile courts in those three years were
black, and 67% of juveniles prosecuted as adults were black.60 In other
words, white teenagers who are arrested by the police are less likely than
blacks to be prosecuted in juvenile court, and much less likely to be prose-
cuted in felony court as adults.
All but one of the juvenile exonerees in our database were convicted,
as adults, of rape or murder. For these two extremely serious crimes, the
racial winnowing of juvenile offenders is severe. In 1990-94, 59% of mur-
der defendants injuvenile court were white and 36% were black; but among
juvenile murder defendants who were tried as adults the proportions were
more than reversed: 69% were black and only 25% were white. 6 1 For rape,
the proportion of blacks went from 44% of juveniles arrested,62 to 53% of
those prosecuted for rape in juvenile court, to 72% of juvenile rape defen-
dants prosecuted as adults.6 3 There are, no doubt, false convictions among
the cases that remain in juvenile court, and a substantial proportion of them
may involve white defendants. Like other false convictions with compara-
tively light sentences, these are errors that are unlikely to ever be corrected
by formal exoneration. None appear in this database.
Even so, 25% of juvenile murder defendants prosecuted in adult courts
64
in the early 1990s were white, as were 28% of juvenile rape defendants -
but only 9% of juvenile exonerees. The disparity could be due to chance;
the number of cases is not large. It could also be due to systematic racial
differences in the process of investigation. Black juvenile rape defendants,
like all black rape defendants, face a special danger of cross-racial misiden-
tification. In many of the juvenile murder exonerations (and in some of the
rapes) the primary evidence against the defendant was a false confession.
Eight-five percent of the juvenile exonerees who falsely confessed were Af-
rican American (11/13). It may be that police officers are more likely to
use coercive interrogation tactics on black juveniles than on white juve-
niles-that would explain the high proportion of blacks among the innocent
tent with the sample in a 1998 Bureau of Justice Statistics report on juvenile defendants in
criminal courts. See FED. BUREAU OF INVESTIGATION, supra note 15, at 252 tbl.43.
60 GERARD A. RAINVILLE & STEVEN K. SMITH, BUREAU OF JUSTICE STATISTICS, JUVENILE
of two such windows. Rapes are vastly over represented among exonera-
tions because DNA identification enables us to detect errors in rape convic-
tions that were obtained before that technology became available. Death
sentences are over represented-by an even greater margin-in part be-
cause we work hard to detect and correct errors in judgments that could lead
to the execution of innocent defendants. That poses a terrible question:
How many additional hundreds or thousands of false convictions would we
have discovered if we had worked just as hard to find them among non-
capital murders, or among non-homicidal felonies?
The extraordinary rate of exoneration of death-sentenced defendants
also raises deep questions about the accuracy of our system for determining
guilt in capital cases. Exonerations from death row are more than twenty-
five times more frequent than exonerations for other prisoners convicted of
murder, and more than 100 times more frequent than for all imprisoned fel-
ons. This huge discrepancy must mean that false convictions are more
likely for death sentences than for all murder cases, and much likely than
among felony convictions generally-an unavoidable and extremely dis-
turbing conclusion.
Finally, the frequency of exonerations from death row is a chilling re-
minder of the consequences of these false convictions. If we managed to
identify and release 75% of innocent death-row inmates before they were
put to death, then we also executed twenty-five innocent defendants from
1989 through 2003. If, somehow, we have caught 90% of false capital con-
victions, than we only executed eight innocent defendants in that fifteen-
year period. Is it conceivable that a system that produces all these horren-
dous errors in the first place could also detect and correct 90% of those er-
rors, after the fact? And considering the number of mistakes in capital tri-
als, even an unlikely 90% exoneration rate would be disturbingly low.
Worse yet, the high rate of death-row exoneration is limited to defen-
dants who have been sentenced to death. Approximately half of all defen-
dants who are convicted at capital trials are sentenced to life imprisonment
instead; and of those who are sentenced to death most are resentenced to
life imprisonment at some point in the process of review; about 40% have
their convictions or sentences reversed on their first appeal, and most of
them are ultimately resentenced to life. 65 In other words, the bulk of de-
fendants at capital trials are subject to the frightening risk of error that
plagues capital prosecutions-they are as likely as other capital defendants
to be convicted of murders they did not commit-but they get little or none
65 Andrew Gelman, James S. Liebman, Valerie West, & Alexander Kiss, A Broken Sys-
tem: The Persistent Patterns of Reversals of Death Sentences in the United States, 1 J.
EMPIRICAL LEGAL STUD. 209, 214 (2004).
2005] EXONERA TIONS IN THE UNITED STA TES 553
APPENDIX
INDIANA MASSACHUSETTS
Charles Smith, 1991 Harold Sullivan, 1989
James Cameron, 1993 Louis Santos, 1990
Dwayne Scruggs, 1993* Marvin Mitchell, 1997*
Jerry Watkins, 2000* Donnell Johnson, 2000
Richard Alexander, 2001 * Neil Miller, 2000*
Larry Mayes, 2001* Marion Passley, 2000
John Jeffers, 2002 Eric Sarsfield, 2000*
Ulysses Rodriguez Charles, 2001*
IOWA Angel Hernandez, 2001 *
Terry Harrington, 2003 Peter Limone, 2001
Joseph Salvati, 2001
KANSAS Louis Greco, 2002
Joe Jones, 1992* Henry Tameleo, 2002
Eddie James Lowery, 2003* Kenneth Waters, 2001*
Shawn Drumgold, 2003
KENTUCKY Dennis Maher, 2003*
William Gregory, 2000*
Larry Osborne, 2002 MARYLAND
Kirk Bloodsworth, 1993*
LOUISIANA Anthony Gray, 1999*
Isaac Knapper, 1991 Michael Austin, 2002
Gerald Burge, 1992 Bernard Webster, 2002*
Roland Gibson, 1993
Hayes Williams, 1997 MICHIGAN
Curtis Kyles, 1998 Robert Farnsworth, Jr., 2000
Clyde Charles, 1999* Dwight Love, 2001
John Thompson, 1999 Eddie Joe Lloyd, 2002*
Shareef Cousin, 1999 Kenneth Wyniemko, 2003*
Albert Burrell, 2000
Michael Graham, 2000
Dwight LaBran, 2001
GROSS ETAL. [Vol. 95
VIRGINA
David Vasquez, 1989*
Walter Snyder, 1993*
Edward Honaker, 1994*
Troy Webb, 1996*
Earl Washington, 2000*
Marvin Anderson, 2001 *
Jeffrey Cox, 2001
Julius Earl Ruffin, 2003*
WASHINGTON
Benjamin Harris, 1997
Mark Clark, 1999
Jeff Schmieder, 1999
Ross Sorrels, 2003
WISCONSIN
Frederic Saecker, 1996*
Anthony Hicks, 1997*
Cornelius Reed, 1997