Hjlpp Cassell Examination of Alleged False Confessions Spring 1999
Download original document:
This text is machine-read, and may contain errors. Check the original document to verify accuracy.
Page 1 LEXSEE 22 HARV .J.L. & PUB. POLY 523 Copyright (c) 1999 Harvard Society for Law & Public Policy, Inc. Harvard Journal of Law & Public Policy Spring, 1999 22 Harv. J.L. & Pub. Pol'y 523 LENGTH: 37719 words ARTICLE: THE GUILTY AND THE "INNOCENT": AN EXAMINATION OF ALLEGED CASES OF WRONGFUL CONVICTION FROM FALSE CONFESSIONS NAME: Paul G. Cassell * BIO: * Professor of Law, University of Utah College of Law (email@example.com). Special thanks to James Agar, Al Alschuler, Patricia Cassell, Gisli Gudjonsson, Lynne Henderson, Richard Leo, Richard Ofshe, Lee Teitelbaum, Welsh White, and the participants in the Virginia Constitutional Law Workshop for assistance in preparing this article. Jonathan Owens, Michael Schwalb, and Pam Vickrey provided valuable research help. I appreciate those who took time to help me locate hard-to-find original trial court records or understand the cases, including Ken Burr, John Cliff, Jr., Rosita Creamer, Darrell Dugan, Chuck Griffith, Phil Harju, Steven Hovani, Darnisa Johnson, Josh Marquis, Mike Mermel, Karen Olio, Jeff Pavletic, David Raupp, Howard Swindle, Elena Tompkins, and Rick Whitt. The conclusions I drew about the cases are my own. This article was supported by the University of Utah College of Law Research Fund and the University of Utah Research Committee. SUMMARY: ... Judge Eisele concluded that "the Court specifically finds that [Fairchild] was not instructed or coached regarding the content of his confessions." ... TEXT: [*524] Given the fallibility of human institutions, the possibility exists that police might obtain a confession from an innocent person to a crime that he did not commit. It is even possible that this "false confession" might, in turn, lead to an erroneous conviction. Despite occasional claims that specific individuals have been wrongfully convicted as the result of false confessions, this specific risk has never been the subject of empirical study. Professors Richard Leo and Richard Ofshe have attempted to fill this void with what is sure to be a widely cited study of sixty cases of alleged police-induced false confessions in the post-Miranda era. n1 According to the authors, in twenty-nine of these cases the false confession resulted in the wrongful conviction of an innocent person. n2 These assertions are not advanced in an effort to right wrongs in individual cases, but rather to justify possibly dramatic changes in how the justice system handles interrogations and confessions. Indeed, Leo and Ofshe conclude that the problem of false confessions threatens the very "quality of criminal justice in America by inflicting significant and Page 2 22 Harv. J.L. & Pub. Pol'y 523, *524 unnecessary harms on the innocent." n3 They accordingly recommend that judges should be empowered to review confessions for "reliability" through close scrutiny of the "post-admission narrative" of suspects. n4 In a previous article, I simply assumed that Leo and Ofshe correctly asserted that all of their cases involved "innocent" persons and discussed what implications might be drawn from this assumption. n5 Yet this linchpin claim about innocence is [*525] worth examining. All of Leo and Ofshe's policy recommendations rest on conclusions drawn from police interrogation gone wrong. If they have studied not interrogation failures but rather its successes, any justification for change disappears. More generally, scrutiny of their cases may shed light on an important methodological question: how can such miscarriages of justice be accurately identified? Leo and Ofshe rely in large measure on secondary sources for the descriptions of the evidence against the defendants in their collection. This approach is understandable. For many cases, court records are available only in the local courthouses where the trial took place, while media accounts are often readily accessible in computerized databases. Relying on secondary sources, however, poses the risk of inaccurate recounting of the evidence. Examining primary sources for the cases in Leo and Ofshe's collection reveals that this is a very real problem. Part I places the issue of false confessions in context. It notes that the innocent are at risk not only from false confessions, but also from "lost" confessions - that is, confessions that police fail to obtain from guilty criminals that might help innocent persons who would otherwise come under suspicion for committing a crime. In comparing these competing risks, a critical issue is the relative frequency of false confessions when compared to lost confessions, a frequency that thus far rests almost solely on the Leo and Ofshe collection. Part II then dives into the individual cases. It begins by narrowing the focus to the twenty-nine persons who were convicted of crimes and then examines nine of these twenty-nine cases in detail. Based on review of original trial court records and other similar sources, the part concludes that each of these nine persons were, in all likelihood, entirely guilty of the crimes charged against them. Part III explores the lessons that might be drawn from the high proportion of guilty criminals in the Leo and Ofshe collection of "innocent" persons. This fact suggests that academic research on miscarriages should not rely on media descriptions of the evidence against defendants. Journalists will all too often slant their reports in the direction of discovering [*526] "news" by finding that an innocent person has been wrongfully convicted. Reliance on second-hand media accounts can also obscure particular problem areas by over-generalizing the false confession problem. When the Leo and Ofshe anthology is whittled down to the handful of undisputed cases of wrongful conviction, the false confession problem is revealed to be not pandemic in the American criminal justice system, but rather concentrated among a narrow and vulnerable population: persons with mental disabilities. Part III concludes by critiquing Leo and Ofshe's proposal that judges should closely scrutinize the "fit" between the "post-admission narrative" of a suspect and the crime facts. Their overbroad suggestion would result in the suppression of many truthful confessions of criminals who may fail to give full accounts of their crimes for various reasons. Instead of suppressing confessions on reliability grounds, we should depend on juries to decide their truth or falsity, just as we ultimately rely on juries to determine the guilt or innocence of criminal suspects. I. The False Confession Problem in Context When the police obtain a false confession from an innocent person, that person is placed at risk of being wrongfully convicted. But this is not the only risk to the innocent posed by police interrogations. The innocent are also jeopardized when police fail to obtain a truthful confession from the true perpetrator of a crime. That truthful confession could prevent suspicion from wrongfully falling on an innocent person and could even exonerate an innocent person who has been wrongfully charged with, or convicted of, a crime. As I argued in my earlier article, Protecting the Innocent from False Confessions and Lost Confessions - And From Miranda, n6 weighing these competing possibilities requires some assessment of the relative frequency of these two risks. The available empirical evidence provides reason to Page 3 22 Harv. J.L. & Pub. Pol'y 523, *526 believe that today the innocent are more at risk from restraints on police that hinder their efforts to obtain truthful confessions than from the lack of additional protections against the comparatively rare risk of false confessions. n7 Moreover, there is good reason to believe that the [*527] Supreme Court's decision in Miranda has exacerbated the risks to the innocent. The Miranda decision has reduced the number of truthful confessions, while at the same time doing nothing about, and probably even worsening, the false confession problem by diverting the focus of courts away from the substantive truth of confessions to procedural issues about how they were obtained. n8 Professors Leo and Ofshe have written a reply to my analysis, challenging these points. n9 A detailed rebuttal of their critique is unnecessary here, but a few responses are in order because they demonstrate the importance of carefully analyzing the validity of the Leo-Ofshe collection of cases. Leo and Ofshe first argue that it is impossible to derive any estimate of the frequency of false confessions because of an obvious lack of precise records and related methodological difficulties. A "humble" answer of "I do not know," they write, is the researcher's only proper response to the question of how frequently false confessions occur. n10 The difficulty is that Leo and Ofshe fail to follow their own suggestion. Leo and Ofshe ultimately argue that it is "well established" that "psychologically-induced false confessions occur frequently enough to warrant the concern of criminal justice officials, legislators, and the general public." n11 In the popular press, they have made even more sweeping claims, such as false confessions happen "all the time." n12 However much Leo and Ofshe would like to disguise the fact, these are not "humble" claims of ignorance, but rather empirical claims about the frequency of false confessions indeed, a claim allegedly strong enough to justify restructuring police interrogations throughout America. n13 [*528] Leo and Ofshe are also surprisingly tight-lipped about what seems to be a straightforward way to gain some understanding of the frequency of false confessions. In my article, I suggested simply drawing a random sample of criminal cases and determining the percentage of false confessions within the sample. Leo and Ofshe eschew any such effort, arguing "the project of quantification is Cassell's, not ours or any other researchers studying ... false confessions." n14 This claim is untrue, as my interest in quantification through sampling is shared by others knowledgeable in the field. For example, Dr. Gisli Gudjonsson, whom Leo and Ofshe describe as one of the world's "leading authorities on false confessions," n15 has undertaken precisely this project in no less than three separate articles, each of which draws a sample with the goal, among others, of determining the frequency of false confessions. n16 While Leo and Ofshe quarrel with the applicability of this research from Iceland to America, n17 they fail to come to grips [*529] with the overarching point that the sampling methodology employed by Gudjonsson could be used to make some estimate of the frequency of false confessions here. For example, the 1994 sample of 173 filed cases from Salt Lake City drawn by Bret Hayman and me contains no evidence of even a single case involving an alleged false confession. n18 Similarly, it seems unlikely that Richard Leo's 1993 sample of 182 interrogations in the San Francisco Bay Area contains many - or perhaps even any - false confessions. n19 Leo and Ofshe do not explain why they are unwilling to reexamine this sample to see if it contains any false confessions. Interrogation researchers in this country have also drawn other samples, apparently without ever encountering any false confessions. n20 The dearth of false confessions in all these samples suggests that false confessions occur quite infrequently, with the result that any effort to determine frequency will necessarily involve a methodology for estimating low probability events. My previous article offered one such approach. To estimate the frequency of false confessions, one should canvass the available empirical evidence for estimates of (1) the number of criminal cases; (2) the error rate (that is, the wrongful conviction rate) in those cases; and (3) the proportion of wrongful convictions attributable to false confessions. Combining these three numbers will produce an estimate of the number of wrongful convictions from false confessions. n21 While I acknowledged that gathering evidence of the error rate is quite difficult, I relied on what appears to be the only plausible published estimate from Professor Ronald Huff and his colleagues. They surveyed criminal justice professionals around the country and asked [*530] them to estimate the error rate in the system. n22 Leo and Ofshe call the resulting estimate of the error rate (below one percent) "empirically orthless" n23 - apparently a new view of the quality of the study, as Leo had previously cited this very estimate to support one of his arguments. n24 If anything, the Huff error rate estimate is probably too high because of the well known human tendency to overestimate the probability of extremely low frequency events. It seems reasonable to rely on the Huff estimate to Page 4 22 Harv. J.L. & Pub. Pol'y 523, *530 generate an upper-bound estimate of the frequency of false confessions. n25 My article then derives a possible range of wrongful convictions from false confessions to provide an order-of-magnitude assessment of the problem, an assessment suggesting that such wrongful convictions are quite rare. n26 Although Leo and Ofshe refuse to offer even a ballpark assessment of the annual number of wrongful convictions from false confessions, they remain confident that the false confession problem dwarfs the lost confession problem. Although my article explained why Miranda might harm the innocent by blocking truthful confessions, n27 they claim that no such problem occurs, avoiding any need to weigh the competing risks. To reach this firm conclusion, they create a caricature of both my argument and the nation's criminal justice system. Miranda could harm innocent suspects, they claim, in only two "scenarios": the "frustrated detective scenario," in which a suspect invokes his Miranda rights and the frustrated detective nonetheless goes on to obtain a false confession from an innocent person, and "the ever-diligent [*531] etective scenario," in which a diligent detective continues to investigate a case after it has been solved and is unable to obtain a confession from the true perpetrator of a crime because of Miranda. These simplistic scenarios, which Leo and Ofshe misleadingly attribute to me, n28 fail to capture the more realistic risks Miranda poses to the innocent. The most fundamental flaw in the scenarios is their simplifying assumption that only one detective in one jurisdiction is investigating one crime. In the real world, many investigators from many different jurisdictions investigate many different cases. In some of these investigations, they will obtain confessions that exonerate innocent persons. Indeed, Leo and Ofshe unwittingly recognize the absolving power of confessions when they acknowledge that "reliable confessions from the true perpetrators are among the leading sources of exoneration of the wrongfully convicted ...." n29 If Miranda impedes police success in interrogation, it is a logical corollary that it will also impede one of the "leading sources" of exoneration. The only remaining question, then, is whether Miranda impedes police interrogation. Evidence of Miranda's harmful effects is mounting. For example, along with various co-authors, I have developed empirical evidence of Miranda's substantial harm to law enforcement. n30 In my most recent articles, I have analyzed the precipitous drop in crime clearance rates that followed immediately on the heels of Miranda and concluded that Miranda severely hampered police effectiveness. n31 Interestingly, [*532] even before publication of this analysis, Leo himself concluded that "Miranda appears to have an effect on the collateral functions of interrogation" such as "clearing crimes." n32 Miranda's adverse effect on clearance rates is of central importance to the innocent, because a reduced clearance rate means that in some cases detectives fail to obtain confessions from the true perpetrators of criminal acts that would allow them to "clear" or solve the case. These lost clearances occur most often when police arrest a suspect for one crime, but are prevented by Miranda from obtaining confessions to other crimes he has committed. n33 For example, if police apprehend an armed robber at the scene of the crime, he may invoke his Miranda rights and prevent police from learning that he has committed five other similar robberies. If an innocent person has been charged, or even convicted, for one of these other robberies, Miranda may well prevent his exoneration. The possibility that Miranda harms the innocent by blocking confessions from criminals finds support in real world observations. Professor Sam Gross's detailed empirical study of wrongful convictions from eyewitness misidentifications explained that before Miranda, the typical way in which a miscarriage was discovered was that "the actual criminal was arrested on an unrelated charged and, after being held in custody for a day or two, she confessed to the perpetration of all the crimes charged to the misidentified suspect." n34 Since that time, Gross concludes, such exonerations through true confessions appear to have declined significantly, with Miranda being a possible cause. n35 [*533] In light of the competing risks to the innocent from false confessions and lost confessions, the important public policy question becomes the relative frequency of false confessions as a cause of wrongful convictions versus true confessions as a cure for wrongful convictions. My previous article not only argued that false confessions rarely cause wrongful convictions, but also cited two studies suggesting that true confessions are among the most common means of exonerating those wrongfully convicted. For example, Gross reported that fifty-four percent of the wrongful convictions from eyewitness misidentification were uncovered when the actual criminal confessed. n36 Similarly, Arye Rattner's more general collection of miscarriages found that the actual culprit's confession was the leading means of exoneration, responsible for forty percent of the exonerations. n37 Leo and Ofshe do not dispute these figures, but argue that I "fail to mention" a more recent study in which, they dramatically proclaim, "0% of the wrongfully Page 5 22 Harv. J.L. & Pub. Pol'y 523, *533 convicted and incarcerated innocent were exonerated by a confession." n38 This contrived datum is meaningless. The study in question, subtitled "Case Studies in the Use of DNA Evidence to Establish Innocence After Trial," n39 was specifically limited to the subject of its subtitle - exonerations through DNA evidence. Thus, by definition, one hundred percent of the exonerations in the study came from DNA, meaning that zero percent came about through any other method. n40 The only meaningful data that bear on the overall proportions come from the two studies I cited, suggesting the prime importance of confessions in exonerating the wrongfully convicted. Because the innocent are today more at risk from lost confessions than false confessions, I proposed replacing Miranda with a system of videotaping interrogations. n41 This [*534] would improve the lot of the innocent, because Miranda reduces the numbers of truthful confessions, reducing the potential for confessions to clear wrongfully convicted innocent persons. At the same time, Miranda does nothing about the false confession problem. It appears to be common ground among those who have studied the issue, including Leo and Ofshe, that Miranda fails to protect against false confessions, since those who are innocent will want to talk to police and waive their Miranda rights. n42 My argument is relatively straightforward - not, as Leo and Ofshe intimate, a "newfound discovery" that has gone "surprisingly unnoticed," n43 but rather a simple development of similar positions long advanced by many other knowledgeable observers. n44 Indeed, my conclusion appears to be reinforced by an essay recently written by Richard Leo, in which he argued that "it is even possible that Miranda - despite its high-minded intentions - has undermined any protection the law might have otherwise offered against the admission of false confessions into evidence." n45 Replacing Miranda with videotaping offers a real chance to identify those rare cases of police interrogation gone bad, while at the same time not impeding police in their efforts to obtain confessions. Instead of contending on the merits of these issues, Leo and Ofshe single-mindedly maintain that their collection of wrongful convictions demonstrates such a serious problem of wrongful convictions from false confessions that we must [*535] restructure the criminal justice system to erect new safeguards against this particular danger. n46 Because of the near-exclusive emphasis Leo and Ofshe place on these particular cases, they should be closely scrutinized. If this examination reveals that the alleged cases of "false" confessions of innocent persons are actually truthful confessions from guilty criminals, Leo and Ofshe's policy recommendations could be seriously flawed. With this premise in mind, then, we can turn to reviewing the Leo-Ofshe collection. II. The Leo and Ofshe Collection of False Confessions Professors Leo and Ofshe are on the right track in attempting to collect empirical evidence on false confessions. Many fundamental issues in the criminal justice system are grossly in need of factual illumination, with police interrogation and confessions being high on the list. Moreover, an evaluation of false confessions has never before been undertaken, and any criticisms of the Leo-Ofshe project should give due regard to its difficulty and its importance. It is hard to determine if police interrogation is often malfunctioning, but if it is, that fact would have important policy consequences. Nonetheless, counting the cases of "false" confessions raises sensitive methodological and evaluative questions that must be handled far more carefully than done by Leo and Ofshe. A. Defining Who Is "Innocent" Discussion about risks to the innocent must first grapple with the question of who qualifies as an "innocent" person. Previous research on miscarriages of justice has generally focused on "wrong-person mistakes - the conviction ... of the factually 'innocent'." n47 Moving beyond the factually innocent to [*536] the legally innocent would raise a host of questions not readily susceptible to empirical analysis: what kinds of state of mind defenses (including insanity and entrapment) were erroneously rejected at trial, when did the quantum of proof dip below the "beyond a reasonable doubt" standard, and so forth. Leo and Ofshe thus wisely avoid this quagmire by focusing on the more discrete and researchable category of wrong person mistakes. Leo and Ofshe claim to have discovered sixty cases in which an innocent person falsely confessed to a crime he did not commit. This article focuses not on this entire collection, but rather the subset of twenty-nine claimed cases of a Page 6 22 Harv. J.L. & Pub. Pol'y 523, *536 false confession leading to a wrongful conviction. For policy purposes, false confessions leading to erroneous convictions are the major point of concern. If a person who has made a false confession is not convicted - because the police do not arrest, the prosecutor does not indict, or the jury does not convict - then the screens in the system have at least worked to prevent the ultimate miscarriage of justice, the conviction of an innocent person. To be sure, false confessions may result in considerable trauma before a determination of innocence, and, in an ideal system, no such false confessions would be obtained. I mean in no way to minimize such concerns, but rather to narrow the focus to the evidence justifying their claim that additional safeguards are needed in criminal adjudication. n48 Again, a venerable tradition supports this approach. n49 Leo and Ofshe also seem to adopt this view implicitly, as the main effect of their policy proposals is not to reduce false confessions per se, but rather to prevent wrongful convictions later in the process through such measures as after-the-fact judicial scrutiny of the credibility of confessions. n50 [*537] B. Determining Who Is In Fact "Innocent" With the subject clearly defined - factually innocent persons who have been wrongfully convicted - the question next arises of how to determine who is "innocent." One could easily take the position that "objective" truth is unknowable and therefore such determinations lie beyond human capacity. Professors Leo and Ofshe refreshingly contend that we can determine whether defendants are truly guilty or innocent. Given that we have a judicial system specifically designed to make such determinations, a problem then arises. One could argue, as Ofshe has elsewhere, that "if a decision is ever to be made about questions of guilt or innocence, it should be made by a jury not by a contributor for or readers of scientific journals." n51 The only way to avoid this problem of the researcher as judge and jury is to confine analysis to cases of undisputed wrongful convictions, an approach some researchers have adopted. n52 Here, however, Leo and Ofshe opt to follow what they describe as "the leading contemporary research" on miscarriages, a study by Professors Bedau and Radelet on allegedly innocent persons convicted of capital crimes. n53 This is a cause for concern, because Bedau and Radelet's catalogue of "innocents" ignores physical evidence of guilt, n54 incorrectly cites sources that in fact indicated defendants were guilty, n55 includes works of fiction as proving innocence, n56 and contains other serious flaws. n57 More recent work by the same authors is even worse. n58 [*538] Leo and Ofshe promise to avoid such concerns by limiting their collection of "innocents" to cases in which "no physical or other significant and credible evidence indicated the suspect's guilt; the state's evidence consisted of little or nothing more than the statement 'I did it;' and the suspect's factual innocence was supported ...." n59 Their methodology improves on Bedau and Radelet's by providing a continuum of the evidence of innocence - from cases of "proven" to "highly probable" to "probable" innocence. But, like the Bedau and Radelet survey, the problem remains that Leo and Ofshe's judgment as to who is innocent is highly subjective and, in more than a few cases, demonstrably wrong. Some concrete examples will illustrate this point. What follows are discussions of nine of the twenty-nine cases, arranged in alphabetical order, in which Leo and Ofshe claim that an innocent person was wrongfully convicted as the result of a "false" confession. n60 In these discussions, I review Leo and Ofshe's claim that no "credible evidence" supported the defendant's guilt. n61 Based on a more thorough description of the cases than Leo and Ofshe provide, the reader can readily see that this claim is untrue and that substantial evidence supported the guilt of each of these defendants. 1. Barry Fairchild Leo and Ofshe have one case in their catalog in which they allege that an innocent man was executed: Barry Fairchild. Fairchild confessed to, and was convicted of, participating in the murder and rape of Marjorie Mason on February 26, 1983. Relying primarily on secondary sources, Leo and Ofshe claim the confession was coerced and that "no independent evidence connected Fairchild to the crime." n62 The judicial opinions in the case give a decidedly different [*539] impression. On the day of the murder, police chased two suspects who were driving the victim's car. The suspects escaped on foot. Police later discovered the Page 7 22 Harv. J.L. & Pub. Pol'y 523, *539 victim's body and, among other things, a baseball cap bearing the inscription "CAT Diesel Power." One witness reported seeing Fairchild wearing such a hat about a week before the murder, and two officers also recognized the hat as having been worn by Fairchild on prior occasions. n63 Police also received a tip that Fairchild and his brother had raped several women in the past. The women reportedly feared coming forward because they felt Fairchild was dangerous. n64 Police received a further tip from a different informant about Fairchild's involvement which contained information that the police had independently confirmed. For example, the informant told police that Fairchild and his brother had escaped the police on foot after the victim's car was stopped. The account was consistent with events occurring during the police chase. n65 Several days after the murder, the police received a report that Fairchild was trying to escape to California by bus. They managed to stop the bus, but Fairchild evaded them. n66 When eventually caught after hiding in the woods for several days, "Fairchild's attitude was: 'You got me!' He was willing to talk." n67 The police took Fairchild back to Little Rock where he gave a videotaped confession. Although Leo and Ofshe claim that the videotape shows Fairchild "looking away from the camera and responding to the prompting of others in the room," n68 this was not the view of the federal district court judge who carefully evaluated this claim in response to Fairchild's habeas petition. Judge G. Thomas Eisele, described as a "moderate and fair-minded" judge even by those who opposed Fairchild's execution, n69 [*540] concluded: I watched Mr. Fairchild making his statements and ... his statements give the feeling of truth to me because particularly when he is using his hands to describe ... 'And we went up this and down this hill,' he's making an uphill or downhill [motion] with his hand just automatically as he talks. All of the incidental body language is corroborative, it seems to me, of what is being said. And what is being said ... did not give me the impression that it had been rehearsed ... Rather, it seems to have the indicia of spontaneity and truth. n70 Judge Eisele concluded that "the Court specifically finds that [Fairchild] was not instructed or coached regarding the content of his confessions." n71 Two years later, on yet another habeas petition, Judge Eisele again rejected such claims: This is a death penalty case. It deserves the most careful and serious consideration possible. Back at the time of the hearing in 1987, the Court carefully observed and listened to the videotaped confessions - not once but several times. In light of the issues now being raised ... the Court has viewed these videotapes yet again and has also reviewed the [earlier] transcript ... It finds no reason to depart from the factual finding that I made then. On the contrary, that review has reinforced the Court's confidence in those findings. The Court is convinced that no reasonable person could listen to the evidence presented at the two-day hearing and view the videotaped confessions and still have any doubt about the involvement of Mr. Fairchild in the rape and murder of Ms. Mason. n72 Fairchild's confession "went into explicit detail concerning the abduction, rape, and robbery of the victim." n73 Details in the confession were corroborated. Fairchild, for example, said his accomplice had a "little old nickel plate" gun that was a .22 or a .25. The bullets removed from the victim were .22 caliber. n74 Leo and Ofshe note that the prosecution acknowledged the [*541] confession "was, in part, not true," n75 apparently referring to the prosecution's stipulation that Fairchild falsely confessed that his accomplice was Harold Green. n76 But it could well have been that Fairchild committed the crime with his brother. n77 If so, or on the reasonable assumption that the accomplice was a friend of Fairchild's, what Leo and Ofshe call a "glaring error of fact" n78 actually is a perfectly understandable deception that, if anything, makes the confession more believable. After confessing, Fairchild guided the officers on a crime scene tour and "gave the directions which brought them Page 8 22 Harv. J.L. & Pub. Pol'y 523, *541 all to the scene where Ms. Mason's body had earlier been found. He pointed out where they left the body." n79 During the tour, the officers asked Fairchild about missing jewelry from the victim. Fairchild asked if a person who had the jewelry would get in trouble. After being assured that she would not if she had no knowledge of the crime, "Fairchild then told the officers that his sister had the [victim's] watch. The officers then drove to Mr. Fairchild's house where they obtained the watch" from Fairchild's sister. n80 At trial, the victim's father testified that the watch, a special mail order Cassio diver's watch tested waterproof to one hundred meters, was "just like" the watch he gave to his daughter for her twenty-second birthday. The victim's mother said the watch had a "unique" kind of strap on it and was "Greta's watch." n81 Reviewing this and other evidence, Judge Eisele concluded that the evidence that the watch was the victim's was "very strong, indeed, overwhelming." n82 Leo and Ofshe concede Fairchild had a watch "similar" to that owned by the victim, but note Fairchild testified that he bought the watch from someone at a pool hall and then sold it to his sister. n83 Fairchild's trial story seems incredible on its face. Fairchild claimed that this person, whom he knew only as "Ham," came up and offered to sell the watch. [*542] The "pool hall" was "just kind of a place." n84 He did not know what street it was on. n85 Moreover, as the district court recognized, it was not possible for the police officers working the case to know when Fairchild confessed that Fairchild's sister would have such a watch, which in turn would be identical to the victim's watch. The officers could only have learned these facts "out of the mouth of Mr. Fairchild" when he confessed. n86 The watch "by itself," the judge concluded, "makes clear the involvement of Mr. Fairchild in the crimes against Ms. Mason far beyond any reasonable doubt." n87 Leo and Ofshe also claim Fairchild was vulnerable to coercion because he was mentally retarded, citing an IQ test conducted by Ruth Luckasson producing a score in the low 60s. n88 The district court concluded, however, that "at some point, Ms. Luckasson began to lose her scientific objectivity and skepticism." n89 The Court recounted problems with her findings, concluding "her critical opinions in this case [are] flawed and unsubstantiated by the evidence." n90 The court thoroughly reviewed all of the evidence concerning Fairchild's mental abilities (in an opinion that, on this issue alone, spans more than sixty pages in the Federal Supplement), reaching the conclusion that Fairchild was "clearly not 'mentally retarded.'" n91 This finding was affirmed by the Eighth Circuit. n92 The Arkansas Supreme Court reached the same conclusion. n93 Leo and Ofshe discount all of the evidence of guilt based largely on a segment aired on ABC's 20/20. In the program, various men claimed to have been pressured to confess to the murder, n94 and Fairchild recited claims of abuse as well. The program conceded that the bulk of the allegations had been [*543] rejected by Judge Eisele n95 but failed to mention that Judge Eisele's rejection came only after a seventeen-day evidentiary hearing, after which he made 133 pages of oral findings from the bench, and entered a 413-page written order on the remaining factual and legal issues! n96 Judge Eisele concluded that the new claims "did not change [his] prior finding that Fairchild's confessions were voluntary." n97 The judge observed "there was no direct evidence presented at the hearing that Fairchild had been forced to confess." n98 Any abuse was improbable because it would have taken place in "an area where many officers were present. And they were constantly coming and going. And the officers present were not under only one command. They were from a variety of police jurisdictions." n99 The judge also found that supporting testimony tracking Fairchild's new story "was not credible and was manufactured to conform with Fairchild's claims." n100 The Court of Appeals, after "careful review of the record," upheld the district court's finding that the confession was voluntary and added "the evidence does not support any other conclusion." n101 Leo and Ofshe give no reason to credit the allegations aired in the 20/20 program over the contrary - and conscientiously determined - findings of the judicial system. n102 2. Joseph Giarratano On February 4, 1979, Barbara Ann Kline and her fifteen year old daughter Michelle were murdered in their apartment in Norfolk, Virginia. Michelle was raped and killed some time before her mother. Police immediately suspected Giarratano, [*544] who was living in the same apartment but had disappeared on the night of the murders. n103 The next day, hundreds of miles away in Jacksonville, Florida, Giarratano approached a uniformed police officer eating breakfast. Giarratano told the officer that he had "killed two people in Norfolk, Virginia, and wanted to turn Page 9 22 Harv. J.L. & Pub. Pol'y 523, *544 himself in." n104 Giarratano explained that he had killed a "lady in Norfolk" in an argument over some money and then raped and strangled the lady's daughter. n105 Giarratano repeated the same account to another patrol officer, except he made no statement about a sexual assault. n106 The following day, Giarratano was interviewed by two detectives from Norfolk who had examined the crime scene. After they told Giarratano that "they had been there," n107 Giarratano admitted his earlier confession "was not the way it was." n108 He proceeded to give a detailed confession to the killings, explaining he had raped the daughter and then strangled her. He left the apartment and later returned. When the girl's mother arrived and unlocked the door to the apartment, Giarratano jumped out. She started screaming, so he stabbed her two or three times. n109 (An autopsy revealed that the victim had died from three knife wounds.) n110 He left two dogs behind in the apartment n111 and headed for Florida. n112 Giarratano said he had stabbed the mother because [*545] she "would know I was the one that killed Michelle and I wanted to keep her from talking." n113 Within the next few weeks, Giarratano was interviewed by a staff psychiatrist at a state hospital concerning the effects of drugs - cocaine, Dilaudid, and possibly alcohol - that Giarratano had consumed at the time of the crimes. The psychiatrist testified that the drugs would "loosen his controls." n114 As for reversal of the time sequences and differing accounts of the crime, the psychiatrist testified that this was attributable to the combination of drugs resulting in "peripheral neuropathy, loss of recent memory" producing "confabulation." n115 The defense psychiatrist reached the same conclusion. In a lengthy interview, Giarratano described the crimes in detail, giving an account that was "substantially consistent with the one which he gave the Norfolk police." n116 Giarratano said he killed the daughter when she infuriated him by resisting his attempt to have intercourse and then killed the mother. n117 The psychiatrist concluded: "Mr. Giarratano was very credible in his description [of the crime] during the Clinic interview." n118 Based on interviews with Giarratano and his family members, the defense psychiatrist concluded Giarratano's "'previously suppressed rage and anger' had been 'reactivated by the sequence of events (with) Michelle' and that the murders were 'symbolic acts' by which the defendant's hatred was discharged against persons he identified in his mind with his mother and sister." n119 It is worth noting that Giarratano had a history of active involvement in drugs and violent outbursts. n120 [*546] To cast doubt on these repeated confessions, Leo and Ofshe rely on inaccurate descriptions of the crime apparently generated by avowed death penalty opponents seeking to overturn Giarratano's capital sentence. For example, while Leo and Ofshe claim that hair samples did not link Giarratano to the crime, n121 in fact one of the pubic hairs "found on Michelle's left hand, stomach and pubic area was consistent in 'race, color and microscopic characteristics' with one of [Giarratano's] pubic hairs." n122 Leo and Ofshe report that one fingerprint found at the scene of the crime matched Giarratano's, n123 but in fact seventeen matching prints were found. n124 Leo and Ofshe also contend police found no blood on Giarratano's clothing, but in fact human blood type O - the same as one of the victim's - was found on the front and side of one of his boots. n125 Leo and Ofshe further claim that the murder was committed by a right-handed person and Giarratano was left-handed with only limited use of his right hand due to childhood neurological damage. n126 Both points are questionable. A belatedly hired defense analyst did opine, after viewing crime scene photographs, that "the sharp edge of the knife ... was upward. This is typical of a right-handed person assaulting the victim from behind." n127 As is obvious, this is a far cry from proving [*547] that a right-handed person committed the crime; the sharp edge of the knife would be "upward" on any number of different attack scenarios, and any additional analysis would be largely guesswork without further information about the precise position of the murderer at the time of the crime. And although Giarratano has told his apologists that the neurological damage in his right hand was long standing, his own medical materials suggest that "right upper extremity sensory deficit" was attributable to a wrist laceration associated with his 1983 suicide attempt - some four years after the murders - during which Giarratano was hospitalized for "having severely slit his right wrist." n128 At the time of the murders, it should be noted, Giarratano was sufficiently dexterous to work on a scallop boat. n129 The feeling one gets reviewing the inaccurate second-hand accounts is that opponents of the death penalty have distorted the record on Giarratano's guilt for their own purposes. Accordingly, one should be exceedingly cautious in relying on their secondary accounts of the "evidence" in the case. n130 Page 10 22 Harv. J.L. & Pub. Pol'y 523, *547 3. Paul Ingram Paul Ingram was arrested on November 28, 1988, following complaints of sexual abuse from his daughters, then eighteen and twenty-two years old. n131 Ingram waived his right to counsel and made incriminating statements that day. Additional incriminating statements followed during an investigation over the next few months. On May 1, 1989, Ingram pled guilty to three counts of third degree rape with each of his daughters. n132 In October of 1989, Ingram retained new counsel and moved to withdraw his guilty plea. He claimed his original plea resulted from "a variety of influences [*548] including deception, brainwashing, religious and familial coercion" at the hands of friends, counselors, and others. n133 At a hearing on the motion to withdraw the plea, Ingram denied participating in any abusive conduct and claimed that the sexual acts he admitted were not the products of conscious memory, but rather were "visualizations." n134 Ingram further testified that, near the end of July, 1989, he suddenly realized that he did not commit these crimes. n135 The supporting expert for the defense was Richard Ofshe, who testified that the incriminating statements resulted from accidental hypnosis and fantasies during questioning. n136 Ofshe, however, admitted specifically that "certain factual questions are beyond the scope of my professional expertise. I do not know nor do I have any opinion as to whether or not Paul Ingram committed one or more sexual assaults on his daughters." n137 Three psychologists testified for the state and "generally agreed that Ingram's statements were real recollections and not the products of any alleged trances or hypnosis." n138 One, a family counselor hired by the defense, submitted a report "that reflected statements by Ingram acknowledging long-term abuse of his children and involvement in 'incest, sodomy, and homosexual activity.'" n139 After six days of testimony, the trial judge noted the obvious - that he was a "neutral person in this controversy." n140 Accordingly, the judge's findings, after hearing directly from Ofshe, provide an opportunity to test Leo and Ofshe's claim that their views would be accepted "by an overwhelming majority of neutral observers." n141 The judge, however, disagreed with Ofshe and found the confession to be true. [*549] After listening to the tape of Ingram answering police questions on the afternoon of his arrest, the judge observed that "not terribly long" into it Ingram "essentially confesses to molesting both of these young women." n142 Given that Ingram was a law enforcement officer, the judge also found it was "highly, highly unlikely that he would be convinced to confess unless he were guilty." n143 Moreover, the admissions that day - made before any contact with psychologists or others who allegedly coerced him - contained "'virtually incontestable evidence of guilt.'" n144 During questioning, "Ingram told officers, among other things, of sex practices used to prevent pregnancy with one of the daughters, of that daughter's abortion ... when she did become pregnant by him, and of having anal intercourse with the daughter during her menstrual period so that 'the bed wouldn't get messed up.'" n145 The abortion also provided supporting physical evidence of sexual abuse, as Ingram did "not suggest that this abortion did not in fact take place or was not verifiable." n146 The trial judge further found that the three state witnesses were "more credible than Dr. [sic] Ofshe" n147 and that Ofshe was "considerably less qualified" than the other witnesses "to give opinions in this area." n148 Ofshe's testimony was not credited, in part, because he was "not a clinical psychologist" and "not an expert in sex abuse." n149 In addition, the judge found that Ofshe conducted an "odd" experiment in an effort to show Ingram was spouting back information fed to him by the police. n150 While Ingram confessed to a scenario fed to him by Ofshe, Ofshe had chosen facts that "came pretty close to what one of the victims had accused the defendant of" rather than [*550] something "totally foreign from anything that could probably be true." n151 With respect to Ofshe's claim (based on "reading a dry record") that Ingram was in an "hypnotic state, or in a trance" when he confessed, the trial judge stated: "I find that to be strange. I wonder if that can be done [i.e., determining that someone is in a hypnotic state from merely reading a transcript]. I have great cause for concern with that." n152 The judge also found it strange that two weeks after Ingram pled guilty, Ofshe contacted him. Ofshe told Ingram, "You're innocent," to which Ingram responded, "No, I'm guilty." n153 A few weeks later, Ingram switched to agree with Ofshe. The judge also concluded that Ingram's new claims were incredible. The judge noted that Ingram's position was that "his mental state and his memory were good" up to his arrest, then until mid-July of 1989 his memory was poor, and Page 11 22 Harv. J.L. & Pub. Pol'y 523, *550 then after [*551] July of 1989 "his memory was good again." n154 "I just don't believe that," the judge concluded, emphasizing "I just have to say that straight out ... I believe his testimony is impeached, and I believe that he is somewhat of a manipulator." n155 The judge further observed that the two daughters had each accused their father of abuse and that "there is no real reason that's been given to me here in this courtroom, why they have or would falsely accuse their father." n156 Indeed, the daughters' allegations developed in a "rather logical" fashion. n157 To be sure, the court noted, there were "inconsistencies in their statements" and "exaggerations" about fantastic satanic sessions that in all likelihood never occurred. n158 Such features, however, are "not an unusual thing" when a person has "been subjected to severe sexual abuse over a period of years." n159 In any event, Ingram was not charged with any form of satanism but rape of his daughters. n160 "I just find that he did it," the trial judge concluded. n161 These findings were upheld by the Washington Court of Appeals, n162 and habeas corpus challenges to them were rejected by the United States District Court for the Western District of Washington n163 and the Ninth Circuit. n164 Leo and Ofshe do not appear to claim any new evidence has surfaced since then, relying primarily on an old article by Ofshe himself. This is a curious supporting citation. Ofshe wrote there that "it is impossible for anyone other than those directly [*552] involved to know whether or not Paul Ingram" sexually abused his daughters and that the point of the present paper is not to argue Mr. Ingram's innocence. No attempt has been made to develop the analysis and marshall [sic] evidence [to that effect] ... If a decision is ever to be made about questions of guilt or innocence, it should be made by a jury not by a contributor for or readers of scientific journals. n165 What appears to be the only new development since then tends to confirm Ingram's guilt; at a 1996 hearing before Washington's Clemency and Pardons Board, Paul Ingram's son made the first public disclosures about his own childhood sexual abuse by his father. n166 The Board also heard from [*553] Ofshe, n167 but denied clemency. n168 4. Richard Lapointe n169 In 1992, Richard Lapointe was convicted of sexually assaulting and murdering his wife's eighty-eight year old grandmother, Bernice Martin, and then setting her apartment on fire. Leo and Ofshe claim, citing a tract prepared by a group called "The Friends of Richard Lapointe," n170 that "no physical evidence ... linked Lapointe to the crime." n171 In fact, as the Connecticut Supreme Court noted in affirming the conviction, "a stain on the victim's bedspread was human semen from a person who was a secretor with Type A blood. [Lapointe] has Type A blood and is a secretor. Also the stain lacked sperm, which is consistent with the semen of a person who has had a vasectomy. [Lapointe] had a vasectomy" before the murder. n172 Other evidence also pointed to Lapointe. When a relative called Lapointe's wife to express concern about the victim on [*554] the night of the murder, Lapointe picked up another phone, without being asked to join the conversation, and volunteered to check on the victim himself. He then took a less-than-direct route to her apartment, n173 where he smelled smoke and felt heat from the door. Although unable to gain access, he telephoned the relative from a neighboring apartment to report everything was fine. When the relative said she was going to check herself, Lapointe promptly returned to the victim's apartment and "discovered" the fire. n174 After the murder, "before any information regarding a possible sexual assault became known to the police or the public, Lapointe stated [to a friend] that 'it was [a] shame they killed an old lady, but they didn't have to rape her, too.'" n175 When later asked how he learned that the victim had been sexually assaulted, Lapointe said "he had been informed by a doctor at the hospital on the night of the murder." n176 The medical personnel involved, however, unanimously testified to the contrary. n177 During a police interview before suspicion centered on him, Lapointe also "exhibited considerable curiosity concerning the results of the autopsy and asked if there had been causes of death other than smoke inhalation." n178 Ofshe and Leo claim it was "virtually impossible" for Lapointe to commit the murder because his wife provided an Page 12 22 Harv. J.L. & Pub. Pol'y 523, *554 alibi for all but thirty to forty-five minutes of the day. But the police reinterviewed Lapointe's wife on the day he confessed. She conceded that Lapointe left their house around the time of the murder, contrary to the story both she and Lapointe previously gave police. n179 On the day he confessed, Lapointe voluntarily came to the police station and waived his Miranda rights. Tests performed later revealed that Lapointe had "a full-scale intelligence quotient (IQ) of 92," within the average range. n180 Within an [*555] hour of questioning, Lapointe "became quiet, slumped down in his chair, sighed and stated, 'I killed her.'" n181 Lapointe then denied making that statement and asked the detective whether it was possible to kill someone and not remember. When the detective said it was possible, Lapointe confessed orally and, in a one-sentence signed statement, admitted killing the victim and then blanking out. n182 The detective tried to elicit more details, and Lapointe gave another brief confession. n183 Next another detective interviewed Lapointe and recounted the reinterview of Lapointe's wife that had destroyed his alibi. Lapointe proceeded to give a more detailed account of his murder. He said, among other things, that he removed the victim's underwear and threw it to the right of the bed, where it was in fact found. n184 He confessed to raping the victim and then ejaculating on the bedspread. The semen was found on the bedspread n185 and no sperm was found on the victim's body. n186 The detective asked whether the victim screamed and Lapointe said she had not. To test Lapointe, the detective falsely said that a neighbor had heard screams. Lapointe adamantly denied the victim had screamed. n187 Lapointe said that he obtained a steak knife with a brown plastic handle and stabbed and strangled the victim. A melted brown plastic knife handle and blade were found near the victim. n188 Relying on second-hand newspaper accounts, Leo and Ofshe claim Lapointe's confession should be discounted because of alleged inconsistencies. For example, Leo and Ofshe claim the confession to penile rape was inconsistent with the evidence because "in fact, the victim was raped with a blunt instrument." n189 The medical examiner, however, did not attribute the victim's injuries to a foreign object, but testified that the victim suffered "blunt trauma." n190 In neither his [*556] report n191 nor his trial testimony n192 did he attribute these injuries to a foreign object. n193 Leo and Ofshe also contend that Lapointe's confession to "killing" the victim on the couch was inconsistent with "medical testimony establishing that she was not killed while on the couch." n194 There is no discernible inconsistency between medical testimony and the confession on this issue. Firefighters found the victim on the floor close to the couch, n195 and the medical examiner testified that she was alive during the fire because the stabs wounds were not the cause of death. n196 There is thus no conflict with Lapointe's confession that he "stabbed Bernice in the stomach while she was laying on the couch," n197 then set the fire and left. In any event, all of the evidence concerning Lapointe's confession and its alleged inconsistencies was presented to the jury. Leo and Ofshe offer no new credible evidence to discredit the verdict. n198 It is worth noting that the continued unfounded suggestions that Lapointe is innocent remain a source of considerable distress for the victim's family. n199 [*557] 5. Jessie Misskelley In 1994, Jessie Misskelley was found guilty of participating in the murder of three eight-year-old boys - Steven Branch, Christopher Byers, and Michael Moore. Police interviewed Misskelley about a month after the boys were found dead in a creek - tied up, beaten, and mutilated. Although Leo and Ofshe assert the confession was "inconsistent with the facts of the case," n200 it was in fact proven beyond a reasonable doubt to be consistent in its most important respect - the identity of the main killers. About a month after the gruesome murders, police asked Misskelley to come to the station house in order to answer some questions. Misskelley admitted that he watched as two of his friends - Damien Echols and Jason Baldwin - beat and abused the young boys, and that he (Misskelley) kept one of the boys (Michael Moore) from escaping. n201 Based on this incriminating statement, police arrested Echols and Baldwin as principals in the murders and Misskelley as their accomplice. The accuracy of Misskelley's identification of Baldwin and Echols as the killers was established by guilty verdicts at a separate trial resting entirely on independent evidence, n202 because Misskelley's confession could not be Page 13 22 Harv. J.L. & Pub. Pol'y 523, *557 used as evidence at that trial. n203 That independent evidence included clothing fibers found on the victims' clothes that were microscopically indistinguishable from items found in the Baldwin and Echols residences, n204 and from various witnesses who heard Echols and Baldwin admit committing the crimes. n205 The capital sentencing phase also included grisly testimony from a defense psychiatric expert who said Echols believed, [*558] among other things, drinking blood of others gave him special powers. n206 At Misskelley's trial, the prosecution acknowledged some discrepancies between Misskelley's confession and the crime scene, n207 but identified a number of consistencies knowable only by someone with first-hand knowledge of the crime. These details included not only correctly identifying Echols and Baldwin as the killers, but also the following: . Misskelley confessed that, when he left the scene, the Byers boy was already dead on the ground; there was evidence that the creek contributed to the deaths of the other two boys, but not the Byers boy. n208 . Misskelley confessed that the Byers boy was cut on his penis; the Byers boy was the only boy found with severe genital mutilations. n209 . Misskelley confessed that one of the boys was cut on the face; one of the boys had facial lacerations. n210 Leo and Ofshe also report that "numerous" witnesses placed Misskelley at a wrestling competition forty miles from the crime scene. n211 This wrestling alibi would, in the colorful phrase of reporters covering the trial, be "pinned for the count." n212 Two of the alibi witnesses had previously given statements to the police that they were unaware of Misskelley's whereabouts on the night of the murder. n213 According to the reporters, the prior statements "destroyed any credible [*559] Misskelley alibi." n214 In any event, at trial, the defense fully explored all of the issues. Although Misskelley did not take the stand, his attorneys presented extended testimony from an expert, none other than Richard Ofshe, about the alleged false confession. n215 The jury nevertheless convicted Misskelley, and nothing in Leo and Ofshe's account suggests that any significant relevant evidence was kept from them. Developments since Misskelley's trial, not discussed by Leo and Ofshe, strongly confirm the verdict's accuracy. After his conviction, Misskelley confessed two more times. On the drive to the state prison, after he was assured that nothing he said could be used against him, Misskelley gave a detailed statement about the crime. n216 Prosecutors then arranged for a judicially-approved interview of Misskelley, over the strenuous objections of his defense attorneys. n217 In the forty-minute recorded interview, Misskelley provided details about the crime. Misskelley said that, when he got off work at dinnertime, n218 he went to a wooded area with his friends Echols and Baldwin. When three boys entered the woods, Echols jumped them. n219 When they resisted, he (Misskelley) and [*560] Baldwin had entered the fray. Misskelley grabbed the one with the blue boy scout uniform (Michael Moore) to prevent his escape. n220 To keep him under control, Misskelley hit him repeatedly on the head. n221 Misskelley then watched as one of the other boys was cut on the penis, recounting that "I seen blood fly." n222 Misskelley also explained how the boys were tied up with shoelaces from their shoes. n223 When first questioned by police, he had said that the boys were tied up with rope to throw the police "off track." n224 Misskelley left before Echols and Baldwin, carrying with him a bottle of whiskey, which he busted under a highway overpass close to the woods. n225 Prosecutors and defense counsel went to the overpass and found a broken bottle at the indicated location. The broken bottle neck matched a bottle of Evan Williams Kentucky Bourbon, the kind Misskelley said he drank the day of the murders. n226 6. Bradley Page Bradley Page, a student at U.C. Berkeley, was convicted of voluntary manslaughter for the 1984 murder of his girlfriend, Bibi Lee. On the morning of the murder, after several fights between Page and Lee left the atmosphere "tense," n227 Page, Lee, and another woman went jogging in a secluded area. Lee separated from the other two; Page later went back to look for Lee by himself. He returned fifteen minutes later seeming "angry," "worried," and "somewhat scared and confused," claiming not to have seen her. n228 Page convinced the other woman to drive back Page 14 22 Harv. J.L. & Pub. Pol'y 523, *560 with him the ten miles to campus, leaving Lee to her own devices, a decision the woman "was not comfortable with." n229 Page did not seem particularly concerned [*561] about Lee during the rest of the day. The next morning, the police, who were summoned by one of Lee's worried roommates, were told by Page that Lee had disappeared while jogging. Page denied that he was upset or angry the morning of the run. n230 A search with bloodhounds and Explorer scouts of the area failed to locate Lee that day. Her body was eventually discovered several weeks later in exactly the same area where Page had gone looking for her while jogging. n231 After discovering the body, police questioned Page, who denied knowledge of the crime. Page then agreed to take a polygraph test. On the third time through the questions, Page began "wailing," making it impossible to continue the test. The polygraph examiner noticed that Page did not exhibit any physical signs of crying. Based on the first two tests, the examiner concluded that Page "tested deceptive for the entire test," and responded particularly when injury to Lee's head was mentioned. n232 Questioning resumed and Page said, "Well, if I did something I must have blacked it out." The investigators indicated they did not believe him. Page responded, "I remember hitting and kicking her, and wailing on her, or going off on her," but claimed not to remember when or where this occurred. n233 Efforts to get further details were unavailing, and questioning stopped. When questioning resumed, one of the investigators made a "direct accusation." Page was silent for a moment, then confessed he saw Lee on his return and tried to kiss her and talk to her. When Lee pulled away, Page became angry and backhanded her, knocking her to the ground. Lee seemed to be unconscious and her nose was bleeding. n234 Page also confessed that he returned that night, had sex with her [*562] body, and then covered her with dirt to give her a "decent burial." n235 The officers then attempted to take a taped statement from Page. He "essentially related the same story" but seemed somewhat "confused, tentative or vague ... [But] despite these indications of imperfect memory, Page was very specific regarding many of the details of the assault." n236 When a district attorney arrived to confirm the confession several hours later, Page attempted to recant the confession, claiming he confessed only because, among other reasons, he felt guilty for leaving Lee behind. When asked to explain how he came up with the details of the assault story, Page said he had made various assumptions or surmises. n237 Relying on second-hand accounts from Page's father, a defense psychologist, and a friend of the victim, n238 Leo and Ofshe conclude that "Page's confession failed to lead to corroboration, and abundant evidence supported the conclusion that he was innocent." n239 The California Court of Appeals viewed matters differently in a unanimous opinion written by Judge Clinton White, a former public defender from Berkeley not known for credulously accepting police accounts of crimes. The court explained that "despite Page's claim to the contrary," the prosecution's case "did not rely solely on his naked confession. In reality, the confession was supported and corroborated by the internal details Page supplied - details that only the killer could know - such as the location of the body, n240 the location of head and nose injuries, n241 and the method of [*563] burial." n242 The Court of Appeals also noted that Page's explanation at trial for the confession "strained the jury's credulity to the breaking point. His explanation was rife with internal inconsistencies, and was also inconsistent with the explanation he gave the officers in his final taped statement." n243 Leo and Ofshe further assert that in 1994 Michael Ihde was identified as Lee's murderer. n244 This bold claim is supported with a citation to "CBS News" (more precisely the "Connie Chung" show) which said that Ihde was a possible suspect in the killing. n245 Noting that Lee was an Asian-American, Leo and Ofshe claim Ihde was the real killer because he admitted "that he killed three San Francisco Area women - one of whom was non-white." n246 In fact, Ihde admitted killing a "black" woman. n247 Ihde's modus operandi was also inconsistent with the Lee murder. n248 Leo and Ofshe claim that Ihde's "appearance was consistent with the reported eyewitness evidence" in Page's trial. n249 This is apparently a reference to testimony from a defense witness, who said she saw a person she thought was Lee being pulled into a van by a man "in his mid-40s, 6 feet to 6 feet 3 inches, 220-225 pounds with a prominent beer belly, [*564] beard, and unkempt curly hair." n250 This testimony was severely undercut at trial, n251 and in any event is not remotely "consistent" with Ihde's appearance. At the time of the murder, Ihde was a "meth freak," approximately six feet four inches, 150 pounds - wiry, as skinny as a rail, with bright red "carrot top" hair. n252 Ihde was also without a car when Lee was killed. n253 It is therefore unclear how Ihde could have driven to the area where Lee was supposedly seen being pulled into a van, which was itself several miles from where she was jogging. It is much Page 15 22 Harv. J.L. & Pub. Pol'y 523, *564 less clear how Ihde would have figured out how to drive back to bury Lee in precisely the same area where she had been jogging with Page. 7. James Harry Reyos James Harry Reyos was convicted in 1983 of murdering a Catholic priest, Patrick Ryan, in a hotel room in Odessa, Texas. Leo and Ofshe list Reyos in the "proven" innocent category. n254 The source for this strong assertion is a letter Dennis Cadra sent to the Texas governor two days before Cadra left the district attorney's office that handled the case. n255 Cadra was a prosecutor in the office that prosecuted Reyos, but did not personally handle the trial. Cadra's letter argues it "was physically impossible for Mr. Reyos to have committed the crime for which he was convicted." n256 The letter recounts that Father Ryan was murdered in Odessa sometime between 7:00 p.m. and midnight [*565] on December 21, 1981. The letter claims, however, that Reyos was in Roswell, New Mexico until 8:00 p.m. on the 21st and got a speeding ticket around Roswell at 12:15 a.m. early the next morning. This leaves only four hours and fifteen minutes to drive 200 miles to Odessa to commit the murder and then return. To do this, the letter concludes, Reyos "would have had to have averaged a driving speed of over 111 miles per hour." n257 This all sounds quite persuasive until one compares the letter's claims with the original trial transcript. n258 To put Reyos in Roswell until 8:00 p.m. on the 21st, the letter relies on testimony from David Myers, a dormmate of Reyos's at college, who testified (according to the account in the letter) that he hung out with Reyos at some point "between the 19th and the 22nd." n259 The letter makes further arguments to exclude all of these days but the 21st, creating Reyos' alibi. Myers actually testified, however, that he had spent time with Reyos "sometime before Christmas of 1981," on a day that was "probably not the 24th." n260 The first few lines of the trial transcript of the prosecutor's cross-examination of Myers proves that the letter's representations about the dates being limited to between the 19th and 22nd are simply inaccurate: Q: Mr. Myers, you said this visit could have been the night of the 23rd? A: Yes, sir. Q: Or the night of the 22nd? A: Right. Q: You are not sure? A: No, I am not. n261 Myers's testimony, given a year-and-a-half after the murder, thus leaves entirely open the possibility that he met Reyos on the 23rd, not to mention the 22nd, n262 leaving Reyos without an [*566] alibi on the 21st. Not only was Myers unsure about the date, but Reyos did not tell police when questioned five days after the murder that he was with Myers on that night. n263 Because it is unclear even what day Reyos was with Myers, the seemingly precise "111 miles per hour" calculation rests on a foundation of sand. The ease of collapsing the supposed "proof" of Reyos's innocence should itself be strong evidence supporting the jury's verdict. But even if Myers firmly testified to hanging out with his old dormmate on the night of the 21st, the jury could have simply credited other trial testimony - none of which is discussed in the letter. Olivia Gonzales, for example, testified that she saw Reyos driving the victim's car by himself the day after the murder. n264 If the jury believed her testimony - and nothing in the cross-examination cast real doubt on it n265 - it would be powerful evidence of guilt. Evidence was also presented that Reyos performed oral sex on the victim the day before the murder, although Reyos gave an incredible account that the victim had "forced" him to do it. n266 Nonetheless, it was undisputed that early on the day of the murder, Reyos went to the victim's house seeking a ride. n267 Page 16 22 Harv. J.L. & Pub. Pol'y 523, *566 The jury also had before it confessions made by Reyos. The letter discounts these based on testimony from a psychologist concluding they were false confessions. n268 However, cross-examination at trial established that this opinion rested heavily [*567] on Reyos's honesty during psychiatric examinations. n269 Moreover, the psychologist relied exclusively on psychological profiles and never compared Reyos's confessions with the physical crime scene evidence. A comparison suggests the confessions were truthful. Reyos first called a 911 operator about the killing and said "you are talking to the killer." n270 When police came to his hotel room to question him about the call, Reyos said he killed the victim "with a razor, but I mostly beat him." n271 The medical testimony established that the victim was beaten to death and had one "linear laceration" that could have been a razor cut. n272 The defense offered no testimony to explain Reyos's knowledge of the manner of death. Reyos later testified at trial that "I don't recall anything after" the police opened the door to talk to him about the 911 call. n273 Reyos was also able to describe features of the hotel where the victim was murdered despite claiming to have never been there. n274 The police confirmed that Reyos bought a case of Coors beer, and some Coors beer cans were found in the hotel room where the murder was committed. n275 At the police station, Reyos reiterated in a "brashful manner" n276 that he killed the victim and wanted to know why it was taking so long for the police to record his statement. n277 A public defender arrived and argued with Reyos, unsuccessfully telling him to "shut up," "don't say anything," and "be quiet." n278 Ultimately, Reyos gave a recorded statement, first confessing to the murder and then repudiating the confession. n279 Although I have been unable to locate a transcript of the confession, the Texas Court of Appeals later rejected Reyos's claims that the confession was inconsistent in relation to the other evidence in the case. n280 [*568] The most compelling fact supporting Reyos's guilt is that all of the alleged exculpatory evidence including the alleged alibi - was capably presented to the jury. No good rationale is offered as to why the presumptively conscientious jurors found Reyos guilty beyond a reasonable doubt when he was innocent. The jury's verdict was upheld in a unanimous opinion from the Texas Court of Appeals, which noted that the "alibi was certainly not established as a matter of law." n281 Cadra's letter was also rejected. After receiving the letter, the Texas Board of Pardons and Parole voted sixteen to zero against a pardon. n282 The same information was pressed in a habeas petition, which the trial court in Odessa rejected on grounds that it offered no new evidence to warrant revisiting the case. n283 8. Linda Stangel Linda Stangel was convicted of killing her boyfriend, David Wahl, by recklessly pushing him to his death from a cliff. On November 12, 1995, Wahl disappeared while on a trip with Stangel to the Oregon coast. Several weeks later a body washed up in Washington, which was ultimately identified as Wahl's. The injuries to the body suggested that his death was caused by a sudden impact, such as falling from an extreme height. n284 Stangel initially told the police that she and Wahl got into a fight. Wahl left to go on a walk by himself around noon at the oceanside Ecola State Park and never came back. n285 She fell asleep in the van for four to five hours. When she awoke, Wahl still had not returned, so she drove the two hours home. Stangel never stopped on the way to seek help in finding Wahl. Finally, at 7:30 p.m., nearly eight hours after Stangel last saw [*569] Wahl, she called the police. n286 Stangel moved to Minnesota after Wahl's death but returned to Oregon in July for Wahl's memorial service. Police officers Travis Hampton and Alan Corson asked if Stangel would accompany them in an effort to run through the last day Wahl was seen. Stangel wondered why police had not done this sooner and indicated that it was appropriate under the circumstances. n287 The two officers and Stangel retraced the route down the coast, eventually arriving at Ecola Park. At the park, Leo and Ofshe claim the two officers "obliged [Stangel] to walk up the narrow, steadily rising bluff trail" where she "broke down in apparent fear of the cliff edge as they climbed the trail" and confessed to accidentally Page 17 22 Harv. J.L. & Pub. Pol'y 523, *569 shoving Wahl off the cliff. n288 The trial judge found a different version of the facts to be credible after hearing testimony directly from officers Hampton and Corson, defendant Stangel, and Richard Ofshe. n289 While sitting in the parking lot with the officers, Stangel repeatedly looked over her shoulder toward the cliff. n290 Hampton asked if something happened to Wahl up there, but Stangel did not reply. Hampton asked if there was something she had not told them, but Stangel again did not reply. Hampton asked if she wanted to go up, and Stangel said yes. They proceeded up the cliffside trail, arriving at a viewpoint. Stangel sat down and said she was afraid of heights. Hampton asked if she wanted to go back, but Stangel said no and led the officers further up the trail. When their way was obstructed by a log, Stangel asked Hampton "out of the blue" n291 if hypnosis could help her remember what happened. Hampton asked her to simply explain what happened. Stangel said, "What if I could tell you [*570] what happened but not where it happened?" n292 She explained that Wahl came at her, that she was scared, and demonstrated a straight arm push. Hampton told her that she was not guilty of murder if it was truly an accident. He asked Stangel to elaborate, and she laughed and said she made the whole thing up. She was silent, momentarily, and then told Hampton that she was afraid to talk because she would be blamed. Stangel told Hampton that she wanted to continue up the trail. n293 After a short hike, they arrived at a viewpoint without a railing next to a 300 foot drop to the beach below. Stangel told the detectives that this was the place where Wahl fell off the cliff. Stangel and the two officers spent the next fifteen minutes at the viewpoint. Stangel described how Wahl went over the cliff and demonstrated the event by moving the officers into Wahl's position. Stangel told the officers that she and Wahl had been talking about their relationship. She wanted to end it, but Wahl wanted to continue it. While they were talking, Wahl gave her a slight push to scare her, and it frightened her, even though she was not in fear of going over the edge. She then withdrew herself, yelled "f****r," and pushed him over the edge. n294 In the police car back at the parking lot, the officers took a recorded statement from Stangel. n295 On the day Wahl died, Stangel told him she "was going back to Minnesota" without him. Wahl went to the beach for fifteen minutes, then returned and asked Stangel to go with him up the trail. They walked slowly discussing their relationship, reaching the edge of the "very steep" cliff. Stangel stated: We were talking about our relationship again. Then I'm like just forget about it. And I proceeded to go near the edge of the cliff. And Dave came up behind me ... And, while we were standing there, he faked pushed me. And I'm afraid of heights, so my natural reaction was to just push him back. I just said, "hey f****r" and pushed him, accidentally. [*571] Wahl was "inches," or "maybe a foot," from the cliff at the time. Wahl "made this loud, annoying scream" as he fell to his death. Stangel made no attempt to get medical attention for Wahl and was sure he was dead. Stangel then went back to her car for a "couple of hours" and drove home without calling for help. She told everyone, including the police, that Wahl just walked off because "I didn't want to be accused and there were lots of fingers pointing at me right away." n296 At 9:00 p.m. the same day, when Stangel was safely back in her hotel room - and presumably no longer needing to "escape the immediate stress of the narrow and terrifying heights" that Leo and Ofshe claim induced the first confession n297 - she waived her Miranda rights and gave another parallel statement. n298 Stangel recounted the same facts leading up to the fatal push and then explained, "He had one hand on my back, one hand like on my stomach. I mean I'm out there in space and he pretends to push and I freaked. So I pushed him back... It was like a get the hell away from me kind of push." n299 At a suppression hearing, Stangel claimed the confessions were "made up to make these officers happy." n300 Ofshe testified for the defense as well, relying in significant measure on information from Stangel n301 to conclude that her fear of heights and the officer's statement that an accident was not a murder led to a false confession. n302 At Page 18 22 Harv. J.L. & Pub. Pol'y 523, *571 the conclusion of the hearing, the court advised the parties how it would rule: "I will accept Detective Hampton and Detective Corson's version of events, and find them to be believable and more believable than the defendant." n303 Similarly, the court concluded, "I do not accept Dr. [sic] Ofshe's opinions regarding the coercion or the coercive nature of the interrogation. I simply didn't believe him, and I do believe that Ms. Stangel gave her statements freely and [*572] voluntarily, and that the detectives did not coerce her into giving the statement." n304 At trial, the jury heard from the two officers and from Stangel and Ofshe. During trial, Linda Stangel was caught telling several lies. n305 After hearing all the evidence, the jury did not credit Stangel's and Ofshe's testimony and returned a guilty verdict. The jury's verdict poses a serious methodological issue for Leo and Ofshe. They promise to include in their collection only cases in which the "overwhelming majority" of neutral observers would find the confession to be false. Yet in this case, all twelve jurors unanimously found to the contrary - beyond a reasonable doubt - after hearing directly from Ofshe and after reviewing all of the available evidence. Leo and Ofshe do not suggest that any direct evidence was withheld from the jury, noting only that the indirect evidence that Stangel "passed" a polygraph test was excluded from the trial. n306 Leo and Ofshe appear to be unaware that Oregon law makes polygraph results absolutely inadmissible, n307 and their treatment of Stangel's allegedly exculpatory polygraph test seems inconsistent with their refusal to credit inculpatory tests in other cases. n308 In any event, the exculpatory polygraph test was an equivocal result from a defective machine - one of its four channels was broken - produced in the defendant's hometown where her mother worked with the police department. n309 All neutral observers in [*573] the judicial system in the Stangel case - the judge and the twelve jurors - found the confession was not coerced, and Leo and Ofshe offer no good reason for relitigating and discounting the findings here. 9. Martin Tankleff Martin Tankleff, a seventeen year old with an IQ of 124, n310 was convicted of murdering the parents who adopted him, Arlene and Seymour Tankleff. Citing only Tankleff's petition for habeas corpus, Leo and Ofshe claim "no evidence linked Tankleff to the crime." n311 Two weeks before the murders, however, Martin Tankleff said that if he could "have a hit" on both of his wealthy parents, he could drive any car he wanted. n312 A few days before the murders Tankleff had an ugly, public argument with his father. n313 On the morning of the murders, police responded to a 911 call and found Tankleff alone at home "soiled with blood," n314 his mother dead and his father grievously injured. n315 The police noted that Tankleff's various stories were internally conflicting and inconsistent with the physical evidence at the crime scene. n316 Tankleff agreed to accompany police to the station house. There, an enterprising detective arranged for a fictitious phone call to come in reporting that Tankleff's father miraculously had regained consciousness and identified Tankleff as the attacker. n317 Tankleff offered various explanations for this, and then quickly shifted to saying "it was like another Marty Tankleff that killed [*574] them." n318 After police administered Miranda warnings, Tankleff voluntarily made what a reviewing court described as "a full confession." n319 Tankleff said that his parents were fighting and accordingly turned to him for attention, with the result that they were "smothering" him. As a result of this and other resentments, he decided to kill them. He went to his parent's bedroom, hitting his mother four or five times with a barbell. Because she was fighting him off, he then went to the kitchen and got a knife and cut her throat and stabbed her. Then he went to his father's den, "just knocked him silly" with the barbell, and slashed his father's throat. He then showered and washed off the barbell and knife. As Tankleff continued to explain the details and the detectives prepared to reduce the confession to writing and videotape, an attorney called the station house asking that questioning of Tankleff cease, a request the police honored. n320 Tankleff's confession came about ten minutes after the fictitious phone call and a little more than two hours after questioning started. n321 Later that day, a detective overheard Tankleff acknowledge to his sister that he committed the crime. n322 Leo and Ofshe refuse to credit Tankleff's confession to killing his parents with a knife and a barbell, which was recovered from his room, even though the confession narrative was quite consistent with medical testimony about the nature of the injuries. n323 Leo and Ofshe argue the confession was inconsistent [*575] with a negative test of the knife and barbell for blood traces, hair and fibers, n324 but these facts tracked Tankleff's confession that he had Page 19 22 Harv. J.L. & Pub. Pol'y 523, *575 "washed off the barbell and the knife" in the shower. n325 Other testimony established that both items were moved within hours of the murders. n326 Leo and Ofshe further state definitively that the injuries to the father "were caused by a hammer." n327 The treating physician, however, testified that the injuries "had been caused by a hammer or an object similar to a hammer" n328 and the possibility of a hammer was not "to the exclusion of any other similar type weapon." n329 In any event, medical testimony established that the mother's injuries were "consistent with having been caused by" the barbell taken from Tankleff's room, n330 and that the father's injuries were caused by a "blunt instrument," with any further specification impossible due to the emergency surgical procedures performed. n331 Leo and Ofshe claim that the time of death of the mother was much earlier than indicated in Tankleff's confession n332 not recognizing that the mother's time of death could not be established with any precision by the medical examiner, but was generally within the time frame indicated by Tankleff. n333 The jury also had the benefit of testimony from a defense expert that Tankleff was "brainwashed," n334 but rejected this view and returned guilty verdicts. III. Lessons from the Cases The nine cases just discussed demonstrate that in a substantial portion of Leo and Ofshe's collection, the allegedly "innocent" person was in all likelihood properly found guilty at [*576] trial or by plea - particularly in cases where the most that Leo and Ofshe are willing to venture is the view that the person was "highly probably" or "probably" innocent, rather than "proven" innocent. The guilt of these nine defendants, for whom original court records were available, suggests that other defendants in the Leo-Ofshe collection may be guilty as well. n335 [*577] To all this, some may reply that I have missed the forest for the trees, n336 that the point is not exactly how many miscarriages from false confessions occur, but rather that they occur at all. But setting the record straight in these cases is important in its own right, in order to avoid cruelly traumatizing the victims' families n337 and unfairly maligning the professionalism of police, prosecutors, judges, and defense attorneys. More important, at least three broader lessons can be learned from these mischaracterized cases of innocence, as I explain in this Part. First, relying on news media accounts of trials to determine "innocence" is dangerous, because of the media's considerable bias towards discovering "news" by finding that the system malfunctioned. Second, removing the spurious cases from the Leo-Ofshe collection and examining the residual "undisputed" wrongful convictions reveals that the false confession problem is not pervasive, but rather is concentrated among a narrow population - those with serious mental problems. Finally, the cases demonstrate that even those who are guilty of crimes will frequently give a confession that is inconsistent with the known [*578] facts of the case. This presents a serious problem for Leo and Ofshe's proposal to suppress confessions whose post-admission narrative fails to closely track the facts of the case. A. Problems in Determining "Innocence" The first issue worth examining is how Leo and Ofshe could have wandered so far astray on some of these cases of alleged "innocence." The question becomes even more puzzling when we realize that Leo and Ofshe, to avoid precisely the type of reevaluation undertaken here, promised to limit their collection of cases to those lacking any "credible evidence" corroborating the defendant's guilt. n338 Yet in all of the nine cases just examined, such evidence plainly existed. While it is possible that Leo and Ofshe have simply played fast and loose with the facts, n339 the distortion in these cases is perhaps more simply attributable to the kinds of secondary sources Leo and Ofshe relied upon. While Leo and Ofshe describe their research as relying "where possible" on "primary" sources such as "trial records," n340 the great bulk of it actually rests on second-hand accounts. Some of these secondary sources were draped in red warning flags that should have been heeded, such as a recounting of a trial from the defendant's father, n341 a book published by "the friends of" a defendant to raise funds for his defense, n342 and a habeas petition designed to win a defendant's release from prison. n343 Many other sources were second-hand newspaper accounts available in computerized databases. No doubt relying on such readily-available information saves considerable time. Having spent innumerable hours collecting original court records on just nine cases, I can attest to the practical advantages of computerized research. But with the advantage of speed comes the danger of error, a danger that becomes unacceptable [*579] in Page 20 22 Harv. J.L. & Pub. Pol'y 523, *579 light of news media biases. While courts and juries legally must be impartial, journalists are not so restricted and have an understandable interest in slanting evidence to suggest wrongful conviction. This is a particularly serious problem in "investigative" reporting because "many journalists deem investigative pieces successful only if they uncover malfeasance, [creating] an incentive to interpret ambiguous facts in the worst possible light." n344 However well intentioned a reporter may be, the fact remains that it is only "news" if an innocent person is convicted. One reporter explained the ease of writing compelling yet biased stories this way: "If you've got a crying mom, you've got a story. If you've got a crying dad, my God, you've got two stories! If you don't get the other side of the story, you've got to be swayed." n345 Even entirely impartial media sources can also be influenced by the agendas of others. This may account for the particularly egregious misrepresentations found in second-hand descriptions of death penalty cases. n346 Family members or supporters of a prisoner may also believe, for understandable reasons, that an innocent person has been wrongly convicted and set out to use the media to influence the case through accounts of the "evidence" that, predictably, put the best possible light on things. The normal checks and balances found in other areas of journalism may operate less effectively here. Prosecutors, and indirectly police agencies, operate under rules of legal ethics that severely restrict the information that can be provided about a case outside the courtroom. n347 On the other side, the enforcement of similar rules against defense attorneys is problematic and, in any event, supporters of the defendant outside the legal profession are free to give their views. To be sure, crime victims can, and in some cases do, respond. But in other cases they may not. Victims often have no stomach for public battles over the crime, believing that the issues will be resolved in the courtroom. One reporter described the resulting [*580] asymmetries in sexual abuse cases as follows: Those accused of sexual abuse have an overwhelming interest in discrediting the child and family members accusing them ... They resort to war by public diplomacy. The other side - therapists, family, friends - cannot answer back in kind. For all those concerned about the victim ... maintaining privacy is crucial ... No one who cares about the human costs of crime to the victim will lightly compound it. As a result, the message carried to the media is often heavily biased in favor of the perpetrator. n348 It should be noted that a disproportionate number of Leo and Ofshe's innocents were men who committed crimes against women or children. Some have argued that the media is structurally biased against accepting claims from or supportive of such victims. n349 All this suggests that relying on second-hand media accounts is not a reliable means of determining the accuracy of convictions, as the media will inevitably manufacture more "innocents" than really exist. At the very least, primary sources such as trial transcripts should be carefully consulted before crediting a news story about a "wrongful" conviction. B. Overgeneralizing the Problem of False Confessions While second-hand media reports inflate the number of reported cases of wrongful conviction from false confession, an important additional question is whether such reports are skewed in any systematic way. After all, the news media might give us a general sense of what is happening, even if it exaggerates the extent to which it is happening. But there is good reason to believe the media not only creates miscarriages but also creates a particular kind. For all the reasons just given, the media's greatest interest is in depicting wrongful convictions in a dramatic it-could-happen-to-you fashion. In an effort to appeal to its audience, the media is more inclined to find "average" persons who have been wrongly convicted, possibly obscuring the fact that more unique populations are especially at risk. [*581] The hypothesis that the Leo-Ofshe collection is skewed could be tested if we could compare it to a more reliably identified collection of wrongful convictions. Professor Sam Gross' research on eyewitness misidentification provides such a means. He restricted his research to cases of "undisputed" misidentifications - those in which there was a clear determination of innocence, preferably from the prosecuting authority that originally charged the defendant. n350 There is undoubtedly considerable wisdom in this conservative approach, as even Richard Ofshe has Page 21 22 Harv. J.L. & Pub. Pol'y 523, *581 acknowledged. n351 The problems with the subjective determination of "innocence" in the Leo-Ofshe collection, like similar problems elsewhere, n352 suggests that reliance on second-hand sources combined with understandable enthusiasm for the enterprise of discovering miscarriages may produce more such cases than really exist. Although using undisputed cases suffers from the limitation of requiring the willingness of officials to admit error, police and prosecutors would seem to have ethical obligations to reverse genuine cases of wrongful conviction and, in a number of cases, have in fact done so. In any event, the problem of official reluctance to admit error likely affects only the number of wrongful convictions, not their nature. Only a relative handful of Leo and Ofshe's cases would satisfy the criterion of undisputed wrongful conviction. It appears that in none of the cases of "probable" or "highly probable" innocence has there been an official acknowledgment of error. Even among the fifteen "proven" cases of wrongful conviction from false confession, many are disputed. Prosecutors believed Lavale Burt was guilty of murder. n353 The [*582] district attorney who prosecuted Steven Linscott specifically declined to declare him innocent. n354 In the case of George Parker, the prosecution apparently continued to press for his conviction, only to be blocked by an appellate court. n355 State police continued to believe Peter Reilly committed the crime after an appellate court overturned the conviction. n356 The guilt of James Reyos has been discussed previously. n357 Prosecutors continue to believe that Earl Washington was guilty of rape and murder. n358 Indeed, in commuting Washington's death sentence as one of his last acts in office, Governor Douglas Wilder said that he was not fully convinced that Washington was innocent and that Washington "had knowledge of evidence relating to the crime which it can be argued only the perpetrator would have known." n359 Governor Wilder also noted, in remarks that could apply to the death penalty cases discussed here, that the capital punishment opponents pressuring him to grant a full pardon did "not enjoy a grasp of the specific facts of the case." n360 As a result, assuming the recitations in secondary sources have accurately recounted the facts in the remaining cases, nine of Leo and Ofshe's twenty-nine wrongful conviction cases (approximately one-third) are undisputed: Bradley Cox, n361 [*583] Ralph Jacobs, n362 William Kelley, n363 Laverne Pavlinac, n364 John Purvis, n365 Melvin Reynolds, n366 Christopher Smith, n367 David Vasquez, n368 and Johnny Lee Wilson. n369 When these nine defendants are examined, a striking common feature emerges - eight of the nine, all but Bradley Cox, are mentally retarded n370 or have other serious mental problems. n371 This means that, after extensive research, Leo and Ofshe have uncovered only one undisputed case in the last quarter century in which a mentally normal person was wrongfully convicted from a false confession, and that case appears to offer few generalizable lessons about police interrogation. n372 More [*584] important, the high concentration of the mentally infirm among the undisputed cases suggests that, for the most part, false confessions are caused not by police questioning techniques in general but rather by the application of those techniques to certain narrow, mentally limited populations. n373 Although more research is clearly needed on this question, there is nothing new in recognizing that those with mental limitations are at special risk of false confessions. As long ago as 1963, the President's Panel on Mental Retardation concluded that "some of the retarded are characterized by a desire to please authority: if a confession will please, it may be gladly given." n374 Other commentators have expressed similar concerns. n375 Consistent with these cautions, American courts "have long recognized that confessions by mentally retarded persons are somewhat suspect." n376 Leo and Ofshe could perform a valuable service if they analyzed interrogation tactics that posed particular risks to this population. But identifying false confessions solely among this constricted population is not the limited project Leo and Ofshe have in mind. Instead, they wish to proceed more broadly and, for example, heap scorn on Fred Inbau and the other authors of the most widely used police interrogation manual. n377 Leo and Ofshe claim these authorities "persist in the self-serving and misguided belief that contemporary psychological methods are not apt to cause an innocent suspect to confess - a fiction that is flatly contradicted by all of the scientific research on [*585] interrogation and confession." n378 I have suggested that, to the contrary, Leo and Ofshe are guilty of rhetorical overstatement in claiming that common police questioning techniques are "apt" to cause false confessions. n379 But even setting that aside, Leo and Ofshe fail to acknowledge that the manual expressly recognizes the special problems that arise in Page 22 22 Harv. J.L. & Pub. Pol'y 523, *585 questioning those with mental problems, warning that "an innocent person with a psychological affliction may seem to be guilty." n380 This admonition is followed with a cautionary example in which a innocent suspect with an "unstable personality" aroused suspicions during interrogation. n381 Moreover, in an article directed to law enforcement authorities, Fred Inbau has specifically written that "special protection must be afforded [to persons of below-average intelligence] ... to minimize the risk of obtaining untruthful admissions due to their vulnerability to suggestive questioning." n382 Questioning should proceed, Inbau instructed, "with all reasonable precautions to guard against untrustworthy admissions, bearing in mind the suspect's particular vulnerability to suggestiveness with respect to possible explanatory conduct." n383 Thus, far from deserving condemnation, Inbau should be commended for alerting law enforcement communities to a particularly perilous area for false confessions. If criticism is to be directed anywhere, it might be towards Leo and Ofshe for suggesting more generally that all police interrogation should be regarded with suspicion. This overbroad indictment will inevitably, and unfortunately, distract at least some attention away from where it apparently [*586] should be focused - the mentally retarded. Precisely what safeguards are needed to protect the mentally retarded is difficult to say. American courts already specially scrutinize confessions from the retarded under the voluntariness test. n384 The miscarriages collected by Leo and Ofshe establish that this scrutiny is imperfect, although they do not establish the extent of the problem. Because of such examples, some have proposed restrictions on police questioning of the mentally retarded. n385 Judicially enforceable rules for police questioning would, however, raise considerable practical difficulties. n386 As a result, police training may be the best preventative medicine and could, perhaps, even be done with materials from Inbau. Oversight of police questioning of the retarded may be best handled through continued judicial scrutiny of the voluntariness of confessions and expert testimony to juries on the peculiar susceptibilities of the retarded to this problem. n387 At the same time, caution is needed in applying these rules. Those with mental illness are particularly likely to fall victim to crime n388 and thus may be especially severely harmed by restrictions on police investigation. Also, in other areas of the law, doctrines benefiting the mentally retarded have led to biased mental testing to prove that defendants suffered from this disability. n389 [*587] And, finally, the retarded are not immune from criminal temptations. n390 Nonetheless, a focus on the possibility of false confessions given by those with mental limitations appears to offer a chance of a targeted - and successful - response to the existing false confession problem. n391 The examination of undisputed cases also reveals one disturbing point about Leo and Ofshe's analysis that, unfortunately, must be made: their accuracy in identifying cases of false confessions is distressingly low. In total, Leo and Ofshe analyzed twenty-nine cases of wrongful convictions from false confessions. Of these twenty-nine, the nine undisputed cases did not require any special identification because all parties conceded the convictions should be overturned. Removing these nine uncontested cases leaves twenty in which the truth of the confession was disputed. The analysis presented here demonstrates than in nine of those twenty cases - Fairchild, Giarratano, Ingram, Lapointe, Misskelley, Page, Reyos, Stangel, and Tankleff - Leo and Ofshe have incorrectly concluded that the defendants were probably innocent. n392 Even giving Leo and Ofshe the generous benefit of [*588] the doubt that they were entirely right on the remaining eleven cases, their success rate in identifying false confessions in the disputed cases can be no better than eleven of twenty (fifty-five percent), barely better than one would expect from flipping a coin to decide a controverted issue. This low batting average raises serious questions about the admissibility of expert testimony resting on Leo and Ofshe's research about false confessions. The admissibility of expert testimony on false confessions is quite controversial - some courts allow it, n393 while others exclude it. n394 Although this is not [*589] the place for an extended discussion of these competing approaches, n395 the clear trend in the cases is that, at the very least, courts should preclude any testimony on the truth or falsity of a particular confession. n396 Moving beyond that, expert testimony might in theory be justified on interrogation conditions that might produce false confessions. On this point, however, a practical difficulty arises. Before allowing such testimony, courts must find that it pertains to "scientific knowledge." n397 In making this determination, courts can consider the acceptance of particular conclusions within a relevant scientific community. n398 It is not at all clear that acceptance of conclusions about false confessions yet exists given the preliminary nature of false confession research. Professor Welsh White has recently noted that the Page 23 22 Harv. J.L. & Pub. Pol'y 523, *589 empirical data on false confessions could be described as "tentative and fragmentary." n399 Professor Saul Kassin, himself a leading researcher on the subject of false confessions, has gone further and concluded that "the topic of confession evidence has largely been overlooked by the scientific community. As a result of this neglect, the current empirical foundation may be too meager to ... qualify as a subject of 'scientific knowledge' [*590] according to" the rules governing expert testimony. n400 Testimony resting on Leo and Ofshe's research, at least concerning how "false" confessions are produced, does not appear to satisfy the requirements to be admissible expert testimony. Before accepting expert testimony, courts should "consider the known or potential rate of error" from a particular technique. n401 Leo and Ofshe's technique results in an error rate - at least forty-five percent - that is unacceptably high. Moreover, the rules of evidence generally permit expert opinion based on "facts or data" that are "of a type reasonably relied upon by experts in the particular field." n402 Experts in the field of miscarriages would certainly look to primary court records first, as even Leo and Ofshe seem to concede. n403 Yet in practice their analysis quite frequently departs from the preferred approach. Their analysis only rarely cites - much less discusses - primary court records, even where judicial opinions are readily available in published reporters. Of course, judicial records are not the only sources that should be consulted; the point is that they certainly ought to be among the main sources consulted. Because of the high error rate and failure to follow accepted research techniques, courts should not allow expert testimony resting on Leo and Ofshe's analysis of any of the particular cases discussed here. n404 Courts certainly need more reliable research on this important subject, and I add my voice to those who are encouraging further investigation of the false confession problem. n405 C. Problems With Post-Admission Narrative Analysis. Rather than focusing on particular subpopulations at special risk for false confessions, Leo and Ofshe propose more sweeping changes to confessions law. Among other things, Leo [*591] and Ofshe suggest that courts should "carefully scrutinize" a confessor's "post-admission narrative" against the known facts of the case. n406 In their view, "the fit between the specifics of a confession and the crime facts determines whether the 'I did it' admission should be judged as reliable or unreliable evidence." n407 They further argue that, if the discrepancies are substantial enough, courts should conclude that the confession is unreliable and suppress it. n408 As a general proposition, there is nothing wrong with examining confessions to see whether they track the crime facts and reveal details about the crime that only the perpetrator would know. No doubt juries routinely make such evaluations in assessing confession evidence. Leo and Ofshe propose, however, to go further and require courts to make a specific determination about "fit," with that determination governing the admissibility of defendants' statements. Such a proposal goes well beyond traditional judicial scrutiny, which focuses on the "voluntariness" of the confession, n409 and would in effect substitute judges for the jury as the trier of these important, and often outcome-dispositive, facts. Because this far-reaching change would govern the admissibility of all confessions, not just those from the vulnerable population of the mentally retarded, its desirability rests on proof that it would especially suppress false confessions but not true ones. n410 There are substantial practical reasons for doubting that the proposal would be so discriminating. n411 To begin with, many guilty suspects do not give a "confession" to the crime, but rather make only an incriminating statement. By definition, such a statement will [*592] fail to fit the facts of the crime, even if it accurately suggests guilt. Consider, for example, a suspect's claim that he was with the victim on the day of the murder, but left shortly before she was killed. Such a statement may be quite important to a successful prosecution of the suspect, but the prosecution's entire theory will be that the suspect did not leave before the murder and that this part of the narrative does not fully track the facts of the crime. This scenario unfolds very frequently. Bret Hayman and I found in Salt Lake City that, among suspects who gave incriminating statements, thirty-six percent provided something less than a confession. n412 An even larger percentage comes from Leo's empirical study in northern California, which found that, among suspects giving incriminating statements, sixty-three percent gave something less than a "full" confession. n413 Both these studies suggest that vast numbers of suspects' incriminating narratives will be at odds with the facts of the case simply because suspects do not confess to the crime in its entirety. Page 24 22 Harv. J.L. & Pub. Pol'y 523, *592 Presumably Leo and Ofshe, if required to elaborate on this point, would respond that mere incriminating statements are not subject to their proposal; instead, only "confessions" of the form "I did it" would be subject to the new rules. n414 In another article, they refer to Black's Law Dictionary definitions of "confession" and "admission." The dictionary defines a "confession" as a statement in which a suspect "acknowledges himself to be guilty of the offense charged, and discloses the circumstances of the act or the share and participation which he had in it." n415 An "admission" is defined as "the avowal of a fact or of circumstances from which guilt may be inferred, but only tending to prove the offense charged, and not amounting to a confession of guilt." n416 If Leo and Ofshe's proposal is triggered by a "confession" but [*593] not by an "admission," then judges will have to reliably distinguish between the two. Scant precedent suggests they could do so. For example, the Supreme Court held in Miranda that, for self-incrimination purposes, "no distinction can be drawn between statements which are direct confessions and statements which amount to 'admissions' of part or all of an offense." n417 Perhaps part of the reason the Court avoided this quagmire is that no distinction can effectively be made. Consider, for example, the common scenario of a suspect refusing to confess to the crime in all its horror. Leo and Ofshe recognize this possibility, providing the illustration of a guilty suspect who confesses that he kidnapped and killed a child, but is unwilling to confess to also raping her. n418 They would apparently allow the jury to hear this statement in the suspect's trial for murder, kidnap, and rape, but do not explain why. They have two possible grounds for admitting this statement - that it is only an "admission" and therefore not subject to their rules, or that it is a "confession" subject to their rules, but sufficiently close to the facts to be admitted. On either theory, numerous problems will develop. On the first theory - that the statement, with its relatively minor deviation from the truth, is treated not as a "confession" but only as an "admission" - then virtually all statements from suspects will fall outside the scope of the new rules. No doubt virtually all confessions contain at least some minor deviation from the truth which, it could be argued, transform them from confessions into admissions. On the second theory - that the statement, while deviating slightly from the truth, is a "confession" but nonetheless admissible because it sufficiently tracks the crime - courts will be forced to decide difficult questions of when confessions hew closely enough to the facts. The hypotheticals are endless and perplexing. Suppose the suspect admits kidnapping, but not murder or rape? Or admits consensual sex in violation of a statutory rape statute, but not forcible rape or kidnapping or murder? Or admits taking the child for a ride, but not with the intent to obtain ransom or commit a felony - typically elements [*594] of kidnapping? n419 Or admits stalking the child, but not giving the child a ride? All of these involve a confession to some element of the charged or an uncharged crime, but it is not clear which of these statements qualify as "confessions" subject to the Leo and Ofshe proposal. If any of these statements are defined as "confessions," they will be at odds with important parts of the crime and, therefore, apparently suppressible under the Leo and Ofshe proposal that judges should scrutinize the "fit" of the confession. The problems only mount when we realize that the guilty suspect, even if "confessing" to committing all of the charged crimes, such as murder, kidnap, and rape, might nonetheless provide a post-admission narrative that deviates from the crime's facts. This is a widely recognized problem. As one knowledgeable observer recognized, "even 'true' confessions are often riddled with half-truths because suspects want to paint the most favorable picture possible." n420 The popular Inbau manual reports that many offenders "will admit guilt to a very serious offense, while at the same time refusing to do so regarding a less related one that was part of the same series of events." n421 For example, a suspect might confess to a murder but deny stealing a crucifix from the victim. n422 Even looking only at the nine cases carefully reviewed here reveals a plethora of reasons for deviations, or apparent deviations, between the facts of the crime and the perpetrator's otherwise truthful confession, including the following: . a desire to cover for family members or friends; n423 [*595] . an attempt to minimize the seriousness of an offense; n424 . a drug-induced memory problem or confabulation; n425 Page 25 22 Harv. J.L. & Pub. Pol'y 523, *595 . an intention to throw the police investigation "off track"; n426 . a desire to give the impression of having "blacked out" around the time of the crime; n427 . an insufficiently precise description of the location of incriminating evidence to permit its recovery; n428 . an inability to complete questioning because of legal restrictions on police interrogation; n429 and . a manipulative suspect. n430 Leo and Ofshe have not yet described how courts would evaluate such legitimate reasons for deviations between confessions and the facts of the crime. Nor have they offered any specific explanation of how tightly the facts of a confession must fit the facts of the crime. In an earlier brief exchange with them, I pressed this point, explaining that evaluating their proposal was difficult because they had not spelled out its operation. n431 They replied that the confession "must fit the facts of the crime to a reasonable degree," n432 but did not provide illustrations or other elaboration of this "reasonable degree" standard. [*596] Their collection of cases provides an opportunity to assess the degree of fit in the context of real cases. This assessment demonstrates that Leo and Ofshe would have courts keep from juries many truthful confessions. They contend that in all nine of the cases I have closely examined here, the confession was not simply inaccurate, but so inaccurate that courts should have excluded it. n433 This means that in all of these cases - and many others like them quite probably guilty suspects would have succeeded in suppressing reliable and highly incriminating statements, with the consequence that most of them would have escaped any punishment for their crimes. The Barry Fairchild case illustrates this point. Leo and Ofshe suggest that Fairchild's confession was so riddled with "glaring errors of facts" n434 that the trial court should have suppressed it. n435 The glaring errors of fact, however, were apparently understandable to the jury, and for good reason. For instance, Fairchild confessed to committing the crime with his accomplice Harold Green, who had an airtight alibi for the murder. This "glaring error" is readily understandable when we realize that Fairchild's actual accomplice was quite possibly his brother. n436 While making much of this supposed "error," Leo and Ofshe simultaneously give short thrift to confirmed parts of the confession. For instance, Fairchild confessed to taking a unique watch from the victim during the murder, a watch that police retrieved, with Fairchild's help, from Fairchild's sister. This corroborated part of the confession and proved, a federal district court concluded, Fairchild's involvement "far beyond any reasonable doubt." n437 This fact is discounted by Leo and Ofshe, however, because Fairchild had some facially implausible explanation about buying the watch from an unidentified person at an unidentified place. n438 If Fairchild is a fair portrait of how tightly a confession must fit the facts to satisfy the "reasonable degree" test, then defense attorneys will have a field day suppressing vast numbers of truthful [*597] confessions. n439 While post-admission narrative scrutiny would suppress many true confessions, false confessions might actually survive it. Consider, for example, police bent on coercing a suspect into signing a "confession" to a crime he did not commit. Presumably police will design such a confession to track the crime scene quite closely, and likely even include details about the crime not available to the public. The post-admission narrative of such confessions will fit the crime to more than a "reasonable degree" and will be unaffected by the new rule. Apart from the rare case of police dishonesty, even truly false confessions can track the crime scene. An innocent suspect may simply describe something that bears a striking resemblance to a fact of the crime. n440 More often, police may reveal the manner in which the crime was committed during questioning, only to have the false confessor report back the same details later. n441 Leo and Ofshe recognize that "contamination" can occur during questioning, n442 a possibility attested to in cases in their catalogue. n443 They provide, however, no explanation as to how courts should decide between the competing possibilities of an innocent defendant regurgitating information fed to him by the police and of a guilty defendant reporting his involvement in the crime. n444 This "regurgitation" problem suggests rather [*598] strongly that the efficacy of the Leo-Ofshe proposal to scrutinize post-admission narratives is inevitably coupled with their additional proposal to Page 26 22 Harv. J.L. & Pub. Pol'y 523, *598 videotape interrogations. Without such recording, it will be impossible to determine whether an alleged consistency of the confession is due to what a suspect learned from the police or knew from committing the crime. But even with a recording scheme in place, in close cases the courts still will have to make difficult determinations. Compounding the difficulties, such regurgitation simply will not take place among innocent suspects. Presumably a fair number of guilty suspects will give statements that conform to what they think the police know, as revealed during questioning. As the Inbau manual reports, "[a] guilty person is always interested in hearing the whole story or in finding out exactly what may be known about him so that an assessment may be made of the situation." n445 No doubt a number of guilty suspects provide statements that fit what the police have, and no more. To gain some idea of the overall efficacy of the Leo-Ofshe proposal, we need to estimate the relative number of true and false confessions that would be excluded under it. Because the proposal's operational details have yet to be spelled out, making authoritative assessments is difficult. But the issues raised here clearly demonstrate considerable room for concern that many guilty suspects would go free because courts would not allow juries to hear incriminating statements from guilty suspects. n446 We need not engage in such an untested approach to presenting wrongful convictions if other mechanisms reliably can separate the guilty from the innocent. Such separation is traditionally the task of juries. Thus, an inseparable part of the case for the Leo and Ofshe proposal is the claim that juries often fail to sort true confessions from false ones. At the outset, this claim seems counterintuitive. Not only [*599] should we be ready to indulge a modest presumption that the foundation of our criminal justice system works, but also that the kinds of issues swirling around false confessions seem readily susceptible to jury evaluation. Did the suspect confess to details only the real perpetrator would know? Did the cops "feed" the information to the suspect? Is there some innocent explanation for the incriminating statement from the suspect? These are not subjects generally beyond the ken of the average jury and indeed, as Leo and Ofshe recognize, there are many cases in which jurors have acquitted possible false confessors. n447 Moreover, even were these subjects beyond juror comprehension, n448 courts have recently begun to allow expert testimony on such subjects. n449 We can, of course, assist juries to accurately make such determinations. Replacing Miranda with a system of videotaping confessions is an example of such an improvement. n450 Giving juries all of the evidence surrounding the taking of statements would undoubtedly improve jury findings. Leo and Ofshe, however, distrust juries for two reasons. First, they allude to mock juror research, which they suggest calls into question the abilities of juries to identify false confessions. n451 The two cited studies have rather limited relevance to the issues discussed here, as they involved mock juror assessments of confessions given only after threats or promises. In both, although jurors tended to reject confessions extracted by threats, n452 they credited those obtained through suggestions of leniency. n453 Even assuming that individual mock [*600] jurors arrive at their decisions with the same care as real world deliberating juries, n454 the application of this research to the false confession issue is problematic. Confessions induced through leniency can obviously be truthful confessions. The studies do not prove that jurors fail to separate the truthful confessions from the false ones because that was not the hypothesis tested. Indeed, the findings can be read to suggest that the mock jurors were accurately assessing the facts. For example, in one study, the jurors rendered guilty verdicts at a rate of nineteen percent in a case without a confession, twenty-five percent when the facts were varied to include a confession obtained by a threat, thirty-eight percent when a confession was obtained by a promise, and fifty-six percent when a confession was obtained without threat or promise. n455 This is precisely the ranking one would hope to find if jurors were focusing on the factual guilt or innocence of a suspect: the fewest guilty verdicts when no confession exists, followed by slightly more guilty verdicts when a confession was obtained by threat, more still when a confession was obtained by a promise, and the most when a confession was obtained without threats or promises. Indeed, it is interesting that the authors of these studies read their research as suggesting, in part, that "people do clearly recognize the risk of false confessions." n456 More on point to the issue of juror abilities in this area would be assessing the accuracy of actual jury verdicts. Leo and Ofshe [*601] attempt such analysis, culling from their case collection the figure that seventy-three percent of the false confessors who proceeded to trial were convicted. n457 Of course, that figure is too high given that a number of Page 27 22 Harv. J.L. & Pub. Pol'y 523, *601 their "innocent" suspects were quite probably guilty. But even setting that point aside, the significance of this percentage is rather limited when we realize, as Leo and Ofshe note, that their cases "do not constitute a statistically adequate sample of false confession cases." n458 Instead, it disproportionately reflects, among other things, cases that are famous or that involved consultation by Leo or Ofshe. n459 Wrongful convictions are more likely to gain notoriety than acquittals, and thus "famous" cases will include a disproportionate number of convictions. Similarly, Leo and Ofshe are presumably asked to consult most often in cases where the defense is seriously concerned about a wrongful conviction from a false confession. As a result of these two biasing features, the actual percentage of false confessors convicted by juries would be far less in a randomly drawn sample. n460 We must await future research on an unbiased sample before reaching firm conclusions about jury capabilities, which should be compared not against a hypothetical, perfect world but real-world, and hence imperfect, alternatives such as enhanced judicial scrutiny. All this suggests that the case against jury evaluation of alleged false confessions has yet to be convincingly made much less the case that adding post-narrative admission scrutiny would improve the reliability of the system. Until these points are proven, there is little reason to abandon our trust in juries and defer to judges to make difficult assessments about the accuracy of confessions. Black letter law recognizes "the province [and] capacity of juries to assess the truthfulness of confessions." n461 More generally, we rely on juries everyday to make difficult assessments about all sorts of complicated issues [*602] in criminal trials, including the ultimate guilt or innocence of criminal defendants. It is no news that juries occasionally make mistakes in some of these determinations, including even convicting an innocent person in isolated cases. But though juries are imperfect, they are deeply rooted in our constitutional structure, n462 which incorporates the principle that "the Trial of all Crimes ... shall be by Jury." n463 In fact, our system probably relies on juries precisely because of the long-standing belief that they can do a good job of determining the truth. n464 In view of these facts, it seems hard to argue with the conclusion that we should permit "a jury rather than a judge to assess the evidentiary value of an appropriately obtained confession." n465 Without a clear case that the alternatives are better, the jury system should govern trials revolving around claims of false confessions no less than those revolving around other vital issues. IV. Conclusion The issue of false confessions deserves careful review. More generally, separating the guilty from the innocent is a critically important business, and the criminal justice institutions involved in this enterprise warrant great scrutiny. But research on such subjects must grapple with the foundational question of how to determine whether a defendant is truly "innocent." A fair-minded approach cannot credulously accept claims of miscarriages offered from whatever source, but instead must employ a more objective means of double-checking the criminal justice system. A detailed examination of the alleged miscarriages from false confessions assembled by Leo and Ofshe reveals that a significant fraction of the "innocent" were, in fact, guilty criminals. The miscategorization of these cases stemmed primarily from reliance on inaccurate second-hand media [*603] reports, suggesting the great value of relying on primary sources for miscarriage research. The miscategorization also tended to obscure the nature of the false confession problem. Distilling down the cases to those where wrongful convictions clearly occurred, the false confession problem is apparently not a pervasive one, but rather one concentrated among the mentally retarded. Efforts to improve the quality of police questioning and judicial review of confessions should focus there, rather than on encouraging a more sweeping, and less likely to be adopted, substitution of judges for juries as the arbiters of the accuracy of confessions. Legal Topics: For related research and practice materials, see the following legal topics: Criminal Law & ProcedureInterrogationMiranda RightsGeneral OverviewCriminal Law & ProcedureDefensesEntrapmentEvidenceScientific EvidenceDNA Page 28 22 Harv. J.L. & Pub. Pol'y 523, *603 FOOTNOTES: n1. 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 (1998). Leo and Ofshe's claims about false confessions were recently featured in page-one articles in prominent newspapers. See Jan Hoffman, Police Refine Methods So Potent, Even the Innocent Have Confessed, N.Y. Times, Mar. 30, 1998, at A1; Thomas H. Maugh II, Glendale Case Raises Issue of Reliability of Confessions; Experts Say Some Arise from Unstable Minds of Coercion, L.A. Times, Apr. 2, 1998, at A1. n2. See Leo & Ofshe, Consequences of False Confessions, supra note 1, at 478 tbl. B2. n3. Id. at 493. n4. Id. at 495-96. n5. See Paul G. Cassell, Protecting the Innocent from False Confessions and Lost Confessions - And from Miranda, 88 J. Crim. L. & Criminology 497 (1998). n6. Id. n7. See id. at 530-32. n8. See id. at 538-52. n9. Richard A. Leo & Richard J. Ofshe, Using the Innocent to Scapegoat Miranda: Another Reply to Paul Cassell, 88 J. Crim. L. & Criminology 557 (1998). n10. Id. at 560. n11. Id. at 561. n12. See, e.g., Maugh, supra note 1, at A1 (quoting Richard Leo in writing that false confessions "happen all the time"); Dateline (NBC television broadcast, Dec. 23, 1997) (when asked how often innocent people are put behind bars because of false confessions, Richard Ofshe answers "We know it happens all the time"); Richard Jerome, Suspect Confessions, N.Y. Times Mag., Aug. 13, 1995, 6, at 28 (quoting Richard Ofshe) ("If it happens just one-half of 1 percent of the time it still means that hundreds, or perhaps thousands, of people each year are being unjustly imprisoned."). n13. Professor Welsh White's recently published essay appears to suffer from a similar problem. It argues that it is "impossible to estimate" the "number of false confessions." Welsh S. White, What Is An Involuntary Confession Now?, 50 Rutgers L. Rev. 2001, 2039 (1998). At the same time, however, it contends that "standard Page 29 22 Harv. J.L. & Pub. Pol'y 523, *603 interrogation methods precipitate a significant number of false confessions." Id. at 2042 (emphasis added). The article also suggests that confessions derived from interrogation methods with a "tangible likelihood" of producing a false statement should be inadmissible. Id. at 2039. It is hard to see how this "likelihood" determination can be made without further quantitative evidence about the numbers of false confessions. n14. Leo & Ofshe, supra note 9, at 562. n15. Id. at 566 (internal quotation omitted). n16. Gisli H. Gudjonsson & Hannes Petursson, Custodial Interrogation: Why Do Suspects Confess and How Does It Relate to Their Crime, Attitude and Personality?, 12 Person. Individ. Diff. 295, 298 (1991); Gisli H. Gudjonsson & Jon F. Sigurdsson, How Frequently Do False Confessions Occur?: An Empirical Study Among Prison Inmates, 1 Psychol. Crime & L. 21, 25 (1994); Jon F. Sigurdsson & Gisli H. Gudjonsson, The Psychological Characteristics of "False Confessors": A Study Among Icelandic Prison Inmates and Juvenile Offenders, 20 Person. Individ. Diff. 321, 324 (1996). n17. Leo and Ofshe argue that low Icelandic figures for false confessions would not translate to America because confessions there must be repeated in front of a judge. See Leo & Ofshe, supra note 9, at 562 n.22. However, this point is misplaced because the Icelandic studies asked whether prisoners claimed "to have made a false confession during a police interview." See, e.g., Gudjonsson & Sigurdsson, supra note 16, at 23 (emphasis added). The studies did not ask whether the false confession was repeated later in court. Leo and Ofshe also argue that, in contrast to American police, Icelandic police are not legally permitted to employ deception during interrogations. See Leo & Ofshe, supra note 9, at 562 n.22. This point leaves open the question of whether the Icelandic legal norm has been translated into a standard police practice. Moreover, even without such a norm in the United States, the scant available data suggest that most police interrogations here do not appear to involve deceptive tactics. See Richard A. Leo, Inside the Interrogation Room, 86 J. Crim. L. & Criminology 266, 278 (1996) (finding thirty percent of interrogations involved "confront[ing a] suspect with false evidence of guilt"). Finally, the extent to which this one difference is material is unclear, as Leo and Ofshe have seemingly concluded that the false confessions here do not stem from any single police tactic. See Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 Denv. U. L. Rev. 979, 1115 (1997) (stating that because false confessions come from the "improper use of interrogation as a whole, no single procedure can be proscribed and thereby adequately protect the innocent"). Rather than rejecting the Icelandic studies out of hand, the better approach is to use them cautiously to illuminate questions about false confessions, particularly in view of the dearth of available American data. n18. See Cassell, supra note 5, at 509 (discussing Paul G. Cassell & Bret Hayman, Police Interrogation in the 1990s: An Empirical Study of the Effects of Miranda, 43 UCLA L. Rev. 839 (1996)). n19. See id. at 508-09 (discussing Leo, supra note 17). See also White, supra note 13, at 2040 n.231. n20. See Cassell, supra note 5, at 509-10 (collecting studies from New Haven, Denver, Washington, and Pittsburgh, none of which mention a false confession). n21. See id. at 513. Page 30 22 Harv. J.L. & Pub. Pol'y 523, *603 n22. See C. Ronald Huff et al., Guilty Until Proven Innocent: Wrongful Conviction and Public Policy, 32 Crime & Delinq. 518, 523 (1986). n23. Leo & Ofshe, supra note 9, at 570. n24. See Jerome H. Skolnick & Richard A. Leo, The Ethics of Deceptive Interrogation, Crim. Just. Ethics, Winter/Spring 1992, at 10 (citing Huff et al., supra note 22). n25. To disparage my calculations, Leo and Ofshe quote a sentence from a letter from Professor Albert Alschuler to me that discussed a preliminary draft of the article. See Leo & Ofshe, supra note 9, at 571 (citing letter from Albert Alschuler to Paul Cassell (Aug. 29, 1997)). This seems odd to me, since the purpose of circulating drafts for comments is to obtain frank suggestions for revisions, not provide ammunition for critics. In any event, the quoted comments in that letter led directly to substantial revisions in my final published article and accordingly have no relevance to what appeared in print. Professor Alschuler has authorized me to report that he agrees that the point on which he was commenting was corrected in the final article. n26. See Cassell, supra note 5, at 520-21. n27. See id. at 538-52. n28. See Leo & Ofshe, supra note 9, at 571 ("According to Cassell, there are two scenarios that describe how Miranda harms the innocent."). n29. Id. at 573. n30. Paul G. Cassell, Miranda's Social Costs: An Empirical Reassessment, 90 Nw. U. L. Rev. 387, 486-98 (1996) [hereinafter Cassell, Miranda's Social Costs]; Paul G. Cassell, All Benefits, No Costs: The Grand Illusion of Miranda's Defenders, 90 Nw. U.L. Rev. 1084 (1996); Paul G. Cassell, Miranda's "Negligible" Effect on Law Enforcement: Some Skeptical Observations, 20 Harv. J.L. & Pub. Pol'y 327 (1997); Paul G. Cassell, The Costs of the Miranda Mandate: A Lesson in the Dangers of Inflexible, "Prophylactic" Supreme Court Inventions, 28 Ariz. St. L.J. 299 (1996). See also Raymond Atkins & Paul H. Rubin, The Impact of Changing Criminal Procedure on Crime Rates (Oct. 23, 1998) (unpublished manuscript on file with author) (finding Miranda increased crime rates by eleven percent). n31. Paul G. Cassell & Richard Fowles, Handcuffing the Cops?: A Thirty Year Perspective on Miranda's Harmful Effects on Law Enforcement, 50 Stan L. Rev. 1055 (1998) [hereinafter Cassell & Fowles, Handcuffing the Cops?]. For discussion of this article, compare John J. Donohue III, Did Miranda Diminish Police Effectiveness?, 50 Stan. L. Rev. 1147 (1998) with Paul G. Cassell & Richard Fowles, Falling Clearance Rates After Miranda: Coincidence or Consequence, 50 Stan. L. Rev. 1181 (1998) [hereinafter Cassell & Fowles, Falling Clearance Rates After Miranda]. Without responding in any way to the specific empirical analysis contained in these lengthy articles, Leo and Ofshe attempt the ad hominem dismissal that they are simply a "rhetorical weapon" in my "highly charged anti-Miranda crusade." Leo & Ofshe, supra note 9, at 575. The ideological nature of my empirical enterprise will come as something of a surprise to at least one of my co-authors - Professor Richard Fowles, who teaches in the Page 31 22 Harv. J.L. & Pub. Pol'y 523, *603 Economics Department at the University of Utah and describes himself a "neo-Marxist." n32. Richard A. Leo, The Impact of Miranda Revisited, 86 J. Crim. L. & Criminology 621, 677-78 (1996). n33. See Cassell & Fowles, Handcuffing the Cops?, supra note 31, at 1064, 1089-90; Donohue, supra note 31, at 1156; Cassell & Fowles, Falling Clearance Rates After Miranda, supra note 31, at 1188. n34. Samuel R. Gross, Loss of Innocence: Eyewitness Identification and Proof of Guilt, 16 J. Legal Stud. 395, 431 (1987) [hereinafter Gross, Eyewitness Identification]. n35. See id. n36. See id. n37. Ayre Rattner, Convicting the Innocent: When Justice Goes Wrong 45-49 & tbl. 8 (1983) (Ph.D. dissertation, Ohio State University) (on file with Univ. Microfilms Int'l). n38. Leo & Ofshe, supra note 9, at 573. n39. Edward Connors et al., Convicted by Juries, Exonerated by Science: Case Studies in the Use of DNA Evidence to Establish Innocence After Trial (1996). n40. See id. at 3 (reporting that the criteria for inclusion in the study included "subsequent exoneration ... resulting from post-trial exculpatory DNA tests"). n41. See Cassell, supra note 5, at 553-56. See also Leo & Ofshe, supra note 1, at 494 (recommending that all interrogations be videotaped in order to provide an exact record to assist the judge or jury in determining whether a confession is voluntary and reliable). n42. See Cassell, supra note 5, at 539-40 (collecting references). n43. Leo & Ofshe, supra note 9, at 576. n44. See, e.g., William J. Stuntz, The Uneasy Relationship Between Criminal Procedure and Criminal Justice, 107 Yale L.J. 1, 47 (1997) (stating that "it is possible that Miranda makes it both harder to get confessions from the guilty and easier to get them from the innocent"); Daniel Givelber, Meaningless Acquittals, Meaningful Convictions: Do We Reliably Acquit the Innocent?, 49 Rutgers L. Rev. 1317, 1379 (1997) (concluding that Miranda is of greatest immediate benefit to the guilty and that it thus "renders it harder for the truly innocent to have their voices heard and acknowledged"); Joseph D. Grano, Confessions, Truth and the Law 215 (1993) (asserting that instead of assisting courts in identifying situations where reliability might be of special concern, "Miranda has induced judges at all levels to split hairs over the meaning of black-letter rubrics"); Louis Michael Seidman, Brown and Miranda, 80 Cal. L. Rev. 673, 744-45 (1992) (stating that in Page 32 22 Harv. J.L. & Pub. Pol'y 523, *603 many cases Miranda has "served to insulate the resulting confessions from claims that they were coerced or involuntary"). Cf. Henry J. Friendly, The Fifth Amendment Tomorrow: The Case for Constitutional Change, 37 U. Cin. L. Rev. 671, 680-81 (privilege against self-incrimination deprives "[a] man in suspicious circumstances but not in fact guilty ... of official interrogation of another whom he knows to be the true culprit ...."). n45. Richard A. Leo, Miranda and the Problem of False Confessions, in The Miranda Debate: Law, Justice and Policing 271, 276 (1998). n46. Other commentators have also relied on Leo and Ofshe's cases as support for changes in rules governing police interrogation. See, e.g., White, supra note 13, at 2042-44. n47. Hugo Adam Bedau & Michael L. Radelet, Miscarriage of Justice in Potentially Capital Cases, 40 Stan. L. Rev. 21, 45 (1987). See also Edwin M. Borchard, Convicting the Innocent: Sixty-Five Actual Errors of Criminal Justice vi (1932); Gross, Eyewitness Identification, supra note 34, at 396; Samuel R. Gross, The Risks of Death: Why Erroneous Convictions are Common in Capital Cases, 44 Buff. L. Rev. 469, 475 (1996) [hereinafter Gross, Capital Cases]; Michael L. Radelet et al., Prisoners Released from Death Rows Since 1970 Because of Doubts About Their Guilt, 13 T.M. Cooley L. Rev. 907, 910 (1996). Cf. Givelber, supra note 44, at 1327 (noting narrowness of focusing on factually innocent). n48. One public policy reform for reducing pre-trial trauma to the innocent is improved speedy trial measures, a reform I have suggested elsewhere for other reasons. See Paul G. Cassell, Barbarians at the Gates? A Reply to Critics of the Victims' Rights Amendment, 1999 Utah L. Rev. (forthcoming) (advocating victims' right to a speedy trial); Paul G. Cassell, Balancing the Scales of Justice: The Case for and Effects of Utah's Victim's Rights Amendment, 1994 Utah L. Rev. 1373,1402-07 (same). n49. See, e.g., Borchard, supra note 47; Gross, Eyewitness Identification, supra note 34. n50. See Leo & Ofshe, supra note 1, at 495 (recommending enhanced judicial scrutiny of confessions); Richard J. Ofshe & Richard A. Leo, The Social Psychology of Police Interrogation: The Theory and Classification of True and False Confessions, 16 Stud. in L., Pol. & Soc'y 189, 213-15 (1997) (same). n51. Richard J. Ofshe, Inadvertent Hypnosis During Interrogation, 40 Int'l J. Clinical & Experimental Hypnosis 125, 151 (1992). n52. See infra notes 350-51 and accompanying text (discussing this approach). n53. Leo & Ofshe, supra note 1, at 432 n.9 (citing Bedau & Radelet, supra note 47). n54. See Stephen J. Markman & Paul G. Cassell, Protecting the Innocent: A Response to the Bedau-Radelet Study, 41 Stan. L. Rev. 121, 128-33 (1988) (discussing James Adams case). n55. See id. at 135 & n.72 (discussing books cited to prove the innocence of Everett Appelgate that in fact believed him guilty). Page 33 22 Harv. J.L. & Pub. Pol'y 523, *603 n56. See id. at 139 n.90 (discussing novel on Joe Hill in which protagonist is actually a guilty murderer). n57. See generally id. at 126-40. But see Hugo Adam Bedau & Michael L. Radelet, The Myth of Infallibility: A Reply to Markman and Cassell, 41 Stan. L. Rev. 161 (1988) (responding generally but not defending their analysis of any particular case). n58. Bedau and Radelet have attempted to deflect criticism of their earlier "subjective" determination of innocence with a new criterion they (misleadingly) call the "Markman-Cassell" criterion: whether a jury has acquitted a defendant or an appellate court has vacated a conviction. Radelet et al., supra note 47, at 914. Under this bizarre standard, one can talk freely about such "innocent" defendants as O.J. Simpson, Dan White (murderer of San Francisco supervisor Harvey Milk), Lemrick Nelson (murderer of Yankel Rosenbaum), and Stacey Koon and Laurence Powell (members of the Los Angeles Police Department who beat Rodney King and were acquitted by one jury). See generally George P. Fletcher, With Justice for Some: Victims' Rights in Criminal Trials (1995) (discussing the miscarriages of justice from these acquittals) Cf . Louis B. Schwartz, "Innocence" - A Dialogue with Professor Sundby, 41 Hastings L.J. 153, 154 (1989) (stating that "defendants are acquitted for many reasons, the least likely being innocence"). Statistics cited by Professor Givelber may suffer from a similar flaw. See Givelber, supra note 44, at 1336-38 (relying on raw data about number of acquittals to apparently reach conclusions about number of innocents convicted). n59. Leo & Ofshe, supra note 1, at 436. n60. For discussion of how these nine cases were identified, see infra note 392. n61. Leo & Ofshe, supra note 1, at 436. n62. Id. at 467. n63. See Fairchild v. Lockhart, 675 F. Supp. 469, 488 (E.D. Ark. 1987). n64. See id. at 489. n65. See id. n66. See id. at 488-89. Cf. California v. Hodari D., 499 U.S. 621, 623 n.9 (1991) (noting "proverbial common sense" that flight suggests guilt and citing Proverbs 28:1, "The wicked flee when no man pursueth"). n67. Fairchild, 675 F. Supp. at 474. n68. Leo & Ofshe, supra note 1, at 467 (citing Execution of Retarded Man is Fought, N.Y. Times, Aug. 31, 1995, at B12). n69. John Brummett, The Fairchild Issues Won't Die, Arkansas Democrat-Gazette, Aug. 29, 1995, at 7B. Page 34 22 Harv. J.L. & Pub. Pol'y 523, *603 Accord 1 Almanac of the Federal Judiciary, at 8th Cir. p. 9 (1997-2) (noting lawyers gave Eisele's legal skills "very high marks" and quoting one lawyer as saying "he was voted one of the best district judges in the United States"). n70. Fairchild v. Lockhart, 744 F. Supp. 1429, 1435 (E.D. Ark. 1989). n71. Fairchild, 675 F. Supp. at 491. The fact that the allegedly "coerced" confession was videotaped casts doubt on the coercion claim. Police bent on coercing a false confession would have found it far simpler to extract a written false confession than a videotaped one. See Fairchild, 744 F. Supp. at 1499. n72. 744 F. Supp. at 1437 (emphasis added). A court-appointed psychiatric expert later looked at the tape and likewise concluded Fairchild appeared to be "spontaneous, relaxed." Id. at 1483. n73. Fairchild v. Norris, 869 F. Supp. 672, 681 (E.D. Ark. 1993). n74. Id. at 674. n75. Leo & Ofshe, supra note 1, at 467 n.316. n76. See Fairchild v. Norris, 21 F.3d 799, 803 n.1 (8th Cir. 1994). Green was out of the state at the time of the crimes. See id. n77. See supra notes 64 and 65 and accompanying text. n78. Leo & Ofshe, supra note 1, at 467 n.317. n79. Fairchild, 675 F. Supp. at 474. n80. Id. See also Fairchild, 744 F. Supp. at 1500-01, 1505. n81. Fairchild, 744 F. Supp. at 1502. n82. Id. at 1501. See also id. at 1501-05 (reviewing evidence connected with the watch). n83. Leo and Ofshe, supra note 1, at 467 n.316. n84. Fairchild, 744 F. Supp. at 1503. n85. See id. at 1504. Page 35 22 Harv. J.L. & Pub. Pol'y 523, *603 n86. Id. at 1505. n87. Id. n88. See Leo & Ofshe, supra note 1, at 467 n.314 (citing forensic evaluation by Professor Ruth Luckasson and Dennis W. Keyes concluding Fairchild had an IQ in low 60's). n89. Fairchild, 744 F. Supp. at 1464. n90. Id. at 1467. n91. Id. at 1496. n92. See Fairchild v. Lockhart, 979 F.2d 636 (8th Cir. 1992). n93. See Fairchild v. Norris, 876 S.W.2d 588 (Ark. 1994). n94. 20/20, Confession at Gunpoint? (ABC television broadcast, Mar. 29, 1991). n95. The district court had previously found Fairchild willing to lie under oath, and indeed recited lengthy transcript passages that showed Fairchild had lied about the same issues he discussed with 20/20. See Fairchild, 744 F. Supp. at 1489. This finding did not appear in the 20/20 program. n96. See Fairchild, 979 F.2d at 638. Unfortunately, I have been unable to locate a copy of these findings, as they now appear to be missing from the district court. Quotations in this paragraph are from the court of appeals' description of the district court's findings. n97. Fairchild, 979 F.2d at 639. n98. Id. n99. Fairchild, 744 F. Supp. at 1498. n100. Fairchild, 979 F.2d at 639 n.4. n101. Id. at 639-40. n102. The program also seems to focus on whether the confession was coerced, not whether it was accurate - at one point suggesting that it was brutality against Fairchild that "cracked the Marjorie Mason case." 20/20, supra note 94. Page 36 22 Harv. J.L. & Pub. Pol'y 523, *603 n103. See Giarratano v. Commonwealth, 266 S.E.2d 94, 96 (Va. 1980). n104. Id. n105. Id. n106. See id. n107. Telephone Interview with Rick Whitt, Norfolk police, one of the original investigators on the case (Oct. 6, 1997) [hereinafter Whitt Interview]. n108. Giarratano, 266 S.E.2d at 97. The details of this confession are recounted at id. at 96-97. n109. See id. at 97. n110. See id. at 95. n111. When police arrived at the scene, access to the girl's bedroom "was barred by the presence ... of two vicious dogs, one being a German Shepherd." Id. Giarratano, however, was well known to the dogs, as he had lived in the apartment for several weeks. See Whitt Interview, supra note 107. n112. Giarratano also said that he threw the knife "out in the yard." Giarratano, 266 S.E.2d at 97. In more recent accounts, Giarratano has told his supporters that he confessed to throwing the knife out in the "backyard." See John Harris, A Widely Watched Date With Death; Virginia Inmate's Plea for Clemency Draws National Attention, Wash. Post, Feb. 17, 1991, at A1, cited in Leo & Ofshe, supra note 1, at 490 n.507. He then claims that the police failure to discover the knife proves his confessions were false. However, Giarratano's statement about the knife did not occur until February 7, two days after police had discovered the bodies. See Giarratano, 266 S.E.2d at 95-97. Thus, the search for evidence that immediately followed the discovery of the bodies was not made with the intent of discovering a knife "out in the yard" but was, instead, "routine neighborhood canvas." Whitt Interview, supra note 107. In any event, the failure to find a knife in an unspecified public area (with a large amount of foot traffic) is unremarkable and, it bears emphasizing, Giarratano's confession that he took the knife out of the apartment is entirely consistent with the fact that the knife was not found in the apartment. n113. Giarratano, 266 S.E.2d at 97. n114. Id. at 98. n115. Id. n116. Psychiatric Evaluation of Dr. C. Robert Showalter, Clinical Director, Forensic Psychiatry Clinic, University of Virginia Hospital at 8 (Aug. 2, 1979) [hereinafter Dr. Showalter Evaluation] (on file with author). Page 37 22 Harv. J.L. & Pub. Pol'y 523, *603 n117. See id. n118. Id. n119. Giarratano, 266 S.E.2d at 102. See also Dr. Showalter Evaluation, supra note 116, at 9-13. n120. See Giarratano, 266 S.E.2d at 101; Dr. Showalter Evaluation, supra note 116, at 13 (noting "chronic and acute" impairment of Giarratano's "capacity to control his behavior" which is reflected in his history of "assaultive behavior"). n121. Leo & Ofshe, supra note 1, at 489 (citing Pamela Overstreet, Rally Scheduled on Behalf of Condemned Killer, U.P.I. Regional News (Feb. 7, 1991)). n122. Giarratano, 266 S.E.2d at 99. See also Certificate of Analysis of June E. Browne to Norfolk Police Dept. at 2 (Mar. 16, 1979). While an additional seven pubic hairs were not consistent with Giarratano's sample, these hairs were recovered in areas that suggested that they might have come from the victim. See Certificate of Analysis, supra, at 2. No pubic hair samples from the victim were submitted for testing. Id. n123. See Leo & Ofshe, supra note 1, at 489 n.503 (citing June Arney, Joseph M. Giarratano; Bloody Boot Prints Led Him To Doubt His Own Confession, Virginian-Pilot & The Ledger Star, June 26, 1994, at A15 (quoting extensively from interview with Giarratano and his defense lawyers)). n124. See Giarratano, 266 S.E.2d at 98-99. Giarratano lived at the apartment until his precipitous flight to Florida after the murders, so it is unclear what relevance the fingerprints have in any event. n125. See id. at 99. Leo and Ofshe further claim that bloody shoeprints did not match Giarratano's boots. See Leo & Ofshe, supra note 1, at 489. But the prosecution never claimed that they did. The shoeprints were found in a pool of blood on the stairway landing, where the stabbing occurred. See Whitt Interview, supra note 107. Presumably the pool of blood spread out after Giarratano committed the murder, so the failure to find his shoeprints would not be surprising. Later a number of people stepped in the pool of blood, including the discoverer of the body, the emergency paramedics who responded, and various other persons responding to the double homicide. Id. n126. See Leo & Ofshe, supra note 1, at 489. n127. Letter from Pat W. Wojtkiewicz to Marle Deans, Mar. 19, 1989 (on file with author). n128. Report of Jeffrey T. Barth, Director, Neuropsychology Assessment Laboratory, on Joseph Giarratano, at 2, 3 (Dec. 1, 1986) (on file with author). n129. See Showalter Report, supra note 116, at 5. Page 38 22 Harv. J.L. & Pub. Pol'y 523, *603 n130. Giarratano himself wrote a law review article on the risk of executing the innocent. Joseph M. Giarratano, "To the Best of Our Knowledge, We have Never Been Wrong:" Fallibility vs. Finality in Capital Punishment, 100 Yale L.J. 1005 (1991). Omitted from the article (among other subjects) was any discussion of the adverse effect that spurious claims of innocence from prisoners like Giarratano have on those legitimately seeking to have their convictions overturned. n131. See State v. Ingram, No. 13613-9-II, slip op. at 1 (Wash. Ct. App. Jan. 22, 1992). n132. See id. at 2. n133. Id. at 3. n134. See id. n135. See id. n136. See id. at 3-4. See also Report of Proceedings at 617, Washington v. Ingram, No. 88-1-752-1 (Thurston Cty. Sup. Ct. 1990) [hereinafter Ingram Proceedings Tr.]. n137. Washington Clemency and Pardons Board, Tr. of Hearing on Paul Ingram (June 7, 1996) at 16 (statement of Gary Tabor quoting report from Ofshe). Accord Ingram Proceedings Tr., supra note 136, at 912 (quoting report from Ofshe that he had "no opinion if the daughters were raped here"); id. at 574 (testimony from Ofshe to this effect). n138. Ingram, No. 13613-9-II, slip op. at 3-4. n139. Id. at 4. n140. Ingram Proceedings Tr., supra note 136, at 901. n141. Leo & Ofshe, supra note 1, at 438. n142. Ingram Proceedings Tr., supra note 136, at 904. n143. Id. at 906. n144. Ingram, No. 13613-9-II, slip op. at 8 (apparently quoting trial judge). n145. Id. Page 39 22 Harv. J.L. & Pub. Pol'y 523, *603 n146. Order Granting Summary Judgment at 5, Ingram v. Riveland, No. C93-5399(WD) FDB (W.D. Wash. May 5, 1994). n147. Ingram Proceedings Tr., supra note 136, at 908. Ofshe's expert opinion in a similar case was also found to be unpersuasive. See Lynne Henderson, Suppressing Memory, 22 Law & Soc. Inq. 695, 727 (1997) (describing case of Lynn Crook). n148. Ingram Proceedings Tr., supra note 136, at 912. n149. Id. at 910-11. See also Daniel Brown et al., Memory, Trauma Treatment, and the Law 607 (1998) (describing another repressed memory case in which Ofshe's expert testimony was rejected on this ground). n150. See Ingram Proceedings Tr., supra note 136, at 911. n151. Id at 911-12. Ofshe's experiment was to tell Ingram that one of the "facts" of the case was that Ingram had made his son and daughter have sex together while Ingram looked on. See Ofshe, supra note 51, at 147. Ofshe conducted the experiment the first time he met Ingram when he "didn't know much about the case." Ingram Proceedings Tr., supra note 136, at 577. Both the daughters, however, had in fact alleged sexual abuse by their brothers and there were additional allegations that Ingram had allowed other people to have sex with his children in his presence. Id. at 579-80, 644. See also infra note 166 (noting recent disclosure by brother of sexual abuse by Ingram). Experts on the sexual abuse of children have also criticized Ofshe's experiment, concluding that the event Ofshe fed to Ingram "was never documented to be an unlikely occurrence." Karen A. Olio & William F. Cornell, Making Meaning Not Monsters: Reflections on the Delayed Memory Controversy, 3 J. Child Sexual Abuse 77, 87 (1994) (hereinafter Olio & Cornell, Making Meaning Not Monsters). See also Henderson, supra note 147, at 729 (raising the question of the ethics of such an experiment and concluding "Ofshe ought not to have used the very subject under investigation - that is, sexual abuse of Ingram's children"); Karen A. Olio & William F. Cornell, The Ingram Case: Pseudomemory or Pseudoscience?, Violence Update, June 1994, at 3 [hereinafter Olio & Cornell, The Ingram Case] (raising questions about the experiment). n152. Ingram Proceedings Tr., supra note 136, at 912. Several leading experts on memory have also concluded that Ofshe made the "logical error" of simply assuming that whatever followed from the hypnosis "must be false." Brown et al., supra note 149, at 411. In general, Ofshe has been charged with trafficking "in a bizarre misreading of critical concepts." Joan C. Golston, A False Memory Syndrome Conference: Activist Accused and Their Professional Allies Talk About Science, Law and Family Reconciliation, 5 Treating Abuse Today 24, 26 (Jan. 1995). n153. Ingram Proceedings Tr., supra note 136, at 912-13. According to an account of the conversation, Ofshe strongly suggested to Ingram that the confession might be fantasies in the following terms: Ofshe: So what you are talking about, what is in those books may very well be nothing more than the same kinds of fantasies that I think are going on in this case. Ingram: But how do you prove that they are fantasies? Page 40 22 Harv. J.L. & Pub. Pol'y 523, *603 Ofshe: Nobody can prove that they are not. State v. Quattrocchi, No. P92-3759, at 947 (Providence Sup. Ct. Apr. 1, 1998) (prosecutor quoting transcript during cross-examination of Richard Ofshe). n154. Ingram Proceedings Tr., supra note 136, at 917. n155. Id. at 917-18. n156. Id. at 902. n157. Id. at 902. n158. Id. at 906. See generally Ofshe, supra note 51 (providing other examples). n159. Ingram Proceedings Tr., supra note 136, at 906. n160. See id. n161. Id. at 907. The judge's finding that Ingram "did it" demonstrates that the plea was upheld not on the narrow procedural grounds that he had failed to establish a basis for withdrawal, but rather because it was true. Indeed, the trial judge was required to consider the supporting evidence for the plea under Washington law, which allows withdrawal of pleas on grounds of "manifest injustice." State v. Taylor, 521 P.2d 699, 700-01 (Wash. 1974). Insufficient evidence to support a plea constitutes manifest injustice. See, e.g., State v. Zumwalt, 901 P.2d 319, 322-24 (Wash. Ct. App. 1995) (reversing trial judge and allowing withdrawal of plea because of insufficient factual basis); State v. D.T.M., 896 P.2d 108, 110-11 (Wash. Ct. App. 1995) (same). n162. See State v. Ingram, No. 13613-9-II (Wash. Ct. App. Jan. 22, 1992). n163. See Ingram v. Riveland, No. C93-599WD (W.D. Wash. May 5, 1994). n164. See Ingram v. Riveland, No. 94-35627, slip op. at 2 (9th Cir. June 26, 1995) (concluding "most of the evidence strongly supports the finding that Ingram was not coerced" into pleading guilty). n165. Ofshe, supra note 51, at 151. Even if Ofshe's article developed the argument for Ingram's innocence, the article itself has encountered a fair amount of criticism from experts on child sexual assault that, in my view, more than a few neutral observers would find persuasive. See, e.g., Brown et al., supra note 149, at 396 (calling the article a "questionable source"); Olio & Cornell, Making Meaning Not Monsters, supra note 151, at 87 (concluding Ofshe's "claims that Ingram's confessions were based on pseudomemories seem doubtful"); Robert M. Reece, Making Meaning - A Pediatrician's View, 3 J. Child Sexual Abuse 119 (1994) (defending Olio and Cornell after attack by Ofshe); Karen A. Olio, Truth in Memory: Comments on Elizabeth Loftus's 'Reality of Page 41 22 Harv. J.L. & Pub. Pol'y 523, *603 Repressed Memory', Am. Psychologist, May 1994, at 442 (concluding Ofshe's experiment was flawed); Karen A. Olio & William F. Cornell, The Facade of Scientific Documentation: A Case Study of Richard Ofshe's Analysis of the Paul Ingram Case (unpublished manuscript, on file with the Harvard Journal of Law & Public Policy) (concluding that Ofshe's paper "lacks empirical data and [contains] serious methodological flaws"); Olio & Cornell, The Ingram Case, supra note 151, at 5 (concluding "Ofshe's findings in the Ingram case are, at best, speculative"). Cf. Verbatim Report of Proceedings at 18, 27, Crook v. Murphy, No. 91-2-0011-2-5 (Super. Ct. Benton County, Washington, Mar. 4, 1994) (trial court finds in an incest case that Ofshe "resolved at the outset to find a macabre scheme of memories progressing toward satanic cult ritual and then creates" such a scheme and that Ofshe's credibility was "limited by his stridency"). Leo and Ofshe's other source about the case is the book Lawrence Wright, Remembering Satan (1994) (based on two New Yorker articles). It is not at all clear why Wright's book, written to provide a lively account of the events, should be preferred over impartial judicial findings. For example, a professor at the Harvard Medical School recounted that when Wright contacted her, his "mind was made up"; she ended the conversation wondering "whether quaint ethical principles like accuracy and impartiality had become obsolete." Judith Herman, Presuming to Know the Truth, 48 Nieman Reports 43, 44 (1994). See also infra note 167 (noting Wright's efforts to obtain clemency for Ingram). Also, to make its storyline more believable, the book inexplicably omits any discussion of the trial court findings concerning the incredibility of Ingram. See Wright, supra note 165, at 187-88 (noting that Ingram filed a motion to withdraw plea, but not discussing hearings and ultimate rejection of that motion). n166. Washington Board of Clemency and Pardons, Tr. of Hearing on Paul Ingram (June 7, 1996) at 18-19 (testimony from Chad Ingram concerning physical and sexual abuse over an extended period of time by his father). Ofshe instantly reached the diagnosis (without ever talking to Chad Ingram) that the charges were attributable not to abuse but to a feeling "there must be some reason why my life is a disaster." Cf. American Psychiatric Association, The Principles of Medical Ethics: With Annotations Especially Applicable to Psychiatry 9 (1998 ed.) ("on occasion psychiatrists are asked for an opinion about an individual who is in the light of public attention ... It is unethical for a psychiatrist to offer a professional opinion unless he/she has conducted an examination and has been granted proper authorization for such a statement"). n167. See Rachael Zimmerman, Son of Deputy Says He Was Sexually Abused: Dramatic Report in Testimony to Clemency Board, Seattle Post-Intelligencer, June 8, 1996, at B1 (noting testimony from Richard Ofshe and Lawrence Wright, who "cast off the typical role of journalist-as-objective-observer to testify"). n168. See Washington Board of Clemency and Pardons, Minutes, Dec. 13, 1996 (three-to-one vote to deny clemency). n169. The Lapointe case is not only discussed by Leo and Ofshe, but also forms the basis for the introduction to Welsh S. White, False Confessions and the Constitution: Safeguards Against Untrustworthy Confessions, 32 Harv. C.R.-C.L. L. Rev. 105, 105-06 (1997). White, however, concedes that "experts disagree" about whether the Lapointe case was a miscarriage of justice. Id. at 131. n170. See Leo & Ofshe, supra note 1, at 459 n.227 (citing Donald S. Connery, Convicting the Innocent: The Struggle of a Murder, A False Confession, and the Struggle to Free a "Wrong Man" ix-xii (1996)). The better part of the tract is a transcript of a day-long public forum designed to "learn how we can help Richard better." Connery, supra note 170, at 60. It forum featured Professor Ofshe as one of the speakers. Id. at 95-108. n171. Leo & Ofshe, supra note 1, at 459. Page 42 22 Harv. J.L. & Pub. Pol'y 523, *603 n172. State v. Lapointe, 678 A.2d 942, 945-46 (Conn. 1996). As Mike Wallace reported in a Sixty Minutes story about the case, Lapointe had Type A blood, which "narrowed it down to 30 percent of the nation's population." Mike Wallace reported, but did not discuss, Lapointe's secretor status or lack of sperm. Sixty Minutes: Did He Do It?, (CBS television broadcast, June 30, 1996). Instead, the story deceptively said that police had unsuccessfully "tried to narrow it further." Id. At trial, Lapointe apparently offered an elaborate theory about how the fire might have "cooked" the semen, removing all traces of sperm. The jury, of course, was not required to accept all the factual premises of this theory. n173. See id. at 945 nn.5-6. n174. See id. at 945. n175. Id. at 946. n176. Id. n177. See id. n178. Id. n179. See id. at 948, 955. In another instance of trying to cover up for her husband, Lapointe's wife falsely told police early in the investigation, in the presence of Lapointe, that his blood was Type O. See id. at 946. n180. Id. at 955. n181. Id. at 949. n182. See id. at 949. n183. See id. at 949-50. n184. See id. at 952 n.23, 944. n185. See id. at 952 n.23, 945. n186. See State's Exhibit 18, Autopsy Report of Dr. Arkady Katsnelson at 5, Connecticut v. Lapointe, No. CR89-107933 (Sup. Ct. Hartford 1992) (on file with author). n187. See Lapointe, 678 A.2d at 951. Page 43 22 Harv. J.L. & Pub. Pol'y 523, *603 n188. See id. at 952 n.23, 944. n189. Leo & Ofshe, supra note 1, at 461. n190. Connecticut v. Lapointe, No. CR89-107933 (Sup. Ct. Hartford May 7, 1992), trial trans. at 69, 118 [hereinafter Lapointe Trial Tr.]. n191. See State's Exhibit 18, Autopsy Report of Dr. Arkady Katsnelson, Connecticut v. Lapointe, No. CR89-107933 (Sup. Ct. Hartford 1992). n192. See Lapointe Trial Tr., supra note 190, at 54-120. n193. Early in the investigation, one of the detectives who questioned Lapointe thought that a foreign object might have been involved, based on his reading of the autopsy report. See Lapointe Trial Tr., supra note 190, at 1312-13 (testimony of Detective Lombardo). However, contrary to Leo and Ofshe's suggestion that Lapointe was simply confessing to the police theory of the crime, the detective recorded Lapointe's confession to penile rape. n194. Leo & Ofshe, supra note 1, at 460. n195. See Lapointe, 678 A.2d at 944. n196. See id. at 945 n.9. n197. Statement of Richard Lapointe, July 4, 1989, at 2 (on file with author). n198. Leo and Ofshe note that a Sixty Minutes story about the case four years after the verdict relied on two witnesses who said they saw a man not matching Lapointe's description running from the apartment building where the crime was committed. Sixty Minutes said the "police discounted these eyewitnesses," but does not explain why. The description of the running man matched that of a man who was involved in a hit-and-run accident at the same time. The police investigated the possibility that he was involved in the murder but were able clear him through blood typing. Telephone Interview with Detective Paul Lombardo, Manchester Police Dept. (Oct. 22, 1997). In any event, a "running" man was not at all unusual at this location, where pedestrians had to move quickly to cross heavy traffic. See id. n199. See Steve Jensen, Detectives in Lapointe Murder Case Speak Out; Police Say Court Ruling Vindicates Them, Hartford Courant, July 21, 1996, at C1 (victim's daughter reports that because of all the "falsehoods about 'poor Richard' we haven't been able to relax for one minute"). n200. Leo & Ofshe, supra note 1, at 461. n201. For a description of the confessions, see Misskelley v. State, 915 S.W.2d 702 (Ark. 1996). The actual Page 44 22 Harv. J.L. & Pub. Pol'y 523, *603 confession transcripts are contained in 3 Abstract and Brief for Appellant Jessie Lloyd Misskelley, Jr., at 479-513, Misskelley v. State, No. CR 94-00848 (Ark. 1995) (vol. 3 at 479-513) [hereinafter Misskelley Trial Abstract]. n202. See Echols v. State, 936 S.W.2d 509 (Ark. 1996). n203. See Bruton v. United States, 391 U.S. 123, 126 (1968) (holding codefendant confession inadmissible against other defendants on hearsay grounds). n204. See Misskelley, 915 S.W.2d at 709; Echols, 936 S.W.2d at 518; Misskelley Trial Abstract, supra note 201, at 290. n205. See Echols, 936 S.W.2d at 518, 519-20. In at least one of the admissions, Baldwin said that Misskelley had "messed everything up." Echols v. State, No. CR-94-00928 (Ark. 1995) (vol. 1 at 380). n206. See Echols, 936 S.W.2d at 521. Ofshe has elsewhere apparently suggested that Damien Echols, too, was wrongfully convicted. See Richard Ofshe, "I'm Guilty if You Say So," in Connery, supra note 170, at 95, 101 (noting that police in the case "fastened on a young man who, unfortunately, called himself Damien" (emphasis added)). n207. See Misskelley, 915 S.W.2d at 708. Some of the discrepancies appear to have been deliberate ploys by Misskelley to make parts of his confession seem less believable. See infra note 224 and accompanying text. n208. See Miskelley, 915 S.W.2d at 708. n209. See Misskelley Trial Abstract, supra note 201, at 485-86 (confession transcript). See also Misskelley, 915 S.W.2d at 708. n210. See Misskelley Trial Abstract, supra note 201, at 211-12; Misskelley, 915 S.W.2d at 708. Misskelley failed a polygraph test before confessing. Misskelley Trial Abstract, supra note 201, at 220-23. n211. Leo & Ofshe, supra note 1, at 462. See also Connery, supra note 170, at 102 (quoting Ofshe arguing "there's no way [Misskelley] could have committed this crime because he was thirty-five miles away ... at an amateur wrestling arena"). n212. Guy Reel et al., The Blood of Innocents: The True Story of Multiple Murder in West Memphis, Arkansas 91 (1995). n213. See Misskelley Trial Abstract, supra note 201, at 340-41, 342-43. n214. Reel et al., supra note 212, at 293. One of Leo and Ofshe's supporting articles for the alibi defense bears the revealing title "Friends Challenged on Misskelley Alibi." Leo & Ofshe, supra note 1, at 462 n.252 Page 45 22 Harv. J.L. & Pub. Pol'y 523, *603 (citing Glen Chase, Friends Challenged on Misskelley Alibi, Arkansas Democrat-Gazette, Feb. 1, 1994, at 1A). This article notes that prosecutors spent the day "picking apart" the alibi testimony. n215. A copy of Ofshe's testimony is available on the Free the West Memphis Three website (visited Jan. 13, 1999) <http://www.wm3.org/html/confession<uscore>analysis.html> [hereinafter Ofshe's Misskelley Testimony], apparently organized by death-penalty opponents trying to block Echols' death sentence. For an interesting competing website with evidence and analysis supporting Misskelley's guilt, see the Midsouth Justice Page (visited Mar. 6, 1999) <http://www.angelfire.com/ar/necromancer/index.html>. n216. Reporters summarized parts of Misskelley's second confession as follows: When they saw the three boys come into the woods, they hid. Echols grabbed Michael Moore and, when Christopher and Steven came to his rescue 'hitting Damien, trying to help their friend,' Misskelley and Baldwin joined the fray. The cops reported that they asked Misskelley how the boys were kept under control. 'They were like puppies,' he told them. 'When you whoop a puppy and tell it to stay, it will.' In his flat drawl, Misskelley then described the removal of Chris Byers' sexual organs. 'Blood went everywhere,' he said. Reel et al., supra note 212, at 305-06. I rely on a secondary source for the description of this confession because the police reports describing it are apparently unavailable. n217. See Statement of Jessie Misskelley, Jr. at 1-5 (Feb. 17, 1994)(transcript on file with author). n218. See id. at 7, 9. This timing fits the crime. See Leo and Ofshe, supra note 1, at 461 (noting that the boys disappeared some time after 5:30 p.m. and arguing that Misskelley's initial confession was at odds with this fact). n219. See Statement of Jessie Misskelley, Jr., supra note 217, at 12. n220. See id. at 12-13. n221. See id. at 13-14. n222. Id. at 20. n223. See id. at 24. n224. Id. at 23. Leo and Ofshe rely heavily on the discrepancy between Misskelley's statement about rope and the shoelaces found at the crime scene to conclude Misskelley is innocent. See Leo and Ofshe, supra note 1, at 461-62. n225. See Statement of Jessie Misskelley, Jr., supra note 217, at 8, 26-27. Page 46 22 Harv. J.L. & Pub. Pol'y 523, *603 n226. See Reel et al., supra note 212, at 315-16 (recounting statement by prosecutor Brent Davis). n227. See People v. Page, 2 Cal.Rptr.2d 898, 899 (Cal. App. 1991). n228. Id. at 900. n229. Id. n230. See id. at 901. n231. Leo and Ofshe claim that Page's later confession was disproven by the failure of the search dogs to pick up a trail of blood near the area during the search the day after the murder. See Leo & Ofshe, supra note 1, at 456. In fact, however, while one dog did not alert to the area, another dog did begin digging wildly at the ground at the site where Lee's body was ultimately discovered. The handler, however, misread the signal. See Page, 2 Cal.Rptr.2d at 901 & n.3. n232. Page, 2 Cal.Rptr.2d at 904. In another article, Leo and Ofshe report incorrectly that Page was "falsely told that he failed the polygraph." Richard J. Ofshe & Richard A. Leo, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 Denv. U. L. Rev. 979, 1041 (1997) (emphasis added). n233. Page, 2 Cal.Rptr.2d at 904. n234. See id. at 905. n235. Id. n236. Id. at 905-06. n237. See id. n238. See Leo & Ofshe, supra note 1, at 455-457 (citing Jack Page, A Question of Justice: A Father's Plea for Bradley Page, East Bay Express, Oct. 12, 1990, at 1; Anthony Pratkansis & Elliot Aronson, The Age of Propaganda: The Everyday Use and Abuse of Persuasion (1991); Melanie Thernstrom, The Dead Girl (1990)). n239. Leo & Ofshe, supra note 1, at 455. n240. Page confessed that he found Lee on the "east side of the Skyline [drive]." People v. Page, No. 81366, trial tr. at 7896 (Alameda Cty. Superior Ct. Apr. 8, 1988) (closing argument playing tape of confession) [hereinafter Page Trial Tr.]. In a later attempt to recant his confession, Page claimed to have simply "assumed" she was on the east side of the road. See Page, 2 Cal.Rptr.2d at 906. Page 47 22 Harv. J.L. & Pub. Pol'y 523, *603 n241. Leo and Ofshe claim that Page's confession "did not fit the known crime facts" because "it was not until days after the interrogation that the coroner determined that Lee had three large breaks at the base of the skull." Leo & Ofshe, supra note 1, at 455. In fact, police knew that the back of Lee's skull had been injured immediately when she was uncovered. What they did not know at that time - and thus did not know when they questioned Page - was that her nose bone and right eye socket had been fractured. The pathologist only later discovered these injuries. See Page Trial Tr., supra note 240, at 7480; Telephone Interview with Alameda County Asst. D.A. Ken Burr (Nov. 7, 1997). He determined that they could have been caused by someone administering a backhand blow to the nose. See Page, 2 Cal.Rptr.2d at 903. n242. Page, Cal.Rptr.2d at 915. Page confessed he had buried Lee's body. See Page Trial tr., supra note 240, at 7907. Later, Page argued that he simply "surmised that since the dogs did not find [Lee], she must have been buried." Page, 2 Cal.Rptr.2d. at 906. Leo and Ofshe also claim that "the blanket" found in Page's car did not link up with the confession to having sex on it. It would be astonishing to find evidence of the murder in Page's car five weeks later; he had plenty of time to dispose of it. The woman who was with Page on the day of the murder testified that a blanket was in Page's car the day of the murder, a blanket that could not be found five weeks later when the police searched the car. See Page Trial Tr., supra note 240, at 7801. n243. Page, 2 Cal.Rptr.2d at 915. n244. See Leo & Ofshe, supra note 1, at 456. n245. The program's producers declined an offer to read the trial transcript before airing the program. Telephone Interview with Alameda County Asst. D.A. Ken Burr (October 3, 1997) [hereinafter Burr Interview]. n246. Leo & Ofshe, supra note 1, at 457. n247. See Burr Interview, supra note 245. Ihde's other two killings could not have been Lee, because they involved a woman in Washington State (Ihde was in prison for this murder when he bragged about the two other killings) and a woman in the San Francisco area (a DNA match with semen found in the victim led to his successful capital prosecution for this crime). See id. n248. Among other differences, Ihde's victims were found naked from the waist down. See Burr Interview, supra note 245. Lee was found with her pants in place. See id. Page confessed to pulling them up after having sex with her body. See id. n249. Leo & Ofshe, supra note 1, at 457. n250. Page, 2 Cal.Rptr.2d at 907. See also Berkeley Police Department Press, Nov. 9, 1984 (information wanted poster with sketch of "van man") (in possession of author); Page Trial Tr., supra note 240, at 7425 (noting weight placed at 270 in an initial call). n251. While driving onto a freeway onramp, the witness saw a woman being pulled into a van. The witness's identification of the woman as Lee was made under very suggestive circumstances and included a description of the woman's clothing that was inconsistent with the clothing worn by Lee at the time of the Page 48 22 Harv. J.L. & Pub. Pol'y 523, *603 murder. The area where the woman was pulled was a steep slope, where a person might need help to get back up. See Page Trial Tr., supra note 240, at 7423-26. n252. Burr Interview, supra note 245. n253. See id. n254. See Leo & Ofshe, supra note 1, at 451. n255. Letter from Dennis Cadra to Gov. Ann Richards (Dec. 31, 1991) [hereinafter Cadra Letter]. The letter was the gist of two articles to the same effect, which in turn form the basis for Leo and Ofshe's description of the case. See Ginny Carroll, True Confessions - Or False?, Newsweek, Sept. 13, 1993, at 41; Howard Swindle, Shadows of a Doubt, Dallas Morning News, July 4, 1993, at A1. n256. Cadra Letter, supra note 255, at 2 (emphasis in original). n257. Id. at 6 (second emphasis added). n258. The transcript is no longer available at the Odessa courthouse or district attorney office. I am greatly indebted to Howard Swindle of the Dallas Morning News, who kindly agreed to provide me with his copy of the transcript. n259. Cadra Letter, supra note 255, at 5. n260. Trial Transcript at 102-04, 109, State v. Reyos, No. A-14,583 (June 8, 1983) [hereinafter Reyos Trial Tr.]. n261. Id. at 110 (emphasis added). n262. The letter attempts to exclude the 22nd because the evidence established "Mr. Reyos [was] already drunk and at a garage until after 6:00 on that date and was later arrested and spent the night of the 22nd in the Roswell jail." Cadra Letter, supra note 255, at 6. However, that would leave ample time for Reyos to meet Myers after leaving the garage around 6:00 p.m., hang out and drink with him for an hour, and then later be arrested. Also, while the defense called numerous witnesses to bolster other parts of Reyos' story, it apparently never (so far as appears in the trial transcript) called a witness to establish either the fact of the arrest that night or its time; instead it relied solely on Reyos' own testimony. See Reyos Trial Tr. (June 8), supra note 260, at 116. n263. See Reyos Trial Tr. (June 9), supra note 260, at 39-40, 44. On a different occasion, Reyos also told the police he had an alibi in Albuquerque. See id. at 20. n264. See Reyos Trial Tr. (June 8), supra note 260, at 21-28. Gonzalez also testified that Reyos was wearing green that day. See id. at 25. Reyos was wearing a green army jacket when arrested. See id. (June 7) at 73. Page 49 22 Harv. J.L. & Pub. Pol'y 523, *603 n265. If anything, the cross examination bolstered the testimony. While the defense attorney suggested the witness "didn't think much about [seeing Reyos]," Gonzales answered: "No, I told my husband about it. I told him right away ...." Id. (June 8) at 32. She also reported the matter to police several days later when the victim's death was discovered. See id. She was "absolutely certain" about the day. Id. at 36-37. n266. See Reyos Trial Tr. (June 8) at 87-88, id. (June 8) at 23-25. Cadra's letter conceded that Reyos lied under oath on this point. See Cadra Letter, supra note 255, at 3. n267. See Reyos Trial Tr. (June 8) at 25-26. n268. See Cadra Letter, supra note 255, at 3. n269. See Reyos Trial Tr. (June 9), supra note 260, at 34-43. n270. Reyos Trial Tr. (June 7), supra note 260, at 62-63. n271. Id. at 81. n272. Id. at 115. n273. Id. (June 8) at 131. n274. See id. (June 9) at 47-48. n275. See id. at 47. n276. Id. (June 9, vol. VI) at 8. n277. See id. (vol. VIII) at 56. n278. Id. at 59. n279. See Reyos v. State, No. 08-83-00266-CR, slip op. at 3 (Texas Ct. Apps. 1984) (unpublished). n280. See id. n281. Reyos, 08-82-00266-CR, slip op. at 6. n282. See Swindle, supra note 255, at 1A. The board includes a large number of members so that it has a broad representation of diverse viewpoints, not simply those inclined to credit government accounts. See Page 50 22 Harv. J.L. & Pub. Pol'y 523, *603 Telephone Interview with Dudley Sharp, Justice for All (Nov. 7, 1997). n283. See Order Denying Writ of Habeas Corpus, State v. Reyos, No. A-14,583-A (Jan. 27, 1995) (denying petition because "there are no controverted, previously unresolved facts material to the legality of [Reyos's] confinement"). n284. See Closing Argument, State v. Stangel, No. CC 96-1278 (Clatsop County Cir. Ct. 1996) (CourtTV videotape) [hereinafter Closing Argument Tape] (prosecutor summarizing evidence). n285. See Order Denying Motion to Suppress at 2, State v. Stangel, No. 96-1278 (Clatsop Cty. Cir. Ct. 1996) [hereinafter Findings of Fact]. Otherwise unattributed material in this description is taken from the court's findings of facts. n286. See Closing Argument Tape, supra note 284 (prosecutor summarizing evidence). n287. See State v. Stangel, No. CC 96-1278 (Clatsop County Cir. Ct. 1996), Suppression Hearing transcript at 23-24 [hereinafter Stangel Tr.]; Findings of Fact, supra note 285, at 2. n288. Leo & Ofshe, supra note 1, at 471. n289. See Findings of Fact, supra note 285, at 3-4. Where specifically noted, some additional material from the testimony of the officers has been added concerning the content of Stangel's statements, material that is presumably not included in the findings of fact because it was not critical to the suppression issue before the court. Cf. infra note 303 and accompanying text (noting court found officers to be credible). n290. See Findings of Fact, supra note 285, at 3. n291. Stangel Tr., supra note 287, at 41. n292. Id. n293. See id. at 41-42. n294. See id. at 43-45. n295. All the material in this paragraph is taken from the first recorded confession of Stangel, which was introduced as an exhibit at the suppression hearing and trial in State v. Stangel, No. CC 96-1278 (Clatsop County Cir. Ct. 1996). I am indebted to Richard Ofshe for supplying me with copies of Stangel's confessions and other related materials. n296. Id. Page 51 22 Harv. J.L. & Pub. Pol'y 523, *603 n297. Leo & Ofshe, supra note 1, at 471 (emphasis added). n298. The following facts are taken from the transcript of Stangel's second recorded confession, State v. Stangel, No. CC 96-1278 (Clatsop County Cir. Ct. 1996). n299. Id. at 10-11. n300. Stangel Tr., supra note 287, at 332. n301. See State v. Stangel, No. CC 96-1278, Ofshe Transcript at 83, 129-30 (Clatsop County Cir. Ct. 1996) (hereinafter cited as Stangel Ofshe Tr.). This is important because if Stangel was not truthful, then Ofshe's conclusion could be affected. n302. See Stangel Ofshe Tr., supra note 301, at 252, 257. n303. Id. at 367. n304. Id. at 368. n305. Stangel claimed to have written a note for Wahl on a Christmas ornament while waiting in the van for him to return, but Wahl's mother saw the ornament the next day without any inscription. See Closing Argument Tape, supra note 284. Stangel also claimed to have a disabling fear of heights, but was confronted at trial with a picture showing her and Wahl enjoying a roller coaster ride. Id. Cf. Stangel Tr., supra note 287, at 255 (concession by Ofshe that a picture showing Stangel "screaming in delight" on a roller coaster would be inconsistent with her professed extreme fear of heights). n306. See Leo & Ofshe, supra note 1, at 472 n.359 (noting exclusion of Stangel's polygraph test and wondering why the police officers were not required to take a polygraph). n307. See State v. Brown, 687 P.2d 751 (Or. 1984) (holding polygraph evidence is not admissible in civil or criminal trials). Cf. United States v. Scheffer, 118 S.Ct. 1261 (1998) (upholding military rule of evidence barring polygraph evidence). n308. See supra note 210 and accompanying text (Misskelley); supra note 232 and accompanying text (Page). n309. See Telephone Interviews with Josh Marquis, Clatsop County District Attorney (Sept. 29 & Nov. 6, 1997) (noting that the results were "equivocal," one of the four channels of the machine was broken, and Stangel's mother had a professional relationship with the police agency). n310. See Respondent's Mem. of Law at 55, Tankleff v. Senkowski, 96 CV 0507 (E.D.N.Y. 1997) [hereinafter Respondent's Mem. of Law] (citing trial tr. at 4246). Page 52 22 Harv. J.L. & Pub. Pol'y 523, *603 n311. Leo and Ofshe, supra note 1, at 458 (citing second-hand recounting of Petn. of Martin Tankleff for Writ of Habeas Corpus, Tankleff v. Senkowski, 96 CV 0507 (E.D.N.Y. 1997)). n312. Respondent's Mem. of Law, supra note 310, at 47 (citing trial tr. at 141, 168-69, 172-73, 202). See also id. at 47 (noting defendant's post-murder statement that once he inherited parent's money he would take friends out in a limousine) (citing trial tr. at 171-72, 194-95). n313. See id. at 59 (citing trial tr. at 4641-44, 46); id. at 48 (noting argument a few days before the murder) (citing trial tr. at 595-96, 613-15). n314. Tankleff, 96 CV 0507 (E.D.N.Y. 1997), slip op. at 4. n315. The father died about a month later, never having regained consciousness. n316. See Respondent's Mem. of Law, supra note 310, at 19, 24 (citing trial tr. at 3440-43, 3445-48, 3446-48, 4936-37; appx. at 1123-24, 1206-07, 1404-05, 1409, 1416-17, 564-68, 1281-82). n317. See People v. Tankleff, 606 N.Y.S.2d 707, 709 (N.Y. App. Div. 1993), aff'd, 646 N.E.2d 805 (N.Y. 1994). n318. 606 N.Y.S.2d at 709. n319. Id. Accord, Tankleff, 96 CV 0507 (E.D.N.Y. 1997), slip op. at 6 (finding Tankleff "made a full confession"). n320. See Respondent's Mem. of Law, supra note 310, at 35-38 (citing trial tr. at 2892-98). n321. See Tankleff, 606 N.Y.S.2d at 709. Accord Tankleff, 96 CV 0507 (E.D.N.Y. 1997), slip op. at 6. Cf. Leo & Ofshe, supra note 1, at 458 (suggesting the confession was obtained "after five and one-half hours of accusatory interrogation"). The appellate court on review also found that the reliability of the confession "was, if anything, enhanced" by the fictitious phone call ploy. Tankleff, 606 N.Y.S.2d at 710. See also Tankleff, 96 CV 0507 (E.D.N.Y. 1997), slip op. at 10 (finding "no credible evidence of physical or psychological coercion"); Tankleff v. Senkowsi, 135 F.3d 235, 245 (2nd Cir. 1998) (rejecting habeas claim that confession was coerced). n322. See Respondent's Mem. of Law at 60, supra note 310 (citing trial tr. at 4694-95). n323. The autopsy revealed that the mother's head had blunt force injuries inflicted before her throat was cut and she was stabbed (including "defensive wounds"). See People v. Tankleff, No. 1290-88 (Suffolk Cty. Ct. 1990), trial tr. at 3942-46, 3965, 4004-5. The father suffered from incised wounds, a depressed skull fracture, and incised neck wounds. See id. at 3993-4000. See also supra note 319-20 and accompanying text (describing the confession). Page 53 22 Harv. J.L. & Pub. Pol'y 523, *603 n324. See Leo & Ofshe, supra note 1, at 458. n325. Respondent's Mem. of Law, supra note 310, at 37 (citing trial tr. at 2892-98). n326. See id. at 49 (citing trial tr. at 733-78) (knife moved); id. at 58 (citing trial tr. at 4394-96) (barbell moved). n327. Leo & Ofshe, supra note 1, at 458. n328. People v. Tankleff, No. 1290-88 (Suffolk Cty. Ct. 1990), trial tr. at 4347. n329. Id. at 4359. n330. Id. at 3975. n331. Id. at 3998-40001. n332. See Leo & Ofshe, supra note 1, at 458. n333. See People v. Tankleff, No. 1290-88 (Suffolk Cty. Ct. 1990), trial tr. at 3955-56 (noting inability to narrow time of death any further than between 3:00 a.m. and approximately 6:00 a.m.). n334. Respondent's Mem. of Law, supra note 310, at 57 (noting testimony of psychiatrist Herbert Spiegel) (citing trial tr. 4305-07). n335. For example, in one other case - Juan Rivera - the evidence strongly suggests guilt. I am, however, deferring final judgment until a full transcript from the recent retrial becomes available. Rivera confessed to the brutal rape and murder of eleven-year-old babysitter Holly Staker. He was convicted in 1993. In 1996, the Illinois Court of Appeal ordered a retrial, while emphasizing "the evidence at trial was sufficient for the jury to conclude that defendant was guilty beyond a reasonable doubt." State v. Rivera, No. 2-94-0075, slip op. at 34 (Ill. Ct. App. 1996) (unpublished opinion). On retrial, the jury again convicted Rivera, and in December 1998 he was sentenced to life in prison. See Phuong Le, Baby-Sitter's Killer Given Life Term: Rivera's Retrial Ends with Same Sentence As Original Trial in '93, Chi. Trib., Dec. 11, 1998, at 8. Although I have not obtained court records from the second trial, the evidence from the first gave ample reason for accepting Rivera's confession. Leo and Ofshe claim the confession was coerced, but the trial judge found that "the whole background of Mr. Rivera indicates a compliance with wanting to speak with the police, being cooperative and ultimately giving incriminating statements." Brief of Plaintiff-Appellee at 45, Rivera (No. 2-94-0075) (quoting Supp. Rec. I, 117). Rivera's confession came only after Rivera gave several conflicting accounts of his whereabouts on the night of the murder. Rivera eventually broke down and admitted he had been in the apartment with Holly. He then claimed that they had consensual sex followed by a rebuff from him. She attacked him with a knife, and he stabbed her in the ensuing struggle. Leo and Ofshe indicate that "prior to signing two police-written confessions, Rivera began to hyperventilate and bang his head against the cell wall so violently that he was medicated." Leo & Ofshe, supra note 1, at 490 n.515. But it was after Rivera's first oral confession that he was found Page 54 22 Harv. J.L. & Pub. Pol'y 523, *603 hyperventilating and "tapping" his head against the cell wall. Rivera, No. 2-94-0075, slip op. at 5. Also, even though a nurse received authorization to medicate Rivera, the drugs were never administered because she found him stable and not in danger of hurting himself. See id. The trial court judge specifically found that this tapping "was not the result of any overbearing psychological coercion by the police, but it seems rather on the basis of the testimony by his own realization that his statements to the police were now putting himself into the very situation where he did not want to be." Brief of Plaintiff-Appellee at 49, Rivera (No. 2-94-0075) (quoting R. 820). Rivera's last confession, given the next morning, was "largely consistent with the known facts of the case." Rivera, No. 2-94-0075, slip op. at 5. Rivera confessed that when Holly taunted him for difficulty in having an erection, he got a knife from the kitchen and stabbed her. See State v. Rivera, trial tr. at 1564-69. Rivera also physically demonstrated various aspects of the commission of the offense. See id. at 1483-92. The police also learned new information about the crime. For example, Rivera told police that, after the murder, he took a mop that was outside the house and used it to knock in the bottom panel of the rear door of the apartment in order to make it look like a home invasion; he then wiped the mop clean of fingerprints with a rag and left the rag there. See id. at 1484-86. There is nothing in the records I have seen indicating that police were aware of the rag. It was only after Rivera's confession that police reexamined a videotape of the crime scene, noticing a rag precisely where Rivera had confessed to leaving it. See Telephone Interview with Mike Mermel, Lake County State Attorney's Office (Feb. 23, 1999). The evidence presented at the second trial was even stronger than that offered at the first. For example, the little girl Holly Staker was baby sitting, who was eight at the time of the retrial, identified Rivera as the murderer. See id. In addition, some ambiguous DNA evidence that Rivera had capitalized on at the first trial had, because of improvements in DNA testing by the time of the second trial, become irrelevant to his defense. See id. Finally, it is important to emphasize that the second trial, like the first, included defense testimony from a clinical psychologist that Rivera's confession was coerced. The jury, however, disagreed, again raising doubts about Leo and Ofshe's claim that the "overwhelming majority" of neutral observers would, after hearing all the evidence, agree with their views. Leo and Ofshe cite not court records but Chicago Tribune articles about the first trial. Shortly before trial the Tribune "scooped" other local papers with an "exclusive" interview with Rivera about his innocence. See Andrew Martin, Staker Suspect: I Never Touched That Girl, Chi. Trib., April 9, 1993, at 1. Coverage favorable to the defense followed, which Leo and Ofshe use. For example, Leo and Ofshe note that Rivera was wearing an electronic leg monitor showing he was home the night of the crime. See Leo & Ofshe, supra note 1, at 491 (citing Andrew Martin & Robert Enstad, Rivera Confession Coerced, Defense Says, Chi. Trib., Feb. 18, 1993, at 8). But ample trial testimony proved the system's unreliability. See Rivera, No. 2-94-0075, slip op. at 8. Two days after the murder the probation officer assigned to Rivera testified that his ankle bracelet was too loose, and a neighbor who saw Rivera testified that the bracelet was loose enough to be circumvented. See Brief of Defendant-Appellant at 18, Rivera (No. 2-94-0075) (citing R. 6185). In any event, Rivera's pre-confession versions of events was that he was at a party (or "hanging around" outside the house where the party was, depending on which version one chooses), not at his home where the monitor showed him being. After the murder, the county scrapped the monitoring system. See Andrew Buchanan, Return to Suspect Monitors Proposed; Waukegan Murder Ended System in '92, Chi. Trib., September 21, 1997, at 1. n336. See Richard A. Leo & Richard J. Ofshe, Missing the Forest for the Trees: A Response to Paul Cassell's "Balanced Approach" to the False Confession Problem, 74 Denv. U. L. Rev. 1135 (1997). n337. See supra note 199 and accompanying text (noting trauma to victim's family from claims of defendant's innocence). n338. Leo & Ofshe, supra note 1, at 436. Page 55 22 Harv. J.L. & Pub. Pol'y 523, *603 n339. Cf. Henderson, supra note 147, at 724 (reviewing book by Ofshe and concluding he "plays loose with the facts in many instances"). n340. Leo & Ofshe, supra note 1, at 435 n.15. n341. See supra note 238 and accompanying text (noting reliance on account of trial from Page's father). n342. See supra note 170 and accompanying text (noting reliance on "Friends of Richard Lapointe"). n343. See supra note 311 and accompanying text (noting reliance on Tankleff's habeas petition). n344. Paul Glastris, Are Press Standards Slipping?, U.S. News & World Rep., July 13, 1998, at 22. n345. Judith Herman, Presuming to Know the Truth, 48 Nieman Reports, Spr. 1994, at 43, 44-45 (quoting reporter Stephen Fried). n346. See supra notes 62-102 and accompanying text (Fairchild); supra notes 103-30 and accompanying text (Giarratano); supra notes 169-99 and accompanying text (Lapointe); supra notes 200-26 and accompanying text (Misskelley). n347. See Model Code of Professional Responsibility DR. 7-107 (1998). n348. Herman, supra note 345, at 43 (quoting columnist Randolph Ryan). n349. See id. See also Barbara Santee, More on "Making Monsters," Society, Nov. 1993, at 4. See generally Lee Madigan & Nancy C. Gamble, The Second Rape: Society's Continued Betrayal of the Victim (1989). n350. See Gross, Eyewitness Identification, supra note 34, at 412. n351. See Ofshe's Misskelly Testimony, supra note 215 ("[A] lot of studies are done on what are called disputed confessions as opposed to undisputed confessions, and the undisputed confessions are more important because it is known whether or not the confession was true of false."). n352. See supra notes 54-58 and accompanying text (discussing problems with Bedau-Radelet research). n353. See Linnet Myers, A Conviction Unravels in Tot Killings, Chi. Trib., Dec. 11, 1986, at 1 (reporting "prosecutors are still convinced that they had the right man"). The secondary sources available on this case also make clear, at the least, that Leo and Ofshe have miscategorized Burt as a "proven" innocent man. Prosecutors had persuaded a judge that Burt had fired the fatal shot in an effort to terrorize a witness in an upcoming trial. See Philip Wattley, Man Held in Murder of Boy, 2, Chi. Trib., Sept. 21, 1985, at 7. In reopening the case and entering a judgment of acquittal, the judge relied on ballistics tests indicating that a newly found gun "could Page 56 22 Harv. J.L. & Pub. Pol'y 523, *603 have been" the murder weapon. Rosalind Rossi, Convicted Man Free on New Facts in Boy's Death, Chi. Sun-Times, Dec. 11, 1986, at 7 (emphasis added). Cf. Michael L. Radelet et al., In Spite of Innocence: Erroneous Convictions in Capital Cases 292 (1992) (claiming that ballistics tests "showed" that the gun was used in the killing). This created the possibility that the mother of the victim, rather than Burt, was the killer (although it also created the possibility that Burt had arranged to have the gun hidden where it was found). The judge concluded that the gun "created enough doubt to acquit Burt." Id. Less than two years later, Burt was arrested for dealing drugs and sentenced to sixteen years in prison. See Gary Marx, Industrious Prisoners Make Money Doing Time, Chi. Trib., Aug. 7, 1994, at 1 (quoting Burt as admitting he had been "selling drugs to feed my family"). n354. See Andrew Fegelman, 12-Year Nightmare Ends for Murder Defendant, Chi. Trib., July 16, 1992, at 1. n355. See Bedau & Radelet, supra note 47, at 150-51. n356. See Connery, supra note 170, at 91-92. n357. See supra notes 254-83 and accompanying text. n358. See Peter Baker, Death-Row Inmate Gets Clemency, Wash. Post, Jan. 15, 1994, at A1. n359. Id. Washington is also serving a 30-year sentence for burglary and malicious wounding in another case. See id. n360. Id. n361. See Huff et al., Convicted but Innocent: Wrongful Conviction and Public Policy 127 (1996) (writing that Cox's confession was "conceded by all parties to be false"). n362. See William J. Booher, Wrongly Imprisoned Man Will Get $ 605,000, Indianapolis Star, Mar. 21, 1995, at C1 (reporting payment of damages to Jacobs for civil rights violations in circumstances suggesting innocence). n363. See Pete Shellem, Jailed Man Set Free After False Confession; Proof of Innocence Approved at Hearing, Harrisburg Patriot & Evening News, Jan. 9, 1993, at A1 (noting prosecutors supported release). n364. See Barry Siegel, A Question of Guilt, L.A. Times Mag., Sept. 1, 1996, at 15 (reporting prosecutors sought release of Pavlinac). n365. See Kevin Davis & Ardy Friedberg, Wrongly Convicted Man Enjoys Freedom From 'Hell,' Fort Lauderdale Sun-Sentinel, Jan. 16, 1993, at 1B. Page 57 22 Harv. J.L. & Pub. Pol'y 523, *603 n366. See Radelet et al., supra note 353, at 14-15 (writing prosecutors believed Reynolds innocent). n367. See Booher, supra note 362, at C1. n368. See Connors et al., supra note 39, at 73-74 (reporting prosecutor seeks pardon). n369. See Sharon Cohen, "Ringmasters" Unlock Truth, Free Man Who Confessed to Murder, L.A. Times, Mar. 31, 1996, at A1 (noting pardon from governor). But cf. id. (reporting prosecutor believes Wilson guilty). n370. I use the phrase "mentally retarded" reluctantly, but find that it tracks other writings in the field. n371. Jacobs and Smith are both "mildly" mentally retarded. See Booher, supra note 362, at C1. Kelly had an IQ of 69 and a history of mental illness. See Shellem, supra note 363, at A1. Purvis was a "diagnosed schizophrenic." Davis & Friedberg, supra note 365, at 1B. Reynolds had an IQ of 75 and a history of mild mental retardation. See Radelet et al., supra note 47, at 13. Vasquez was "borderline retarded." Conners et al., supra note 39, at 73. Wilson was retarded. See Cohen, supra note 369, at A1. While not apparently mentally handicapped, Pavlinac was described as a "confused" and "depressed" person. Siegel, supra note 364, at 15. n372. While I have been unable to locate court records on the Cox case, a second-hand account suggests that Cox decided to falsely confess for his own reasons and that the confession should therefore be characterized as a "suspect-induced" false confession rather than a "police-induced" false confession. See Cassell, supra note 5, at 518-20 (distinguishing between these terms). Cox had skipped bond in North Carolina and, when arrested in Ohio, was apparently facing a bounty hunter. Cox's wish to avoid returning to North Carolina (because he thought the bounty hunter might shoot him) apparently had a major role in leading to his decision to confess to a robbery/rape in Ohio. See Huff et al., supra note 361, at 131-33 (reporting that Cox thought "[a] robbery charge, with all the friendly help that he would get from [the police], did not sound too bad; the alternative was virtual kidnapping by a bounty hunter"). In addition, it is also likely that the trial court should have suppressed Cox's confession as "involuntary," because it seems to have been induced by specific promises from the police that rape charges would not be filed if he confessed to a rape/robbery. See id. at 132. Cf. Wayne R. LaFave & Jerold H. Israel, Criminal Procedure 6.2(c) (2d ed. 1992 & Supp. 1997) (concluding that "courts have often held that a confession is involuntary if made in response to a promise the result will be ... the dropping of some charges, or a certain reduction in the punishment defendant may receive"). n373. This conclusion assumes that the mentally retarded persons are not more likely to appear in a collection of undisputed wrongful convictions than other persons would be. This assumption seems reasonable, given that the mentally retarded may have more difficulty in orchestrating an investigation into their wrongful convictions than persons of more intelligence would have. Indeed, it may well be it is because the mentally retarded have such difficulties in organizing their defense at trial that they are so heavily represented among the wrongfully convicted. n374. President's Panel on Mental Retardation, Report of the Task Force on Law 33 (1963). n375. See, e.g., Steven Wisotsky, Miscarriages of Justice: Their Causes and Cures, 9 St. Thomas L. Rev. 547, 554-55 (1997); White, supra note 169, at 142; White, supra note 13, at 2044-46. Page 58 22 Harv. J.L. & Pub. Pol'y 523, *603 n376. James W. Ellis & Ruth A. Luckasson, Mentally Retarded Criminal Defendants, 53 Geo. Wash. L. Rev. 414, 445 (1985). n377. Fred E. Inbau et al., Criminal Interrogation and Confessions (3rd ed. 1986). n378. Ofshe & Leo, supra note 232, at 983. n379. See generally Cassell, supra note 5, at 507-24 (collecting empirical evidence on low frequency of false confessions). n380. Inbau et al, supra note 377, at 57. n381. Id. at 58. n382. Fred Inbau, Miranda's Immunization of Low Intelligence Offenders, 24 Prosecutor: J. Nat'l Dist. Att'ys Ass'n, Spr. 1991, at 10. Inbau also suggested that juveniles might be at special risk of false confession. See id. There is no way to test this assertion in the Leo-Ofshe collection, which apparently consists of exclusively adult cases. There are, however, some suggestions that juveniles are not uniquely at risk of police-induced false confessions. See Sigurdsson & Gudjonsson, supra note 16, at 322 (finding no false confessors among sample of 108 Icelandic juveniles); Graeme Richardson, A Study of Interrogative Suggestibility in an Adolescent Forensic Population 87 (1991) (unpublished M.Sc. Thesis, Univ. of Newcastle Upon Tyne) (finding 14 "false confessors" among sample of 60 British juveniles, all of whom confessed voluntarily to perform the "service" of covering for an older friend who might be subject to more severe legal penalties). n383. Inbau, supra note 382, at 10. n384. See Ellis & Luckasson, supra note 376, at 445-52. n385. See, e.g., White, supra note 169, at 142. The British have imposed special rules on questioning of the persons who appear to be "mentally ill or mentally handicapped or mentally incapable of understanding the significance of questions put to him." Code of Practice for the Detention, Treatment and Questioning of Persons by Police Officers, paras. 1.4. These rules require police to notify an "appropriate adult" when questioning such a person. Id. paras. 3.9 to 3.11. The appropriate adult then is present during questioning to "advise the person being interviewed and to observe whether or not the interview is being conducted properly and fairly." Id. para. 11.16. It would be interested to see whether such an approach could be transferred to the United States. Curiously enough, the biggest obstacle might be Miranda doctrine, since the "appropriate adult" would arguably be an attorney. Of course, once present an attorney would probably prevent questioning of any sort - no matter how proper and fair. n386. See Cassell, supra note 5, at 534-38. n387. See Commonwealth v. Daniels, 321 N.E.2d 822, 828 (Ma. 1975) (suggesting expert testimony on Page 59 22 Harv. J.L. & Pub. Pol'y 523, *603 mental retardation might be required at confession suppression hearings). n388. See E. Paul Holmes et. al., Learning Street Smarts for an Urban Setting, 20 Psychiatric Rehabilitation Journal 64 (1997). n389. See supra notes 89-90 and accompanying text (discussing psychiatrist who lost "her scientific objectivity and skepticism" to render an opinion that a defendant sentenced to death was mentally retarded and therefore ineligible for the death penalty). n390. See, e.g., supra notes 200-26 and accompanying text (noting involvement of Misskelley, who had low IQ, in crime). The available research, however, suggests that overall very low IQ's are disproportionately underrepresented in the criminal population. See James Q. Wilson & Richard j. Herrnstein, Crime and Human Nature 154 (1985). n391. For discussion of a more general response to the problem of wrongful convictions from false confessions, see Cassell, supra note 5, at 538-56. n392. This may be the appropriate point to explain briefly how I came to examine nine particular cases in detail. When preparing a response to the Leo and Ofshe's article, I had initially intended to use a single case to illustrate problems associated with their subjective determination of innocence. Since it could be argued that such an approach seized on isolated cases, I decided to expand my inquiry. To that end, I began collecting information on as many of the 29 cases of wrongful conviction in the Leo and Ofshe collection as possible, searching for primary court records or other information that would allow me to make an informed determination about the accuracy of the juries' verdicts. This turned out to be an extraordinary difficult and time consuming task - much more difficult, I should add, than the Leo and Ofshe approach of collating a few readily accessible newspaper articles about these cases. In most of the cases, I had to personally telephone individual prosecutors to obtain court records. In one case, original court records were no longer available from the court or prosecutor, but I found them by tracking down a newspaper reporter who had saved the old records. See supra note 258 (Reyos). As the months wore on, it became increasingly clear that obtaining primary records on all of the 29 cases in question would require what seemed to be an inordinate expenditure of time. For example, some of the cases involved allegations of wrongful conviction that were sourced in such a way as to make identification of primary sources difficult. In one case (Warney), the no-doubt busy prosecutor simply declined my generous invitation to have him copy the trial transcript and send it to me, and in another case (Washington) exploratory contacts with law enforcement sources indicated that access to primary records would be difficult. Four more cases were from 1980 or earlier (Carmen, Knapp, Parker, Reilly), and the prospects of finding original records nearly twenty years later seemed slim. Accordingly, I stopped further investigation after concluding that at least 9 of the 29 cases were misclassified by Leo and Ofshe and that a further 9 appeared to be undisputed wrongful convictions. At that point, having analyzed almost two-thirds of the Leo and Ofshe collection (the 9 misclassified cases and the 9 undisputed cases), I thought the analysis was sufficiently comprehensive to reach some conclusions. It should be recalled that the Leo and Ofshe collection is itself in no way a "random" sample of cases, as Leo and Ofshe admit. See Leo & Ofshe, supra note 1, at 436. n393. See, e.g., United States v. Hall, 93 F.3d 1337 (7th Cir. 1996) (overturning decision to exclude Ofshe testimony about false confessions), on remand, 974 F. Supp. 1198 (C.D. Ill. 1997), aff'd, <uscore><uscore> F.3d <uscore><uscore>, 1999 WL 16777 (7th Cir. 1999); United States v. Raposo, 1998 WL 879723, at *5 (S.D.N.Y. 1998) (admitting testimony about how persons with certain psychological profiles are more susceptible to Page 60 22 Harv. J.L. & Pub. Pol'y 523, *603 making false confessions); State v. Buechler, 572 N.W.2d 65, 73 (Neb. 1998) (reversing exclusion of expert testimony on effects of drug withdrawal and mental disorders on suggestibility); State v. Miller, 1997 WL 328740, at *6-8 (Wash. Ct. App. 1997) (reversing denial of funds to indigent defendant to obtain expert testimony by Ofshe as to why someone would give a false confession). See also United States v. Shay, 57 F.3d 126, 129, 133-34 (1st Cir. 1995) (finding district court erred in excluding expert testimony by a psychiatrist on "pseudologia fantastica" (also known as Munchhausen's Disease), a mental condition involving an extreme form of pathological lying); People v. Gilliam, 670 N.E.2d 606, 619 (Ill. 1996) (affirming trial court decision to allow expert testimony on the defendant's mental state but not on the circumstances surrounding the voluntariness of his confession); State v. Baldwin, 482 S.E.2d 1, 5 (N.C. Ct. App. 1997) (noting it was improper to exclude expert testimony about psychological factors on grounds that it was prohibited character evidence, but not otherwise ruling on admissibility of testimony), rev. granted, 485 S.E.2d 299 (N.C. 1997), rev. dismissed, 492 S.E.2d 354 (N.C. 1997); State v. Wells, 1994 WL 497745, at *1 (Ohio App. 1994) (noting trial court decision to allow testimony about defendant's mental traits in connection with confession); Lenormand v. State, 1998 WL 852849 (Tex. Ct. App. 1998) (noting trial court decision to allow expert testimony on psychological profile rendering defendant suggestible). n394. See, e.g., United States v. Griffin, 1997 WL 517002, at *3-4 (A.F. Ct. Crim. App. 1997) (affirming exclusion of expert testimony about coerced false confessions), rev. granted, 49 M.J. 150 (Ct. App. Armed Forces 1998); United States v. Koslowsky, 1995 WL 580889, at *3 (A.F. Ct. Crim. App. 1995) (affirming exclusion of expert testimony on false confessions); Beltran v. State, 700 So.2d 132, 133-34 (Fla. Dist. Ct. App. 1997) (affirming exclusion of expert testimony on false confessions and questioning whether expert assessment of confession involuntariness is ever admissible); Bullard v. State, 650 So.2d 631, 632 (Fla. Dist. Ct. App. 1995) (affirming exclusion of expert testimony on coercive effect of alleged threats during questioning); State v. MacDonald, 718 A.2d 195, 197-98 (Me. 1998) (finding unreliable testimony about children of alcoholics suffering from a syndrome that might cause them to falsely confess); Bixler v. State, 582 N.W.2d 252, 254-55 (Minn. 1998) (affirming exclusion of expert testimony on personal characteristics that might render an individual susceptible to pleasing authority figures); People v. Green, 1998 WL 894854, at *2 (N.Y. App. Div. 1998) (affirming exclusion of expert testimony concerning defendant's interrogative suggestibility that made him susceptible to providing a false confession); People v. Lea, 534 N.Y.S.2d 588, 590 (N.Y. App. Div. 1988) (affirming exclusion of expert testimony that defendant's personality rendered him deferential to the wishes of others and rendered confession involuntary); Kolb v. State, 930 P.2d 1238, 1241-42 (Wyo. 1996) (affirming exclusion of testimony on "False Confession Syndrome" on grounds or unreliability). See also State v. Monroe, 711 A.2d 878, 889 (N.H. 1998) (upholding trial court's decision to deny funds for an expert witness in the area of false confessions). n395. Many of the cases admitting expert testimony appear to involve suspects with peculiar mental disabilities. This fits more comfortably with traditional uses of expert testimony on psychiatric subjects than generalized evidence about false confessions, and is also consistent with the findings in this article that the false confession problem is concentrated among persons with mental problems. See supra notes 370-76 and accompanying text. n396. In addition to the cases cited in note 394, excluding expert testimony, see, for example, Hall, 974 F.Supp. at 1205; Wells, 1994 WL 497745, at *2; Lenormand, 1998 WL 852849. See also Miller 1997 WL 328740, at *7 (noting testimony limited to general discussion of false confession); Ofshe's Misskelley Testimony, supra note 215 (Ofshe concedes "I don't know that I've ever said that something was a false confession. I know I've testified as to whether something was coerced or not."). Page 61 22 Harv. J.L. & Pub. Pol'y 523, *603 n397. Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579, 590 (1993). n398. See id. at 594. n399. White, supra note 13, at 2043 (internal quotation omitted). n400. Saul M. Kassin, The Psychology of Confession Evidence, Am. Psychologist, Mar. 1997, at 221, 231. n401. Daubert, 509 U.S. at 594. n402. See, e.g., Fed. R. Evid. 703, cited in Daubert, 509 U.S. at 595. n403. See Leo & Ofshe, supra note 1, at 435 n.15 (representing that "where possible" their research relied on "primary" sources such as court records). n404. Ofshe's theories about coercive persuasion have also been rejected in other contexts. See United States v. Fishman, 743 F. Supp. 713, 719 (N.D. Cal. 1990) (stating that Ofshe's "theories regarding the coercive persuasion practiced by religious cults are not sufficiently established to be admitted in evidence in federal courts of law"). n405. See, e.g., Kassin, supra note 400, at 231 ("further research [on false confession] is sorely needed"). n406. Leo & Ofshe, supra note 1, at 495. n407. Id. at 438. See also Ofshe & Leo, supra note 232, at 1119 (arguing that "the reliability of a confession statement can usually be objectively determined by evaluating the fit between a post-admission narrative and the crime facts"). n408. See Leo & Ofshe, supra note 1, at 496 (urging courts to "insist on a minimal indicia of reliability before admitting confession statements into evidence"). See also Ofshe & Leo, supra note 232, at 1118-19; Leo & Ofshe, supra note 336, at 1142. n409. See Grano, supra note 44, at 59-118 (describing voluntariness doctrine and collecting references). n410. See Cassell, supra note 5, at 540-42. n411. Professor Welsh White reached a similar conclusion to mine in an article published just as my piece was nearing publication. See White, supra note 13, at 2028 ("the safeguards proposed by Ofshe and Leo do not provide a clear, administratively feasible means of evaluating the trustworthiness of specific confessions."). n412. See Cassell & Hayman, supra note 18, at 869 tbl. 4. Page 62 22 Harv. J.L. & Pub. Pol'y 523, *603 n413. See Leo, supra note 17, at 280 tbl. 7 (data derived by dividing "full confession" category by all three incriminating statement categories). The utility of this data in Leo's informative study is somewhat impaired by the broad definitions of the various incriminating statement categories. See Cassell & Hayman, supra note 18, at 929-30. n414. See Leo & Ofshe, supra note 1, at 495 (suggesting this answer). If so, a minor nomenclature point: it would be better to describe their proposal as assessing the post-confession narrative of a suspect, not the post-admission narrative. n415. Black's Law Dictionary 296 (6th ed. 1990) (quoted in Ofshe & Leo, supra note 232, at 991 n. 42). n416. Id. at 48 (quoted in Ofshe & Leo, supra note 232, at 991 n.42). n417. Miranda v. Arizona, 384 U.S. 436, 476 (1966). n418. See Leo & Ofshe, supra note 1, at 440. n419. See Model Penal Code 212.1. n420. George C. Thomas III, Telling Half-Truths, Legal Times, Aug. 12, 1996, at 20. n421. Inbau et al., supra note 377, at 90-91. n422. See id. n423. See supra notes 76-78 and accompanying text (Fairchild). In my previous article, I suggested that such "voluntary" false confessions will constitute a "significant proportion" of all false confessions, particularly because "common sense suggests that suspects will more often 'confess' for understandable reasons (such as protecting a loved one) than because police have somehow convinced them that they actually committed a crime." Cassell, supra note 5, at 519. Leo and Ofshe disagree, arguing that my suggestion is "clearly contradicted by the empirical research literature on the social psychology of false confessions." Leo & Ofshe, supra note 1, at 561 n.22 (citing their own article) Yet their view seems at odds with the statement they offered just one year earlier that "little is known about the frequency or risks of a miscarriage of justice attributable to voluntary unreliable confessions." Ofshe & Leo, supra note 50, at 210. Moreover, Leo and Ofshe's article sheds little light on the relative frequency of voluntary false confessions because it focuses on "police-induced false confessions." Leo & Ofshe, supra note 1, at 433. It does not focus on suspect-induced - i.e., "voluntary" confessions. For instance, they chose not to include even a single example of such a confession in their collection of cases. The decision to exclude such confessions is perhaps understandable, given their interest in focusing on police-induced false confessions. But this hardly constitutes a "clear contradiction" that such confessions are a sizeable proportion of the universe of false confessions. Finally, while citing their own articles, Leo and Ofshe do not dispute any of the other empirical research that finds "voluntary" false confessions to constitute a significant proportion of the false confession problem. See, e.g., Gudjonsson & Sigurdsson, supra note 16, at 23 (forty-eight percent of false confessions in Iceland stemmed from the desire to "protect a significant other"); Richardson, supra note 382, at 87 (fourteen British juveniles reported making a false Page 63 22 Harv. J.L. & Pub. Pol'y 523, *603 confession, all to protect a friend). n424. See supra notes 104-13 and accompanying text (Giarratano); supra notes 288-99 (Stangel). See also Ingram Proceedings Tr., supra note 136, at 584 (testimony from Ofshe agreeing that it is common for guilty persons to minimize their involvement); Leo & Ofshe, supra note 1, at 440 (providing illustration of true confession from child murderer unwilling to admit also to raping the child). n425. See supra note 115 and accompanying text (Giarratano). n426. See supra note 224 and accompanying text (Misskelley). n427. See supra note 182 and accompanying text (Lapointe); supra notes 233-36 and accompanying text (Page). n428. See supra note 112 and accompanying text (Giarratano). n429. See supra note 320 and accompanying text (Tankleff). n430. See supra notes 154-55 and accompanying text (Ingram). n431. See Paul G. Cassell, Balanced Approaches to the False Confession Problem: A Brief Comment on Ofshe, Leo, and Alschuler, 74 Denv. U. L. Rev. 1123, 1127 (1997). n432. Leo & Ofshe, supra note 336, at 1142. n433. Leo & Ofshe, supra note 1, at 495 (concluding that "in each of the recorded false confessions studied here, the account the suspect offered ... was significantly at odds with the crime facts"). n434. Leo & Ofshe, supra note 1, at 467 n.316. n435. See supra note 408 and accompanying text. n436. See supra notes 64-65, 77 and accompanying text. n437. Fairchild v. Lockhart, 744 F. Supp. 1429, 1505 (E.D. Ark. 1989). n438. See supra notes 83-87 and accompanying text. n439. See White, supra note 13, at 2028 (stating that if rigorously enforced, the Leo-Ofshe proposal "would be unfair to law enforcement because, in some cases, an otherwise trustworthy confession would be excluded Page 64 22 Harv. J.L. & Pub. Pol'y 523, *603 simply because the suspect ... refused to provide a sufficiently detailed explanation of the crime"). A related problem is that Leo and Ofshe have not explained what parts of the crime can be scrutinized for inconsistencies. For example, Jessie Misskelley confessed to running down a little boy and handing him back to his captors (Damien Echols and Jason Baldwin), who later killed the boy. See supra notes 201, 220 and accompanying text. Although nothing in the physical evidence contradicted Misskelley's confession to corralling the little boy, Leo and Ofshe nonetheless argue that Misskelley's description of the later murder by Echols and Baldwin was inaccurate. See Leo & Ofshe, supra note 1, at 461-62. But since Misskelley's criminal conduct was running the boy down, not actually killing him, there is no inconsistency between his confession and the "crime" facts of the charged crime. n440. See Roger Parloff, 1993: False Confessions, Am. Law., Dec. 1994, at 33 (giving illustration of false confessor providing description of jewelry taken from the victim that closely resembled missing jewelry). n441. See id. (providing illustrations of this phenomenon). n442. See Leo & Ofshe, supra note 1, at 438 n.24. n443. See, e.g., Booher, supra note 362, at C1 (reporting that, in a case involving the "undisputed" wrongful conviction of Ralph Jacobs, detective testified that false confessors "knew intimate details of the crime only a few detectives and the perpetrators of the crime would know"). n444. See White, supra note 13, at 2025 (stating that Leo and Ofshe "are not clear as to how strong a showing the government must make" that "the suspect was not likely to have learned of the independent evidence from some other source or to have recounted it as a result of a fortuitous guess"). n445. Inbau et al., supra note 377, at 147. n446. One promising direction for future research would be to examine how closely the post-admission narrative tracked the crime facts in cases of "undisputed" false confessions. Unfortunately, detailed information about interrogations in the "undisputed" cases of wrongful conviction examined here are not readily available, as most of the cases are simply described in brief newspaper accounts. n447. See Leo & Ofshe, supra note 1, at 476-77 (collecting illustrations). n448. The effect of interrogation on mentally retarded suspects may be an example. n449. See United States v. Hall, 93 F.3d 1337, 1341-44 (7th Cir. 1996) (reversing district court's exclusion of expert testimony by Professor Ofshe on the susceptibility of the defendant to false confessions) (convicted obtained on retrial). n450. See generally Cassell, supra note 5, at 552-56; Cassell, Miranda's Social Costs, supra note 30, at 486-98; Cassell & Fowles, Handcuffing the Cops?, supra note 31, at 1130. Page 65 22 Harv. J.L. & Pub. Pol'y 523, *603 n451. See Leo & Ofshe, supra note 1, at 491 (citing Saul M. Kassin & Lawrence S. Wrightsman, Coerced Confessions, Judicial Instructions, and Mock Juror Verdicts, 11 J. Applied Soc. Psychol. 489 (1981) [hereinafter Kassin & Wrightsman, Coerced Confessions]; Saul M. Kassin & Lawrence S. Wrightsman, Prior Confessions and Mock Juror Verdicts, 10 J. Applied Soc. Psychol. 133 (1980) [hereinafter Kassin & Wrightsman, Prior Confessions]. n452. See Kassin & Wrightsman, Prior Confessions, supra note 451, at 143-44; Kassin & Wrightsman, Coerced Confessions, supra note 451, at 497. n453. The studies are possibly flawed because they involved mock juror assessment of a "coerced" confession produced by a suggestion of leniency that was not clearly so coercive as to render the resulting confession involuntary. In the studies, the imaginary suspect was told that, if he confessed "he would be treated well during his detention and that the judge would surely be a lot easier on him - maybe even a suspended sentence." Kassin & Wrightsman, Prior Confessions, supra note 451, at 138. Some courts might view this single suggestion of better treatment as insufficient to produce an involuntary confession. See LaFave & Israel, supra note 372, at 297 (reporting that "the cases go both ways on the question of what the result should be when a confession has been obtained in response to a police assertion that cooperation would facilitate prompt release on bail or would mean that the defendant would fare better in subsequent proceedings"). n454. But cf. Kassin & Wrightsman, Prior Confessions, supra note 451, at 145 (cautioning that "the external validity of these experiments is limited by the fact that we assessed the judgments of individual, nondeliberating subjects"); Kassin & Wrightsman, Coerced Confessions, supra note 451, at 504 (noting same concern); Free v. Peters, 12 F.3d 700, 705-06 (7th Cir. 1994) (en banc) (rejecting mock jury research because there is little reason to think it can "offer insight into the ability of a real jury, which has spent days or weeks becoming familiar with the case and has had the benefit of oral presentations by witnesses, lawyers, and the judge, and which renders a verdict after discussion"). n455. Kassin & Wrightsman, Prior Confessions, supra note 451, at 139. See also Kassin & Wrightsman, Coerced Confessions, supra note 451, at 495. n456. Saul M. Kassin & Lawrence S. Wrightsman, Confession Evidence, in The Psychology of Evidence and Trial Procedure 67, 88 (Saul M. Kassin & Lawrence S. Wrightsman eds. 1985). n457. See Leo & Ofshe, supra note 1, at 483 tbl.B3. n458. Id. at 436. n459. See id. at 435-36. n460. The risk, such as it is, is also probably declining because of improvements in scientific technology, expanded recording of interrogations, and improved capability of defense attorneys to explain false confessions. See Cassell, supra note 5, at 525-26. n461. Lego v. Twomy, 404 U.S. 477, 485 (1972). See also Colorado v. Connelly, 479 U.S. 157, 164 (1986) Page 66 22 Harv. J.L. & Pub. Pol'y 523, *603 (concluding reliability determinations for confessions not set by the Constitution). n462. See Gacy v. Wellborn, 994 F.2d 305, 313 (7th Cir. 1993) ("Juries act in ways no reasonable person would act ... Yet for all of this, courts do not discard the premises of the jury system, postulates embedded in the Constitution and thus, within our legal system, unassailable."). n463. U.S. Const. Art. III, 2, cl. 3. See also U.S. Const. amend VI. n464. See generally Akhil Reed Amar, The Constitution and Criminal Procedure: First Principles 116-24 (1997) (developing this point). n465. Albert W. Alschuler, Constraint and Confession, 74 Denv. U. L. Rev. 957, 959 (1997).