Indiana Law Review Flipping a Coin: A Solution for the Inherent Unreliability of Eyewitness Identification Testimony Noah Clements* Introduction By most accounts, mistaken eyewitness identification is the leading cause of wrongful convictions in the United States. 1 This phenomenon is not new but seems to be a timeless aspect of criminal procedure. "Centuries of experience . . . have shown that convictions based solely on testimony that identifies a defendant previously unknown to the witness is highly suspect. Of all the various kinds of evidence it is the least reliable . . . ." 2 Justice Frankfurter once said: What is the worth ofidentification testimony even when uncontradicted? The identification of strangers is proverbially untrustworthy. The hazards of such testimony are established by a formidable number of instances in the records of English and American trials. These instances are recent—not due to the brutalities of ancient criminal procedure.3 The Supreme Court has placed the blame squarely on government suggestive nature when examining witnesses4 but has allowed even tainted identifications when the court is satisfied the identification is otherwise reliable.5 All proposals to improve the reliability ofeyewitness identifications have focused on removing * Associate, Sidley Austin L.L.P. B.A., University ofTexas; J.D., Duke University School of Law. The author would like to thank Robert Mosteller, Jim Coleman, Barak Richman, Ralf Michaels, Erica Beecher-Monas, Justice Scott Blister, John Karr, Alex Key, Scott Edson, Cassandra Robertson, Heather Holmes, Shane Shelley, and most of all, my lovely wife and children for their patience. 1. See, e.g., Edwin M.Borchard, Convictingthe Innocent, atxiii-xv (1932) (claiming forty-four out of a case study of sixty-five innocent defendants were convicted primarily on the basis ofmistaken identification); Innocence Project, Causes & Remedies ofWrongful Convictions, http://www.innocenceproject.org/ causes/index.php (last visited Jan. 8, 2007) [hereinafter Causes] (stating mistaken eyewitness identification played a major part in more than two-thirds of the project's first 130 post-conviction DNA exonerations). 2. Jackson v. Fogg, 589 F.2d 108, 112 (2d Cir. 1978) (granting habeas corpus relief to prisoner convicted solely on the basis of testimony of four eyewitnesses, after finding that the eyewitnesses' identifications were unreliable). 3. Felix Frankfurter, The Case of Sacco & Vanzetti 30 (1927), quoted in United States v. Wade, 388 U.S. 218, 228 (1967). 4. See, e.g., Wade, 388 U.S. at 228-29 (calling governmental suggestion a "major factor contributing to the high incidence of miscarriage ofjustice from mistaken identification"); Stovall v. Denno, 388 U.S. 293, 301-02 (1967) (holding that identification testimony should not be admitted if it "was so unnecessarily suggestive and conducive to irreparable mistaken identification that [the defendant] was denied due process of law"). 5. Manson v. Brathwaite, 432 U.S. 98, 1 14 (1977). 272 INDIANA LAW REVIEW [Vol. 40:27 1 the influential nature of governmental suggestion. For example, by using double blind sequential lineup procedures where lineup participants are shown one at a time and the officer conducting the lineup does not know which participant is the suspect. 6 However, governmental suggestion is not the only problem. Regardless if suggestion has played a part in the identification, eyewitness identification is inherently unreliable. 7 The only solution for this unreliability is to exclude the use of eyewitness identification testimony at trial unless the witness is acquainted with or otherwise familiar with the suspect. This Article examines the unreliability ofeyewitness identification testimony and proposes its exclusion. It argues that what at first may seem a radical idea, in fact, would make convictions much more reliable with a minimal negative impact on the criminal justice system. Part I discusses and provides actual examples of misidentification. Part II discusses the absence of any features by which courts and juries could use to judge the reliability of the identification in any particular case. Part HI concludes by showing that excluding identification testimony would not overly burden the criminal justice system. I. The Substantial Risk of Misidentification A. Witnesses Are Likely to Mistakenly Identify the Wrong Person The bulk of the research on eyewitness identification has been carried out since 1980, 8 well after a series of significant Supreme Court identification due process cases. 9 Thus, the Court has not had an opportunity to review new evidence on eyewitness identification reliability or to decide whether admitting inherently unreliable testimony that is as prejudicial as eyewitness testimony comports with due process. 10 The data does not paint a pretty picture. In one early study, seventy-three unwitting convenience store clerks were subjected to memorably bizarre behavior by "customers" in 146 tests. 11 Two hours later, in only 34.2% of the tests, were the clerks able to correctly identify the customer from a non-suggestive 6. Causes, supra note 1. 7. See Brian L. Cutler & Steven D. Penrod, Mistaken Identification: The Eyewitness, Psychology &theLaw 10-14 (1995) (summarizing the results of several respected identification studies). 8. Id. at 68. 9. E.g., Manson v. Brathwaite, 432 U.S. 98 (1977); Neil v. Biggers, 409 U.S. 188 (1972); Stovall v. Denno, 388 U.S. 293 (1967). 10. There was one study, however, conducted in 197 1 that merely measured subjects' ability to accurately identify a target face only eight minutes after seeing the target's picture. After ten seconds of exposure to the picture, there was 47% accuracy, and after thirty-two seconds of exposure, 75% accuracy. C. Ronald Huff et al., Convicted but Innocent 89 (1996) (citing Kenneth R. Laughery et al., Recognition ofHuman Faces, 55 J. APPLIED PSYCHOL. 477 (1971)). 11. Cutler & Penrod, supra note 7, at 11 (citing John C. Brigham et al., Accuracy of Eyewitness Identifications in a Field Setting, 42 J. PERSONALITY & SOC PSYCHOL. 673 (1982)). 2007] FLIPPING A COIN 273 photoarray; twenty-four hours later, the clerks could do so only 7.8% of the time. 12 In a similar study, where the identification time period of two or twenty-four hours was chosen at random and when the customer was in the photoarray, 41% of the clerks correctly identified him. 13 However, when the customer was not in the photoarray, 34% ofthe clerks mistakenly identified someone else. 14 The false identification rate when the customer was in the photoarray was not recorded. Later studies that used different time periods and situations arrived at similar results. 15 In one remarkable study, 30% of "witnesses" who had not actually witnessed an event, but who had engaged in discussions about it, later testified that they had recalled the incident and identified a person from a lineup as the culprit. 16 In summarizing these studies, Cutler and Penrod reported that the average rate of correct identifications in these simple experiments was 41.8%, while the rate of false identifications was 35.8%. 17 Witnesses who have received training for eyewitness situations do not appear to fare any better. 18 In one recent study of 509 Navy and Marine officers in survival training, subjects were interrogated for forty minutes in high-stress and low-stress simulations and asked to identify their interrogators twenty-four hours later, using various identification procedures. 19 Remarkably, these trained 12. Id. 13. Id. (citing Carol Krafka & Steven Penrod, Reinstatement ofContext in a Field Experiment on Eyewitness Identification, 49 J. PERSONALITY & Soc. PSYCHOL. 58 (1985)). 14. Id. 15. Id. at 1 1-12 (citing Melissa A. Pigott et al., A Field Study of the Relationship Between Quality ofEyewitnesses ' Descriptions and Identification Accuracy, 1 7 J. POLICE SCI. & ADMIN. 84 (1990) (finding four to five hours later, 47.8% correct identifications and 37.5% false identifications when culprit was not in photoarray) and Stephanie J. Platz & Harmon M. Hosch, Cross- Racial/Ethnic Eyewitness Identification: A Field Study, 18 J. APPLIED Soc. PSYCHOL. 972 (1988) (finding 44.2% correct identifications two hours later)). 16. MichaelSeng&William Carroll, Eyewitness Testimony: Strategies & Tactics § 2.43 (2d ed. 2003) (citing Henry B. Brown, An Experience in Identification Testimony, 25 J. Crim. Law & Criminology 621 (1934)). 17. Id. at 12. 18. In the Pigott study, supra note 15, 77% ofthe bank tellers had received training. CUTLER & PENROD, supra note 7, at 12. One study involving police trainees found a 51% false identification rate when presented with a "blank" photoarray. Elizabeth F. Loftus & James M. Doyle, Eyewitness Testimony: CrviL & Criminal § 4-8, at 85 (3d ed. 1997 & Supp. 2004) (citing John C. Yuille, Research and Teaching with Police: A Canadian Example, 33 INT'LREV. Applied Psychol. 5 (1984)). Police training was one of the factors given very heavy weight in Manson v. Brathwaite, 432 U.S. 98, 115 (1977), pertaining to the witness' degree of attention. 19. Abram Katz, U.S. Navy Study: Eyewitnesses Unreliable, NEW HAVEN REG., June 21, 2004, available at http://www.truthinjustice.org/navy-study.htm; see also Charles A. Morgan III et al., Accuracy of Eyewitness Memory For Persons Encountered During Exposure To Highly Intense Stress, 27 INT'L J.L. & PSYCHIATRY 265, 267 (2004). The subjects had spent an average of 4.2 years in the service prior to the training. Id. 274 INDIANA LAW REVIEW [Vol. 40:27 1 officers should have had plenty of opportunity to view their interrogators, often in very close proximity. 20 Yet, in a live lineup, subjects could correctly identify only 30% of the high- stress interrogators and 62% of the low-stress interrogators.21 Using a standard police-type photo spread but without elements of suggestion, only 32% of the high-stress subjects correctly identified their interrogators while 68% made incorrect identifications. 22 Using sequential photos, a technique often proposed to increase reliability by decreasing the influence of "relative similarity," 23 the high-stress group still had only 49% accuracy while the low-stress group's accuracy dropped to 76%. 24 Furthermore, as in previous studies, 25 there was absolutely no correlation between confidence or certainty of the eyewitness and accuracy of the identification in either the low-stress or the high-stress group. 26 B. The Risk ofMisidentification Is Not a Theoretical One There is no way to know for certain how many convictions are based on mistaken identification testimony. Estimates range as high as 5%. 27 One conservative study believes that as few (or as many) as 0.5% of convicted felons are actually innocent. 28 Accepting the conservative figure would mean that as 20. While many details of the study are classified, there is every indication that subjects were "man-handled" during the high stress interrogations. Katz, supra note 19. 21. Id. 22. Id. The subjects that experienced thirty minutes of low-stress interrogation, however, were able to accurately identify their interrogators 88% of the time from a photo spread. Id. 23. Gary L. Wells et al., Eyewitness Identification Procedures: RecommendationsforLineups and Photospreads, 22 LAW & HUM. BEHAV. 603, 617 (1998). 24. Katz, supra note 19. The report indicated that an identification was made in every case, showing that the high-stress group made more mistaken identifications than correct ones in every case, and even the low-stress group of trained military officers made an unacceptable number of mistaken identifications, ranging from 12% for the photo spread to 38% for a live lineup. Id. The low stress group's relatively low false identification rate should not be considered typical of eyewitnesses in general because these witnesses had an extraordinary opportunity to view the target: forty minutes in close proximity without physical stress. Rather, one should note that even under these circumstances, which should lead to very reliable identifications, only twenty-four hours after the event, the subjects made 38% false identifications in a live lineup and 25% false identifications in a sequential photo test. Morgan et al., supra note 19, at 272. 25. See, e.g., LOFTUS & DOYLE, supra note 18, § 3-12; SENG & CARROLL, supra note 16, § 2.4. 26. Katz, supra note 19. 27. LOFTUS & DOYLE, supra note 18, § 4-1, at 77. 28. HUFFET AL., supra note 10, at 59-62. The authors tried to be reasonably conservative. The authors took the research from an early study by Kalven and Zeisel, H. KALVEN & H. ZEISEL, The American Jury (1966), which found that in 4% of criminal cases studied, the jury convicted where the judge would have found the defendant not guilty. Because the jury could be expected to be more accurate than the judge at least some of the 2007] FLIPPING A COIN 275 many as 10,000 people a year are convicted of crimes they did not commit.29 It is impossible to know how many of these convictions are based on mistaken identification testimony, but one can extrapolate based on how many exonerations involved mistaken eyewitness identification testimony. Studies ofthe causes of wrongful convictions show that mistaken eyewitness identification testimony accounts for between one-half to two-thirds of these errors. 30 These studies encompass a wide range of crimes. 31 Notably, these mistakes were not confined to the cases where there was only one eyewitness and included instances where the witnesses had ample time to view the perpetrator. In one celebrated case of mistaken identification which occurred in 1979, seven store clerks were robbed at gunpoint by a "gentleman bandit" who pointed a chrome-plated handgun at them. 32 These seven clerks identified Father Bernard Pagano, a Roman Catholic priest, as the robber, and few doubted that he would have been convicted if another man had not confessed before the trial ended. 33 Another infamous example of misidentification involved the 1984 rape of Jennifer Thompson, a twenty-two-year-old college student with a 4.0 grade point average. 34 According to Thompson, during her ordeal she "studied every single time, perhaps as much as half the time, the authors halved this estimate and then assumed another half of these cases appealed successfully. Id. at 60. The authors combined this figure with their own survey that questioned Ohio prosecutors, judges and public defenders about their perception of wrongful conviction rates. Id. at 61. Because the majority of respondents believed that the rate was something less than 1%, the authors again halved the 1% rough estimate based on the Kalven & Zeisel study, and arrived at a 0.5% rate, which they believed was conservative. This figure does not account for cases where both thejudge and the jury incorrectly believed in the defendant's guilt. 29. Id. at 62. Peter Neufield and Barry Scheck believe that this number is probably much higher, based on the 25% DNA exoneration rate in sexual assault cases. EdwardConnors ET al., Nat'l Inst, of Justice & Dep't of Justice, Convicted by Juries, Exonerated by Science: Case Studies in the Use ofDNAEvidencetoEstablish Innocence After Trial, at xxviii-xxxi (1996). Because there does not seem to be anything inherent in sexual assault cases that would make eyewitnesses more prone to mistakes than in robberies or other serious crimes where the crucial proof is eyewitness identification, it naturally follows that the rate of mistaken identifications and convictions is similar to DNA exoneration cases. 30. See, e.g. , BORCHARD, supra note 1 , at xiii-xv (citing forty-four out of sixty-five innocent defendants equaling 67.7%); Huff et al., supra note 10, at 64 (finding of 205 wrongful convictions, including 54 from Borchard, that 52.3% were due to eyewitness misidentification); Causes, supra note 1 (stating more than two-thirds of first 130 innocent defendants exonerated by DNA). 31. Ofthe 205 cases studied in the Huff study, 45% involved murder or manslaughter, 30.5% involved robbery, and 12.5% involved rape. HUFFET AL., supra note 10, at 64. 32. Seng & Carroll, supra note 16, § 1.2. 33. Id. 34. Jennifer Thompson, / Was Certain, But I Was Wrong, N.Y. TIMES, June 18, 2000, available at http://tinyurl.com/4qbea. Cutler & Penrod's meta-analysis of studies involving more than 16,000 subjects shows that there is no correlation between intelligence and accuracy. Cutler 276 INDIANA LAW REVIEW [Vol. 40:27 1 detail on the rapist's face." 35 After the rape, she immediately went to the police department and worked on a composite sketch. 36 She then identified Ronald Cotton in a photo array a few days after the rape and still later in a lineup. 37 Cotton was convicted in January 1985.38 Ronald Cotton was not the man who raped Thompson.39 After an appeals court overturned the conviction because of improper exclusion of exonerating evidence, Cotton was granted a new trial in 1987, this time for two rapes because a "second victim decided that Cotton was her assailant."40 Although there was evidence that the actual rapist, Bobby Poole, confessed in prison, the judge refused to admit that evidence at trial. 41 When Bobby Poole was brought into court during a pre-trail hearing, Thompson was asked if she had ever seen him.42 She said, "I have never seen him in my life."43 But she was mistaken. In 1995, DNA tests proved that Bobby Poole was the rapist.44 By that time Ronald Cotton had spent more than ten years in prison. 45 Not only have people been falsely imprisoned, some have surely been executed based on false eyewitness identification. 46 One hundred and twenty- three death row prisoners have been exonerated since 1973. 47 Mistaken & PENROD, supra note 7, at 81-82. 35. Thompson, supra note 34. Research shows that Thompson's efforts to study and remember the details of her attacker's face should have led to more accurate memory retention and identification later. Seng & Carroll, supra note 16, § 2.33. The Supreme Court in Neil v. Biggers placed a great deal of emphasis on a similar degree of attention in assessing the reliability of the victim's identification under due process. 409 U.S. 188, 200 (1972). 36. Thompson, supra note 34. 37. Id. 38. Innocence Project, Case Profiles: Ronald Cotton, http://www.innocenceproject.org/case/ display_profile.php?id=06 (last visited Jan. 8, 2007) [hereinafter Ronald Cotton]. 39. Id. 40. Id. 41. Id. 42. Thompson, supra note 34. 43. Id. 44. Id. 45. Ronald Cotton, supra note 38. 46. See generally Michael L. Radelet & Hugo Adam Bedau, The Execution ofthe Innocent, 61 Law & Contemp. Probs. 105 (1998). 47. Death Penalty Information Center, Innocence: List of Those Freed from Death Row, http://www.deathpenaltyinfo.org/article.php?scid=6&did=110 (last visited Jan. 8, 2007). DPIC describes the criteria for inclusion on the exoneration list in the following manner: For Inclusion on DPIC's Innocence List: Defendants must have been convicted, sentenced to death and subsequently either—a) their conviction was overturned AND (i) they were acquitted at a re-trial or (ii) all charges were dropped; b) they were given an absolute pardon by the governor based on new evidence of innocence. Id. Some claim that the list is misleading because exonerees have not been proven innocent. See, e.g., Florida Commission on Capital Cases, Case Histories: A Review of 24 Individuals 2007] FLIPPING A COIN 277 eyewitness identification testimony played a major role in forty-nine of these cases. 48 These numbers should given a base of 1060 executions since 1976, 49 imply a more than 10% innocence rate and an extremely high 5% mistaken identification rate. While it is almost impossible to definitively prove innocence after a prisoner has been executed because of the lack of judicial review, the purported victim was later found alive in thirty-two cases between 1900 and 1985. 50 But states do not willingly participate in this inquiry — "some states candidly admit that their policy is never to confess error." 51 The role of mistaken identification in the death penalty innocence cases is much smaller than the two-thirds rate seen in the innocence project cases,52 however. This may be because eyewitness testimony plays a smaller part in death penalty cases than in other serious crimes. According to a Houston Chronicle study, which included a survey of Texas defense lawyers and prosecutors as well as an examination of several capital murder cases tried in Houston, "the vast majority of death penalty trials has [sic] no eyewitness Released from Death Row 5 (Sep. 10, 2002), available at http://www.floridacapitalcases. state.fl.us/Publications/innocentsproject.pdf ("A defendant is found guilty or not guilty, never innocent."). These claims ignore the substantial burdens required by appellate courts to set aside convictions, especially when the issue raised is one of innocence, and misplaces the subsequent burden of proof. 48. Author's hand-count based on DPIC list of exonerated prisoners. Death Penalty Information Center, Cases of Innocence 1973-Present, http://www.deathpenaltyinfo.org/article. php?scid=6&did=109 (last visited Jan. 8, 2007). The Center on Wrongful Convictions (CWC) performed an earlier, more thorough study of the first eighty-six legally exonerated death row prisoners since 1973, and found that forty-six, or 53.5%, involved faulty eyewitness testimony. Rob Warden, Center on Wrongful Convictions, How Mistaken and Perjured Eyewitness Identification Testimony Put 46 Innocent Americans on Death Row, May 2, 2001, http://www.law.northwestern.edu/depts/clinic/wrongful/Causes/ eyewitnessstudy01.htm. The CWC list included some freed prisoners as exonerated who did not appear on the DPIC list: for example, William Jent and Earnest Miller. 49. Death Penalty Information Center, Executions by Year, http://www.deathpenaltyinfo.org/ article.php?scid=8&did=146 (last visited Jan. 30, 2007). 50. David Margolick, 25 Wrongfully Executed in U.S., Study Finds, N.Y. TIMES, Nov. 14, 1985, atA19. 51. Edward Lazarus, Why States Don't Confess Error in Death Penalty Cases, CNN.COM, June 14, 2000, http://archives.cnn.eom/2000/LAW/06/columns/lazarus.saldano.6.12/index.html. 52. The Innocence Project is a non-profit legal clinic affiliated with the Benjamin N. Cardozo School ofLaw at Yeshiva University and created by Barry C. Scheck and Peter J. Neufeld in 1992. The project is a national litigation and public policy organization dedicated to exonerating wrongfully convicted people through DNA testing and reforming the criminal justice system to prevent future injustice. As a clinic, law students handle case work while supervised by a team of attorneys and clinic staff. Innocence Project, About the Innocence Project, http://www.innoncenceproject.org/about (last visited Feb. 8, 2007). 278 INDIANA LAW REVIEW [Vol. 40:27 1 testimony." 53 One recent execution where the conviction rested mainly on questionable eyewitness identification testimony was the case of Gary Graham. 54 The only witness who claimed to see Graham commit the murder in a supermarket parking lot "claimed to have seen a total stranger ... 9:30 at night, in the dark, from a distance of 40 feet away for two seconds."55 There were many problems with this identification. When presented with a photoarray containing Gary Graham' s picture two weeks after the murder, Bernadine Skillern said that Graham' s photo resembled the murderer but that the murderer's complexion was darker and his face thinner. 56 The next day, she picked Graham out of a lineup—he was the only subject in both the photoarray and the lineup. 57 Six other eyewitnesses failed to identify Graham as the murderer, who they said was much shorter than Graham. 58 Two witnesses who saw the killer in the supermarket checkout lane, including one who stood next to him, emphatically declared that Gary Graham was the wrong man. 59 Graham was executed on June 22, 2000.60 53 . Steve Brewer, Murder Trial Eyewitnesses 'A Luxury, ' HOUSTONCHRON. , June 22, 2000, at A17. Contrary to the findings of researchers, see infra notes 126-30 and accompanying text, the experienced capital defense lawyers surveyed stated that eyewitness testimony would actually create an opportunity for them to create reasonable doubt, because eyewitness testimony is so unreliable. Id. 54. CNN.COM, Execution Loomsfor Texas Inmate Convicted on Testimony ofOne Witness, June 15, 2000, http://archives.cnn.eom/2000/LAW/06/15/graham.execution/#l. 55. Id, 56. Interview with Larry Marshall, The New Abolitionist (Sept. 2000), available at http://www.nodeathpenalty.org/newab016/larryMarshall.html [hereinafter New Abolitionist]; see also http://www.quixote.org/ej/grip/reasonabledoubt/chart-Gary%20Graham.html (last visited Feb. 9, 2007). 57. Id. Nathan Sobel contends that the use of photographs before a lineup reduces identification reliability and raises fairness issues. NATHAN R. SOBEL, EYEWITNESS Identification: Legal & Practical Problems § 10.2 (2d ed. rev. 2004) 58. New Abolitionist, supra note 56. Graham's problems may have been compounded by poor representation by Ron Mock, who did not call two witnesses listed in the police report. Ron Mock was infamous in Texas for losing more death penalty cases than any other lawyer. See Rick Casey, Mock Gone, Not Mockery, HOUSTON CHRON., Dec. 3, 2004, at Bl. 59. New Abolitionist, supra note 56; see also Statement of Richard Burr, Texas Execution: a NewsHourwith Jim Lehrer Transcript, June 22, 2000, available at http://www.pbs.org/newshour/ bb/law/jan-june00/execution_6-22a.html. 60. Lou Jones, Gary Graham, FINAL EXPOSURE: PORTRAITS FROM Death Row (1996), http://www.fotojones.com/editorial/deathrow/inmates/graham.html. 2007] FLIPPING A COIN 279 II. Current Responses and Proposals Do Not Meaningfully Lessen the Risk of Misidentification A. The Neil v. Biggers Reliability Factors Do Not Sufficiently Ensure That Only Reliable Eyewitness Identifications Are Admitted In determining whether an identification should be excluded as violating Due Process, the Supreme Court has established a threshold question of whether the pre-trial identification procedures contained elements of suggestion. 61 The Court has singled out impermissible suggestion because it believes that suggestive procedures "increase the likelihood of misidentification." 62 The reliability of a particular eyewitness identification is only examined once impermissible suggestion is found, but suggestion per se does not violate a defendant's Due Process rights. 63 One Assistant U.S. Attorney has stated that allowing suggestive procedures "is analogous to creating one piece ofevidence, the identification that results from the procedure, and destroying another piece of evidence, the identification, or failure of identification, that would have resulted from a correctly conducted process." 64 Suggestive identification procedures may well be unfair in and of themselves, 65 but there is some indication that if the actual culprit is included in the lineup or photoarray, suggestion does not affect the accuracy of the identification. 66 However, if the actual culprit is not present, even subtle suggestion results in a misidentification rate of up to 90%, as compared to a misidentification rate of 45% if suggestive procedures are not used.67 Although there is some indication that the effect of suggestive procedures is less in real crimes than in staged ones, 68 even the non-suggestive false identification rates (45%) must be seen as unacceptable. Once suggestion is found, courts will determine whether the identification is nonetheless reliable by weighing the factors set out in Neil v. Biggers 69 against 61. Neil v. Biggers, 409 U.S. 188, 198 (1972). 62. Id. 63. Id. at 198-99; accord Manson v. Brathwaite, 432 U.S. 98, 1 13-14 (1977) (characterizing the Due Process right to identification procedures free from suggestion as merely an "evidentiary interest"). 64. Benjamin E. Rosenberg, Rethinking the Right to Due Process in Connection with Pretrial Identification Procedures: An Analysis and a Proposal, 79 Ky. L.J. 259, 292 (1991). 65. Id. 66. Cutler & Penrod, supra note 7, at 116-17 (citing Brian Cutler et al., The Reliability of Eyewitness Identifications: The Role ofSystem and Estimator Variables, 1 1 Law & HUM. BEHAV. 223 (1987)). 67. Id. at 117. 68. Id. at 119 (citing G. Kohnken & A. Maass, Eyewitness Testimony: False Alarms on Biased Instructions?, 73 J. APPLIED PSYCHOL. 363 (1988)). 69. 409 U.S. 188, 199-200(1972). 280 INDIANA LAW REVIEW [Vol. 40:27 1 »70 the "corrupting effect of the suggestive identification itself. These [factors] include the opportunity of the witness to view the criminal at the time of the crime, the witness' degree of attention, the accuracy of his prior description of the criminal, the level of certainty demonstrated at the confrontation, and the time between the crime and the confrontation. 71 The problem is that this reliability test "is not a satisfactory method of measuring reliability." 72 1. The Opportunity of the Witness to View the Criminal.—The amount of time that a witness views a criminal, the lighting conditions, and the proximity of the witness are all relevant to the reliability of an identification. Lighting conditions and the witness's distance from an event have a particularly great influence on a witness's ability to perceive objects and people. 73 However, even in perfect perception conditions where accuracy is highest, identifications are unreliable and mistakes are rampant. 74 The effect of the exposure time is a bit more complex. Curiously, the amount of time a witness views an event does not correlate with accuracy. Although "[c]ommon sense tells us that the amount of time available for viewing a perpetrator is positively associated with the witness's ability to subsequently identify him," 75 this turns out not to be the case. Instead, Cutler and Penrod's meta-analysis of research involving more than 16,000 subjects shows slightly decreasing marginal improvement in recognition as exposure time grows, and this improvement is relatively small. 76 This seems to contradict a 1971 study which showed that eight minutes after an event, eyewitness recognition rate ranged from 47% accuracy after ten seconds of exposure to 75% accuracy after thirty-two seconds of exposure.77 This study can be explained by the fact that most identifications occur more than eight minutes after an event. The difference in accuracy will decrease as more time elapses before the eyewitness is asked to identify the perpetrator. 78 The aforementioned Navy study may be particularly instructive in this 70. Manson v. Brathwaite, 432 U.S. 98, 1 14 (1977). 71. Id. 72. Rosenberg, supra note 64, at 276. 73. LOFTUS & DOYLE, supra note 18, § 2-4, at 15. 74. See supra Part I. 75. Cutler & Penrod, supra note 7, at 101 . 76. Id. The meta-analysis was an effort to normalize all of the disparate eyewitness studies available before 1995. They account for statistical variations by putting the studies together and arriving at a vast set of data for many different witnessing variables. 77. HUFFET AL., supra note 10, at 89 (citing K.R. Laughery et al., Recognition ofHuman Faces: Effects ofTarget Exposure Time, Target Position, Pose Position, and Type ofPhotograph, 55 J. Applied Psychol 477 (1971)). 78. Cutler & Penrod, supra note 7, at 101 . 2007] FLIPPING A COIN 28 1 regard. 79 The officers were subjected to interrogations of forty minutes, much longer than the short seconds or minutes that most crimes encompass. 80 Yet, the reliability of the identifications made by these highly trained military officers was not significantly greater than in other studies with shorter event durations. 81 Another problem is that courts will analyze the first factor on the basis of the witness' s recollection ofthe circumstances, which is subject to an overestimation bias. 82 In one study, witnesses overestimated the duration of a thirty-four-second event by a factor of two-and-one-half times. 83 The confidence bolstering effect of suggestion may have even greater effects that insulate this factor from meaningful review. 84 Recent research indicates that as the confidence of mistaken eyewitnesses is inflated, they report that the lighting was better, they were closer to the action, and the event took longer. 85 2. The Witness ' Degree ofAttention.—Research shows that efforts to study and remember the details of an event or facial features should lead to more accurate memory retention and identification later. 86 Courts have placed emphasis on witnesses' training in order to show that they would pay close attention to the person and event. 87 Courts also tend to believe that a person in danger will pay greater attention to detail than otherwise. 88 However, studies have shown that police officers specially trained in facial recognition are no better than the overall population at either making correct identifications or refraining frommaking false ones. 89 Furthermore, many studies have shown that violence or other stressful situations greatly decrease the ability of a witness to make accurate identifications.90 The Navy study is probably the most accurate one to date on the effects of stress on subsequent identifications. Most other researchers are reluctant to subject their witnesses to the kinds of stress experienced during violent crimes. The effect of stress in that study was 79. Katz, supra note 19. 80. Morgan et al., supra note 19, at 268. 8 1 . Compare id. at 272, with CUTLER & PENROD, supra note 7, at 1 1-12. 82. Rosenberg, supra note 64, at 278-79. 83. LOFTUS & DOYLE, supra note 18, § 2-5, at 16 (citing R. Buckhout, Eyewitness Identification and Psychology in the Courtroom, 4 CRIM. Def. 5 (1978)). 84. Id. § 1-3, at 1. 85. Id. (citing Gary L. Wells & Amy L. Bradfield, "Good, You Identified the Suspect": Feedback to Eyewitnesses Distorts Their Reports of the Witnessing Experience, 83 J. APPLIED Psychol. 360(1988)). 86. Cutler & Penrod, supra note 7, at 88; Seng & Carroll, supra note 16, § 2.33. 87. See, e.g., Manson v. Brathwaite, 432 U.S. 98, 115 (1977) ("[A]s a specially trained, assigned, and experienced officer, he could be expected to pay scrupulous attention to detail "). 88. See, e.g., Neil v. Biggers, 409 U.S. 188, 200 (1972) ("She was no casual observer, but rather the victim of one of the most personally humiliating of all crimes."). 89. Cutler & Penrod, supra note 7, at 86 (citing M. M. Woodhead et al., On Training People to Recognize Faces, 22 ERGONOMICS 333 (1979)). 90. See LOFTUS & DOYLE, supra note 18, § 2-7. 282 INDIANA LAW REVIEW [Vol. 40:27 1 enormous. 91 3. The Accuracy of a Prior Description.—There is no agreement whether identifications preceded by a detailed description objectively matching the person later identified are more accurate than other identifications. 92 This inconsistency extends to the question of a relationship between the ability to describe faces and accuracy in identifying faces. 93 Furthermore, even where Cutler and Penrod found that subjects who had a high ability to describe faces could make more accurate identifications, there was a very low correlation between consistent descriptions and accuracy. 94 Unfortunately, both courts and jurors value description consistency. 95 To a non-psychologist evaluating the conflicting results, the evidence should be considered inconclusive at best. 96 4. The Certainty ofEyewitness Identification.—If there is one thing that the research is virtually unanimous on, it is this: there is no correlation whatsoever between eyewitness certainty and accuracy. 97 ''Experienced judges understand that the most positive witness is not always the most reliable." 98 Even if confidence did correlate with accuracy, 99 the fact that confidence is malleable and often bolstered by police and prosecutors should raise doubt in the predictive power of the confidence. 100 Consequently, many courts have begun to place very 91. See supra notes 21-24 and accompanying text (showing false identification rates up to 68%). 92. See CUTLER & PENROD, supra note 7, at 93 (comparing Pigott et al.'s, supra note 15, 1990 study showing no significant correlation between description accuracy, completeness and congruence with Wells's, supra note 23, conflicting 1985 study). 93 . Id. at 83 (citing Peter N. Shapiro & Steven Penrod, Meta-analysis ofFacial Identification Studies, 100 PSYCHOL. BULL. 139 (1986)); see also Rosenberg, supra note 64, at 277 (citing Wells & Murray, What Can Psychology Say About the Neil v. Biggers Criteria for Judging Eyewitness Accuracy?, 68 J. APPLIED PSYCHOL. 347 (1983)); Goldstein et al., Does Fluency of Face Description Imply Superior Face Recognition?, 1979 BULL. PSYCHONOMIC SOC'Y 13, 15-18. 94. Cutler & Penrod, supra note 7, at 93. 95. See, e.g., Gregory-Bey v. Hanks, 332 F.3d 1036, 1050 (7th Cir. 2003); State v. Cheeseboro, 552 S.E.2d 300, 308 (S.C. 2001); Cutler & Penrod, supra note 7, at 183 (summarizing studies jurors' beliefs in the predictive power of consistent description of peripheral details); SOBEL, supra note 57, § 6.7 (summarizing many cases where description accuracy and consistency played a major role). 96. Compare CUTLER & PENROD, supra note 7, at 83 (summarizing studies where eyewitnesses with a high ability to describe faces made more correct identifications, but it was unknown whether they made fewer false identifications), with Rosenberg, supra note 64, at 277 (suggesting no such relationship between ability to describe faces and accurate identifications). 97. See, e.g., CUTLER & PENROD, supra note 7, at 94-95; LOFTUS & DOYLE, supra note 1 8, § 3-12; Seng & Carroll, supra note 16, § 2.4. 98. SOBEL, supra note 57, § 6. 12, at 6-50. 99. Ebbe B. Ebbesen & Vladimir J. Konecni, Eyewitness Memory Research: Probative v. Prejudicial Value, 5 Int'l Dig. Hum. Behav. SCI. & LAW 2 (1996), available at http:// www.psy.ucsd.edu/~eebbesen/prejvprob.html. 100. See Cutler & Penrod, supra note 7, at 186-90; Loftus & Doyle, supra note 18, § 3- 2007] FLIPPING A COIN 283 little reliance on witness confidence. 101 5. The Time Between the Crime and the Confrontation.—The amount of time passed before a witness or victim identities the perpetrator is undoubtedly an important factor in determining the reliability of an identification. Memory retention apparently drops off in a sharp "forgetting curve" after an event, eventually stabilizing into an extremely low rate of accurate identification, a rate approaching chance in some studies. 102 Cutler and Penrod's survey of studies that manipulated retention intervals demonstrated that fewer correct identifications (51% vs. 61%) and more false identifications (32% vs. 24%) were associated with longer delays. 103 Very short intervals may be particularly important when it comes to reducing false identifications. In one study looking at very short intervals (two hours versus twenty-four hours) in low stress situations, false identifications increased from 15% to 52%, while correct identifications decreased less dramatically from 43% to 39%. 104 There is some evidence, however, that this factor is much less relevant than the opportunity to view. 105 This may be because the process of forgetting does not seem to occur in a predictable passive decay mechanism, but rather in a more complex manner where new experiences, and even older ones, interfere with the process of reliable memory. 106 In some cases memories may be affected by unconscious transference, where a person seen in another context is identified as the criminal. 107 Furthermore, in some cases, violent events may make people engage in a process of "motivated forgetting," where the subconscious mind will block aspects of the event from memory. 108 The problem is not that the courts rely too heavily on reliability factors that do not accurately predict reliability, such as the degree of attention, but that even if every factor pointed to a more reliable identification, the corresponding degree of reliability is still unacceptable. Even the inclusion of other, more valid reliability factors, such as whether the testimony involved cross-race identification, would not make eyewitness identification testimony sufficiently 12. 101. E.g., Krist v. Eli Lilly & Co., 897 F.2d 293, 296-97 (7th Cir. 1990) (questioning probativity of witness certainty in any context); Brodes v. State, 614 S.E.2d 766, 771 (Ga. 2005) (reversing a conviction because the jury charge on eyewitness reliability included element of certainty); Commonwealth v. Santoli, 680 N.E.2d 1116, 1121 (Mass. 1997) (disapproving jury instructions on witness certainty); State v. Ramirez, 817 P.2d 774, 781 (Utah 1991) (adopting alternative reliability criteria without certainty factor); SOBEL, supra note 57, § 6.12. 102. LOFTUS & DOYLE, supra note 18, § 3-2(a). 103. Cutler & Penrod, supra note 7, at 106. 104. Id. (citing Krafka & Penrod, supra note 1 3). 105. SOBEL, supra note 57, § 6.13. 106. Seng & Carroll, supra note 16, § 2.38. 107. Loftus & Doyle, supra note 18, § 4-10. 108. Id. § 3-3. 284 INDIANA LAW REVIEW [Vol. 40:27 1 reliable. 109 The baseline accuracy rates (assuming that every factor pointed to a "more reliable" identification) range from 50-60% even in non-stressful witnessing situations which is not too much more reliable than a coin toss. Courts acknowledge this, though, and rely on the jury to get it right. The Supreme Court has placed a great deal of confidence in the adversarial jury system as a means of ferreting out mistaken eyewitness identification testimony: It is part of our adversary system that we accept at trial much evidence that has strong elements of untrustworthiness .... While identification testimony is significant evidence, such testimony is still only evidence .... Counsel can both cross-examine the identification witnesses and argue in summation as to factors causing doubts as to the accuracy of the identification including reference to both any suggestibility in the identification procedure and any countervailing testimony such as [sic] alibi. 110 B. Juries Cannot Meaningfully Determine Whether Eyewitness Identification Is Accurate "Few moments are more dramatic than when a courtroom witness, upon prompting from the prosecutor, outstretches an arm, extends a finger, and declares with rock-solid certainty that the accused is the person she saw fleeing the scene of the crime with bloodied hands." 111 Studies have shown that jurors overwhelmingly believe eyewitness identification testimony. Rare is the My Cousin Vinny moment, where the defense lawyer can show that the eyewitness is blind or viewing the event through filthy windows. 112 In a couple of studies, even this sort of discrediting information (the eyewitness had very poor eyesight and was not wearing glasses at the time) resulted in only a 4% lower conviction rate. 1 13 Cutler and Penrod state directly, "[T]here are more convictions than there are accurate identifications." 114 Jurors believe in eyewitnesses "despite impeachment, despite aggressive cross-examinations, and despite cautionary instructions." 115 Jurors have an "implicit faith" in eyewitness identification 109. See CUTLER & Penrod, supra note 7, at 104 (discussing Penrod & Shapiro's meta- analysis that showed that cross-race identifications were less accurate (57% versus 63%) and subject to more false identifications (22% versus 18%)); LOFTUS & Doyle, supra note 18, § 4-9 (discussing a study that found 55% false identification rate for cross-race identification versus 35% for same-race). 1 10. Manson v. Brathwaite, 432 U.S. 98, 1 13 n.14 (1977) (quoting Clemons v. United States, 408 F.2d 1230, 1251 (D.C. Cir. 1968)). 111. LOFTUS & DOYLE, supra note 1 8, § 4- 1 , at 75. 1 12. My Cousin Vinny (Twentieth Century Fox 1992). 113. Cutler & Penrod, supra note 7, at 1 9 1 . 114. Id. at 186. 1 15. Loftus & Doyle, supra note 18, § 9-1, at 200. 2007] FLIPPING A COIN 285 testimony and "tend to dispose ofinformation that challenges that faith." 1 16 Even if jurors were disposed to question the accuracy of an identification because eyewitness identification testimony is inherently unreliable, jurors would be simply unable to distinguish correct identifications from false ones. 117 In one study involving mockjurors, whether leading or open questions were used and whether the witness was accurate or not, between 73-86% of "jurors" believed the eyewitness identification. 118 The criteria by which jurors judge the reliability of a witness do not correlate with accuracy. Truth is not at issue—we can assume that the victims and other eyewitnesses making the identification are being truthful, even when mistaken. The eyewitness is usually "sincerely convinced of the accuracy of his or her testimony." 119 Jurors tend to evaluate eyewitnesses by three criteria: witness confidence, consistency, and memory of specific details. 120 None of these criteria correlate with identification accuracy. For example, in one study, "jurors" predicted an 83% probability that a "completely certain" eyewitness would correctly identify a culprit, compared with a 28% probability that a "somewhat uncertain" witness would do so. 121 However, studies have found that there is very little correlation between witness confidence and accuracy. 122 There is similarly little correlation between witness consistency or memory of specific details and accuracy. 123 In fact, memory of peripheral details will increase the likelihood of a witness making an identification with confidence, but is inversely correlated with accuracy. 124 Even experienced defense attorneys are unable to effectively counterjurors' propensity to believe eyewitness testimony. Cutler and Penrod found that an attorney's degree of experience and presumed skill at cross-examination did not significantly influence verdicts, even when correlated with known mistaken identifications. 125 "Cross-examination, a marvelous tool for helping jurors 116. Id. 117. See generally Steven I. Friedland, On Common Sense and the Evaluation of Witness Credibility, 40 CASE W. RES. L. REV. 165, 170-88 (1990). 118. CUTLER & Penrod, supra note 7, at 181-82 (citing Gary L. Wells et al., Accuracy, Confidence, and Juror Perceptions in Eyewitness Identification, 64 J. APPLIED PSYCHOL. 440 (1979)). 1 19. Loftus & Doyle, supra note 18, § 10-l(a). 120. Cutler & Penrod, supra note 7, at 181-90, 200-03. Some commentators say that the criteria that jurors use to evaluate eyewitness identification is the same as that used for all witness testimony: perception, sincerity, and memory—the difference may in the end be semantic. See Friedland, supra note 117, at 181. 121 . CUTLER & PENROD, supra note 7, at 178 (citing Gary Wells, How Adequate Is Human Intuition for Judging Eyewitness Testimony?, in EYEWITNESS TESTIMONY: PSYCHOLOGICAL Perspectives 256 (Gary Wells & Elizabeth Loftus eds., 1984)). 1 22. Id. at 94-95 ; Loftus & Doyle, supra note 1 8, § 3- 1 2; Seng & Carroll, supra note 16, § 2.4, § 2.40. 123. See supra notes 92-96 and accompanying text. 124. Cutler & Penrod, supra note 7, at 94. 125. Mat 186. 286 INDIANA LAW REVIEW [Vol. 40:27 1 discriminate between witnesses who are intentionally deceptive and those who are truthful, is largely useless for detecting witnesses who are trying to be truthful but are genuinely mistaken." 126 Normally, cross-examination serves to expose an insincere or dissembling witness. 127 However, with eyewitness identification testimony, an aggressive cross-examination only serves to highlight the witness's sincerity. 128 The attorney can question the witness about any factors, such as stress or cross-race identification, that would lead to more erroneous identifications, 129 but jurors do not tend to credit these factors in assessing the witness's credibility. 130 Nor could expert witnesses help jurors to determine whether a particular witness has made an accurate identification or not. While expert testimony will tend to increase the amount of time that juries spend in deliberation discussing the eyewitness testimony (from 10% of the total deliberation time to 28%) and decrease the conviction rate by up to 20%, 131 expert testimony cannot help the jury determine whether any particular identification is accurate or not. 132 Expert testimony can informjurors about the factors which would make an identification particularly unreliable, and also decrease jurors' reliance on witness confidence. 133 However, when not presented with the particular factors which would make the identification more unreliable than the baseline, jurors will place even greater weight on the identification. 134 This reliance is unwarranted considering that most studies place the baseline reliability rate of eyewitness identification around 50%. 135 C. Current Proposals May Improve Reliability but Not Enough There have recently been a few proposals to improve the reliability of 126. Wells et al., supra note 23, at 609. 127. LOFTUS & DOYLE, supra note 18, § 10-l(a). 128. Id. 129. Id. § 10-2. 130. Cutler &PENROD, supra note 7, at 197-209 ("[T]he effectiveness of cross-examination as a safeguard is still questionable in light of the lack ofjuror sensitivity to factors that are known to be diagnostic of eyewitness reliability."). 131. Mat 218-21. 132. Ebbesen & Konecni, supra note 99, at 4 (arguing that not only is there no theory which would allow an expert to predict the accuracy of a particular identification, but also that the effect of combining the various reliability factors is unknown). But see LOFTUS & DOYLE, supra note 18, § 11-11 (advocating practitioners' use of trace evidence analogy to convincejudges that "expert testimony does not necessarily threaten only a rise in the jurors' general level of skepticism about eyewitnesses, but actually points to specific factors in this specific case that are diagnostic of reliability or error") (emphasis added). 133. Cutler & Penrod, supra note 7, at 227. 134. Id. at 227-30. 135. See supra notes 1 1-26 and accompanying text. 2007] FLIPPING A COIN 287 eyewitness identification evidence 136 by incorporating the recommendations of Gary Wells and his collaborators for improving identification accuracy. 137 These proposals seek not only to eliminate any inadvertent suggestion 138 which may taint the reliability of an identification but also to counteract the tendency for witnesses to choose the person in the lineup or photoarray 139 who merely looks the most similar to the culprit, a tendency known as "relativejudgment." 140 These proposals range from improved questioning techniques by investigating officers 141 and simply informing the witness that the perpetrator may not be in the lineup 142 to changing lineup procedures themselves by using blank 143 or sequential lineups. 144 However, even if the proposal is seen as the most effective of these procedures, the use of double blind sequential lineups do not make eyewitness identifications reliable enough. Many psychologists propose double blind sequential lineups to improve reliability—some studies have shown that they can reduce the false identification rate by more than 50%. 145 What makes psychologists so enthusiastic about this procedure is that it has appeared to reduce false identification rate in experiments without adversely affecting the ability to obtain correct identifications when the culprit is in the lineup. 146 However, two large-scale real-world tests have cast some doubt on the efficacy of this technique. In addition to the Navy study, where the low stress interrogation group made fewest false identifications with 1 36. See generally NAT'LlNST. OFJUST., Dep'T OFJUST., EYEWITNESS EVIDENCE: AGUIDE FOR Law Enforcement (1999), available at www.ncjrs.gov/pdffilesl/178240.pdf [hereinafter The Guide]. 137. Wells et al., supra note 23. 1 38. The Guide assumes that law enforcement is acting in good faith. The Guide, supra note 136, at 2. 139. The term "lineup" will be used to refer to both photoarrays and live lineups. 140. Wells et al., supra note 23, at 613-15 ("[M]ost of the 54% who identified the culprit in a culprit-present lineup would simply have identified someone else if the culprit had not been present . . . eyewitnesses tend to select whomever looks most like the perpetrator regardless of whether the actual perpetrator is in the lineup."). 141. The Guide, supra note 136, at 13-16, 21-25. 142. Id. at 32. One study found that simply telling a witness that the culprit may not be in the lineup reduced the false identification rate when the culprit was not in the lineup from 78% to 33%, while having no adverse effect on a witness's willingness to make a positive identification when the culprit was in the lineup. Wells et al., supra note 23, at 615. 143. Wells et al., supra note 23, at 616. 144. The Guide, supra note 136, at 34, 36. 145. Cutler & Penrod, supra note 7, at 128. One commentator stresses the importance of making sure that sequential lineups are double blind, if done at all, since the effect of inadvertent suggestive clues may be greater than in simultaneous lineups. Wells et al., supra note 23, at 634. 146. Cutler & Penrod, supra note 7, at 128. Cutler & Penrod have found that sequential presentation may eliminate the effects of some types of suggestiveness (subtle clothing clues, for example), reducing false identification rates from 84% to 25%. Id. at 133. 288 INDIANA LAW REVIEW [Vol. 40:27 1 the use of a photoarray rather than sequential photos (12% vs. 25%), 147 a field test by three Illinois police departments found that sequential procedures led to more rather than fewer false identifications. 148 Another problem is that when an identification is not made through the sequential method, witnesses are often given a second chance to identify a suspect through a simultaneous lineup, more than erasing any advantage. 149 The biggest problem, however, is that sequential lineups and other procedures are not sufficient to make eyewitness identifications reliable. Law enforcement organizations who adopt these positive steps are to be commended, 150 but they are using a band-aid on a gaping wound. The best achievable rates for false identifications is around 20%, 151 and any imperfect witnessing condition can result in false identifications rates ranging from 51% to68% 152 to90%! 153 HI. Eliminating Eyewitness Identification Testimony Would Not Unreasonably Burden the Criminal Justice System Because current proposals do not sufficiently lessen the risk of misidentification, the only rational response to such a high rate of false identifications is to eliminate these identifications from trial. DNA tests can only exonerate those suspects and defendants for whom there is such physical evidence. The vast majority of cases do not involve DNA evidence. There has been some movement to institute new safeguards for the use of eyewitness testimony in capital and other murder cases, but these proposals only seek to remove the elements of suggestion and relative similarity by incorporating Gary 147. Morgan et al., supra note 19, at 272. This counterintuitive result may be anomalous. Multiple repeated tests would be needed to determine whether this result is representative. This needed type of repetition is very rare in the world of eyewitness tests. 148. Report to the Legislature of the State of Illinois : The Illinois Pilot Program on Sequential Double-Blind Identification Procedures 41 (2006), available at www.chicagopolice.org (follow "Illinois Pilot Report on Eyewitness Identification Methods") (statement of Sheri Mecklenburg, Program Director, Illinois State Police). 149. Cutler & Penrod, supra note 7, at 129. 150. See, e.g., The GUIDE, supra note 136; Drv. CRIM. Just., N.J. OFF. OF ATTORNEY Gen., Attorney General Guidelines for Preparing and Conducting Photo and Live Lineup IdentificationProcedures (200 1 ), available at http://www.state.nj .us/lps/dcj/agguide/photoid. pdf. 151. CUTLER & PENROD, supra note 7 , at 1 29 ; Gunter Koehnken et al . , Forensic Applications of Line-Up Research, in PSYCHOLOGICAL ISSUES IN EYEWITNESS IDENTIFICATION 205, 227 (Siegfried Sporer et al. eds., 1996). Even in the low stress group for the Navy study, where the subjects viewed the interrogators for 40 minutes at close proximity, the false identification rates more commonly seen were 25% and 38%. Morgan et al., supra note 19, at 272. 152. Morgan et al., supra note 19, at 272. 153. Cutler & Penrod, supra note 7, at 1 16-17. 2007] FLIPPING A COIN 289 Wells's recommendations. 154 As discussed in the previous section, these proposals improve identification accuracy, but not enough. Furthermore, there is no reason to believe that false identifications are more of a problem in murders than in other violent crimes; in fact, the opposite may be the case. Victims make up the vast majority of eyewitnesses, and live victims are rare in most homicide cases. 155 This helps to account for the lower number of wrongful capital convictions attributed to false identification. 156 Generally, victims making an identification have been subjected to an extreme amount of stress, which greatly increases the false identification rate. 157 For about half of all violent crimes, however, eyewitness testimony is extremely reliable because the crime was committed by someone known to the witness, such as a relative. 158 This is why eyewitness identification testimony should only be excluded in those cases where the defendant is a stranger to the witness, the witness is an accomplice, 159 and only in FBI Crime Index felony cases. 160 The Crime Index is simply chosen as a proxy for serious crimes for which the cost of further investigation is outweighed by the benefit of fewer wrongful convictions. The burden on the criminal justice system would not be all that great. A survey of prosecutors in thirty states resulted in an estimate that only 3% of felony cases are based on eyewitness identifications, 161 and another estimate pegged the number at 5%. 162 If this number is halved to account for those crimes in which the witness previously knew the defendant, only 1 .5-2.5% of these cases are based on suspect identifications. Police and other investigators should, of course, be allowed to continue to use identifications as an investigative tool. Surely investigators can find other evidence in those 1.5-2.5% of cases, so that 154. See, e.g., Illinois Governor's Office Report of the Governor's Comm'n on Capital Punishment 31-40 (2002), available at http://www.idoc.state.il.us/ccp/ccp/reports/ commission_report/chapter_02.pdf. 155. Samuel R. Gross, Lost Lives: Miscarriages of Justice in Capital Cases, 61 LAW & Contemp. Probs. 125, 137(1998). 156. See supra notes 46-52 and accompanying text. 157. See supra notes 19-24 and accompanying text. 158. Gross, supra note 155, at 137. 1 59. Perjured eyewitness testimony from accomplices accounted for 1 5, or 32.6%, ofthe forty- six erroneous identifications discussed by the Center on Wrongful Convictions report. Warden, supra note 48. Professor Gross claims that witness perjury is a "far more common cause of error in murders and other capital cases than in lesser crimes." Gross, supra note 155, at 139. He attributes this factor to absence of a live victim to contradict the perjurer. Id. at 137-39. 160. FBI, Crime in the U.S. 2003 3 (2004) [hereinafter Crime]. 161. CUTLER & PENROD, supra note 7, at 8 (citing A.G. Goldstein et al., Frequency of Eyewitness Identification in Criminal Cases: A Survey of Prosecutors, 27 BULL. PSYCHONOMIC SOC'Y 71-74 (1989)). 1 62. Id. (citing Wallace D. Loh, Psycholegal Research: Past & Present, 79 MlCH. L. Rev. 659, 686(1981)). 290 INDIANA LAW REVIEW [Vol. 40:27 1 these cases can be prosecuted with more reliable evidence. Murder cases are the least dependent on eyewitness identification testimony, yet they have the highest "clearance" (or arrest and charge) rate of all Crime Index crimes at 62.4%. i63 Police were able to find the evidence needed even without witnesses. The next highest clearance rates were for aggravated assault and rape, both crimes where in many cases, the victim was acquainted with the culprit. 164 And where the crime was committed by a stranger, the police should continue to use the victims' and other witnesses' accounts for investigations. With new investigative techniques, such as the use of DNA testing, police may be able to solve many of these crimes more easily than ever before. Certainly, there are many cases, including homicides, where there is no other evidence but eyewitness identification. Sometimes, there is not even that. When the crime is especially serious, the pressure to find and convict a culprit is especially high. 165 But where there is no other evidence, the added pressure to get justice for the community may lead to mistakes. These mistakes can be fatal. Conclusion Eyewitness identification testimony is known to the courts and to psychologists to be extremely unreliable. However, there is great resistance to excluding this type of evidence at trial. The commonsense belief that "seeing is believing" is hard to overcome. The problem is not just that people are being convicted of crimes they did not commit, but that for every wrongful conviction there is a guilty party left to wreak havoc on the public. 166 The only effective way to fix this problem is to exclude eyewitness identification testimony from trials. After all, the goal is not just to convict someone, but to convict the actual perpetrator of the crime. Current proposals would greatly reduce the number of false identifications but not by enough. When even under the best of circumstances, victims or witnesses who experienced stress will make a mistaken identification around 50% of the time andjuries cannot discern the accurate from the false, there is something inherently wrong with using this unreliable identification to convict someone. Identifications are fine as an investigative tool, but if eyewitness identification is all that these cases hang on, we cannot be sure that the real culprits pay for these crimes. The very possibility that such dangerously unreliable evidence is causing innocent people to be executed or imprisoned should counsel against allowing its use in court. 163. Crime, supra note 160, at 255. 164. U.S. Dep't of Just., Bureau of Just. Statistics, Criminal Victimization in the United States, 2002, Statistical Tables, Table 6 (2004), available at http://www.albany. edu/sourcebook/pdf/t3 17.pdf. 165. See Gross, supra note 155, at 135. 166. HUFFET AL., supra note 10, at 150.