Safety in numbers? deciding when DNA alone is enough to convict

New York University Law Review

Volumn 85, Issue 4, 2010, Pages 1130-1185

  Roth, Andrea ^a  

^a STANFORD LAW SCHOOL   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 78649347674     PISSN: 00287881     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (241)

1. The mother and nanny stalked by Baker lived at 1240 Sixth Avenue in San Francisco; the nurse-decedent lived at 1279 Sixth Avenue. Appellant's Opening Brief at 81, People v. Puckett, No. A121368 (Cal. Ct. App. Apr. 18, 2009).
2. Chris Smith, DNA's Identity Crisis, S.F. MAG., Sept. 2008, http://www.sanfranmag. com/story/dna's-identity-crisis. The blood-stained notice was destroyed by the government at some point before trial. Appellant's Opening Brief, supra note 1, at 85.
3. Edward Humes, Guilt by the Numbers: How Fuzzy Is the Math that Makes DNA Evidence Look So Compelling to Jurors?, CAL. LAW., Apr. 2009, at 20, 20-22.
4. Appellant's Opening Brief, supra note 1, at 56.
5. See Humes, supra note 3, at 22 ("What the jurors didn't know, though, and what the judge didn't think they needed to know, is that there's another way to run the numbers.").
6. See infra note 214 (calculating posterior odds of Puckett being source as only about thirty-three percent); see also David H. Kaye, Rounding Up the Usual Suspects: A Legal and Logical Analysis of DNA Trawling Cases, 87 N.C. L. REV. 425, 490-92 (2009) (discussing probability of another source in Bay Area sharing Puckett's DNA profile).
7. By defining "pure cold hit" in this way, I mean to isolate those cases in which the government's evidence is almost entirely quantifiable, thus raising the issues addressed in this Article related to the limits of purely probabilistic proof of guilt. For example, while the government in Puckett's case introduced not only the profile match but also proof of Puckett's connection to the Bay Area, this added evidence at most limits the size of the likely suspect population to others with a similar connection. Because the chance of another suspect sharing the profile is still relatively high, the case is qualitatively different from cases with other significant individualized evidence of guilt. See discussion infra Part III (discussing difficulty of finding guilt beyond reasonable doubt in "cold hit" cases). And while a prior conviction might make the suspect statistically more likely to be the perpetrator, courts may not view that fact alone as enough to create a legally sufficient case against the suspect if there are others who also match the profile. See infra notes 210-18 (discussing role of evidence of prior conviction in meeting sufficiency standard generally and in Puckett). Others have similarly used the term. See, e.g., Simon A. Cole & Michael Lynch, The Social and Legal Construction of Suspects, 2 ANN. REV. L. & SOC. SCI. 39, 49-50 (2006) (describing innocent attorney Brandon Mayfield as "pure cold hit suspect" in Madrid train bombing case where only evidence against him was database match and other details law enforcement found suspicious, for example, that he converted to Islam and had not left country). Still others have used the related term "naked trawl" case to denote a prosecution based on a DNA cold hit (the result of a "database trawl") and no other individualized evidence of guilt. See, e.g., Kaye, supra note 6, at 472, 490.
8. See discussion infra Part I.B (discussing recent rise in cold hit prosecutions).
9. See infra notes 50-53 and accompanying text (listing cold hit cases).
10. See infra notes 54-60 (listing cases and their dispositions).
11. Dist. Attorney's Office for the Third Judicial Dist. v. Osborne, 129 S. Ct. 2308, 2327 (2009) (Alito, J., concurring) (internal citation omitted).
12. ANDREI SEMIKHODSKII, DEALING WITH DNA EVIDENCE: A LEGAL GUIDE 136 (2007).
13. See infra note 84 and accompanying text (discussing popularity among scholars of Bayes' Theorem, one method for determination).
14. Cole & Lynch, supra note 7, at 50-51 (quoting Laurence H. Tribe, Trial by Mathematics: Precision and Ritual in the Legal Process, 84 HARV. L. REV. 1329, 1372 (1971)). I discuss Tribe's argument more fully in Part III.B. See infra text accompanying notes 153-57.
15. Scientists typically refer to the conflation of the random match probability (RMP) and source probability as the "fallacy of the transposed conditional." See infra Part II for further discussion. The RMP of 1 in 1.1 million in Puckett does not signify that there is only a 1 in 1.1 million chance that someone other than Puckett is the source of the DNA. Rather, it means that a person randomly selected from the population has a 1 in 1.1 million chance of matching or, equivalently, that we would expect 1 in every 1.1 million people to match the profile. See infra text accompanying note 94 (explaining confusion in use of RMP). In a population of about two million men, we would expect about two men to match. In such a case, the probability that the suspect might not actually be the true source might even be greater than fifty percent, a far cry from 1 in 1.1 million.
16. The "Wells Effect" refers to jurors' demonstrated tendency to impose liability more readily in cases involving impressionistic evidence such as eyewitness testimony rather than naked statistical evidence. See infra notes 157-62 and accompanying text (citing work by Gary Wells on this phenomenon).
17. The suspect's DNA is sent to the laboratory for testing, either simultaneously or after the evidence sample is tested, depending on the laboratory's practices and the timing of the suspect's arrest. Cf. Erin Murphy, The Art in the Science of DNA: A Layperson's Guide to the Subjectivity Inherent in Forensic DNA Typing, 58 EMORY L.J. 489, 497-98 (2008) (describing how crime scene DNA samples are taken).
18. See JOHN M. BUTLER, FORENSIC DNA TYPING: BIOLOGY, TECHNOLOGY, AND GENETICS OF STR MARKERS 23-24 (2d ed. 2005) (explaining genetic principles behind DNA testing).
19. ATTG is a sequence of four units called nucleotides that make up the genetic strand. Here, the sequence includes A (adenine), T (thymine), and G (guanine). See id. at 18-19.
20. See Murphy, supra note 17, at 494-96 (describing practice of STR testing); cf. BUTLER, supra note 18, at 30 (discussing advantages of STR markers).
21. See Murphy, supra note 17, at 494-96 (describing testing).
22. United States v. Yee, 134 F.R.D. 161, 181 (N.D. Ohio 1991).
23. The categories are African American, Caucasian, Southeast Hispanic, and Southwest Hispanic. See People v. Venegas, 954 P.2d 525, 537 (Cal. 1998) (discussing use of categories). Thus far, no court has upheld an admissibility challenge to DNA match statistics based on race-based limitations of the allelic frequency tables.
24. For example, the estimated chance of a member of the Afro-Caribbean population having a 23 allele at the locus FGA is, according to the FBI's tables, 0.234, or 23.4%. Chantal J. Fr ́ egeau et al., Population Genetic Characteristics of the STR Loci D21S11 and FGA in Eight Diverse Human Populations, 70 HUM. BIOLOGY 813, 837 tbl.9 (1998).
25. Bruce Budowle et al., Source Attribution of a Forensic DNA Profile, FORENSIC SCI. COMM., July 2000, http://www.fbi.gov/hq/lab/fsc/backissu/july2000/source.htm ("The average random match probability for unrelated individuals for the 13 STR loci is less than one in a trillion, even in populations with reduced genetic variability....").
26. In low copy number (LCN) cases-involving a very small amount of recovered DNA-typing a full thirteen-loci profile may be impossible. See, e.g., NUFFIELD COUNCIL ON BIOETHICS, THE FORENSIC USE OF BIOINFORMATION: ETHICAL ISSUES 19, 23 (2007) [hereinafter NUFFIELD REPORT] (discussing problems with LCN analysis). At least one court has ruled LCN testing results inadmissible on the ground that the method has not gained "general acceptance" in the relevant scientific community as required in some jurisdictions- including California-under the Frye test. Court Reporter's Notes at 3, People v. Espino, No. NA076620 (Cal. Super. Ct. Mar. 18, 2009) (ruling LCN results inadmissible because of lack of consensus in scientific community). The number of such cases is increasing. See Cole & Lynch, supra note 7, at 43-44 (discussing potential risks of enhanced ability to use LCN samples).
27. See, e.g., Humes, supra note 3, at 23 (discussing 5.5 loci profile due to degradation in Puckett); Solomon Moore, Damaged DNA Evidence Shrinks Serial Killer Case, N.Y. TIMES, May 22, 2009, at A14 (discussing partial matches resulting from damaged DNA sample).
28. Ray Wickenheiser, General Guidelines for Categorization and Interpretation of Mixed STR DNA Profiles, 39 CAN. SOC. FORENSIC SCI. J. 179, 214 (2006).
29. See, e.g., id. at 199-207 (explaining example of mixed sample and finding that correct match can be more accurately deduced from single-source sample); William C. Thompson, The Potential for Error in Forensic DNA Testing, GENEWATCH, Nov.-Dec. 2008, at 5, 6 (noting that chance of coincidental match will increase if sample is mixed compared with single-source sample).
30. The power of mtDNA in identifying suspects is limited because it is inherited only through the mother, sequence types tend to cluster geographically, and population databases are very limited. See Frederika A. Kaestle et al., Database Limitations on the Evidentiary Value of Forensic Mitochondrial DNA Evidence, 43 AM. CRIM. L. REV. 53, 66-67 (2006) (identifying problems with mtDNA database). As a result, mtDNA match statistics are currently much less discriminating than RMPs associated with nuclear DNA typing. See SEMIKHODSKII, supra note 12, at 68-71 (describing different calculations necessary to interpret mtDNA evidence).
31. See Budowle et al., supra note 25 ("[A]mong African Americans, Chinese, and Caucasians, the most common conditional probability for a 13 STR locus profile is expected to occur with a frequency no more than one in 40,000 among full siblings.").
32. As a practical matter, the nature of the evidence in most pure cold hit cases makes mounting an innocent presence defense at trial untenable. See, e.g., Yun S. Song et al., Average Probability that a "Cold Hit" in a DNA Database Search Results in an Erroneous Attribution, 54 J. FORENSIC SCI. 22, 23 (2009) ("In cases of sexual assault, courts have reasoned that the intimate nature of the sample forecloses arguments that it might have been left accidentally or inadvertently.").
33. See Melendez-Diaz v. Massachusetts, 129 S. Ct. 2527, 2531-32 (2009) (holding that drug analyst's hearsay report was inadmissible under Confrontation Clause, but noting that not all chain-of-custody witnesses or equipment technicians need testify). Even before Melendez-Diaz, several lower courts had recognized that this practice violated the rule of Crawford v. Washington, 541 U.S. 36, 68 (2004), which prohibited the government from introducing hearsay affidavits in lieu of live testimony absent a showing of the live witness's unavailability and a prior opportunity for cross-examination. See, e.g., Roberts v. United States, 916 A.2d 922, 938-39 (D.C. 2007) (holding unconstitutional, but affirming due to lack of plain error, government's introduction of serologist's hearsay report through DNA expert).
34. See, e.g., Appellant's Opening Brief, supra note 1, at 11-13 (describing DNA analyst's testimony about testing process); Official Transcript of Proceedings at 3-79 to 3-80, State v. Derr, No. 08-K-04-000930 (Md. Cir. Ct. June 28, 2006) (testimony of FBI Analyst Jennifer Luttman) [hereinafter Luttman Testimony] (describing DNA-testing process).
35. See, e.g., Appellant's Opening Brief, supra note 1, at 17 (describing analyst's testimony regarding RMP calculation).
36. Budowle et al., supra note 25; see also D.J. Balding, When Can a DNA Profile Be Regarded as Unique?, 39 SCI. & JUST. 257, 258 (1999) (discussing whether sufficiently small RMP can justify calling defendant's profile unique).
37. See Luttman Testimony, supra note 34, at 3-79 to 3-80 (testifying that when probability is "exceedingly rare," meaning "smaller than 1 in 280 billion in all four of the population databases that we examine," she "will say the person is the source of the DNA to a reasonable degree of scientific certainty").
38. See, e.g., Smith, supra note 2 (noting that government called genetics and statistics expert Dr. Ranajit Chakraborty in Puckett case to rebut defense expert).
39. See William C. Thompson et al., Evaluating Forensic DNA Evidence, Part 2, CHAMPION, May 2003, at 24, available at http://www.nacdl.org/public.nsf/freeform/championmag? OpenDocument (roll over "Champion Magazine" and follow "Search All Articles" hyperlink to find article) (describing ease of providing laboratory files on compact discs for defense attorney review).
40. See Jonathan J. Koehler, Fingerprint Error Rates and Proficiency Tests: What They Are and Why They Matter, 59 HASTINGS L.J. 1077, 1094-98 & 1096 n.68 (2008) (discussing design of blind proficiency testing in DNA laboratories while noting difficulty in conducting enough tests to report accurate laboratory-specific error rates); see also infra note 204 (discussing importance of false positive rate as counterpoint to RMP).
41. See Bruce Budowle & Tamyra R. Moretti, Genotype Profiles for Six Population Groups at the 13 CODIS Short Tandem Repeat Core Loci and Other PCR-Based Loci, 1 FORENSIC SCI. COMM. (1999), http://www.fbi.gov/hq/lab/fsc/backissu/july1999/budowle.htm (presenting genotype frequency tables for thirteen core loci by population group).
42. See infra note 203 (discussing criticisms of reported RMPs and reports that matches are less rare than claimed).
43. See, e.g., Andrea L. Roth & Edward J. Ungvarsky, Data Sharing in Forensic Science: Consequences for the Legal System, 2009 AM. STAT. ASS'N, PROC. JOINT STAT. MEETING 469, 470-71 (discussing need for outside assessment of government claims); Erin Murphy, Give Scholars Access to the National DNA Database, S.F. CHRON., Feb. 24, 2009, at A11 (arguing that outside review is needed based on "broad reliance on the accuracy of DNA matches"); D.E. Krane et al., Time for DNA Disclosure, 326 SCI. 1631, 1631-32 (2009) (calling for greater scholarly access to DNA databases); see also Song et al., supra note 32, at 22, 24 (noting justice concerns about increased use of cold hits and calling for continued testing of assumptions behind government model). Even Professor David Kaye, who predicts that researchers will likely not find anything in the FBI's databases to contradict the FBI's RMP estimates, has called upon the FBI to allow access and end the controversy once and for all. David H. Kaye, Trawling DNA Databases for Partial Matches: What Is the FBI Afraid of?, 19 CORNELL J.L. & PUB. POL'Y 145, 170-71 (2009) (encouraging government to make anonymized version of data available to researchers to increase confidence in use of data).
44. DNA Identification Act of 1994, Pub. L. No. 103-322, §§ 210,301-06, 108 Stat. 1796, 2065-71 (codified as amended in scattered sections of 42 U.S.C.), authorized the FBI to establish a national DNA database for identification purposes. The FBI's National DNA Index System (NDIS) became operational in October 1998, and the Combined DNA Index System (CODIS), enables information exchanges between NDIS, state, and local authorities. See FED. BUREAU OF INVESTIGATION, U.S. DEP'T OF JUSTICE, THE FBI'S COMBINED DNA INDEX SYSTEM PROGRAM-CODIS 1 (2000), available at http://www.rootsecure.net/ content/downloads/pdf/fbi_codis.pdf.
45. Compare NAT'L RESEARCH COUNCIL, THE EVALUATION OF FORENSIC DNA EVIDENCE 134 (1996) (noting that "more than 20 suspects have been identified" through offender database searches), with Fed. Bureau of Investigation, CODIS-NDIS Statistics, http://www.fbi.gov/hq/lab/codis/clickmap.htm (last visited Aug. 16, 2010) [hereinafter CODIS-NDIS Statistics] ("As of June 2010, CODIS has produced over 120,300 hits assisting in more than 117,800 investigations.").
46. As of last year, thirteen states collected DNA samples from arrestees. In addition, federal arrestees are now subject to DNA testing regardless of whether they are ultimately convicted of the charged offense. Randal C. Archibold, Justice Department Details Program for Collecting DNA from People in Federal Custody, N.Y. TIMES, Apr. 19, 2008, at A11. And in May 2010, the House of Representatives passed a bill providing financial incentives to states that collect DNA from arrestees for certain serious crimes. Katie Sepich Enhanced DNA Collection Act of 2010, H.R. 4641, 111th Cong. (2010). The United Kingdom now permits DNA sampling of arrestees for recordable offenses. See CAROLE MCCARTNEY, FORENSIC IDENTIFICATION AND CRIMINAL JUSTICE 195 (2006) (discussing change in U.K. practices). Also, as of 2008, twenty-eight states collected DNA samples from juveniles adjudicated delinquent. Natalie A. Bennett, A Privacy Review of DNA Databases, 4 I/S: J.L. & POL'Y FOR INFO. SOC'Y 821, 840 (2008). Indeed, some politicians and legal scholars have called for the establishment of population-wide citizen DNA databases. See id. at 837-38.
47. See CODIS-NDIS Statistics, supra note 45 (describing success of CODIS program).
48. See NAT'L RESEARCH COUNCIL, STRENGTHENING FORENSIC SCIENCE IN THE UNITED STATES: A PATH FORWARD 71-72 (2009) (describing initiative).
49. Id. at 72.
50. See, e.g., State v. Hunter, 861 N.E.2d 898, 901 (Ohio Ct. App. 2006) (noting in rape case that "literally no other evidence linked appellant to the crime" other than DNA match with RMP of 1 in 756 trillion (emphasis added)); State v. Toomes, 191 S.W.3d 122, 129 (Tenn. Crim. App. 2005) ("The only-and we emphasize the only-evidence connecting [Toomes] to the victim's rape is the DNA results."); R v. Lashley [2000] EWCA (Crim) 88, [3] (Eng.) (describing jury instruction explaining that Lashley was 1 of 7-10 males in United Kingdom expected to share this DNA profile, and noting that jury convicted even though there was no other evidence against defendant); R v. Adams, [1996] 2 Crim. App. 467, 469-70 (Eng.) (finding no inculpatory evidence against suspect except DNA match with disputed RMP between 1 in 2 million and 1 in 200 million); see also Mike Redmayne, Rationality, Naturalism, and Evidence Law, 2003 MICH. ST. L. REV. 849, 879-80 [hereinafter Redmayne, Rationality] (discussing Lashley and noting that RMP was 1 in 4 million); Mike Redmayne, Appeals to Reason, 65 MOD. L. REV. 19, 21 (2002) (discussing Lashley); cf. State v. Davis, 698 N.W.2d 823, 826-27 (Wis. Ct. App. 2005) (describing evidence for warrant issued against Davis, accused of 1994 rape, as being cold hit found in 1998-1999 and laboratory's conclusion that "only reasonable scientific explanation" for match "was that Davis was the source of the semen in [the victim's] underwear"). Other apparently "pure cold hit" cases have not yet been reviewed by an appellate court. See, e.g., Preliminary Hearing Minutes, People v. Moore, No. YA062921 (Cal. Super. Ct. May 30, 2007) (basing case against defendant solely on DNA match with RMP with denominator in quintillions). Of course, even in those cases described by courts as having only DNA match evidence, undisputed facts presumably exist that would narrow the potential suspect popu lation. See Kaye, supra note 6, at 487 (noting in hypothetical "naked trawl" case that "some men would have been in the hospital or have been too feeble to have carried [the victim] away; many would have airtight alibis; many would be psychologically incapable of the alleged conduct").
51. See, e.g., State v. Raines, 857 A.2d 19, 21-22 (Md. 2004) (stating that only evidence was that both perpetrator and Raines were black males of medium height); Roberson v. State, 16 S.W.3d 156, 160-63 (Tex. App. 2000) (providing no evidence except cold hit and match to blood type common to 9-10% of all men); R v. Watters [2000] EWCA (Crim) 89 (noting no inculpatory evidence except DNA match linking defendant to cigarette butts found at five apparently related burglaries in Birmingham, England, evidence that defendant was smoker and that he lived in Birmingham, and fact that he was male, coupled with assumption that most people who crack safes are men); Appellant's Opening Brief, supra note 1, at 38, 56-57, 102-03 (describing primary evidence as cold hit with RMP of 1 in 1.1 million, prior sexual assault conviction, and Puckett's connection to Bay Area); Official Transcript of Proceedings at 4-12, State v. Derr, No. 08-K-04-000930 (Md. Cir. Ct. Aug. 31, 2006) (recording defendant's motion for acquittal based on evidentiary insufficiency since only evidence identifying defendant was cold hit and very general physical description of perpetrator). In other cases, the additional information is still general but limits the suspect population more narrowly to a neighborhood or those seen around the crime area. See, e.g., Riggs v. State, 809 N.E.2d 322, 324 (Ind. 2004) (mentioning that only evidence against Riggs for 1985 rape-murder was 2000 cold hit and fact that Riggs lived near location where victim went missing); State v. Abdelmalik, 273 S.W.3d 61, 63 (Mo. Ct. App. 2008) (describing no inculpatory evidence except DNA match with RMP of 1 in 1 quintillion and fact that defendant lived within two miles and worked within one mile of crime scene at time of 1980 killing); People v. Rush, 630 N.Y.S.2d 631, 632 (Sup. Ct. 1995), aff'd, 672 N.Y.S.2d 362 (App. Div. 1998) (noting that only evidence permissible in court was cold hit and fact that defendant had been seen in "area" of robbery-rape three days before incident); cf. People v. Soto, 981 P.2d 958, 961 (Cal. 1999) (describing only inculpatory evidence as DNA match with RMP of 1 in 189 million in Hispanic population and fact that victim thought masked rapist's voice might be defendant's because defendant had come over earlier that day, while noting that defendant (Latino, black hair, dark complexion) did not match victim's physical description of rapist (white, blond hair, olive complexion)); Springfield v. State, 860 P.2d 435, 449 (Wyo. 1993) (providing no evidence other than DNA match, fact that defendant was consistent with general description of perpetrator, and that victim said at trial that defendant "definitely resembles" her attacker, though she could not identify him).
52. See, e.g., Appellant's Opening Brief, supra note 1, at 38 (describing prosecution's case in Puckett, including its assessment of defendant's RMP).
53. See, e.g., Abdelmalik, 273 S.W.3d at 66 (relying on RMP of 1 in 1 quintillion). None of these cases has discussed what the source probability might be, given the RMP and the population of likely suspects. Instead, nearly all of these courts have engaged in the fallacy of the transposed conditional, see infra note 95 and accompanying text (distinguishing between RMP and source probability), and treated the RMP as the source probability. While low RMPs on the order of 1 in 1 quintillion necessarily imply astronomically high source probabilities as well, higher RMPs may well be consistent with a source probability of less than 99.9%. See infra Part V (examining cases discussed for whether they would be resolved at sufficiency stage because of high RMP).
54. See, e.g., Hunter, 861 N.E.2d at 901 (affirming conviction after trial though "literally no other evidence linked appellant to the crime" other than DNA match with RMP of 1 in 756 trillion (emphasis added)); Toomes, 191 S.W.3d at 129 (affirming conviction after trial though "[t]he only-and we emphasize the only-evidence connecting [Toomes] to the victim's rape is the DNA results").
55. See, e.g., Motion To Compel Government To Produce Statistics on Matching and Near-Matching Profiles Maintained in the CODIS DNA Database at 1, United States v. Rue, Crim. No. F-3817-05 (D.C. Super. Ct. Dec. 3, 2005) ("The sole basis for the indictment of Mr. Rue was a DNA database 'cold hit.' . . . Prior and subsequent to the DNA search, to the best knowledge of defense counsel, no other evidence has been alleged to link Mr. Rue to the complaining witness."); Response to Defendant's Motion To Compel Production of Statistics on Matching and Near-Matching Profiles Maintained in the CODIS Database at 2, United States v. Rue, Crim. No. F-3817-05 (D.C. Super. Ct. Jan. 6, 2006) (referring to match between defendant and DNA sample as "cold hit"). Rue entered a plea of guilty on January 11, 2006. See Court File, United States v. Rue, Crim. No. F- 3817-05 (D.C. Super. Ct. Jan. 11, 2006).
56. Tim O'Neil & William C. Lhotka, DNA Is Just a Start in Two Old Murders: Expert Says He Doesn't Know of a Single Rape or Murder Conviction Based on DNA Alone, ST. LOUIS POST-DISPATCH, Feb. 13, 2007, at A1 (noting acquittal of rape defendant on twentyseven counts resting on DNA evidence alone).
57. See, e.g., R v. Lashley, [2000] EWCA 88, [3], [14]-[16] (setting aside conviction based on insufficiency of evidence when DNA could have matched "five or six men in the United Kingdom"); see also Redmayne, Rationality, supra note 50, at 880 (discussing Lashley).
58. For example, police investigating a burglary in Bolton, England, recovered blood off a window and got a cold hit match to Raymond Easton, a man who lived two hundred miles away and was in the British database by virtue of a previous domestic dispute. Though Easton could barely walk because of Parkinson's disease and had a corroborated alibi, police insisted he was the culprit based on the RMP of 1 in 37 million. Easton was exonerated after police conducted more advanced DNA testing at additional loci. Cole & Lynch, supra note 7, at 48; see also William C. Thompson, The Potential for Error in Forensic DNA Testing (and How That Complicates the Use of DNA Databases for Criminal Identification), 2008 COUNCIL FOR RESPONSIBLE GENETICS NAT'L CONF. 9, http:// www.councilforresponsiblegenetics.org/pageDocuments/H4T5EOYUZI.pdf (noting coincidental cold hit match between two unrelated crime scene DNA samples); id. at 33 (noting cold hit tying Nottingham boy to Northern Ireland terrorist bombing case that appeared to be coincidental match).
59. For example, police investigating a 1969 Michigan murder recently conducted DNA testing on stains on the victim's clothing, which yielded database hits to two men: Gary Leiterman, a sixty-two-year-old nurse with a forgery conviction, and John Ruelas, a fortyyear- old convicted murderer. People v. Leiterman, No. 265821, slip op. at 2 (Mich. Ct. App. July 24, 2007). Though Ruelas would have been only four years old at the time of the murder, prosecutors insisted he was a "chronic nose-bleeder" and, while not the murderer, had bled at the scene. Cole & Lynch, supra note 7, at 48. A more likely explanation for the hit to Ruelas is contamination; both men's samples were processed in the same laboratory at the same time as the sample from the 1969 case. Id.; see also William C. Thompson, Tarnish on the "Gold Standard": Understanding Recent Problems in Forensic DNA Testing, CHAMPION, Jan./Feb. 2006, at 10, 13 (discussing Washington "false 'cold hit[]'" where juvenile respondent's database sample was used as "training sample" in laboratory at same time another kit was analyzed for rape committed when juvenile was a young child); id. at 13-14 (discussing Australian toddler-murder case where young mentally disabled woman, herself a victim of sexual assault, was falsely inculpated by cold hit when her own DNA, found on outside of condom used in sexual assault against her, was analyzed at same time as toddler's clothing); Thompson, supra note 58, at 9 (noting coincidental cold hit in Chicago burglary case to woman incarcerated at time of offense); id. at 25 (citing examples of false cold hits due to likely contamination between samples); id. at 28-29 (discussing admitted false cold hit due to suspected contamination in New Jersey case).
60. See, e.g., Thompson, supra note 58, at 31 (noting Sacramento rape case in which police got cold hit match to man who lived in local area and when further investigation suggested his innocence, laboratory reviewed analyst's work and found interpretive error); cf. id. at 33 (discussing near-exclusion in 2005 of Missouri man whose CODIS profile matched evidence sample at twelve of thirteen loci, when police realized that man's profile had been mislabeled at non-matching locus and in fact all thirteen loci matched).
61. See supra notes 48-49 and accompanying text (discussing DNA Initiative).
62. See, e.g., Song et al., supra note 32, at 22 ("[C]old-hit cases raise justice-related concerns, especially since mounting a defense to a crime that occurred in the past becomes increasingly difficult as time progresses.").
63. Appellant's Opening Brief, supra note 1, at 4 & n.6, 9 (noting that victim's landlady, investigating police officers, and medical examiners were all dead or incapacitated by time of trial).
64. Cf. Solomon Moore, Progress Is Minimal in Clearing DNA Cases, N.Y. TIMES, Oct. 25, 2008, at A9 ("An audit... found that 217 backlogged cases [in Los Angeles] involved sexual assaults so old the 10-year statute of limitations had lapsed."); Cole & Lynch, supra note 7, at 46 (noting high number of backlogged and older cases waiting for analysis and database searches).
65. See generally Meredith A. Bieber, Meeting the Statute or Beating It: Using "John Doe" Indictments Based on DNA To Meet the Statute of Limitations, 150 U. PA. L. REV. 1079 (2002) (describing new practice of issuing "John Doe" indictments to allow police to "continue their efforts to identify the suspect").
66. See, e.g., MCCARTNEY, supra note 46, at 63, 183 (discussing concern that "faith in forensic science has been too easily used to shore up falling confidence in police investigative competence" and explaining that once investigators recover fingerprints or DNA from crime scene, both police and defense attorneys tend to neglect follow-up investigation); Cole & Lynch, supra note 7, at 44 ("Forensic investigation is positioned to displace much of the 'good, old-fashioned detective work' of canvasing [sic] neighborhoods, questioning witnesses, interrogating suspects, and cultivating informants." (quoting MARK FUHRMAN, DEATH AND JUSTICE 222 (2003))).
67. See MCCARTNEY, supra note 46, at 32 (quoting official as acknowledging that: "You are not going to get admissions [from suspects] these days, the use of informants is not as effective as it was, even societal factors such as the public not relating to the police as they once did... as other things have fallen away, forensic evidence is what is left").
68. See, e.g., Adam Liptak, Study of Wrongful Convictions Raises Questions Beyond DNA, N.Y. TIMES, July 23, 2007, at A1 (noting that two hundred prisoners have been shown innocent because of DNA evidence); The Innocence Project, http:// www.innocenceproject.org/know/Browse-Profiles.php (last visited July 26, 2010) (listing and linking to information on over two hundred cases in which convicted defendants were exonerated through DNA testing).
69. Given the clear differences in using DNA as a tool of exclusion versus inclusion, this perception may not be justified. See discussion infra note 145 (discussing differences between DNA exoneration and DNA proof of guilt).
70. See, e.g., MICHAEL LYNCH ET AL., TRUTH MACHINE: THE CONTENTIOUS HISTORY OF DNA FINGERPRINTING 184 (2008) ("[I]n some cases the cold hit is the sole, or main, item of criminal evidence."); Cole & Lynch, supra note 7, at 47 (observing "some very interesting cases" in which DNA "was not corroborated and was even contradicted by other evidence"); Kaye, supra note 6, at 490-92 (discussing Puckett as database trawl case based on defendant's regional proximity); Erin Murphy, The New Forensics: Criminal Justice, False Certainty, and the Second Generation of Scientific Evidence, 95 CAL. L. REV. 721, 741 (2007) ("[I]n some cases, the government may proceed on the sole basis of genetic evidence or marginally probative additional evidence, such as the suspect's proximity to the scene of the offense."); Song et al., supra note 32, at 22 ("Cold-hit cases have prompted courts to confront the question of whether a genetic match constitutes sufficient evidence to uphold a conviction.").
71. See generally Boaz Sangero & Mordechai Halpert, Why a Conviction Should Not Be Based on a Single Piece of Evidence: A Proposal for Reform, 48 JURIMETRICS J. 43, 45 (2007) ("[T]he possibility of a laboratory error... is much more likely than a random match."); Brooke G. Malcom, Comment, Convictions Predicated on DNA Evidence Alone: How Reliable Evidence Became Infallible, 38 CUMB. L. REV. 313 (2008) (discussing concerns about human error and fraud with regard to forensic testing in laboratories).
72. See, e.g., Song et al., supra note 32, at 22-24 (discussing probability of erroneous cold hit matches); Jason Felch & Maura Dolan, DNA: Genes as Evidence, L.A. TIMES, July 20, 2008, at A1 (noting problems with using matches obtained through database searches due to potential for incorrect matches); Humes, supra note 3, at 20-22 (using Puckett as example where possibility exists of coincidental match); Smith, supra note 2 (describing existence of data suggesting that coincidental "low-level matches" are much more common than previously thought); Thompson, supra note 58, at 10 (suggesting probability of coincidental matches is higher from cold hits than from profile matching).
73. See, e.g., MCCARTNEY, supra note 46, at 101 (arguing that courts "need to demand independent corroborative evidence" before convicting defendant on basis of forensic identity evidence alone, though not specifying why); id. at 145 (reporting that Sir Alec Jeffries, the father of modern DNA testing, had qualified his early support for comprehensive national DNA database in Britain by stating that cold hit alone should not be considered conclusive evidence of guilt and had expressed concern about potential discrimination in database samples); NUFFIELD REPORT, supra note 26, at 71 (arguing that because misleading DNA evidence could be given great weight in courtroom, it is "vital that defendants are not convicted on a DNA match alone").
74. Cole & Lynch, supra note 7, at 44; see also Murphy, supra note 70, at 723 (describing "new generation of forensic sciences," including DNA typing, as "stak[ing] a central and indispensable role in the future administration of criminal justice").
75. See Jackson v. Virginia, 443 U.S. 307, 316 (1979) (describing reasonable doubt standard as necessary to protect due process of law); see also Herrera v. Collins, 506 U.S. 390, 402 (1993) (reaffirming that "the Jackson inquiry is aimed at determining whether there has been an independent constitutional violation-i.e., a conviction based on evidence that fails to meet the Winship standard," requiring proof beyond reasonable doubt on each element of offense as matter of due process).
76. See Jackson, 443 U.S. at 320 (holding that sufficiency requires more than "mere modicum" of evidence).
77. See, e.g., Richardson v. United States, 468 U.S. 317, 318 (1984) (citing instances of defendant's motions for acquittal).
78. See, e.g., Jackson, 443 U.S. at 324 (articulating standard for habeas relief based on insufficiency of evidence); see also WAYNE R. LAFAVE ET AL., CRIMINAL PROCEDURE § 24.6(c) (3d ed. 2000) (describing sufficiency standard).
79. Jackson, 443 U.S. at 320 n.14; see also Keith A. Findley & Michael S. Scott, The Multiple Dimensions of Tunnel Vision in Criminal Cases, 2006 WIS. L. REV. 291, 348 (noting that Jackson ruling protected "constitutional requirement of proof beyond a reasonable doubt").
80. See, e.g., People v. Brun, 872 N.Y.S.2d 188, 191 (App. Div. 2009) (reversing conviction for "criminal use of a firearm" on sufficiency grounds where government failed to prove firearm was operable, which is one element of offense).
81. See, e.g., Goodsell v. State, 289 S.W.3d 534, 535-36 (Ark. Ct. App. 2008) (reversing sexual assault conviction for insufficient evidence where alleged victim recanted prior testimony and government failed to corroborate defendant's confession with other substantive evidence of assault).
82. See, e.g., Berry v. State, 212 P.3d 1085, 1094 (Nev. 2009) (holding that government failed to prove that pellet gun met definition of "deadly weapon").
83. In nearly all cases I have found in which an appellant raised a sufficiency claim related to DNA evidence, the appellant's argument was that the DNA is uncorroborated and insufficient on its own. For examples, see infra note 90. In one case, a litigant argued that, notwithstanding corroborative evidence of guilt, DNA evidence was prejudicial because it turned the case into the type of "trial by mathematics" condemned in People v. Collins, 438 P.2d 33, 38-42 (Cal. 1968). See United States v. Wright, 215 F.3d 1020, 1028 (9th Cir. 2000) (rejecting appellant's argument on grounds that "Collins is not a DNA case" and is "simply inapplicable"). In Collins, the court reversed the conviction of one of the defendants due to prejudicial testimony given by a mathematician who improperly used the product rule to estimate the probability that the defendants were guilty based on various characteristics they shared with the victim's and a witness's description of the suspects. 438 P.2d 33. To bring an argument in a pure cold hit case based on Collins, a defendant would have to argue that the match statistic as calculated was somehow unreliable because of a lack of independence among variables. Should future studies show that currently reported RMPs are indeed unreliable, such arguments may gain traction.
84. Scholars appear to be in agreement on this point. See, e.g., David J. Balding & Peter Donnelly, How Convincing Is DNA Evidence?, 368 NATURE 285, 285-86 (1994) (using version of Bayes' Theorem to illustrate calculation of source probability); Kaye, supra note 6, at 491-92 (explaining how to determine source probability using simplified version of Bayes' Theorem); Thompson, supra note 58, at 11 n.2 (noting that size of unrelated population multiplied by probability of finding at least one additional individual with same genetic profile gives simplified estimate of chance of finding at least one match). Professor Richard Friedman offers a straightforward statement for the non-mathematician regarding the relevant terms used in Bayesian analysis: A simple statement of Bayes' Theorem uses three terms. One is the prior odds of a proposition-that is, the odds as assessed before receipt of the new evidence. The second is the posterior odds of the proposition-that is, the odds that the proposition is true as assessed after receipt of the new evidence. And the third is the likelihood ratio. Simply defined, the likelihood ratio of a given body of evidence with respect to a given proposition is the ratio of the probability that the evidence would arise given that the proposition is true to the probability that the evidence would arise given that the proposition is false. Richard D. Friedman, Commentary, A Presumption of Innocence, Not of Even Odds, 52 STAN. L. REV. 873, 875 (2000).
85. California Quick Facts from the U.S. Census Bureau, http://quickfacts.census.gov/ qfd/states/06000.html (last visited July 24, 2010).
86. Absent non-DNA evidence of guilt, our initial estimate of the chance that the defendant is the source is the inverse of the suspect population.
87. Using Bayes' Theorem, the odds that the defendant is the source (or posterior odds) would be the prior odds that the defendant is the source multiplied by the likelihood ratio, which is the inverse of the RMP-that is, the chance of the match assuming the defendant is the source (one), divided by the chance of the defendant matching, assuming he is not the source (the RMP). In the example above, the posterior odds would be: While the posterior "odds" express the ratio of the probability the defendant is the source to the probability he is not, the posterior "probability" would be equal to (odds)/(odds1) = 10/11 = 91%. See NAT'L RESEARCH COUNCIL, supra note 45, at 132 (1996) (explaining conversion of odds to probability when using Bayes' Theorem). Note that this calculation assumes a uniform prior probability throughout the suspect population. While this is not entirely realistic, since various persons in the suspect population will be more or less likely than the defendant to be the source of the DNA, the government will likely be unable to prove more favorable prior odds without information about those in the suspect population. See Kaye, supra note 6, at 492 (noting assumptions underlying this "simplified version" of Bayes' Theorem); Song et al., supra note 32, at 24 (noting that while assumption that "a priori, each individual in the population is equally likely to have committed the crime" is not precise, it is acceptable estimate to make results "conservative"); Thompson, supra note 58, at 11 n.2 (noting possibility of using uniform prior probability distribution among suspect population). Note also that this formula assumes a zero false-positive rate in the DNA-typing process.
88. See, e.g., McDaniel v. Brown, 130 S. Ct. 665, 670 (2010) ("It is further error to equate source probability with probability of guilt, unless there is no explanation other than guilt for a person to be the source of crime-scene DNA.").
89. As noted previously, the nature of the evidence in most pure cold hit cases makes these defenses untenable. See supra note 32 (discussing sexual assault cases in which type of evidence available forecloses possibility of innocent presence).
90. See People v. Soto, 35 Cal. Rptr. 2d 846, 859 (Ct. App. 1994), aff'd, 981 P.2d 958 (Cal. 1999) (describing defendants' attempts to argue that cold hit is legally insufficient); State v. Abdelmalik, 273 S.W.3d 61, 65 (Mo. Ct. App. 2008) (same); People v. Rush, 630 N.Y.S.2d 631, 632 (Sup. Ct. 1995) (same); State v. Hunter, 861 N.E.2d 898, 901 (Ohio Ct. App. 2006) (same); Roberson v. State, 16 S.W.3d 156, 159 (Tex. App. 2000) (same).
91. See Soto, 35 Cal. Rptr. 2d at 859 (describing defendant's argument that DNA evidence should have to be corroborated because it only brought Soto "within a class of potential suspects"); Abdelmalik, 273 S.W.3d at 66 (describing defendant who argued that DNA match alone did not provide "substantial evidence of identity"); Rush, 630 N.Y.S.2d at 632 (recounting testimony of DNA expert who distinguished fingerprinting from DNA matching in terms of former's ability to result in "absolute identification," contrasted with latter's lack of ability to do so); Roberson, 16 S.W.3d at 159 (describing appellant who argued "that DNA evidence is not an inclusionary tool but one of probability," given that "DNA testing can exclude someone as the donor of semen, but it cannot definitively link someone to a crime without other evidence"). In Hunter, the appellant argued "that the jury clearly lost its way in giving credence to the DNA results when literally no other evidence linked appellant" to the charged stranger-rape. 861 N.E.2d at 901. His claim appeared to be not that the DNA was insufficient because the probability of a coincidental match was too great, but that the DNA was not sufficiently reliable because of alleged gaps in the chain of custody. Id.
92. Cf. infra Part III.C (arguing that very high source probabilities can justify conviction).
93. For cases rejecting sufficiency challenge, see, for example, Abdelmalik, 273 S.W.3d at 66, Rush, 630 N.Y.S.2d at 634, Hunter, 861 N.E.2d at 901, State v. Toomes, 191 S.W.3d 122, 131 (Tenn. Crim. App. 2005), and Roberson, 16 S.W.3d at 171.
94. See McDaniel v. Brown, 130 S. Ct. 665, 670 (2010) (acknowledging this fallacy as "the prosecutor's fallacy"); Brief of 20 Scholars of Forensic Evidence as Amici Curiae in Support of Respondent, at 12 & nn.7-9, McDaniel v. Brown, 130 S. Ct. 665 (2010) (No. 08- 559) [hereinafter McDaniel Amici Brief] (noting frequency of "fallacy of the transposed conditional" and difference between that fallacy and similar "prosecutor's fallacy," where government mistakes RMP for chance that defendant is innocent); William C. Thompson, Letter to the Editor, The Prosecutor's Fallacy in George Clarke's Justice and Science: Trials and Triumphs of DNA Evidence, 54 J. FORENSIC SCI. 504, 504 (2009) (noting literature documenting "well-known prosecutor's fallacy" and pointing out Judge Clarke's own fallacious statements conflating RMP and chance that observed match is coincidental).
95. The tendency of courts, litigants, and journalists to misunderstand the RMP and source probability may also be related to their near-exclusive focus in the last few years on the so-called "database search controversy," a dispute among statisticians as to how best to communicate the strength of cold hit evidence to a jury by using the RMP and/or the chance of finding at least one match in the offender database searched (the "database match probability," or DMP). See, e.g., People v. Nelson, 185 P.3d 49, 62-64 (Cal. 2008) (recounting four different methods of analyzing statistical significance of match); United States v. Jenkins, 887 A.2d 1013, 1019-20 (D.C. 2005) (describing "raging debate" among scientists over "which method of probability determination is the most significant for expressing a cold hit"). Ultimately, it is the chance of finding the profile in the potential suspect population, not the offender database, that is the relevant probability for determining the probative value of a cold hit. The concepts of the RMP and DMP (approximately the RMP multiplied by the size of the database), and the difference between those two probabilities, are not well understood by litigants. See generally David H. Kaye, Case Comment, People v. Nelson: A Tale of Two Statistics, 7 LAW PROBABILITY & RISK 249 (2008) (portraying Nelson court's difficulty in comprehending difference in probative value of DMP versus RMP). Ironically, in a pure cold hit case, the DMP may well be a less favorable statistic to the defendant than the source probability.
96. See McDaniel Amici Brief, supra note 94, at 14-15 (distingushing between RMP and source probability by noting that "no meaningful conclusions" about source probability may be drawn "from DNA evidence alone").
97. S.W.3d 122, 129 (Tenn. Crim. App. 2005).
98. Id. at 131. The court noted that it was not "announcing an iron-clad principal [sic] that DNA evidence, without corroboration, is always sufficient to support a conviction," and noted the "unusual feature" in Toomes's case that "two separate DNA comparisons were performed," one of Toomes and one of his twin brother, the man first implicated by the database cold hit. Id. at 131 n.4.
99. See Brief of Appellant Darrell Toomes at 4-6, State v. Toomes, 191 S.W.3d 122 (Tenn. Ct. App. 2005) (No. W2005-00517-CCA-R3-CD) (raising sufficiency claim on grounds of disputed aggravating factor, but not DNA).
100. 861 N.E.2d 898, 901 (Ohio Ct. App. 2006).
101. 16 S.W.3d 156, 161, 172 (Tex. App. 2000).
102. Id. at 167.
103. 35 Cal. Rptr. 2d 846, 859 n.27 (Ct. App. 1994), aff'd, 981 P.2d 958 (Cal. 1999). While the court did "take judicial notice" of the fact that there were only 30.4 million people in California at the time of the offense, id. at 851 n.13, an apparent sign that it understood the relevance of the number of potential suspects, it did not discuss why the potential suspect population would be limited to California residents, or why an RMP of 1 in 189 million among Hispanics would conclusively show that profile was unique in the relevant population.
104. 273 S.W.3d 61, 66 (Mo. Ct. App. 2008).
105. 630 N.Y.S.2d 631, 632, 634 (Sup. Ct. 1995). In at least one other case, a court did not explicitly commit the fallacy of the transposed conditional but appeared to reject a sufficiency claim with little analysis, and with no discussion of the RMP or the size of the suspect population. In Springfield v. State, 860 P.2d 435 (Wyo. 1993), a member of the Crow tribe was accused of a stranger rape in Sheridan, Wyoming. The court rejected a sufficiency claim without analysis based on the defendant fitting a general description of the perpetrator and a DNA match with an RMP of 1 in 221,000 for the American-Indian population and 1 in 17 million for the African-American population. Id. at 447-49. Springfield was three-fourths Crow and one-fourth African American. Id. at 437. Springfield was originally identified based on a lead related to the victim's missing camera rather than a cold hit, and the only evidence from the government at trial regarding the camera was exculpatory of Springfield. Id. at 437-38. While the victim testified at trial that Springfield "definitely resembles" her attacker, she could not identify him before trial, and her statement may merely have signified that he matched the description. Id. at 449. The court concluded that "the evidence was sufficient for reasonable and rational individuals to conclude that the appellant was the perpetrator." Id. at 449 (internal citation omitted).
106. In British courts, as in the United States, the prosecutor bears the burden of proving sufficient evidence for every essential element of the crime. If the prosecution fails to meet this burden, the court should direct a verdict of acquittal. ADRIAN KEANE, THE MODERN LAW OF EVIDENCE 37 (7th ed. 2008).
107. [2000] EWCA (Crim) 88, [3]; see also MCCARTNEY, supra note 46, at 95 (discussing Lashley); Redmayne, Rationality, supra note 50, at 880 (asserting that RMP of defendant DNA match was around 1 in 4 million).
108. [2000] EWCA (Crim) 89, [8], [10].
109. Id. at [8], [9], [11], [12].
110. Id. at [18]. Interestingly, the court later noted that even assuming the accuracy of a post-appeal recalculation putting the RMP for the brothers at 1 in 29,000, the evidence would still be insufficient: "[A]t the end of the day, greater though those odds are, they do nothing to eliminate the possible brother." Id. at [24]-[25].
111. [1996] 2 Crim. App. 467, 468, 470; see also Kaye, supra note 6, at 484 (discussing Adams). The crime occurred in a town of just over 150,000 men between the ages of eighteen and sixty, and there seemed to be little dispute over treating the potential suspect population as at least 200,000, on the assumption that one could not state that there was more than a 75% chance that the perpetrator came from the town itself. Adams, [1996] 2 Crim. App. at 472-74; see also Kaye, supra note 6, at 484 (describing source population issues).
112. Adams, [1996] 2 Crim. App. at 468; see Peter Donnelly, Appealing Statistics, SIGNIFICANCE, Mar. 2005, at 46, 46 (discussing Adams).
113. Adams, [1996] 2 Crim. App. at 467.
114. Id. at 470.
115. The appellate court quashed Adams's first conviction, based on concerns that the trial court had failed to properly instruct the jury. Id. at 482. In an appeal from Adams's subsequent second conviction on the same evidence, the court upheld the conviction. R v. Adams, [1997] EWCA (Crim) 2474.
116. Donnelly, supra note 112, at 48 (noting that Court of Appeal now "advocate[s] that judges should summarise cases in the... way" described above).
117. See, e.g., State v. Skipper, 637 A.2d 1101, 1106, 1108 (Conn. 1994) (concluding that "[t]he utilization of Bayes' Theorem by the prosecution," specifically admission of evidence that prior probability of defendant's paternity of victim's child in sexual assault case was fifty percent, violated presumption of innocence and was reversible error); State v. Spann, 617 A.2d 247 (N.J. 1993) (reversing sexual assault conviction based on prejudicial admission of arbitrary fifty percent prior probability of paternity and, while not precluding potential discussion of prior probabilities on retrial, noting controversy surrounding "the evidentiary use of Bayes' Theorem at all," and Professor Kaye's suggestion that Bayes' Theorem be excluded from criminal cases).
118. See Friedman, supra note 84, at 873-74 (noting "continuous battle" between frequentists, or "Bayesioskeptics," and Bayesians); F. Taroni & A. Biedermann, Inadequacies of Posterior Probabilities for the Assessment of Scientific Evidence, 4 LAW PROBABILITY & RISK 89, 92 (2005) (discussing concern that while "Bayesian view has secured its position as a coherent framework for evaluating evidence in forensic science," experts might usurp jury function by suggesting prior probabilities).
119. Some disputes should be easy to resolve in the government's favor, such as in Regina v. Adams. See supra note 111 and accompanying text (discussing facts of Adams that court found sufficient to justify finding of proof beyond a reasonable doubt). Others might be more difficult and may preclude resolution of the case at the sufficiency stage. See discussion infra Part V (discussing such cases).
120. See, e.g., Commonwealth v. Sullivan, 482 N.E.2d 1198, 1200 (Mass. App. Ct. 1985) (holding that judge's charge requiring jurors to find only "above fifty percent" chance of guilt was error because "[t]he idea of reasonable doubt is not susceptible to quantification; it is inherently qualitative"); McCullough v. State, 657 P.2d 1157, 1159 (Nev. 1983) (holding that judge's charge equating reasonable doubt with 7.5 on scale from 0 to 10 was reversible error and noting that "[a]ny attempt to quantify [the concept of reasonable doubt] may impermissibly lower the prosecution's burden of proof, and is likely to confuse rather than clarify"); Rita James Simon & Linda Mahan, Quantifying Burdens of Proof: A View from the Bench, the Jury, and the Classroom, 5 LAW & SOC'Y REV. 319, 329 (1971) (quoting judge's view that "[p]ercentages or probabilities simply cannot encompass all the factors, tangible and intangible, in determining guilt-evidence cannot be evaluated in such terms"). But see Jack B. Weinstein & Ian Dewsbury, Comment on the Meaning of "Proof Beyond a Reasonable Doubt," 5 LAW PROBABILITY & RISK 167, 172-73 (2006) (supporting adoption of jury instruction that would suggest possible quantification of proof beyond reasonable doubt as ninety-five percent).
121. Commonwealth v. Rosa, 661 N.E.2d 56, 63 (Mass. 1996).
122. James Franklin, Case Comment-United States v. Copeland, 369 F. Supp. 2d 275 (E.D.N.Y. 2005): Quantification of the "Proof Beyond Reasonable Doubt" Standard, 5 LAW PROBABILITY & RISK 159, 159-60 (2006) (listing objections to quantification of reasonable doubt as prologue to thesis that courts should at least set floor, such as 80%, below which evidence should be deemed insufficient); see also 9 JOHN HENRY WIGMORE, EVIDENCE IN TRIALS AT COMMON LAW § 2497 (James H. Chadbourn ed., rev. ed. 1981) ("[N]o one has yet invented or discovered a mode of measurement for the intensity of human belief.").
123. See 32A C.J.S. Evidence § 1627 (2008) ("The preponderance-of-the-evidence standard of proof requires that the factfinder determine whether a fact sought to be proved is more probable than not....").
124. See Franklin, supra note 122, at 159 (noting objections to quantifying reasonable doubt but advocating floor of 80% below which jurors should be instructed to acquit); see also discussion infra Part IV (proposing 99.9% source probability threshold before case is sent to jury).
125. See, e.g., Laura I. Appleman, The Lost Meaning of the Jury Trial Right, 84 IND. L.J. 397, 406 (2009) (noting juries were selected from "the immediate neighborhood" and were "chosen for their knowledge of the crime or their ability to find out").
126. See Barbara J. Shapiro, "To a Moral Certainty": Theories of Knowledge and Anglo- American Juries 1600-1850, 38 HASTINGS L.J. 153, 155-56 (1986) (discussing shift toward juries as third parties and introduction of credibility standards); see also Appleman, supra note 125, at 406 (noting that juries began to lose their ability to personally investigate cases when society grew complex).
127. Shapiro, supra note 126, at 156-57.
128. Id.
129. Elisabeth Stoffelmayr & Shari Seidman Diamond, The Conflict Between Precision and Flexibility in Explaining "Beyond a Reasonable Doubt," 6 PSYCHOL. PUB. POL'Y & L. 769, 770 (2000) (discussing origin of "moral certainty" as strategic counterpoint to intellectual challenges to "absolute certainty" and "metaphysical certainty").
130. Shapiro, supra note 126, at 157-58.
131. ARTHUR P. WILL, A TREATISE ON THE LAW OF CIRCUMSTANTIAL EVIDENCE 303 (Philadelphia, T. & J.W. Johnson & Co. 1896) (contrasting "mathematical" with "moral certainty").
132. Robert C. Power, Reasonable and Other Doubts: The Problem of Jury Instructions, 67 TENN. L. REV. 45, 65 (1999); see also Victor v. Nebraska, 511 U.S. 1, 10-11 (1994) ("Moral evidence has for its subject the real but contingent truths and connections, which take place among things actually existing.... With regard to moral evidence, there is... real evidence on both sides. On both sides, contrary presumptions, contrary testimonies, contrary experiences must be balanced." (quoting 1 WORKS OF JAMES WILSON 518-19 (James DeWitt Andrews ed., Chicago, Callaghan & Co. 1896))).
133. Shapiro, supra note 126, at 158 (emphasis added) (quoting JOHN WILKINS, OF THE PRINCIPLES AND DUTIES OF NATURAL RELIGION 8 (London 1675)).
134. Consider, for example, the instruction in Commonwealth v. Webster: [E]very thing relating to human affairs, and depending on moral evidence, is open to some possible or imaginary doubt. It is that state of the case, which... leaves the minds of jurors in that condition that they cannot say they feel an abiding conviction, to a moral certainty, of the truth of the charge.... [T]he evidence must establish the truth of the fact to a reasonable and moral certainty; a certainty that convinces and directs the understanding, and satisfies the reason and judgment.... This we take to be proof beyond reasonable doubt.... 59 Mass. (5 Cush.) 295, 320 (1850). See generally Victor, 511 U.S. at 11-12 (describing moral certainty as "equivalent" to reasonable doubt and discussing history of concepts).
135. WILLIAM WILLS, AN ESSAY ON THE PRINCIPLES OF CIRCUMSTANTIAL EVIDENCE 262 (Sir Alfred Wills ed., 5th ed. 1905).
136. Steve Sheppard, The Metamorphoses of Reasonable Doubt: How Changes in the Burden of Proof Have Weakened the Presumption of Innocence, 78 NOTRE DAME L. REV. 1165, 1197 (2003) (discussing THOMAS STARKIE, A PRACTICAL TREATISE OF THE LAW OF EVIDENCE, AND DIGEST OF PROOFS IN CIVIL AND CRIMINAL PROCEEDINGS 513 (Benjamin Gerhard ed., Philadelphia, T. & J.W. Johnson, Law Booksellers 1842)). To be sure, philosophers "dubbed this sort of certainty 'moral' not because it had anything to do with ethics or morality but to contrast it with 'mathematical' certainty of the sort traditionally associated with rigorous demonstration." LARRY LAUDAN, TRUTH, ERROR, AND CRIMINAL LAW: AN ESSAY IN LEGAL EPISTEMOLOGY 33 (2006). When I describe a juror who lacks moral certainty I do not mean one who feels the evidence or prosecution itself is immoral or unethical, though a juror could feel that way and nullify. Rather, I mean a juror who is not convinced by the evidence to the point that he has an actual belief, in accord with his understanding and conscience, in the guilt of the accused.
137. WILLS, supra note 135, at 6 (defining such "belief" as underlying concept of "moral certainty").
138. Sheppard, supra note 136, at 1197 (quoting THOMAS STARKIE, A PRACTICAL TREATISE OF THE LAW OF EVIDENCE, AND DIGEST OF PROOFS IN CIVIL AND CRIMINAL PROCEEDINGS 477 (Benjamin Gerhard ed., Philadelphia, T. & J.W. Johnson, Law Booksellers 1842)).
139. See, e.g., Petitioner's Brief on the Merits at 25, Sandoval v. California, consolidated with Victor v. Nebraska, 511 U.S. 1 (1994) (No. 92-9049) (noting that Webster's defines "moral certainty" as "based on an inner conviction" (citing WEBSTER'S THIRD INTERNATIONAL DICTIONARY 1468 (1986))).
140. This is certainly the manner in which the moral certainty requirement is characterized in the existing literature. See, e.g., Power, supra note 132, at 65 (noting philosophers after 1600 recognized that "absolute knowledge based on moral evidence was impossible" and developed concept of moral certainty to "provide[] the highest degree of conviction possible from human... sources"); Shapiro, supra note 126, at 158 (same). Ironically, "moral certainty" has become the focus of litigation by defendants who argue that the phrase causes jurors to convict, in spite of a weak case, because of the moral gravity of the charged offense. See, e.g., Victor v. Nebraska, 511 U.S. 1, 13-14 (1994) (discussing defendant's argument "that the phrase 'moral certainty' has lost its historical meaning, and that a modern jury would understand it to allow conviction on proof that does not meet the beyond a reasonable doubt standard").
141. Simon A. Cole & Rachel Dioso-Villa, CSI and Its Effects: Media, Juries, and the Burden of Proof, 41 NEW ENG. L. REV. 435, 468 (2007); see also Susan Haack, Irreconcilable Differences? The Troubled Marriage of Science and Law, 72 LAW & CONTEMP. PROBS. 1, 12 (2009) (noting modern view that "scientific inquiry is by nature tentative and thoroughly fallibilist").
142. While the phrase "moral certainty" has been removed from many modern jury instructions out of a concern that it might be misunderstood to the detriment of criminal defendants, the Supreme Court has recognized the phrase as synonymous with "beyond a reasonable doubt." Victor, 511 U.S. at 11-12 (describing these phrases as "equivalent").
143. Sheppard, supra note 136, at 1210-11 (quoting State v. Jefferson, 10 So. 199, 200 (La. 1891)); see also State v. Gould, 395 So. 2d 647, 656-57 (La. 1980) (quoting and applying Jefferson); WILLS, supra note 135, at 6 (noting that belief "may be of various degrees, from moral certainty, the highest, to that of mere probability, the lowest").
144. See, e.g., Brandon L. Garrett & Peter J. Neufeld, Invalid Forensic Science Testimony and Wrongful Convictions, 95 VA. L. REV. 1, 5 & n.9 (2009) (discussing scandals involving faulty work from preeminent crime labs).
145. See The Innocence Project, Facts on Post-Conviction DNA Exonerations, http:// www.innocenceproject.org/Content/351.php (last visited July 26, 2010) (providing statistics on post-conviction DNA exonerations). Any arguments for caution in allowing a DNA match statistic to be the sole basis for conviction does not apply to the use of DNA to exonerate suspects. While interpretive errors in assuming number of contributors and testing errors such as allelic dropout could render a reported exclusion unreliable, an exclusion would not be challenged on the basis of the strength or weakness of the match statistic. Only a match needs a corresponding statistic to explain its probative value; an exclusion is simply the absence of a match. See generally Murphy, supra note 17, at 492-93 (discussing difference between using DNA as evidence of inculpation and evidence of exculpation).
146. See, e.g., Jonathan J. Koehler & Daniel N. Shaviro, Veridical Verdicts: Increasing Verdict Accuracy Through the Use of Overtly Probabilistic Evidence and Methods, 75 CORNELL L. REV. 247, 250 (1990) (suggesting that factfinder motivated solely by verdict accuracy would embrace overtly probabilistic methodology); Daniel Shaviro, Commentary, Statistical-Probability Evidence and the Appearance of Justice, 103 HARV. L. REV. 530, 532-33 (1989) (suggesting that trial-by-mathematics critique values appearance of justice over verdict accuracy). Many jurisdictions prohibit convictions based on uncorroborated confessions, but this rule is related to the corpus delicti doctrine and is "designed to prevent the conviction of the coerced and mentally unstable for fictitious crimes," rather than concerned with the accuracy of the verdict. David A. Moran, In Defense of the Corpus Delicti Rule, 64 OHIO ST. L.J. 817, 817 (2003).
147. See, e.g., Charles R. Nesson, Reasonable Doubt and Permissive Inferences: The Value of Complexity, 92 HARV. L. REV. 1187, 1192-93 (1979) (presenting prison yard hypothetical). L. Jonathan Cohen offers a similar hypothetical where 1000 people are seated at a rodeo but only 499 have paid. Cohen argues that "our intuitions of justice revolt against" imposing liability on any one patron for nonpayment, though the chance of nonpayment is greater than fifty percent for any given patron. L. Jonathan Cohen, Subjective Probability and the Paradox of the Gatecrasher, 1981 ARIZ. ST. L.J. 627, 627.
148. While these are the essential facts of Smith v. Rapid Transit, Inc., 58 N.E.2d 754 (Mass. 1945), the case has come to be referred to in the literature as the "blue bus hypothetical" and is commonly referenced in the trial-by-mathematics context. See, e.g., Nesson, supra note 147, at 1194 (citing Smith in discussion of puzzling argument that, even in civil contexts, "high likelihood" but "starkly numerical" cases should be thrown out); Tribe, supra note 14, at 1340-41 (discussing Smith rule approvingly); id. at 1372-73 (describing imposition of liability based on purely probabilistic evidence as "intrinsically immoral").
149. Nesson, supra note 147, at 1194.
150. Smith, 58 N.E.2d at 755 (citation and internal quotation marks omitted).
151. WILLS, supra note 135, at 6.
152. Charles Nesson, The Evidence or the Event? On Judicial Proof and the Acceptability of Verdicts, 98 HARV. L. REV. 1357, 1361 (1985) (providing this example); see also Gary L. Wells, Naked Statistical Evidence of Liability: Is Subjective Probability Enough?, 62 J. PERSONALITY & SOC. PSYCHOL. 739, 749 (1992) (noting Nesson's playing card example and fact that, in second scenario, person who guessed incorrectly based on statistics "was not to blame," having "made the correct decision," as "it was mere chance that produced the improbable outcome").
153. Laurence H. Tribe, An Ounce of Detention: Preventive Justice in the World of John Mitchell, 56 VA. L. REV. 371, 386 (1970).
154. Shaviro, supra note 146, at 540.
155. Tribe, supra note 14, at 1373.
156. Nesson, supra note 147, at 1196; see also David Kaye, The Laws of Probability and the Law of the Land, 47 U. CHI. L. REV. 34, 40 (1979) ("[W]e would prefer not to advertise the fact that we are willing to sacrifice one innocent person in order to secure the conviction of nineteen guilty ones."); Koehler & Shaviro, supra note 146, at 252 (observing that verdicts based on purely probabilistic evidence "make[] the risk of error explicit"); Shaviro, supra note 146, at 533 (noting objection based on "clarity of the risk of error").
157. By a "high subjective certitude" of guilt, I mean that the juror herself views the evidence as establishing a high probability of guilt. See Wells, supra note 152, at 741 (noting Laurence Tribe's different but related hypothesis that jurors might have only, say, 50% subjective certitude of culpability based on base-rate evidence suggesting 80% chance of culpability).
158. See id. at 741-48 (describing experiment).
159. Id. at 741.
160. Id. at 742-43.
161. Id. at 744.
162. See, e.g., Kevin Jon Heller, The Cognitive Psychology of Circumstantial Evidence, 105 MICH. L. REV. 241, 245 (2006) (discussing Wells Effect). See generally Keith E. Niedermeier et al., Jurors' Use of Naked Statistical Evidence: Exploring Bases and Implications of the Wells Effect, 76 J. PERSONALITY & SOC. PSYCHOL. 533 (1999) (describing subsequent testing of Wells Effect).
163. Niedermeier et al., supra note 162, at 533, 541-42.
164. Id. at 533 (emphasis omitted).
165. Id. at 542 (discussing Nancy Pennington & Reid Hastie, The Story Model for Juror Decision Making, in INSIDE THE JUROR: THE PSYCHOLOGY OF JUROR DECISION MAKING 192-221 (Reid Hastie ed., 1993)).
166. Id.
167. See, e.g., Heller, supra note 162, at 270 (noting that "abstract" probabilistic evidence "does not help [jurors] imagine how the defendant actually committed the crime"); RANDOLPH N. JONAKAIT, THE AMERICAN JURY SYSTEM 239 (2003) ("Jurors-and no doubt judges as well-often have difficulty integrating technical evidence into commonsense narratives.... This is particularly true of probabilistic information.").
168. See, e.g., Heller, supra note 162, at 283 (explaining "certainty effect" whereby individuals put less weight on merely probable outcomes).
169. LYNCH ET AL., supra note 70, at 345.
170. Again, the RMP is not equivalent to the source probability, and conflation of the two would be the so-called "fallacy of the transposed conditional." But if the RMP in a given case is lower than 1 in 6 trillion or so, the resulting estimate of the source probability- even assuming that the potential suspect population is the world's population of six billion- will exceed 99.9%. See discussion infra Part V (arguing for 99.9% source probability as possible starting point for legal sufficiency).
171. LYNCH ET AL., supra note 70, at 345.
172. See Edward F. Wright et al., Factors Affecting the Use of Naked Statistical Evidence of Liability, 136 J. SOC. PSYCHOL. 677, 685 (1996) (discussing study results supporting idea that high levels of objective probability moderate reluctance to rely upon naked statistics).
173. Id.
174. Id. at 685-86.
175. Id. at 686; see also Heller, supra note 162, at 301 ("[A]lthough jurors are extremely sensitive to deviations away from certainty, research indicates they are generally willing to convict on the basis of probabilistic evidence that... establishes a 0.995 likelihood of the defendant's guilt.").
176. Jonathan J. Koehler, The Psychology of Numbers in the Courtroom: How To Make DNA-Match Statistics Seem Impressive or Insufficient, 74 S. CAL. L. REV. 1275, 1280 (2001) (emphasis omitted).
177. Heller, supra note 162, at 283; cf. United States v. Veysey, 334 F.3d 600, 605 (7th Cir. 2003) (noting that in "obvious cases of fingerprint and DNA evidence," astronomically high match probabilities weaken "the case against allowing 'naked' statistical evidence to carry" burden of proof).
178. JUDITH JARVIS THOMSON, RIGHTS, RESTITUTION, AND RISK: ESSAYS IN MORAL THEORY 247-48 (William Parent ed., 1986).
179. See, e.g., Karen E. Jenni & George Loewenstein, Explaining the "Identifiable Victim Effect," 14 J. RISK & UNCERTAINTY 235, 237-41 (1997) (reporting causes of "identifiable victim effect" where people are willing to expend greater resources to save identified lives than statistical lives). The "identifiable victim effect" would explain a juror's decision to acquit in the prison-yard example even if the yard had 10,000 prisoners (where the chance of any particular defendant being the one known innocent defendant is only 1 in 10,000), but to convict in a case in which there is no known innocent being falsely accused, but the chance of the defendant on trial potentially being falsely accused given the match evidence is 1 in 10,000.
180. D. Michael Risinger, The Irrelevance, and Central Relevance, of the Boundary Between Science and Non-science in the Evaluation of Expert Witness Reliability, 52 VILL. L. REV. 679, 708 n.86 (2007); cf. Seth F. Kreimer & David Rudovsky, Double Helix, Double Bind: Factual Innocence and Postconviction DNA Testing, 151 U. PA. L. REV. 547, 609 (2002) (describing post-conviction DNA exoneration evidence as "provid[ing] morally certain proof" of innocence).
181. See, e.g., Nesson, supra note 152, at 1379 ("[G]iven the evidence's statistical nature, the public need not defer to the conclusion on the ground that the factfinder is in a better position to evaluate a witness's demeanor."); Nesson, supra note 147, at 1196 (arguing that "nothing presented to the jury puts it in any better position to judge" purely probabilistic cases, thereby decreasing confidence in need for, and efficacy of, jury in such cases).
182. Nesson, supra note 147, at 1196; see also Eleanor Swift, Abolishing the Hearsay Rule, 75 CAL. L. REV. 495, 504 (1987) ("One explanation of cases like Smith [v. Rapid Transit] is that only particularistic proof justifies the use of the elaborate system of adjudicative factfinding. Particularistic proof permits decisions to be based on the trier's knowledge and experience, rather than on the indiscriminate application of general probabilities.").
183. Note that the constitutional standard for determining legal sufficiency is the same in state and federal criminal trials. See Jackson v. Virginia, 443 U.S. 307, 315-16 (1979) (establishing legal sufficiency standard as due process standard under Fourteenth Amendment); see also supra notes 75-76 and accompanying text.
184. Cf. Franklin, supra note 122, at 165 (noting that difficulties in deciding upon particular numerical threshold for reasonable doubt should not preclude legal community from imposing floor of 80%).
185. See supra note 82 (citing case in which pellet gun was held not to meet definition of deadly weapon).
186. See, e.g., Craig R. Callen, Cognitive Science and the Sufficiency of "Sufficiency of the Evidence" Tests, 65 TUL. L. REV. 1113, 1121 (1991) ("[C]ourts also have an interest in using directed verdicts ... as tools to promote uniformity or predictability in the enforcement of legal rules.").
187. Presumably for this reason, judges in some jurisdictions direct a verdict of not guilty in operating-under-the-influence (OUI) cases where the breathalyzer result is close to the legal limit because of the machines' margin of error. See, e.g., Randy S. Chapman, Successfully Defending an OUI Case, in 2 MASSACHUSETTS BASIC PRACTICE MANUAL § 13.5.3(c) (3d ed. 2009 & Supp. 2010) (noting that some judges in Massachusetts "will allow a motion for a required finding of not guilty as to the per se theory" when breath test results are close to state's legal limit).
188. Tribe, supra note 14, at 1374. Tribe further explains why systemic legitimacy in this context matters: Methods of proof that impose moral blame ... on the basis of evidence that fails to penetrate or convince the untutored contemporary intuition threaten to make the legal system seem even more alien and inhuman than it already does to distressingly many. There is at stake not only the further weakening of the confidence of the parties and of their willingness to abide by the result, but also the further erosion of the public's sense that the law's fact-finding apparatus is functioning in a somewhat comprehensible way, on the basis of evidence that speaks, at least in general terms, to the larger community that the processes of adjudication must ultimately serve. Id. at 1375-76. See also Nesson, supra note 152, at 1367 (arguing that public "acceptance" of verdicts is critical to "promot[ing] our assimilation of the behavioral message embodied in the underlying substantive rule").
189. Eric Magnusson, Incomprehension and Miscomprehension of Statistical Evidence: An Experimental Study, 1993 LAW MED. & CRIM. JUST. CONF., AUSTL. INST. CRIMINOLOGY 3, http://www.aic.gov.au/events/aic%20upcoming%20events/1993/~/media/confer ences/ medicine/magnus.ashx; see also Heller, supra note 162, at 284 n.237 (noting significance of this result in Magnusson's study).
190. See Koehler, supra note 176, at 1293-96 (discussing experiment). The author does not make clear whether the attorneys in the mock trial presented the jury with only the RMPs or attempted to suggest a source probability based on the RMP and the size of the likely suspect population. Given that Houston is a city of several million, the source probability even given an RMP of 1 in 1 billion may have been lower than 99.9%. It is surely more likely that many jurors conflated the RMP with the source probability and yet still believed an objectively high source probability, absent other evidence of guilt, was less than fully compelling.
191. Koehler cited as possible explanations for this phenomenon the Wells Effect and a few jurors' curious belief that a low RMP suggested a low source probability. Id. at 1295-96.
192. At least one scholar has called for fingerprint analysts to move away from the individualization model and testify-in line with DNA analysts-in probabilistic terms, based on population databases and match statistics. See Simon A. Cole, Forensics Without Uniqueness, Conclusions Without Individualization: The New Epistemology of Forensic Identification, 8 LAW PROBABILITY & RISK 233, 249-50 (2009).
193. See Ronald J. Allen & Michael S. Pardo, The Problematic Value of Mathematical Models of Evidence, 36 J. LEGAL STUD. 107, 126 (2007) ("With the lab error rate statistics in particular, perhaps the jurors intuited the limits of the data and wanted better and more appropriate information about this lab.").
194. Budowle et al., supra note 25.
195. See, e.g., NAT'L RESEARCH COUNCIL, DNA TECHNOLOGY IN FORENSIC SCIENCE 92 (1992) ("Regardless of the calculated frequency, an expert should-given the relatively small number of loci used and the available population data-avoid assertions in court that a particular genotype is unique in the population."); Song et al., supra note 32, at 23 ("[T]he lack of certainty makes claims of uniqueness improper to make in the presentation of DNA evidence in court.").
196. See, e.g., Balding, supra note 36, at 260 ("One of our assumptions was a 99.9% criterion for uniqueness, chosen arbitrarily."); Luttman Testimony, supra note 34, at 3-79 to 3- 80 (noting that FBI's uniqueness threshold is RMP equal to about one thousand times U.S. population).
197. I use the term "scientific community" in the same respect courts use it in the Frye context; that is, those scientists qualified "with sufficient training and expertise to permit them to comprehend" the scientific matter at issue. Blackwell v. Wyeth, 971 A.2d 235, 252 (Md. 2009). Here, the scientific matter would be the statistics involved in estimating the probability that a profile is unique. The relevant community in a matter involving DNA match statistics would include those in the derivative sciences of statistics, population genetics, and molecular biology, and not merely-or even primarily-forensic scientists. See, e.g., United States v. Porter, 618 A.2d 629, 634-35 (D.C. 1992) (rejecting limitation of scientific community to forensic scientists for purposes of Frye test).
198. B.S. Weir, Forensics, in 2 HANDBOOK OF STATISTICAL GENETICS 848-49 (2d ed. 2003); see also NAT'L RESEARCH COUNCIL, supra note 45, at 33 ("The definition of uniqueness is outside our province."); Budowle et al., supra note 25 (acknowledging in FBI publication that "[t]he size of the population and the appropriate confidence level to use" in making claims of profile uniqueness are, at bottom, "policy decisions"); Michael J. Saks & Jonathan J. Koehler, The Individualization Fallacy in Forensic Science Evidence, 61 VAND. L. REV. 199, 205 (2008) ("The concept of 'individualization' ... exists only in a metaphysical or rhetorical sense. It has no scientific validity ...."); id. at 218 n.94 ("Although the DNA typing model has much to offer the traditional forensic sciences, offering source identifications at trial for sufficiently low probabilities would not be an implication of the science but an evasion of it in the service of advocacy.").
199. This is especially true given that phrases like "reasonable degree of scientific certainty" are amorphous terms that some experts take to mean merely "more probable than not" and others take to mean, more stringently, that "there was 'no reasonable or practical possibility that someone' other than the defendant" could be responsible. Molly Gena, Comment, Shaken Baby Syndrome: Medical Certainty Casts Doubt on Convictions, 2007 WIS. L. REV. 701, 716 (citations omitted) (discussing standards used to analyze bite marks).
200. See, e.g., ALA. CODE § 26-17-505(a)(1) (LexisNexis 2009) (creating rebuttable presumption with test result indicating 99% probability of paternity); IND. CODE § 31-14-7- 1(3) (West 2008) (setting presumption of paternity at 99% probability); LA. REV. STAT. ANN. § 40:34(E)(5) (Supp. 2009) (setting conclusive presumption at 99.9% probability for purposes of child support); MICH. COMP. LAWS ANN. § 712.11(3) (West Supp. 2010) (setting presumption of paternity at 99% probability); 23 PA. CONS. STAT. ANN. § 4343(c)(2) (West 2001) (setting rebuttable presumption at 99% probability); TENN. CODE ANN. § 24- 7-112(b)(2)(C) (2010) (setting rebuttable presumption at 99% probability); UTAH CODE ANN. § 78B-15-505(1)(a), (2) (LexisNexis 2008) (creating presumption with test results indicating 99% probability of paternity, rebuttable only by contradictory genetic test); WYO. STAT. ANN. § 14-2-705(a)(i) (2009) (setting rebuttable presumption at 99% probability).
201. See Rivera v. Minnich, 483 U.S. 574, 575 (1987) (upholding preponderance of evidence standard for paternity tests). Some states require "clear and convincing evidence" for a posthumous claim of paternity. See, e.g., LA. CIV. CODE. ANN. art. 192 (2008) ("The mother shall prove by clear and convincing evidence both that her former husband is not the father and that her present husband is the father.").
202. See, e.g., Lawrence M. Solan, Refocusing the Burden of Proof in Criminal Cases: Some Doubt About Reasonable Doubt, 78 TEX. L. REV. 105, 126-27 (1999) (noting that judges' responses to survey asking them to quantify reasonable doubt ranged from 80% to 95% and higher, and jurors' responses ranged from upwards of 61%); Weinstein & Dewsbury, supra note 120, at 171 ("Were the parties to agree, we would like to add to our charge: 'In my opinion, a probability of guilt of no less than 95% should be necessary to support a conviction.'").
203. A Stanford mathematician has called currently reported RMPs "ludicrous" and "well beyond the bounds of reality." Keith Devlin, Damned Lies, Devlin's Angle, Mathematics Ass'n of Am. (Oct. 2006), http://www.maa.org/devlin/devlin_10_06.html. Court-ordered studies of the Arizona, Maryland, and Illinois DNA databases suggest that nine-loci matches or even thirteen-loci matches may not be as rare as claimed. Jason Felch & Maura Dolan, How Reliable Is DNA in Identifying Suspects?, L.A. TIMES, July 20, 2008, at A1. The FBI continues to resist such studies of the CODIS database. Berkeley researchers have called for more RMP studies upon determining that, assuming independence of the STR loci, the average chance of a coincidental attribution in a cold hit case is 1 in 3.4 million. Song et al., supra note 32, at 22; see also Murphy, supra note 43 (calling for increased researcher access to CODIS).
204. A false positive rate-even as low as 1 in 10,000 or 100,000-will dwarf a typical RMP in any source probability calculation. See William C. Thompson et al., How the Probability of a False Positive Affects the Value of DNA Evidence, 48 J. FORENSIC SCI. 47, 52 (2003) (showing that with RMP of 1 in 1 billion and prior odds of guilt being 1 in 1000, false positive rate as low as 1 in 10,000 still yields posterior odds of only 10 to 1 in favor of suspect being source of DNA); see also Balding & Donnelly, supra note 84, at 286 ("When extremely small match probabilities are claimed, it seems naive, at best, to ignore the possibility of false-positive results through human error."); Sangero & Halpert, supra note 71, at 45 (arguing against convictions based on single piece of evidence because of possibility of laboratory error).
205. Cole & Lynch, supra note 7, at 47; see also McDaniel Amici Brief, supra note 94, at 15 ("[I]f the other evidence points to someone else, a jury could reasonably conclude that the match ... was just a coincidence."); Balding, supra note 36, at 258, 260 (noting that non-DNA exculpatory evidence may trump even very small RMPs).
206. See [2000] EWCA (Crim) 88, [v], [3], [5], [16] (describing facts in case).
207. See Redmayne, Rationality, supra note 50, at 880 (discussing Lashley).
208. See [2000] EWCA (Crim) 89, [9], [25], [29] (discussing facts and holding retrial not necessary).
209. In resolving factual disputes about the size of the suspect population, courts should be careful not to limit the population to a particular class, such as a particular race or ethnicity, merely because the defendant is a member of that class. See, e.g., People v. Wilson, 136 P.3d 864, 868 (Cal. 2006) (confirming that RMP should not be based on race in absence of evidence that perpetrator was of that race); D.H. Kaye, Logical Relevance: Problems with the Reference Population and DNA Mixtures in People v. Pizarro, 3 LAW PROBABILITY & RISK 211, 211 (2004) ("One strangely persistent fallacy in the interpreta tion of DNA evidence is that the relevant ethnic or racial population in which to estimate a DNA profile frequency necessarily is that of the defendant."). See generally Jonathan Kahn, Race, Genes, & Justice, 74 BROOK. L. REV. 325 (2009) (discussing problems with use of race in DNA evidence analysis presentation). Indeed, in some pure cold hit cases, the defendant is of a different ethnicity than the described perpetrator. See, e.g., People v. Soto, 981 P.2d 958, 961 (Cal. 1999) (comparing victim's original description of rapist to defendant).
210. For example, in the Brandon Mayfield fingerprint case, the absence of records showing that Mayfield had left the country was initially deemed by FBI investigators as evidence of concealment rather than innocence. Simon Cole notes that the bias inherent in having already targeted Mayfield may have contributed to officials' interpretation of his travel history as inculpatory. See Cole & Lynch, supra note 7, at 49-50 (discussing Mayfield story); see also D. Michael Risinger, Cases Involving the Reliability of Handwriting Identification Expertise Since the Decision in Daubert, 43 TULSA L. REV. 477, 594 (2007) ("[I]t is not clear, but is to be expected, that [the handwriting expert] generated his conclusions in the shadow of knowledge of the DNA 'match' variable which undermines reliability quite dramatically.").
211. Kaye, supra note 6, at 491.
212. Id. Multiplying the size of the suspect population by the RMP is, in essence, a "much simplified version" of Bayes' Theorem. Id. at 492; see also Thompson, supra note 58, at 11 n.2 (noting that population multiplied by frequency provides simplified estimate of chance of finding at least one match). This calculation assumes that the prior probability that the defendant is the source is one over the number of individuals in the potential suspect population- a "uniform prior probability distribution over most of the male population in the geographic region." Kaye, supra note 6, at 492. While the assumption that "a priori, each individual in the population is equally likely to have committed the crime" is "obviously not true," Song et al., supra note 32, at 24, it is presumably justified absent evidence from the government about other potential suspects.
213. Kaye, supra note 6, at 491-92.
214. If the prior odds of Puckett being the source were about 1 to 2 million, multiplying by the inverse of the RMP (1.1 million) would yield posterior odds of around 1 to 2, or only about a 33% source probability.
215. See, e.g., FED. R. EVID. 413(a) (providing that such evidence "may be considered for its bearing on any matter to which it is relevant"); CAL. EVID. CODE § 1108(a) (West 2009) (providing that evidence of commission of another sexual offense is "not inadmissible").
216. Courts are split as to whether the jury may properly consider a non-testifying defendant's courtroom demeanor as evidence. Compare United States v. Schuler, 813 F.2d 978, 981 (9th Cir. 1987) (holding prosecutor's comment about non-testifying defendant's demeanor improper), with Commonwealth v. Smith, 444 N.E.2d 374, 380 (Mass. 1983) (allowing prosecutor to comment on defendant's smirking during trial).
217. See, e.g., U.S. CONST. art. III, § 3, cl. 1 (stating that conviction for treason requires two witnesses to overt act or defendant's confession in open court); Moran, supra note 146, at 817 (discussing corpus delicti rule prohibiting conviction based solely on uncorroborated defendant's confession).
218. While I make no definitive normative claim in this Article as to whether evidence such as prior convictions and courtroom demeanor should be enough to send an otherwise "pure cold hit" case to the jury, one could imagine compelling arguments to the contrary. For example, a juror's evaluation of demeanor after learning of a cold hit match could be tainted by confirmatory bias. While the juror might otherwise see a defendant's ability to look directly at jurors as suggestive of innocence rather than guilt-induced evasiveness, a tainted juror might view the same behavior as menacing. And while an extremely similar prior sexual act might suggest a propensity to commit similar acts, the use of the mere existence of a prior conviction to suggest guilt would seem to rest on general recidivism statistics that themselves might invoke appearance-of-justice concerns. Moreover, at least one publicized false cold hit pointed toward a man who, while innocent of the charged crime, had a prior conviction for murder. See supra note 59 (explaining that lab contamination likely led to false hit).
219. State v. Toomes, 191 S.W.3d 122, 127, 129 (Tenn. Crim. App. 2005).
220. Id. at 124, 126.
221. For example, if the suspect population were 5 million and the RMP were somewhere between 1 in 5 billion (for African Americans) and 1 in 22 million (for Caucasians), given that the suspect population is not entirely African-American, the chance of another person matching, using the simplified form of Bayes' Theorem, would be some number greater than (5 million x (1/5 billion)) = 1 in 1000. If the suspect population were only 4000, however, then even if the RMP were close to 1 in 22 million, the source probability would be well over 99.9%.
222. See Table DP-1: Profile of General Demographic Characteristics: 2000, Geographic Area: Ripley, TN, http://censtats.census.gov/data/TN/1604763340.pdf (last visited July 25, 2010).
223. See Table DP-1: Profile of General Demographic Characteristics: 2000, Geographic Area: Memphis, TN, http://censtats.census.gov/data/TN/390474920.pdf (last visited July 25, 2010).
224. See Tennessee QuickFacts from the U.S. Census Bureau, http://quickfacts.census. gov/qfd/states/47000.html (last visited July 25, 2010); Arkansas QuickFacts from the U.S. Census Bureau, http://quickfacts.census.gov/qfd/states/05000.html (last visited July 25, 2010).
225. See supra note 209 (urging caution in courts' decisions to limit population to particular class simply because defendant is member of that class).
226. 630 N.Y.S.2d 631, 631-32 (Sup. Ct. 1995). The court viewed the complainant's failure to identify Rush at trial, and her identification of a courtroom spectator instead, as exculpatory evidence. Id. at 631-32. But the complainant did identify Rush as her attacker in a pretrial identification procedure. If the court had engaged in a pretrial sufficiency analysis, this additional information would have taken the case out of the subset of pure cold hit cases.
227. Using the simplified version of Bayes' Theorem, the chance of non-uniqueness would be about 680,000 x (1/500 million) = 1/735.
228. 861 N.E.2d 898, 901 (Ohio Ct. App. 2006) (emphasis added). There was also evidence that the perpetrator had stubble and smelled of alcohol, but no evidence that Hunter fit this description. Id. at 899-900. If Hunter had fit the description, the case would presumably be removed from the "pure cold hit" category unless an upper bound on the number of potential suspects meeting the description could be quantified. Given the lack of evidence that Hunter fit the physical description of the suspect, if the DNA match statistics suggested other matching profiles in the suspect population, these additional details- which could be just as true of nearly all other males aged eighteen to sixty-five with a matching DNA profile-presumably would not stop a court from granting a motion for judgment of acquittal.
229. Using the simplified version of Bayes' Theorem, and taking 3 billion as an upper bound on the number of possible male suspects in the world, the chance of non-uniqueness would be less than 3 billion x (1/756 trillion) = 1 in 252,000.
230. [1996] 2 Crim. App. 467.
231. Using the simplified version of Bayes' Theorem, the chance of non-uniqueness would be about 200,000 x (1/200 million) = 1 in 1000.
232. Statistician Peter Donnelly, testifying for the defense, offered a possible quantification of the exculpatory evidence in the case and then calculated the posterior odds of guilt, assuming an RMP of 1 in 200 million, as being only 55 to 1, meaning that the posterior probability of the defendant being guilty based on the evidence was 55/56, or 98.2%. See Adams, 2 Crim. App. at 476-77 (citing and discussing Donnelly testimony); see also Kaye, supra note 6, at 484 (describing Donnelly's application of Bayes' Theorem). The Court of Appeal admonished the defense for using Bayesian reasoning, though it appears the court's true complaint was with the attempt to quantify inherently impressionistic evidence in a way that may have confused the jury. See Adams, [1996] 2 Crim. App. at 481 ("[T]he theorem's methodology requires, as we have described, that items of evidence be assessed separately according to their bearing on the accused's guilt .... That in our view is far too rigid an approach to evidence of the type that a jury characteristically has to assess ....").
233. See Memorandum in Support of Defendant's Motion for Judgment of Acquittal at 1-4, State v. Derr, No. K04-930 (Md. Cir. Ct. June 29, 2006) (arguing that defendant must be acquitted because failure to present error rate to jury undermines sufficiency of DNA evidence); see also Sangero & Halpert, supra note 71, at 47 (arguing that single piece of evidence should never be sufficient to convict).
234. See Sangero & Halpert, supra note 71, at 54-55 (presenting table of probabilities of guilt as calculated under Bayes' Theorem).
235. See, e.g., Peter Gill & Amanda Kirkham, Development of a Simulation Model To Assess the Impact of Contamination in Casework Using STRs, 49 J. FORENSIC SCI. 485, 486 (2004) ("The most probable outcome of a contamination event is a false exclusion.").
236. See Thompson et al., supra note 204, at 53 ("[T]he particular circumstances of database searches would seem to rule out, or at least greatly reduce, the likelihood of some types of errors, such as those arising from switching or cross-contaminating samples .... However, other ... errors, such as those arising from misinterpretation of test results, might still produce false matches.").
237. The 1996 National Research Council Report rejected calls for requiring the type of blind proficiency testing presumably needed to estimate a laboratory-specific error rate or for using an industry-wide estimated error rate, on grounds that both measures would be infeasible and would not be representative of the chance of error in any particular case. See NAT'L RESEARCH COUNCIL, supra note 45, at 85-87.
238. LYNCH ET AL., supra note 70, at 340.
239. Julie A. Seaman, Black Boxes, 58 EMORY L.J. 427, 474 (2008).
240. LYNCH ET AL., supra note 70, at 191.
241. Id. at 346.