Sense and "sensitivity": epistemic and instrumental approaches to statistical evidence.

• Author: Enoch, David
• Position: III. Solving Some Doctrinal Puzzles through Conclusion, with footnotes, p. 585-611

3. SOLVING SOME DOCTRINAL PUZZLES

   Evidence in the form of statistical assessment has extended far into the legal universe and plays a growing role in court. (87) Statistical inferences based on genetic profiling and the demographic characteristics of populations, as well as cause-and-effect inferences derived from experimental studies and analyses of sample surveys, are all used in the legal arena on a daily basis and unfold in a host of legal settings. (88) DNA is often used for purposes of identification (89) in criminal trials (90) and family law disputes alike. (91) Statistical assessments are used in discrimination cases, most notably in the employment context to substantiate disparate impact claims by pointing to a discrepancy between the proportion of minority group members hired by the employer and the proportion of minorities among the group of qualified people in the relevant market. (92) Courts also allow statistical assessments of a similar nature to be admitted in voting, (93) jury selection, (94) and constitutional and human rights cases. (95) In tort law cases, statistical evaluation is used to prove causation. (96) It is also used for the calculation of economic damages--for example, when courts use base rates that identify incomes lost by similarly situated individuals. (97) Such evidence is likewise used in the context of competition law. (98)

   But we are not interested in all instances in which statistical assessments are used as evidence in courts of law. What we wish to highlight, rather, is the use of statistical evidence in contexts similar to the Blue Bus and gatecrasher hypotheticals: namely, cases in which the statistical evidence provides a base rate for the defendants' liability (99) as opposed to a factual characteristic that is isolated from the ultimate legal question of guilt or liability) and establishes a conclusion that is itself nonstatistical. More specifically, we will focus on situations in which the defendant's liability-triggering, or guilty, conduct is inferred based on reference to membership in a particular population or reference class. (100)

   We begin with the two extreme points: DNA evidence, which courts tend to endorse, and propensity-for-crime evidence, which courts tend not to admit at the guilt phase of trial. In what follows, we will demonstrate how our theoretical framework can explain the different legal regulation of these types of statistical evidence under prevailing law.

   1. DNA Evidence

      DNA evidence is an interesting illustration of the adaptability of our theory to the legal arena. Despite the statistical nature of DNA evidence, which identifies the frequency with which genetic profiles occur in reference populations, courts seem rather happy to rely on it. (101) The theoretical model presented above can shed some light on this notable exception to courts' general resistance to statistical evidence. We start out by describing the legal doctrines governing the admissibility of DNA evidence and then show how our account of the distinction between statistical and individualized evidence can explain and, to an extent, justify this exception.

      DNA evidence first surfaced in American courtrooms in the 1980s and has since emerged as the most important forensic scientific breakthrough of the twentieth century, (102) leading to numerous convictions and hundreds of postconviction exonerations. (103) DNA testing was depicted by one court as "the single greatest advance in the 'search for truth'... since the advent of cross-examination" (104) and has been analogized to "the finger of God." (105) Like other courts around the world, American courts--at both the federal and state levels--sweepingly admit DNA evidence (106) in paternity suits (107) and in criminal trials. In 1988, Florida's Fifth District Court of Appeal was the first appellate court in the United States to uphold the admission of DNA evidence in a criminal proceeding. (108) This was followed by the United States v. Jakobetz case, (109) which marked the first approval of the admission of DNA evidence by a federal appellate court. By 1990, thirty-eight states had admitted DNA evidence, (110) and by the mid-1990s, most states were allowing DNA test results to be admitted as evidence in criminal trials. (111) Whether the Frye standard, Frye-Kelly standard, or Daubert standard is applied, DNA evidence is currently almost universally accepted in both federal and state courts. (112)

      Another dimension of the role of DNA evidence in court is the evidentiary weight ascribed to it. In this context, too, courts have shown a general tendency to endorse DNA evidence, viewing this technology as bringing an unprecedented degree of certitude to the courtroom. For reasons that will be specified below, (113) courts and juries are more prone to convict on the basis of DNA evidence when it is corroborated by other types of evidence, but most courts do not rule out the possibility of convicting on DNA alone. (114) Indeed, when the probabilities of guilt given the DNA match have been sufficiently high, courts have convicted solely on the basis of DNA evidence. (115) Moreover, at least one court has declared that DNA evidence alone is a sufficient basis for conviction even in the face of conflicting eyewitness evidence. (116)

      The theoretical foundations presented further on can offer some insight into the doctrinal treatment of DNA evidence. Before proceeding, however, the following three preliminary points should be made: First, our discussion will not relate to the scientific foundations of DNA evidence. Rather, it will assume a low probability of a false positive match. Indeed, we are going to restrict our attention to just those cases in which the probability that the accused is guilty, given that there is a DNA match, is extremely high, though not quite 1. (117) Second, the discussion will focus on the hard cases of "cold hit" DNA--namely, where DNA is the only evidence and where it was obtained without some prior suspicion. (In other words, DNA evidence was obtained from the scene of the crime and run against some database, and a match was found; the suspect was not first pinpointed and then tested for a DNA match.) And third, we will be restricting our attention to the use of DNA evidence as evidence for the prosecution (in a criminal case). (118) With these stipulations in place, then, can anything be said in favor of using DNA evidence, especially given the background of suspicion toward statistical evidence in general? How do we solve this doctrinal puzzle?

      One obvious feature that distinguishes DNA evidence from most other kinds of statistical evidence is the extremely high level of probability underlying it. This suggests one reasonable, albeit unexciting, solution to the doctrinal puzzle, in terms of the relative value of accuracy: although it may be the case that the same objections to statistical evidence are no less applicable to high-probability evidence like DNA, the value of accuracy is much greater in cases in which the probability is so high. This very high probability underlying the evidence is the core of the difference between the ruling in Kaminsky v. Hertz Corp. (in which the appellate court found a triable issue of fact where market share evidence showed that the Hertz Corporation owned ninety percent of all yellow trucks bearing a Hertz logo) (119) and the ruling in Smith v. Rapid Transit, Inc. (in which the court rejected the market share evidence, which pointed to a smaller market share held by Rapid Transit). (120) In Kaminsky, as in DNA cases, considerations related to the value of accuracy outweighed the standard reasons for not relying on statistical evidence. (121)

      Another possible account for the preferential treatment of DNA evidence is the incentive story. Recall our generalization of Sanchirico's theory, according to which relying on statistical evidence will create inefficient incentives for, say, the Blue Bus Company as well as its competitor, the Red Bus Company. (122) This line of reasoning relies on the assumption that both companies would be in a position to know that their chances of being found liable are unrelated to their relevant conduct (because liability is determined by their market share). But perhaps in DNA cases--certainly, in most DNA cases--the potential offender has no access to such knowledge. Most people possess little knowledge regarding their genome sequence or DNA profile and its frequency in the relevant population. So unlike readily available statistical evidence, the incentive story arguably does not apply here, and there is no incentive-based reason to ignore genuinely probative statistical evidence.

      Despite the plausibility of this explanation, there is room to question whether it captures the full picture. Suppose, for instance, that in addition to DNA, we can also check for DNA* matches. DNA* shares with DNA its incentives-relevant properties (things like what knowledge is and is not available ex ante) but is much less effective probabilistically, so that the probability that the accused is guilty given a DNA* match is, let's say, around seventy percent. In such a case, too, the incentive story collapses, for the effect of the evidence ex post on the imposition of punishment is relatively low, and therefore its ability to disincentivize engagement in the relevant criminal conduct ex ante is similarly low. Yet the intuitive reluctance to rely on statistical evidence is still as strongly present. (123)

      Yet both the high-probability-level account and the incentive explanation hold, without a doubt, at least some of the relevant truth about the treatment of DNA evidence. And in fact, we are committed to using such nonepistemic stories as guides to legal policy. But the Sensitivity requirement can add an additional dimension to the explanation of the DNA conundrum, for DNA is a unique type of statistical evidence from the Sensitivity perspective...