Statistical Evidence: In Search of a Principle

The use of statistical evidence has long been controversial.1 There are numerous cases in which the use of statistical evidence is unobjectionable and routine. Perhaps the most obvious example is DNA evidence.2 Even scholars who object to the use of some types of statistical evidence see DNA evidence as unobjectionable.3 Nevertheless, a series of real and hypothetical cases raises difficult questions about whether and how statistical evidence should be used in court. For example, Sally Clark’s conviction for murdering her two babies was based in part on expert evidence that the probability that the two babies died naturally from Sudden Infant Death Syndrome (“SIDS”, also known as cot death) is 1 in 73 million.4 The way in which statistical evidence was used in this case is highly contentious. In addition, philosophers and evidence scholars have been concerned with hypothetical cases such as the gatecrasher paradox and the red/blue buses scenario,5 in which the use of statistical evidence seems intuitively to be wrong. However, explaining why this use is wrong is difficult. Interestingly, even enthusiastic supporters of the use of statistical evidence in court do not advocate basing legal decisions solely on statistical 1 See, for example, Tribe’s discussion of the Dreyfus case (described at §1.1 below) at the end of th