In the decade that has passed since the first edition of this book appeared, the crest of the wave of interest in statistical evidence has broadened and moved beyond its origins in civil rights law. Significant new developments, reflected in this edition, include, for example, DNA evidence (Sections 2.1.1, 3.1.2, and 3.2.2), epidemiologic studies in toxic substance litigation (Chapter 10), statistical models for adjusting census counts (Section 9.2.1), and vote-dilution cases (Section 13.2.3). It is emblematic of the importance of statistics in the pantheon of scientific evidence that the leading Supreme Court precedent on such evidence—the Daubert1 case—involved toxic substance claims in which epidemiologic studies played a key role. In Daubert, the Court rejected the old Frye test of general acceptance in the scientific community as the basis for admissibility, and explicitly imposed on federal trial judges a gatekeeping function: they must now assess whether the proffered evidence is both relevant and reliable. The new formulation puts judges in the awkward position not only of counting scientific noses, but also of understanding and appraising the scientific basis of what an expert proposes to say, or calculate, on the witness stand. Fortuitously, about a year after Daubert, in 1994, the Federal Judicial Center issued and distributed to all federal judges a Reference Manual on Scientific Evidence, which is largely a primer on the applications of statistical methods. A new edition of the Manual, which unexpectedly turned out to be a best seller, is due to appear this year. Those who find this book heavy going may wish to consult the Manual as a useful introduction to at least some subjects.
But new, case-driven applications of statistics are only part of the development. Perhaps even more important, in the long run, is the continuing flow of statistical studies of the legal system itself. Studies of this sort can offer insights that sometimes challenge commonly held views of venerable legal institutions. Section 5.6.3 gives an example of such a study, involving peremptory challenges of prospective jurors, in which the authors analyze data and find that most peremptory challenges are “guesses.” For another example, as this is being written, the media are prominently reporting a large-scale statistical study of the death penalty, undertaken at Columbia Law School, which paints a startling picture of the high rate of serious errors in criminal trials leading to death sentences. The study will almost certainly influence pending legislation and promises to provide important data in the debate over capital punishment itself. One must note that in both these studies it is the statistical pattern, emerging from the details of individual cases, that tells the most compelling story.
As in the first edition, much of the material for the new portions of this second edition was collected from lawyers, statisticians, or economists who were involved in the cases.We thank them all for their generosity in assisting us. In this connection, we would like to acknowledge in particular Orley Ashenfelter, David Baldus, William Fairley, David Freedman, and Sol Schreiber for their help in furnishing us with their materials and consulting with us on matters of interpretation.