In most endeavors concerned with the acquisition of knowledge, quantitative information is welcomed. In law, however, it appears sometimes that scientific or numerical evidence makes cases harder, not easier. Nevertheless, there are many cases and administrative proceedings, in such areas as environmental law, food and drug regulation, and civil rights, in which statistical data obtained by observation or experiment are readily accepted as assisting in the proper resolution of disputed issues of fact.' When courts or administrators confront scientific and statistical evidence in these proceedings, they are not always certain of how to weigh the evidence or whether they should, or must, rely on the standards for proof that scientists apply in evaluating statistical hypotheses. Two decisions of the Supreme Court illustrate this uncertainty. In Castaneda zv Partida, 2 a grand jury discrimination case, the Court, acting as its own statistician, computed a statistic known as the deviation.3 The Court found this computation highly probative of discrimination in light of the general rule that if the difference between the expected value and the observed number is greater than two or three standard deviations, the hypothesis that the jury drawing was random would be suspect to a social scientist.4 One would not have thought that this reference, in a single footnote, to a standard of proof popular among social scientists would be read as commanding that the same standard be determinative when evaluating statistical evidence in court. Yet, in Hazelwood School District v. United States,5 an employment discrimination case decided the same year, the Court dropped the qualifying language about social science and noted that, under the precise methodology delineated in Castaneda, a disparity of slightly less than two standard deviations was not suspect.6 In the wake of these opinions,
Read full abstract