Article by Craig R. Callen
I am glad to have this opportunity to return to some of the points I made earlier, in an effort with Professors Allen and Shaviro to explore more fully some of the issues that the use of statistics, or of statistical models, raises about evidence. Each of us is dissatisfied with some of the earlier analyses of these issues. In order to make the following response more clear, and because my piece is the final one in this dialogue, I will attempt to describe the current state of our three-cornered debate.
Professor Shaviro originally argued that a general rule prohibits verdicts based on what he calls “entirely” statistical evidence; he contends the rule is based on misconceptions about statistics, the process of proof, and the function of litigation.
Professor Allen argues, as do I, that the idea of “bare,” “naked,” or “entirely” statistical evidence makes no sense, because statistics standing alone prove nothing. Professor Shaviro now says he agrees that a statistic alone has no probative value, which leaves him in a peculiar position—the terminology that he uses does not embody a meaningful concept. As to the meaninglessness of the terms, I rely chiefly on what I said earlier.
Professor Shaviro nevertheless continues to refer to some rule that prohibits relying on “naked statistical evidence,” although he ascribes no new meaning to the phrase. In regard to the supposed rule prohibiting reliance on “naked statistical evidence,” I made the point earlier, relying on Professor Allen among others, that there is no general rule prohibiting reliance on statistical evidence. Courts rely on statistics in torts, anti-trust, and employment discrimination cases—there are no general substantive limitations. Professor Shaviro seems to concede, at least implicitly, that the general rule does not exist. He offers no new evidence of the general rule, choosing instead to talk about his perception that the courts are unduly hostile to statistical evidence.
In regard to that perception, he makes certain claims about the nature of the burden of proof, and particularly the burden of persuasion. He argues that the courts, or scholars, set standards for the adequacy of statistical evidence that are too exacting. In making that argument, he relies on premises that, if consistent, essentially address the burden of persuasion in all cases. In a civil case, Professor Shaviro's own analysis would merely require a comparison of the likelihood of the parties' contentions rather than a minimum level of sufficiency for the burden-bearing party beyond the introduction of some admissible evidence. He says that a court “should not systematically favor either party, by ostensibly declining to decide a case, simply due to the lack of a satisfying amount of evidence.”
Professor Allen, on the other hand, argues that courts are too willing to direct verdicts on any sort of evidence—at least if one believes that their formal standards reflect their practice. He manifests no particular preference between quantified evidence or non-quantified evidence, but he believes that my theory of the role of burdens of proof would make it too easy to obtain a directed verdict or summary judgment on any sort of evidence.
In response to their arguments, this piece first briefly recapitulates my analysis of burden of proof requirements in light of research from cognitive science. Then it explains how that analysis relates to an ordinary task of inference from statistics. Finally, it discusses the evaluation of errors in adjudication, with special attention to modes of inference that might result in errors. Specifically it shows how reliance on paltry evidence can result in errors when the effects of such reliance are measured against the goals and functions of adjudication. Along the way, it probes the main criticisms that Professors Allen and Shaviro have made of my analysis.
About the Author
Craig R. Callen. Professor of Law, Mississippi College School of Law. B.A. 1971, University of Iowa; J.D. 1974, Harvard Law School.
Citation
65 Tul. L. Rev. 1113 (1991)