Article by Ronald J. Allen
Five years ago, I analyzed the uneasy relationship that exists, if error reduction is the primary goal of civil litigation, between conventional probability theory (and its inductivist competitor) and the conventional conception of trials. That analysis generated the conclusion that the analytical problem is inherent in the conception of trials rather than in the various conceptions of probability. The conventional conception of civil trials involves comparing the probability of a plaintiff's case to its negation. My proposal substituted for this a comparison of the plaintiff and defendant's stories, with a verdict entered for whichever party advances the more plausible account (however plausibility is understood). I further argued that the debate over statistical evidence was somewhat beside the point for two reasons. First, errors are reduced by entering verdicts for the more plausible story, however implausible the more plausible one is and regardless of the nature of the evidence at trial, so long as there is a direct relationship between plausibility and accuracy. Second, the debate over statistical evidence overlooks the fact that statistical evidence will be analyzed by a conventional human reasoner who must filter that data through the same web of experience and knowledge that will be applied to any evidence produced at trial. Thus, the assumption that there are two qualitatively distinct types of evidence, statistical and non-statistical, is essentially false.
About the Author
Ronald J. Allen. Professor of Law, Northwestern University.
Citation
65 Tul. L. Rev. 1093 (1991)