The classic treatises on evidence note that the court or jury must weigh the evidence, and upon weighing it, determine whether the plaintiff or the defendant prevails. Some courts, however, have indicated that statistical evidence should not be admitted unless it is subjected to a procedure known as 'hypothesis testing.' There are many rather mechanical procedures for performing these tests and a number of judges, attorneys, and law professors have suggested that hypothesis testing provides an objective, scientific means of settling disputed questions on which statistical evidence is brought to bear. Yet, many circumstances arise in which courts or administrators have puzzled over the meaning of hypothesis tests.
This article examines the status of significance testing in litigation. It describes the case law on the need for the procedure, explains the nature and terminology of hypothesis testing as used in court, enumerates some of the problems that arise in these forensic applications, and pursues one such problem - that of selecting a 'significance level.' These sections show that explicit hypothesis testing is poorly suited for courtroom use. Statements as to what results are or are not 'statistically significant' should be inadmissible.
The article also suggests the use of other statistical tools and terms that do not 'test' hypotheses but can better aid the finder of fact in judging the probative value of the statistical evidence. It is a plea to leave the task of decision to the trier of fact, and not to rely on superficially impressive methods whose seeming objectivity does not withstand analysis. It is a call for using, where suitable, those statistical tools that will aid these decision-makers in the process if inference.
David H. Kaye, Is Proof of Statistical Significance Relevant?, 61 Wash. L. Rev. 1333 (1986).