Law and Statistical Disorder: Statistical Hypothesis Test Procedures and the Criminal Trial Analogy

Abstract

Virtually all business and economics statistics texts start their discussion of hypothesis tests with some more-or-less detailed reference to criminal trials. Apparently, these authors believe that students are better able to understand the relevance and usefulness of hypothesis test procedures by introducing them first via the dramatic analogy of the criminal justice system. In this paper, we argue that using the criminal trial analogy to motivate and introduce hypothesis test procedures represents bad statistics and bad pedagogy. First, we show that statistical hypothesis test procedures can not be applied to criminal trials. Thus, the criminal trial analogy is invalid. Second, we propose that students can better understand the simplicity and validity of statistical hypothesis test procedures if these procedures are carefully contrasted with the difficulties of decisionmaking in the context of criminal trials. The criminal trial discussion provides a bad analogy but an excellent counter-example for teaching statistical hypothesis procedures and the nature of statistical decision-making.