Originally from Dispute Resolution Journal

Nothing is more obnoxious to Americans than a system of punishment that is overinclusive in nature. Notwithstanding the Supreme Court's holding in Korematsu V. United States in 1944 that it was permissible to imprison in relocation camps persons of Japanese ancestry, our system of criminal justice is premised on punishing only the guilty. As a society we are willing to make errors that may favor the guilty in order to minimize the probability of convicting the innocent.

In statistical terms, if the null hypothesis is "Anthony is not guilty," our criminal justice system, with its standard of proof "beyond a reasonable doubt," attempts to minimize Type I errors (rejecting Ihe null hypothesis when it is true) or, in our case, rejecting that Anthony is not guilty when, in fact, he really is innocent. Society can always increase its chances of convicting the guilty, as in Korematsu, but more Type I errors will be made. Simply stated, we would rather commit Type M errors (accepting the null when it is false) by incorrectly concluding a defendant is not guiity than unreasonably risk convicting the innocent.

There are parallels in the industrial world. Not infrequently, an employer knows that misconduct has taken place by an employee or a group of employees but, for whatever reason, is unable to identify the guilty party. A likely response might be to forget about Type I errors and punish the entire group if the guilty person cannot be identified. Most, but not all, arbitrators have ruled that it is unfair to take such an overinclusive approach by punishing the entire group. The few reported decisions in this area are worth examining.