Significance tests and decision rules in communication research: A new method for reducing type II error

Abstract

Previously three methods have been suggested for increasing power: increasing sample size, increasing effect size, or increasing alpha. Sometimes the first two alternatives are not practical and the third is not desirable. This paper proposes a fourth alternative, the partitioning of the decision region into three parts. This procedure is consistent with the Neyman and Pearson (1933) decision theory approach to significance testing. A sample power table is provided. Utilization of this procedure results in an unambiguous interpretation of nonsignificant results and leads to increased power.