Directional and nondirectional alternative hypotheses

Abstract

The problem of one- vs two-tailed tests is discussed. The use of the terms “one tailed” and “two tailed” can lead to confusion, for example, with ANOVA procedures. It is suggested that these terms be replaced with “directional” and “nondirectional” alternative hypotheses. Furthermore, directional alternative hypotheses appear to contradict the operation of the null hypothesis. Thus, it appears that nondirectional alternative hypotheses should be used in all cases. © 1977, The Psychonomic Society, Inc.. All rights reserved.