Effect size and practical importance: a non-monotonic match

摘要

As an alternative to statistical testing, effect size has a non-monotonic linkage with practical importance. Besides random variance and systematic bias, a proper interpretation of effect size hinges on its implication to outcomes of deductive and/or inductive enquiries. Consequently, a small effect size might suggest an important finding, and the same effect size could represent results of different importance. The research community could miss important findings if large effect size values are always expected for professional publication.