Abstract

Chow's one-tailed null-hypothesis significance-test procedure, with its rationale based on the elimination of chance influences, is not appropriate for theory-corroboration experiments. Estimated effect sizes and their associated standard errors or confidence limits will always suffice.