Note on the Denominator of ω 2 in Factorial Designs

Abstract

The denominator of ω² in a factorial design is inflated owing to rectangularity of the distribution of the factors in analysis of variance, while many quantitative psychological variables are normally distributed. The magnitude of the bias depends on the correlation between the dependent variable and the relevant factor and is therefore relative to each experiment. Corrections of ω² for range are discussed.