Calculating and reporting effect sizes on scientific papers (3): Guide to report regression models and ANOVA effect sizes

In the first issue of the Portuguese Journal of Behavioral and Social Research, the importance of calculating, indicating, and interpreting the effect sizes for the differences of means of two groups (d family of effect sizes) was reviewed. Effect sizes are standard metrics that allows the comparison of the results of statistical analyzes of different studies. Effect sizes also report on the impact of a factor on the variable under investigation and the association between variables. After reviewing the effect sizes for the mean differences between two groups (Espirito-Santo & Daniel, 2015) and most of the r family (Espirito-Santo & Daniel, 2017), the review of effect sizes for analysis of variance was lacking. Analysis of variance can be understood as an extension of the d family to more than two groups (ANOVA) or as an r subfamily in which the proportion of variability is attributable to one or more factors. In the r subfamily reviewed in this study, we analyse the change in the dependent variable that results from one or more independent variables. This analysis is focused on general linear models, including regression models and ANOVA. This article provides the formulas for calculating the most common effect sizes by reviewing the basic concepts of the statistics and providing illustrative examples computed in the Statistical Package for the Social Sciences (SPSS). The guidelines for the interpretation of effect sizes are also presented, as well as the cautions in their use. Also, the article is accompanied by an Excel spreadsheet to facilitate and expedite the calculations for interested readers.