Effect Sizes for Research: Univariate and Multivariate Applications

by Robert J. Grissom

The goal of this book is to inform a broad readership about a variety of measures and estimators of effect sizes for research, their proper applications and interpretations, and their limitations. Its focus is on analyzing post-research results. The book provides an evenhanded account of controversial issues in the field, such as the role of significance testing. Consistent with the trend toward greater use of robust statistical methods, the book pays much attention to the statistical assumptions of the methods and to robust measures of effect size. Effect Sizes for Research discusses different effect sizes for a variety of kinds of variables, designs, circumstances, and purposes. It covers standardized differences between means, correlational measures, strength of association, and confidence intervals. The book clearly demonstrates how the choice of an appropriate measure might depend on such factors as whether variables are categorical, ordinal, or continuous; satisfying assumptions; the sampling method; and the source of variability in the population. It emphasizes a practical approach Intended as a resource for professionals, researchers, and advanced students in a variety of fields, this book is an excellent supplement for advanced courses in statistics in disciplines such as psychology, education, the social sciences, business, management, and medicine. A prerequisite of introductory statistics through factorial analysis of variance and chi-square is recommended.

272 pages, Hardcover

First published March 1, 2005

Reviews

[G]

A modern tradition that no one can really understand has built a gold-standard method to produce scientific evidence. The core idea of such method relies on decisions guided by probabilities, which are known as p-values in Mathematical Statistics. Such p-values are tools designed to avoid spurious explanations. If a set of observations can be attributed to random processes, then no explanatory model can be simultaneously proposed. That´s all. It sounds poor because it is poor. To overcome this situation, effect sizes have been created. Effect sizes are mathematical techniques aimed to produce more sophisticated evidence. Effect sizes are focused on the actual magnitude of the phenomena of interest. Grissom and Kim describe in this useful book the most often required effect sizes. Magnitude estimators both for observational and experimental studies are reviewed and clearly explained. Since non-Gaussian vectors of observations are frequently found in everyday research practice, maybe more pages focused on non-parametric effect sizes would have been welcomed. However, this book is a nice introduction to the logic, meaning, and calculation of effect sizes for research.