Effect Sizes

Effect size is a way of reporting the strength of a relationship between two or more variables. In terms of quantitative comparisons, it is simply the extent to which two groups differ from each other concerning the grouping variable. Whether results are significant or not depends on the sample size; however, effect size of the results is independent of the number of research units. Thus, effect size is not influenced by the size of the samples.

Consider an intervention, such as a new book on research methods that is intended to be more useful than an earlier volume on the same topic. The questions arise whether this book is actually better than the old one, and if so, how much better. These questions are important when considering the costs of introducing a new book (e.g., the cost of printing, distribution, and marketing a new volume). Not only publishers but also lecturers and buyers will want to know whether it is worth printing, selling, and purchasing the book or whether continuing using the old book may be just as valuable. To test this, one could compare two groups that learn with the two different books (either the old volume or the new volume). Although the significance testing and the p value will indicate whether there is a difference, the effect size will indicate the size of the difference. The question changes from “does it work” to “how good does it work” and can inform decisions about publishing, reading, and buying the new book. Thus, effect size reports the “effectiveness,” which is important for both scientific research and applied research for the business world. The size of the effect provides information about the expectations of whether or not the difference observed represents something important about the application. For example, knowing that mass-media content that provides positive role representations of homosexual persons reduces levels of homophobia by 50% provides a basis for determining the value of the practice. An effect size can establish the impact of a particular communicative practice.

This entry discusses why the usage of effect size is necessary before and after conducting a study and which kind of effect sizes are the most common in statistical reports regarding the strength of the relationship between variables.