Effect size classification - alex-strobel/DPP-LabManual GitHub Wiki
At AG.DPP, the effect sizes of interest are most often correlations r, sometimes also explained variance (partial) η2, and even less often Cohen's d. To classify effect sizes, researchers often refer to Cohen's guidelines given in Cohen (1988). This has been criticised in recent years, and efforts have been made to provide empirically informed classification schemes, mainly based on the percentiles of correlations (and in some instances also Cohen's d) that can be found in meta-analyses in subdisciplines of psychology, which mainly boils down to .10, .20, and .30 as thresholds for small, medium, and large correlations. Yet, this has also been criticised by, e.g., Schäfer & Schwarz (2019) who observed differences in effect sizes between
- different subdisciplines of psychology
- between- and within-subjects designs
- preregistered and non-preregistered studies
Therefore, if we must classify an effect, we at AG.DPP use the thresholds .10, .20, and .30 to classify correlations as small, medium, and large, and cite Gignac and Szodorai (2016) as reference because they derived their classification from studies from individual differences research. A more detailed overview over the topic of effect size classification is given in this manual:
References
- Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillside, NJ: Lawrence Erlbaum Associates.
- Gignac, G. E., & Szodorai, E. T. (2016). Effect size guidelines for individual differences researchers. Personality and Individual Differences, 102, 74–78. https://doi.org/10.1016/j.paid.2016.06.069
- Schäfer, T. & Schwarz, M. A. (2019). The meaningfulness of effect sizes in psychological research: Differences between sub-disciplines and the impact of potential biases. Frontiers in Psychology, 10, 813. https://doi.org/10.3389/fpsyg.2019.00813