The New Statistics for Neuroscience Majors: Thinking in Effect Sizes.

An ongoing reform in statistical practice is to report and interpret effect sizes. This paper provides a short tutorial on effect sizes and some tips on how to help your students think in terms of effect sizes when analyzing data. An effect size is just a quantitative answer to a research question. Effect sizes should always be accompanied by a confidence interval or some other means of expressing uncertainty in generalizing from the sample to the population. Effect sizes are best interpreted in raw scores, but can also be expressed in standardized terms; several popular standardized effect score measures are explained and compared. Reporting and interpreting effect sizes has several benefits: it focuses on the practical significance of your findings, helps make clear the remaining uncertainty in your findings, fosters better planning for subsequent experiments, fosters meta-analytic thinking, and can help focus efforts on protocol optimization. You can help your students start to think in effect sizes by giving them tools to visualize and translate between different effect size measures, and by tasking them to build a 'library' of effect sizes in a research field of interest.