Why effect sizes? A tutorial (especially for Neuroscientists)
The New Statistics emphasizes effect sizes, confidence intervals, meta-analysis, and Open Science. There’s a lot of momentum to adopt this change of focus. For example, the APA recently released new guidelines for reporting quantitative research and throughout it emphasizes reporting of “effect sizes and confidence intervals” (Appelbaum et al., 2018). It is well worth reading the entire set of guidelines.
Although there is strong progress towards adopting a New Statistics approach, there is still lots of work to be done spreading the good word. To that end, I (Bob) have been writing a series of perspective pieces for the Journal of Undergraduate Neuroscience Education, helping those in my teaching community understand the new approach and why it is so useful. I’ve just drafted the second for this series–it’s a tutorial on effect sizes: what they are and why it is so important to report and interpret them. The examples are all from neuroscience, but I think the general idea will be useful for anyone already using p values to summarize data. The pre-print is here (https://psyarxiv.com/zvm9a ) – the paper hasn’t been submitted yet so please send any feedback/comments my way.
Also, in the paper I compare different ways of standardizing effect sizes. I find this figure, which is in the paper, especially useful: