This blog is about basic statistics applied to neuroscience & psychology data. Despite a trend for more complex designs and analyses, a typical dataset undergoes massive dimensionality reduction, such that group analyses and illustrations tend to rely on a few tools, for instance bar graphs, Pearson’s correlation & ANOVAs on means. Here I focus on robust, informative & intuitive alternatives to these classic, often misunderstood, and outdated tools. I will describe problems I often encounter in behavioural & MEEG studies I review or edit, and suggest solutions. I might suggest rather unorthodox ideas: for instance, in some situations involving small sample sizes, I don’t see the benefits of statistical tests at all – meaningful illustrations and robust and informative measures of effect size often suffice.
After years of practicing robust frequentist statistics, I’ve reached a stage where my papers are pretty much free of p values, because the goal has become to illustrate and quantify, not to reach arbitrary binary decisions – here is a recent example.
Finally, take everything in this blog with a pinch of salt: I’m not a statistician; I work mostly on visual perception in young and older adults.
All pages and posts are © Guillaume A. Rousselet and garstats.wordpress.com, 2016-2018.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.