Demonstrates that almost all of the stats 101 tests are special cases of (simple) linear models, including “non-parametric” tests to some approximation. To establish this, I simulated the rank-correspondence between parametric tests and non-parametric equivalents:
A simple and scalable three-step procedure to apply utility theory to regression.
A cheat sheet and a bunch of shiny applets to show various reaction-time friendly distributions and how to fit them using brms. I installed a Shiny server on Google Cloud to get it running.