How to not get it wrong: observations of easy-to-avoid fallacies from neuroscience and every-day life
This blog is a labor of love for me (=Chris), a neuroimaging analyst, and I use it to clarify my own thoughts, or just rant about amusing observations in academic every-day life. It’s admittedly therapeutic. Also, these morsels are not peer-reviewed contributions, and contain a good amount of opinion (with a tad of whimsical showmanship). Please keep this in mind. They concern easy-to-avoid mistakes in quantitative reasoning. Of course, many great people have plowed this territory already. A few names come to mind: John Allen Paulos, Gerd Gigerenzer, Daniel Kahneman, N. Nassim Taleb, Rolf Dobelli, and others. I recommend their stuff wholeheartedly. It’s great fun to read, might even have some practical relevance for you, and it’s for everybody, not just mathsy types.
(*** marks posts that are more specific to neuroimaging.)
Post 2: Transitivity of correlations – a trap that is easy to fall into
Post 4: Gaussian variables – not a requirement for linear regression to work
Post 5: Adjusting for confounders in linear regressions – it’s not the panacea you think it is
Post 6: Prior probability or base rates – they determine how accurate a medical test is
Post 7: Double dipping, over-fitting, etc. – the impressive feats of the Monday morning quarterback
Post 8***: Method memes – “I don’t know what it means, but I used DEEP LEARNING!”
Post 9: Inductive negatives and universal model skepticism – the flip side to method mania
Post 10: Variance shmariance - stop the fetishism!
Post 11***: Multivariate vs. univariate - RIP a quaint historical cultural divide