Statistics Done Wrong: The Woefully Complete Guide

You can read this book on three levels –

1. Statistics is easy to screw up.

2. Here are all the ways you can screw them up.

3. Here is the actual math that proves it’s being screwed up.

#1 is absolute – you cannot come away from this book without knowing that statistics is a messy science. I got about half of #2 – I have a better idea of how statistics can be wrong. #3 is rough – especially in the beginning, there’s some math that was just lost on me.

But this is still a great book, because society puts far too much faith in statistics and unreplicated scientific studies in general. In the last third of the book, the critiques become more general, and thus more interesting. You learn about the pressure that academics are under to publish, and why journals like to publish more interesting/extreme results, and how open access and digital-only journals are changing the nature of publishing.

You also learn about the sometimes sketchy things that scientists do. They hide their data, they hide their calculations, they “torture” data until it “confesses” something interesting, etc.

If you read it, suffer through the math and some of the tedious esoteric stuff in the beginning. It gets more general, and more useful, later on.