This is a good intro for non-statisticians to avoid some common pitfalls and misconceptions that the public often has. It’ll help you understand what it means when something is statistically significant, why sometimes studies contradict each other, and how to avoid believing in patterns that don’t exist. You’ll understand the following biases and fallacies:
- confirmation bias
- gambler’s fallacy
- hot hand fallacy
- “hindsight is 20/20”: the illusion of hindsight for explaining historical performance/events
One mistake I want to point out:
When pollster’s tell the media that a poll’s margin of error is plus or minus 5%, they mean that if they were to repeat the poll a large number of times, 19/20 (95%) of those times the result would be within 5% of the correct answer.
This is wrong. Other samples aren’t guaranteed to be within 5% of the correct answer, they’re guaranteed to be within 5% of this sample. This book does a great job with a lot of common statistical fallacies, so I’m surprised he didn’t spend any time on this one, which I think is extremely important.
The subtitle, “how randomness rules our lives”, was covered when he talks about how in many situations the results are almost entirely determined by random chance rather than intervention. (Company leadership, investment performance, and sports are prominent examples.)
He talked about determinism in several places, but never centrally addressed the assumptions of probability theory and where they break down. I also would have expected a more thorough discussion of the assumptions of Bayesian analysis, and why it can sometimes be problematic. See the podcast episode Against Bayesianism for the topic I was hoping he would touch on.