The publisher No Starch Press was kind enough to send me a review copy of Statistics Done Wrong by Alex Reinhart. I enjoyed book and recommend it for everyone to read. It’s a relatively short read of about 130 pages and the kindle version is only $9.99.
Alex starts off on the cover page with my favorite quote from Feynman, “The first principle is that you must not fool yourself and you are the easiest person to fool.” This is my favorite quote because it encapsulates the essence of science more than anything else I’ve read.
In the introduction chapter I disagree with Alex’s assertion that the problem with scientists & statistics is poor statistical education. I think the misalignment of incentives is more important. Modern academic scientists are not rewarded for discovering new knowledge that is true but instead are rewarded for publishing.
I’m personally guilty of this in a paper on river flow forecasting, Forecasting River Runoff through Support Vector Machines. We used the best known practices in the paper, e.g. independent training and test data sets. But I’m not confident that our results are true. Why didn’t we continue investigating until we were certain our methods work? Because we wanted to publish! Publishing was the incentive. Feynman’s fantastic famous essay, Cargo Cult Science, captures this sentiment exactly.
Statistics Done Wrong isn’t bad in the sense that it doesn’t address this problem as much as I’d like. The book limits its scope more to bad statistics and less to why scientists don’t have a burning desire to avoid doing bad statistics.
It’s a fun read and only takes about an afternoon to finish. He avoids difficult math and is a good writer.