Currently reading

Paul Hanley
A Month in the Country
Michael Holroyd, J.L. Carr
A Tale of the Dispossessed: A Novel
Laura Restrepo, Dolores M. Koch
Mesabi Pioneers
Jeffrey Smith, Russell Hill
The Crusades Through Arab Eyes
Jon Rothschild, Amin Maalouf
Island of a Thousand Mirrors
Nayomi Munaweera

Truth or Truthiness: Distinguishing Fact from Fiction by Learning to Think Like a Data Scientist

Truth or Truthiness: Distinguishing Fact from Fiction by Learning to Think Like a Data Scientist - Howard Wainer Wainer's stated purpose in writing this book is to help his readers develop habits of mind that allow them to distinguish truth from things that feel right but actually aren't (truthiness). His annotated table of contents is very clear and well-organized, and he makes some very good points. He walks us through what kinds of experiments would be necessary to prove certain claims, talks about what we should then do if those experiments aren't possible (they often aren't), and concludes with some very interesting examples, including the controversial teacher tenure and testing that are mentions in the book blurb.

Wainer worked for the Educational Testing Services for years, and his examples involving testing are enlightening and worth pulling out for discussion in and of themselves. A few of his case studies in the later chapters are pretty well-researched and take into account history and context in guessing at why the statistics say what they do. Really good stuff. I'm glad I stuck it out and got to them, because I found the opening chapters of the book, in which Wainer is laying out how to think correctly, a bit off-putting.

The book blurb talks about "his trademark verve and irreverence", but mostly I would call his overall voice snarky. And sometimes that's amusing. We especially like to let old men get away with it. If you're in the mood to hear faux nostalgia for the good old days before Nate Silver when statisticians were uncool and he could be left in peace on an airline trip, this is the book for you. And he would love to show you his son's Princeton acceptance letter -- yep, printed in right in there, with a statistic on how rare they are just in case you didn't know. He makes snide comments about cheaters, sure, but also about almost everyone else his stories run across. So you have to be in the mood to be amused by a curmudgeon. Also you have to let him get away with being dismissive of any guessing or reasoning in case studies in different directions that clearly don't interest him much. I almost refrained from saying we'd never let a woman write like this, but there it sneaked in.

But if you can get past that, this is really a good outline for a few tests to put statistics through before you believe them. That's a good tool to arm yourself with when venturing out into the internet. And a good chapter on what to look for in a graphic and how data can hide in plain site. He also ends with a chapter titled "Don't try this at home", which must be some kind of humor I don't understand, because he very much encouraged us to try this at home and shares stories of people (strangely focusing on one very unusual family) who do investigate statistics with impressive results.

All in all, I learned a lot, and that's a great thing, but I was looking for something that I could unequivocally recommend to my science-major students, and I'm not sure this is it. But if you're ready to sit back and hear what this character has to say and learn what you can from him, you won't regret it.

I got a free copy of this from Net Galley.