I was asked by Sophie Roell, an editor at The Browser, where every day they ask an expert in a field to recommend the top five books, not by them, in their subject. I was asked to recommend five books on how Americans vote.
The trouble is that I’m really pretty unfamiliar with the academic literature of political science, but it seemed sort of inappropriate for a political scientist such as myself to recommend non-scholarly books that I like (for example, “Style vs. Substance” by George V. Higgins, “Lies My Teacher Told Me,” by James Loewen, “The Rascal King” by Jack Beatty, “Republican Party Reptile” by P. J. O’Rourke, and, of course, “All the King’s Men,” by Robert Penn Warren). I mean, what’s the point of that? Nobody needs me to recommend books like that.
Instead, I moved sideways and asked if I could discuss five books on statistics instead. Roell said that would be fine, so I sent her a quick description, which appears below.
The actual interview turned out much better. Readable and conversational. I give Roell credit for this, keeping me from rambling too much. The interview includes the notorious hookah story, which should provoke a wince of recognition from anyone who’s ever served on an NIH panel.
Below is my original email; the full interview appears here.
I was originally asked for five books on how Americans vote. I’m sure that five excellent books on that topic exist but I don’t know what they are, so instead I’m offering five books on statistics–broadly defined, since I don’t think your readers are looking for a bunch of graphs and equations.
So here goes. Five books, going from the most to the least statistical, but all brilliant, thought-provoking, and both fun and challenging to read (a rare combination).
1. The Bill James baseball abstracts, from 1982 to 1986. He mixes in stories and goofy-statistics with in-depth analysis of questions such as, At what age are baseball players most productive, Which is more important: speed or power, and Are the Cubs hindered by playing so many day games? When somebody asked him to do a study that he didn’t feel like doing, he replied, Hey, I’m not a public utility: if this is so important to you, do the analysis himself. James’s abstracts went downhill after 1986–he started falling in love with his own voice and offering more opinions and less analysis–and he wisely discontinued the series two years later.
2. Judgment under uncertainty: heuristics and biases, edited by Daniel Kahneman, Paul Slovic, and Amos Tversky. The best edited book ever (at least, in the nearly two thousand years since Matthew, Luke, and the others laid down their pens). This collection of works on cognitive illusions (now sometimes called “behavioral economics,” but that’s misleading since economics is only one of the many many areas of applications of these ideas) is just amazing. Chapter after chapter of incredible findings on anchoring and adjustment, overconfidence, the illusion of control, and other concepts that are absolutely essential to understanding how humans think.
3. How animals work, by Knut Schmidt-Nielsen. Physics. How birds can keep themselves in the air, how dogs cool themselves by panting (no, it’s not as simple as you think), how elephants can be so large, and so forth. It’s not really statistical–except that many of the findings are data-based, and the book has lots of amazing graphs. A wonderful example of the interplay between substantive modeling, data collection, and statistical analysis. And, like the other books on this list, a great read.
4. The honest rainmaker, by A. J. Liebling. Stories about an old-time character who went by the name of Colonel John R. Stingo. Apparently Liebling made much of it up himself. In any case, it’s one of my favorite books, and the chapters on Stingo’s moneymaking schemes have a fair amount of statistical content. (You can’t talk about rainmaking without thinking about probability.)
5. How to talk so kids will listen and listen so kids will talk, by Adele Faber and Elaine Mazlish. I read this book long before I had kids–it’s incredibly helpful for interactions with adults as well. It’s a book that really changed my life. No statistical content at all–but, implicitly it’s all about statistics, about what works and what doesn’t. I imagine that there’s a whole scholarly literature in which ideas such as described by Faber and Mazlish have been evaluated statistically. Maybe they don’t even work, I dunno. But the book seems great to me, and I recommend that every statistician read it.