Monday, April 20, 2009

My problem with "The Logic of Life"

I'm currently reading "The Logic of Life" by this economist, Tim Harford. I don't read a lot of nonfiction in my own time. I guess I tend to feel like if I'm not at work, I don't want to work at reading anything. I stick to sci fi/fantasy with the occasional book on computers or statistics if I'm feeling a lot of energy. But this book is not work, it's an easy read, and well written. And it's the first book I've ever read by an economist. What is interesting about it is that Harford is part of this "new breed" of economists who try to use economic principles to understand all human behavior, not just, well, economics. So he uses the idea of "rational people" to explain all sorts of irrational behavior. Such as prostitution, divorce, screwed up bosses, and teenage oral sex. Yup, he's wide ranging. He's not explaining all of that in terms of dollars and cents, it's not that kind of economics. Instead he's explaining it in terms of rational choices that maximize benefits and minimize losses. And he doesn't do a bad job at it - it makes you think!

As I type the premise it sounds either blatantly obvious or really silly, but as he writes it it is neither. Well written, as I said. My main problem with it is not so much the theories being put forth, which I find interesting, but some of the ways the theories are justified. What he, and some of the other economists he cites, seem to do is to look at large group behavior (say the behavior of teenage girls in a city, state, or country) and use that as proof or disproof of their theories. That may seem logical but it has a kind of post-hoc reasoning that sets my teeth on edge. The problem with post-hoc reasoning is that you take a look at an observed fact and then you say "ah, well, it's obviously because of X!" But it's typically possible to come up with a series of other, perhaps equally plausible, explanations, let's say theories A, B, and C. So now you don't know is observation Y due to X, or is it A, B, or C, or (most likley), is it a little bit of all of those?

I tend to view individual behavior as multiply determined, meaning that we can usually find a number of plausible and even real explanations for any single action. I'm writing this blog because I want to communicate, and because I want to achieve a goal of how many posts I write a month, and because I want to work out some of what bugs me about this book, and because I want to avoid working on this boring stuff I've got to do. All of those are true, so it's multiply determined. If that is true of an individual action, how much more multiply determined might a behavioral trend within a population be? And how many more plausible explanations, beyond or even in contradiction of X, can we come up with for the observations we make of that population? And given that, how can we really conclude that we've proven theory X from even a string of interesting observations from populations. Another problem with that line of reasoning is that it's easy to "cherry pick." You'll say "Let me tell you about the six different population trends I've found that prove theory X!" And all fit really well with the theory. But then you either don't know, or even worse know but fail to report, the 456 other observations that fail to confirm theory X. I'm taking this to an extreme here, but hopefully it illustrates the point.

The funny thing is that this problem with his reasoning doesn't necessarily invalidate all of the points that are made. I still agree or at least find myself interested in many of his assertions. It just annoys me! At first I was thinking that perhaps economics just has a different way of hypothesis testing than I've been trained in, and that's certainly plausible. At the same time, I'm having to remind myself that this is an economics book for a wide audience, a "pop economics" book, if you will. I know that books written in my field for popular consumption tend to dumb things down and oversimplify to an extent that I sometimes find appalling. So I won't generalize to all of economics based on one book, and I won't discount all of what I'm reading based on one line of evidence. But I'm still annoyed.

No comments: