Tuesday, February 27, 2007

Freakonomics

I finally got around to reading Steven Levitt and Stephen Dubner's Freakonomics; it's a quick and interesting read with lots of fascinating case studies, although frustratingly inconsistent. I think the main takeaway point is that a lot of "conventional wisdom" floating around out there in the ether is either plain wrong or grossly simplified. To which I say: fair enough.

The weird thing is that Levitt is portrayed as some dis-interested oracle of pure truth, slaying the sacred cows and dispelling common misperceptions. The book makes almost no mention of the fact that many of the ideas Levitt debunks are the product of ongoing social science research, not just the uninformed opinions of pundits and taste-makers. He's clearly a talented and original thinker who has written a number of surprising and influential papers -- but I doubt he's written the final word on such complicated sociological puzzles like the causes of crime or the black-white testing gap. There is also a lack of explanatory power in some of the ideas presented; Freakonomics often seems far better at debunking old ideas than it is in backing up its clever, simplistic alternatives or forging connections between its disparate topics.

The big controversial finding presented in the book is of course Levitt's assertion that the precipitous drop in crime during the early 1990s is best explained by the advent of Roe v. Wade twenty years earlier, and not by any of the dozen more common explanations (the boom economy, gun laws, better policing, etc.). The idea here being that mothers often make decisions about abortion based on their personal and financial ability to raise the child at that time in their life, and apparently that such factors are also strong predictors of 'criminal behavior' in the child. This is the book in a nutshell: bold, clever, unexpected, disturbing ... but true?

It is a somewhat troubling assertion no matter your stance on abortion. But even ignoring for now the merits of the argument it doesn't seem to fit with the rest of the book. Levitt is fundamentally asserting that a child's nurturing environment (or lack thereof) is key in predicting that child's outcome later in life (at least with regards to crime). So it's puzzling that a big chunk of the book is devoted to arguing that parenting style (i.e. reading to your child or taking them to museums or letting them watch TV) has essentially no impact on how well a child does in school. I'm probably missing some obvious unstated argument here, but this seems like a fairly big disconnect to me. Or maybe I'm hoping for a larger theory or organizing principal of human behavior that just isn't there.

Anyway, there's a lot to like in the book -- the chapter on the economics of drug dealing is really cool and their occasional column in the NY Times is also great -- but take it with a grain of salt. Provocative ideas are not always the full story.

2 comments:

Anonymous said...

I think my favorite line from this book is in the Intro: "Morality, it could be argued, represents the way that people would like the world to work—whereas economics represents how it actually does work." As to whether the correlations drawn to support the conclusions are valid or not remain to be seen. One might also be able to correlate the drop in crime to longer periods of bad weather just as well as to Roe V Wade...

Tim said...

Thanks for the comment, Anon.

The line you quote actually struck me as kind of strange. "Economics" represents how the world actually does work? If they had said "science in general" I might buy it, but I think economists often have certain biases that color how they view such problems (as do we all). Ask a psychologist or a sociologist or a linguist about the topics addressed in the book and I suspect you will get a very different answer. Which is perfectly fine, of course, and worth wishing different fields talked to each other more.

For example, there is a lot of fascinating ongoing research about 'stereotype threat' and the psychological underpinnings of learning (for good discussions see here and here) that provides an interesting counterpoint to D&L's discussion.