April 21, 2005

KICS: Keep It Complicated, Smartie!

One amusing aspect of the controversy over economist Steven D. Levitt's abortion-cut-crime assertion is how many academic economists seem to feel that, even though Levitt ducks answering my simple challenges to his theory, they should take his arguments for his theory (based on complex multiple regression studies) on faith because they are more complicated than my easy-to-understand graphs and comparisons.

My statistical training came at MBA school and on the job in the for-profit marketing research industry, where the maxim KISS: Keep It Simple, Stupid! is much admired. Academic economists, however, seem to be followers of KICS: Keep It Complicated, Smartie. The more moving parts required to support a theory, the better!

A research psychologist at a university writes:

These economists, with their faith in regression analyses, amaze me. If you do lots of clever follow-ups you can sometimes get some fairly good idea what is going on using regression, but even that is never definitive -- but just to do some very molar-level coarse regression and declare victory -- it's just crappy social science.

Of course, several academic economists have written long multiple-regression based analyses undermining Levitt's theory. Both Ted Joyce and the team of John R. Lott and John Whitley have sent me their latest, unpublished responses to Levitt-Donohue's response to their first responses (got that?). (Here are the abstracts from the two newest papers.)

(Lott and Levitt notoriously don't like each other. I don't know whether Levitt first attacked Lott's more-guns-less-crime theory before or after Lott attacked Levitt's abortion-cut-crime theory, but in any case, Levitt's Freakonomics includes a half page of ad hominem abuse of Lott while attempting to dismiss Lott's more-guns-less-crime theory.)

One problem with highly complex models is that they make it easier to hide your assumptions. Levitt, in effect, assumes that abortion could not have caused the rise in crime among people born soon after legalization but could have caused the fall in crime after the crack years. So, hesto-presto, he gets the results out of his model that he pre-ordained in the first place.

Joyce assumes that the crack crime wave was a random event and works very hard to see what the effect of abortion was on non-crack related criminality. He finds no effect.

Lott and Whitley, in contrast, are open to the possibility tended to get started first in high abortion places like NYC is not a coincidence. They find a positive relation between the abortion rate and the subsequent murder rate.

Still, what strikes me about both papers is how Levitt's academic critics lack the instinct for the jugular that is inculcated in the business world. Due to the methodologically shoddy way Levitt assembled his theory, it's easy to show graphically that several of his claims about recent social history are 180 degrees false, but academic economists seem to have a prejudice against using killer graphs. The cumulative persuasiveness of both papers is great, but you can see how many observers would, rather than carefully read all the papers, instead just throw up their hands and assume that the fashionable young superstar from the U. of Chicago is right on reputation alone.

I hope economists, of all people, also aren't too offended when I point out that they also have a self-interested reason for hoping Levitt iseen as winning this debate: Freakonomics remains the #2 bestseller on Amazon, which is an extraordinarily strong performance for a book by an economist. I'm sure that dozens of economics professors are preparing book proposals at this very moment.

Colby Cosh explains the ideological reasons both Pro-Choice and Pro-Life alike seem to prefer Levitt's theory:

"Meanwhile, Steve Sailer is quietly destroying the suddenly-ubiquitous Steven Levitt. Freakonomics' hypothesis that the demographic effects of Roe v. Wade ultimately served to drive American crime rates down in the 1990s. It feels like Levitt's theory has been embraced by both pro-choicers, who would like to believe that legalized abortion has positive social effects, and by pro-lifers, who just want to convince people that it has some sort of second-order social effect we are obliged to consider. Unfortunately for both sides, the theory had already been reduced to a rapidly drying heap of bones before Levitt put it in book form."

There is also a personal reason for the popularity of the abortion-cut-crime theory. After 40 million or so legal abortions, there are a whole lot of people with uneasy consciences, and Levitt's theory lets them say to themselves, "I wasn't really sliding out of a sticky situation, I was ... fighting crime! Yeah, that's the ticket!"

Levitt, on the other hand, appears to be personally Pro-Life. He and his wife have had three children themselves, one of whom tragically died, and adopted another. He's an activist in educating people about the dangers to children of swimming pools, saying they owning a swimming pool is times more dangerous to kids than owning a gun.

My published articles are archived at iSteve.com -- Steve Sailer

No comments: