Battlepanda: My (minor) beef with Freakonomics


Always trying to figure things out with the minimum of bullshit and the maximum of belligerence.

Wednesday, May 18, 2005

My (minor) beef with Freakonomics

Steven Sailer himself (supposedly) directed me to his rebuttal of one of Freakonomic's most attention-grabbing conclusions -- that the legalization of abortion led to the fall of crime a generation later. I found his article to be an interesting, if hyperbolic ("pre-emptive executions", indeed) objection to the way Steven Levitt sliced and diced his numbers. Levitt defends his hypothesis to my satisfaction in his blog, but I wish that he did acknowledge the spike in murders that run contra to his theory in the book itself.

I must be almost alone in thinking this, but I don't find Levitt's abortion/crime hypothesis very controversial. If you accept that sub-optimal conditions increases the possiblity that a child grows up to be a criminal, and that most mothers contemplating abortion are unlikely to provide the ideal conditions for raising the child, it would be more difficult to explain why legalizing abortion would not lower the crime rate. Whether or not this would constitute "pre-emptive executions" is a separate argument.

I actually found Freakonomic's chapter on parenting much more problematic. According to Levitt, your child's lot in life is, statistically speaking, cast as soon as the child is born -- who you are as a parent is more important than what you do for your child. Now, I have no doubt that Levitt crunched the numbers correctly, and that even after all the proper statistical controls are taken, being born in a household with highly educated parents is a far better predictor of your eventual academic success than whether your parents read to you every night as a child. But I find the comparison entirely unfair because it holds up a measure that is predictive of a whole array of child-rearing practices against a single facet of childrearing. Of course reading to your child every night is not a silver bullet with an effect large enough to show up in statistical analysis, but that is not the same thing as saying it has no real benefit. There could be no statistical correlation between, say, one's guacamole consumption and one's weight because many people of different lifestyles, tastes and habits like guacamole. This does not mean guacamole is not a fattening food.

I would go along with Levitt's assertion that modern parents are way too neurotic about doing everything exactly right for their little darlings. If he stopped there I'd be completely with him. If you don't read to your child every night, you shouldn't feel guilty as long as you're doing something else to stimulate the child's learning. But he goes on to imply that everything parents consciously do to develop their child's mind is basically bunk. That is a conclusion that I don't think is supported by his data.

But hey, those are quibbles. Freakonomics is an awesome book, and I look forwards to hearing more from Levitt and those who follow his train of thoughs in the future.