Monday, January 30, 2006

Correlation and Causation

I just finished reading Freakonomics,, (Thanks Thomas for turning me onto it).

I have to highly recommend this book to everyone for a few reasons.
1) Change of perspective. The authors do an excellent job of looking at the world in a different way or posing questions in a unique manner.

2) Sound explanations. They do an excellent job of keeping the language simple while applying solid metrics. Their use of correlation and causation is amazing. They were always very explicit about what their data found.

Now onto my real point, a wonderful topic, abortion. Specifically the Levitt (the economist author) wrote a paper, detailing how abortion caused the drop in crime in the early 90's. The interesting point is that both the left and right decried this as horrible. The right's argument was naturally abortion couldn't have such a benefit, while the left was angered by the fact that he applied this metric to the poor and black segments of society.

From what I read I can't disagree with his assesment. The data backs up his argument. Now I am wondering is this because I am pro-choice and don't want to find flaw with his methodolgy? So what I am asking of other people is to look over this if they have the time and find a flaw I have missed.

If there is an error, what is it?

If there is not an error, what does this say about attempts to make abortion illegal?
Is there an alternative that is more palatable for people?
How about mass distribution of contraceptives? It seems that would generate the same effect. The core of the problem is children raised in "bad" environments are more likely to commit crime. Would a vaild anti-crime policy be to encourage the use of contraceptives?