Oops-onomics

The December 7th issue of Economist had this article posted http://www.economist.com/finance/displaystory.cfm?story_id=5246700

Let me start by giving a big thumbs up to Steven Levitt for his unorthodox take on everyday issues in his book "Freakonomics".

I am huge fan of the book, it makes interesting reading, but does not always make logical conclusions. I found Levitt's thought process fascinating and his approach refreshingly simple for a Chicago economist. Levitt has come up with some bold theories, but we should just leave it at that, they are just theories, open to debate.

The issue that is by far the most controversial, is Levitt's take that Roe V Wade was directly responsible for crime reduction. Messrs Foote and Goetz (two economists at the Federal Reserve Bank of Boston) in their paper have shown flaws in the statistical model employed by Levitt and Donohue, I am no economist, but as a student of engineering, keeping 600 possible variables constant in a study of this kind seems highly questionable. Professor Ted Joyce from Baruch College says quite accurately that "this robs the data, of most of their variety, and of much of their ability to explain anything."

In grad school, I recall trying to convince my professor, as to why I had kept one of the terms in a mathematical relation constant. Finally, after close to an hour of discussion, I had to accept that it was at best an assumption, without any physics behind it. I think, the same is true for Levitt's theory, its provocative, but there is no sound basis for some of his conclusions.

Talking about controversial research findings, a paper by Prof. David Grier (Physics Department, University of Chicago) comes to mind. Using a combination of experimental manipulations and statistical mambo jumbo he had managed to prove that like charges "attract". This was eventually proved wrong by his peers, but it shows that if the laws of physics can be interpreted incorrectly, under certain conditions, with the right statistical model and design of experiment, a rather undefined correlation between abortion and crime rate can be formed, as long as the degrees of freedom tend to infinity!