Day By Day

Sunday, December 04, 2005

Freakonomics Challenged

In conversations with Democrat activists over the past year the word "Freakonomics" has appeared with distressing regularity. The reason is that Steven Levitt and John Donohue, the co-authors of Freakonomics put forth a scientifically based utilitarian argument in favor of abortion. To left-wing devotees of the cult of science this was conclusive and lifted the entire debate over abortion out of the moral realm. Donohue, and particularly Levitt who designed the study, became mega-celebrities on the left and heroes to the "pro-choice" movement.

Levitt's argument, simply put, is that the legalization of abortion, by reducing the number of unwanted kids being born to poor, single women, dramatically lowered the level of street crime. For libs this was a two-fer. It provided "scientific" and utilitarian sanction for abortion on demand, and it undermined the argument of prominent conservatives, like Rudy Giuliani, that "get-tough" policies were responsible for the drop in crime statistics at the end of the Twentieth Century

Well not so fast guys.

Christopher Foote and Christopher Goetz, economists at the Federal Reserve, have gone back over Levitt's work and found some glaring errors in his methodology. Their work was presented in a paper" “Testing Economic Hypotheses with State-Level Data: A comment on Donohue and Levitt (2001)”. Federal Reserve Bank of Boston working paper. November 2005. [available here] It is summarized in the Economist here.

Basically, Foote and Goetz found that Levitt committed three important errors. 1) His computer program did not include controls that he claimed to have used. Once appropriate controls are included much of the effect he was trying to measure disappears. 2) Levitt was using outdated data. When data were updated the effect was further diminished. 3) Sloppy design resulted in inappropriate measures being applied in crucial tests. When appropriate data sets were used the effect Levitt claimed to have found disappeared completely.

Once again an attempt to provide "scientific authority" to justify public policy crashes and burns.

Sorry, guys..., back to the drawing boards. Better luck next time.

Parenthetically: One conversation I had was particularly disturbing. A prominent attorney described to me a meeting he and other lawyers had with Steven Levitt in which they queried him on ways in which his work could usefully be applied in courtroom arguments. In recent decades we have seen horrific results from the application of junk science to the law. Given the unreliability and malleability of "scientific authority" it would be wise to limit its applicability to matters of law and public policy.

No comments: