12 November 2004

Rolled up papers at fifty places (Keith Olbermann)

November 12, 2004

SECAUCUS— You know it's bad when the two sides start throwing professors at one another.

Two conflicting scholarly studies on the variance between the national exit polling and the presidential election results, are flying across the Internet, eating up your e-mail storage capacity.

One, from the University of Pennsylvania, reminds us that exit polls are used as 'audits' on the elections in places like Germany and Mexico, and suggests the actual statistical odds that the exit polling was that wrong in the battleground states were 250,000,000 to one.

The other, from a voting project managed by CalTech and MIT, says that while the incorrectness of national exit polling can't be explained by the proverbial 'margin of error,' on a state-by-state basis, it actually was within that margin.

Craig Crawford joins us tonight to try to make political sense of the theory-laden scholarly research.


No comments: