Professor Freeman Stands By Stats

Freeman concluded in a paper on Nov. 10 - a paper, by the way, he has revised twice since - that the odds of exit polls in the critical states of Ohio, Florida and Pennsylvania all being so far off were roughly 662,000 to 1.

Three weeks later, Freeman says he stands by his findings. At the same time, he admits being flabbergasted by the response: More than 1,000 e-mails - a large percentage of which were shockingly mean-spirited, he says - and about 100 requests for interviews, most of which he's had to turn down.

(He's grateful, by the way, for the many positive e-mails he received from faculty members and students at the University of Wisconsin-Madison. Which, he acknowledged, is a big reason he granted my request for an interview. "Madison," he suggested, "is a lot like Philadelphia. I mean, the truth of it is, I don't know a single Bush supporter at this university.")

Freeman says his intent from the beginning was just "to generate some attention" about the discrepancy and to raise a legitimate question: What caused it? "But I didn't know it was going to generate such a reaction."

Granted, "every statistician is going to come up with a slightly different number," he adds. "But I really thought I was just stating the obvious. I don't think there's much to disagree with."

In fact, Freeman says, he expects that a second paper he'll be releasing this week will be far more controversial. That paper, he says, analyzes the different explanations given for the discrepancy - such as the contention by Republicans that Bush supporters disproportionately refused to respond to the pollsters.

That may well be true, Freeman says. "But, you know, nobody's provided any evidence of that."

Whatever the case, Freeman was feeling pretty darn good on Monday. Although the major media virtually ignored his paper when it first came out, CNN, USA Today and The Washington Post have all requested interviews in recent days, he says.