Random exit poll thought
If a significant portion of Bush voters are the way they have been portrayed: resentful of "culture", skeptical of mass media types, preferring the isolated world of their church (or whatever works for them), why is it not reasonable to assume that they would sneer at exit pollsters, rather than answering their 30-page questionnaire? Let's face it--there are significant differences between Kerry voters and Bush voters. Maybe that difference manifests itself by skewing the exit poll sample a handful of points here and there. The Penn statistician seems to indicate this is the only reasonable explanation, short of voter fraud. His analysis, leading to the conclusion that it's a 250 million to 1 shot, is admitedly based on the assumption that there was a totally random sample. Maybe the group of folks who refuse to participate is a growing one, and maybe they are more likely to vote one way than another?