When 22% of respondents say they voted a certain way because of a nebulous concept like "moral values" the press extrapolates wildly about massive conservative turnout and the new ascendancy of the religious right.
But when the exit polls in every swing state predict a Kerry victory, and then Bush wins the election, the press rushes to claim that the exit polls were all wrong.
It seems many reporters engage in "wishful polling." They develop a story angle first and then go find the statistics they need to reinforce it. This is nothing new. However this is the first time I've seen the media almost unanimous in questioning the entire polling discipline.
A new analysis by Dr. Steven F. Freeman of the University of Pennsylvania makes a strong case for the continued accuracy of exit polls. [LINK] Freeman concludes that the natural odds of the actual vote counts varying so dramatically from the exit poll results this year are about 250 million to one. Freeman's full report is available online. [LINK]
Right now the exit polls are one of the few relatively independent checks against widespread vote manipulation. The Republicans are now trying to pressure the news agencies to stop exit polling altogether. [LINK] Let's hope the agencies don't cave in to Republican pressure on this one!