Originally posted by: Pabster
Originally posted by: ProfJohn
At the same time the exit polls in 2004 were a total shame. As some have said there is no way you miss the final results that much using exit polls. Something went seriously wrong is their polling methods. I believe he most likely cause was female voter turnout, they "guessed" the number of female who would turn out as being higher than it actually was and since Kerry took a larger portion of the female vote the exit polls showed him doing better than he actually did.
It was no accident. And expect to see a repeat in 3 weeks.
I wouldn't bet on it. The site
www.electoral-vote.com, a site that averages many polls together (a far more accurate approach), has done some major reworking of their approach since 2004 (as have major polling agencies, I imagine). They applied their new method to the polls collected during 2004, and the results were accurate in every state except Iowa, a state that truly was a statistical tie for all intents and purposes and well within the margin of error. Every other state, believe it or not, was called correctly, with the exception of WI and NM, which were not predicted as they were too close to call. In other words, their current approach seems very accurate and well thought out.
So what do they make of this election? Well, they seem to think that the Senate will be almost tied, with a possible majority seat swinging one way or the other (the two likely Independent Senators will most likely be Democrats in all but name, at least for most purposes). In the House, they seem to think that the Reps are done for, the Dems will end up with a majority of several seats, at least 226.
But hey,
www.electoral-vote.com was WRONG in 2004, right? Plus the guy who runs it leans left. So how about the competition,
www.electionprojection.com, a site run by a red-blooded Republican voting Bush supporter? Not only is he in your political camp, but he was much more accurate in 2004...he did what EV.com was only able to do after the fact, predict the results in every state except one...he got Iowa wrong as well. But his 2004 approach was much more solid than EV.com, he has a great track record of accurate predictions.
So what's he saying this time around? Well, in the Senate he's saying basically a tie, with 50 Republicans, 48 Dems and 2 Independents. Since Lieberman and Sanders are more likely to side with Dems than Reps on most issues (especially Sanders), it's again, essentially a tied Senate. In the House, things aren't quite like EV.com is saying, but the Republicans are still done, with the projection of 222, a majority of 9 seats. Not earth shattering, but it would essentially be a reversal of the current House makeup.
Individual polls (that the Republicans flogged for MONTHS after 2004) can be wrong to a greater or lesser extent, but sources like the two mentioned above have a lot of solid theory behind how they do their predictions, and in both cases, especially the latter site, are quite accurate. You can scoff at predictions all you want, but I think you're leaning a little too much on a story that's been blown WAY out of proportion. The facts here are against you, the site that came within 1 state of prediction 2004 is telling you that the Republicans are going to lose their majorities in both houses and become the minority in at least the House...far from casting doubt on the results, 2004 supports their case.