Honestly, I think it's because the sample size in most polls is so small as to be meaningless. "Scientific" or not, I find it difficult to believe that a "random" sampling of a few hundred voters can provide a meaningful snapshot of the electorate as a whole. I truly believe that this will turn out to be true this time around.
We have a record 44 polls today. If this keeps up we are soon going to have more polls than there are states. Florida wins with seven new polls today. By Tuesday we may be able to make a pretty decent Gaussian distribution just from the daily Florida polls. The most recent Florida poll, from Zogby puts Kerry ahead there 47% to 45%, with a margin of error of 4%. Yesterday he was behind by 1% there, so this is a jump of 3% in one day for Kerry. Of the seven Florida polls, four put Bush ahead and three put Kerry ahead. It is as close as it could be there.
I like this site because they do a very honest job of explaining their sources and statistical methods, and present the data in an attractive well thought out manner. Apart from that admirable effort, the site is neither conservative nor objective.
I've also been following electoral-vote.com and I agree with you that it's a fascinating site that doesn't pick sides. Well worth visiting, and I hope the guy who designed it has made something for his troubles. Understandibly, traffic today is heavy and I'm unable to get a good reading.
Take their projected results with a grain of salt, however. At one point they predicted that South Carolina would go for Kerry, and I think we can pretty well guarantee that this *won't* happen.
I never realized that the gap back then was that big. And yet Bush and Gore ended up tying the popular vote. Interesting.
Is it because the polls accidentally oversample Republicans, do you think, or do that many undecideds come off the fence in the final three days?
Honestly, I think it's because the sample size in most polls is so small as to be meaningless. "Scientific" or not, I find it difficult to believe that a "random" sampling of a few hundred voters can provide a meaningful snapshot of the electorate as a whole. I truly believe that this will turn out to be true this time around.
If this is the result from a single poll, I'm inclined to agree with northstar (get the cat!)
You can find more comprehensive analysis of multiple polls (and more significantly, the trends) at:
http://www.electoral-vote.com/
I like this site because they do a very honest job of explaining their sources and statistical methods, and present the data in an attractive well thought out manner. Apart from that admirable effort, the site is neither conservative nor objective.
I've also been following electoral-vote.com and I agree with you that it's a fascinating site that doesn't pick sides. Well worth visiting, and I hope the guy who designed it has made something for his troubles. Understandibly, traffic today is heavy and I'm unable to get a good reading.
Take their projected results with a grain of salt, however. At one point they predicted that South Carolina would go for Kerry, and I think we can pretty well guarantee that this *won't* happen.