Replies

Can you explain the difference between "internal polls" and the mainstream polls we see that are published every day. What are the different criteria that they use that makes them, supposedly, more reliable?

Also, I have a question on the "Bradley effect". Since so many of the published polls have Obama at an unbelievable advantage (I am specifically talking about the double digit polls) wouldn't even a "normal" Obama win of 1-3% seem to bolster the Bradley effect argument? In other words, regardless of the outcome of this election, many Dems are repeating the double digit win, so when it comes out less than that (as every election in history has) they will continue to say "It must have been the Bradley effect", instead of admitting that their polls are crap.

Sure. The main thing with polling is to get the best sample group of likely voters. Our arguments here all along have been that Zogby, Rasmussen, and most others have been waaaaayyy oversampling Dems because they just "think" the country is "more Democratic." We can't rely on 2006, because off-year elections are notoriously skewed. So you have to look at the voter makeup in 2004, which is roughly even (differs by state, but nationally it was about 39/39).

So, if you go through every major poll, you can predict within one point how much they will be off by how much they oversample Dems. If you want a 10-point lead in MN, sample 10% more Dems, even if the 2004 breakdown was only, say, 2%.

The internal polls generally can't afford that crap---they have to be right NOW, not on election day, because their strategy is based on what they see NOW.

That said, it's entirely POSSIBLE that the major pollsters are right, that Obama wins an unbelievable blowout, and that our analysis is sadly wrong.

But remember this: just because the pollsters have been close in a couple of elections, don't forget that they were hideously wrong in others. In 1994, NONE of them caught the Republican landslide. In 1996, ALL of them were off in Clinton's direction by as much as 8%---but the fact that none of them were off in Dole's direction suggests there is more to this than just "error," that there is some inherent bias that tends to lead them to certain conclusions. Neither the McCain nor Obama camp can afford those kinds of biases.