I don't believe the polls

There, I said it.

I don't think that the pollsters know how to get a good reading of what's going to happen. Certainly the 2000 and 2004 elections weren't an advertisement for their accuracy.

This year hasn't been so hot, either. Remember Obama ahead by 12 in New Hampshire? Remember how Clinton won by double digits there? Oops.

Anyone who really wants to understand what's going on with the polls needs to read Stolen Thunder, who sounds like he really knows the internals of the industry. In particular, his dissection of weighting is a must read.

I also think that the Bradley Effect is in play here. Not so much with Republicans (who quite understandably won't vote for him because of his politics), but with Democrats and Independents. It will be worse because of Joe the Plumber - expect more (maybe a lot more) people being unwilling to talk to the press or pollsters after this.

About the only election prediction I'll make are the following:

• It's going to be a lot closer than people thing, or at least are saying. If there's Bradley Effect in play with Democrats, then Democratic Get-Out-The-Vote efforts will add McCain votes.
  
• The exit polls will be entirely unreliable. Jim Democrat will say he voted for Obama, even if he voted for McCain.
• Levels of fraud will be the worst anyone's every seen. If the race is close, this will make it hard for the winner to govern.
  

What this means for post-election, I have no idea. Politics seems very polarized, although the population in general seems to hate that. Presidential and Congressional approval ratings are in the toilet, along with the Press.

UPDATE 19 October 2008 20:07: Clayton Cramer has an interesting post on this, and raises the same point that Stolen Thunder makes. Namely, the pollsters make assumptions that are not historically valid, and this makes the polls less accurate.