Plus, as in 2004, the polls are a bit flaky..of course, in that election, we realized that several of the polls where actively skewing toward Kerry…if that is true this time…and ‘The One’ and his team know it…then ACORNs work in ‘electioneering’ will pay off big.

Seems like the Dems are over-represented in this poll:

49% dem
36% rep
14% other

and all they can muster is a two-point lead??!!?

jdpaz on October 17, 2008 at 3:53 PM

And the AP/Yahoo poll… 46-Obama, 44-McCain that sampled 57% Dem and 43% Rep. Like the article crying “conspiracy theory” on the intrade buys of McCain. When you look at polls like these and you see McCain trading at less than 25, how can you not make a buy… Yet there was no mention of “conspiracy” when the article came out about the Irish bookie deciding to go ahead and pay off Obama bets. Yeah, right…. I’m sure that is legit. This entire election is beginning to smell so bad I need a gas mask.

CC – BHO: “my Muslim faith”

I do not think Obama is gonna win Pa or Ohio. I love it! :)

It is acceptable to me. Of course I will claim this as my own work, you understand? – Bishop on October 17, 2008 at 3:31 PM

I would have it no other way, Your Grace.

Ideally, you want a large (over 1000) sample of likely voters within a relatively short period of time (that’s why the most valid knock on this AP poll is the 10-day period).

That’s why I’m not too keen on either the Hotline or Battleground polls (even though one of them has the good sense to avoid weekend polling). The daily samples are so small that you get a lot of volatility.

As for Gallup’s “expanded” LV model, ask yourself how realistic it is that turnout will increase 8%. RCP uses both Gallup LV numbers and splits the difference, which is more realistic.

So if you look at the big trackers — Rasmussen, Zogby (eek) and Gallup — you end up with about a 4 to 4.5% lead for BO — and dropping. You can see it even in the overall RCP average now, but most of the major trackers individually show Maverick gaining over the past week or so.

tneloms on October 17, 2008 at 3:57 PM

Although one might question the amount of time taken to conduct the poll, Nate is otherwise talking through his hat.

This is not the typical online poll. It is a blend that selects a sample the traditional way, but questions via computer.

And contrary to Nate, AP-Yahoo! has used this poll to do horserace analysis, though the AP also uses Gfk.

The national pollsters are deliberately skewing their samples to inflate the Obama numbers and discourage Republicans. But if we can get out the vote at least as well as 2006, this is winnable for McCain/Palin.

Steve Z on October 17, 2008 at 3:55 PM

Let me just add that 2006 was a bad republican turnout year, so the numbers could be even better now.

Exit answer: Probably not.

ALLAHPUNDIT PREDICTS OBAMA WINS THE ELECTION!

This will be my new headline for every AP thread from here on out.

Thanks bunches for the information and polls Allah. We appreciate it.

What, no doom and gloom? I’ve lost faith in you, Allah.

It’s Vintage, Duh on October 17, 2008 at 3:04 PM

You missed the last paragraph. Plenty of AP’s patented D&G to be had there.

The reason the “bombshell” result isn’t in the article is that the AP doesn’t use this poll to determine its horse race numbers because it’s not reliable for that purpose. Didn’t you guys notice that you’d never heard of this poll before this was posted on Drudge? Didn’t that make you think that there was a reason that it disagreed with all the other polls out there?

Dem – Rep percentage: 7, 9, 8, 7, 11, 12, 12, 13.

What happened between Waves 4 and 5 so that they suddenly increased the Dem percentage by 4 – 5 points? Were they not getting enough Obama advantage, so they had to spike the sample with more Democrats?

Rasmussen has been weighting their polls with a 39 D – 33 R assumed partisan split. Assuming that is correct, if AP/Yahoo has been undersampling Republicans by about 5-6%, if we assume that all the extra Republicans vote McCain, then assume half the Independent voters taken out of the sample would also vote McCain, McCain would pick up a net 2.5%, meaning that McCain is actually LEADING by a small margin.

Over at http://www.hedgehogreport.com, they’ve got bloggers who pick apart the internals of most of the major polls, and re-apply them to the partisan split of the 2006 electorate (when Democrats won both houses of Congress). The results are striking: the national polls come out about even, and McCain is leading all state polls in FL, OH, VA, and NC, with MO, NH, and PA as tossups.

The national pollsters are deliberately skewing their samples to inflate the Obama numbers and discourage Republicans. But if we can get out the vote at least as well as 2006, this is winnable for McCain/Palin.

49% dem
36% rep
14% other

and all they can muster is a two-point lead??!!?

Here’s an idea for a poll: immigrants other than Hispanics.

We never see polls like these because immigrants from repressive countries vote Republican. My Filipino neighbor–from the island largely taken over by AQ–said that as soon as she heard Obama say he would meet Iran without preconditions, she went for McCain. “You can’t talk to these people.” Eastern Europeans, Asians, Vietnamese–they know what radicals can do and will vote for the tough guy.

Don’t forget Cubans buddy, they are hispanic and know what the beast looks like too.

I’m convinced that the difference in party identification will be at most 5 points based on that fact that it hasn’t moved much between 2000, 2004 and 2006. I’m giving the dems 1 percentage point for fraudulent votes.

This is, of course, assuming McCain doesn’t get burned with an October surprise, make a grevious mis-statement or do something really dumb like suspending his campaign again.

It is acceptable to me. Of course I will claim this as my own work, you understand?

SuperCool on October 17, 2008 at 3:25 PM

I’ve commented several times over the past few months that it a Tortoise vs. Hare race.