Final Pew national: Obama +6

[Grad Students are the Worst]

Likely voters:

Obama 49%
McCain 42%

With undecideds allocated:

Obama 52%
McCain 46%

Registered voters:

Obama 50%
McCain 39%

http://people-press.org/report/468/obama-leads-mccain-in-final-days

[Lief]

Oh Pew, playing with our emotions for a few weeks with crazy numbers and then releasing polls right before the election that make sense.

[J. J.]

A tightening since the last poll.

[Iosif]

A tightening since the last poll.

LOL

I refuse to believe you're real.

[Lunar]

A new national poll is going beyond saying Obama is ahead: It's predicting that Obama will win by six points.

The final Pew poll predicts that the outcome will be Obama 52%, McCain 46%, with the remainder split among third-party candidates. The top-line result among likely voters right now is Obama 49%, McCain 42%, with a ±2.5% margin of error. Last week, Obama had a much wider lead of 53%-38%.

In an interview, Pew's director of survey research Scott Keeter explained to us how they arrived at that 52%-46% predicted outcome. Pew performed a statistical analysis looking at the remaining undecideds on the basis of their demographics and their answers to issue questions, in order to project how they'll vote -- if they'll vote at all, that is.

"This is an unusual group of people, anybody who could get to this point in the campaign and not have made a decision," Keeter said. "So you have to think some of these people are not even going to vote."

Pew predicts a very narrow break of undecideds to McCain, and it won't be enough to overcome Obama's lead.

The same methodology was right on the money in 2004, predicting Bush at 51% of the vote to Kerry's 48%, and in the very-close 2000 election was slightly off in giving Bush 49% to Gore's 47%. But in an election that doesn't seem as close as close as it was in 2000, Keeter doesn't think any remaining uncertain factors will be enough to undo Obama's lead.

/Greg Sargent

[cinyc]

A tightening since the last poll.

LOL

I refuse to believe you're real.

Huh?  The last Pew poll had Obama up by 15.  So it is a significant tightening versus Pew's last poll.

[kevinatcausa]

Technically the poll result here is Obama +7, though it's still a significant tightening. 

The +6 is just hand waving and guesswork on the undecideds. 

[cinyc]

Technically the poll result here is Obama +7, though it's still a significant tightening. 

The +6 is just hand waving and guesswork on the undecideds. 

True.  The tightening is due in part because Pew's last poll was such an outlier that it wasn't believable.  Their final poll is at least plausible (at Obama +6 or +7).

[TheGlobalizer]

Now's the time when the pollsters play CYA.

[Reds4]

28% GOP 38% Dems in this one it looks like..

[All life is a blur of Republicans and meat.]

Nice save Pew.

This looks about right... allocating udecideds is dangerous though.

[kevinatcausa]

What some people (e.g. ARG) use as a measure of a pollsters accuracy after the election is the absolute difference between the final poll's result and the final election result. 

For example, if the final result turns out to be 51-47 Obama-McCain, then a pollster who predicted 54-45 would be (54-51)+(47-45)=5 off, while one whose last poll showed 48-44 would be (51-48)+(47-44)=6 off, so the 54-45 prediction is deemed "closer" than the 48-44 one. 

This is absurd, but anyone who uses this sort measurement gives a pollster a strong incentive to just make their best guess at assigning the undecided voters, regardless of how accurate that guess is. 

[So the Heroes Fall]

A tightening since the last poll.

LOL

I refuse to believe you're real.

Remember when he believed a gain of 0.07% for McCain in one of the tracking polls was the start of a trend? LOL.

[InsaneTrollLogic]

JJ is either deluded or can see into the future. We will know in 48 hours.

[Kalimantan]

JJ is either deluded or can see into the future. We will know in 48 hours.

Oh I think we know already

[Mr.Phips]

I think this poll is wrong about the undecideds.  I think the undecideds will go about 4 to 1 for McCain. 

[Lunar]

I think this poll is wrong about the undecideds.  I think the undecideds will go about 4 to 1 for McCain. 

They did demographic and other analysis, you're just making blind assumptions.  4 to 1 is ridic.

Here's another poll's breakdown: