New Pew Poll: 052%, M46% [From 053%, M38% less than one week ago]

Pew ^

[rangeryder]

"New Pew Poll: 052%, M46% [From 053%, M38% less than one week ago]"

"Pew" is a good name for this smelly liberal poll.

[beagleone]

Pew? More like P.U.

[kesg]

I can’t make any sense out of this poll. For starters, the actual result was 49-42. Pew gets to 52-46 by splitting up the 9% undecideds almost evenly (in a way with which I disagree, but that’s beside the point.)

So, I’ll do with this poll what I typically do with the TIPP poll. For starters, here are the new results with the D37, R37, I26 distribution of 2004:

Obama 48.0
McCain 45.3
Und 6.7

Hmm....not so good. What about marital status? Are they using the lonely singles trick to get more single people into their sample? When I redo the Pew numbers using the 2004 distribution for marital status while holding everything else constant (i.e. using Pew’s original party ID), the new result is this:

Obama 47.9
McCain 42.9
Und 9.2

Well, Obama’s number stays the same, while McCain goes up by about a point. Okay, there are some lonely singles out there who got to spill out their life story to a pollster. But this isn’t yet getting me excited.

Okay, let’s see if they did the city slicker trick. When I reweighted the numbers by area type (urban, rural, or suburban), but without reweighting the poll by party ID, the breakdown looks like this:

McCain 45.2
Obama 44.8
Und 10.0

Ah hah! Pew tried to get away with the city tricker trick! Pew oversampled urban voters who favored Obama (54-35 in the Pew poll, as opposed to 54-45 Kerry in 2004) and/or undersampled surburban voters (47-43 McCain in Pew, 52-47 Bush in 2004) and the rednecks in the sticks (54-37 McCain in Pew, 57-42 Bush in 2004).

Note the further implications. We can infer from the 2004 numbers that the remaining city slickers lean mostly to McCain, the remaining suburbanites will essentially split or lean slightly to McCain, and the remaining rednecks in the woods will essentially split their votes or lean slightly to Obama. That is, of the remaining undecideds, 75% of them live in areas where they will break slightly for McCain and the remaining 25% live in areas where they will break slightly for Obama. Once again, I see 50-48 or maybe 51-48 in our near future.

One other quick comment. According to this poll, McCain leads 47-45 among white Catholics. This is significant because the voting populations of NH and WI are almost entirely white and more Catholic than the national average. PA and OH also have more than the usual share of white Catholics.

[Clintonfatigued]

Nearly all polls show the same trend- the race is tightening, McCain is climbing.

[Chet 99]

Pew provides no explanation on how McCain could gain 8% in just a matter of days from its last poll.

Well, I think we know the answer. They included more Republicans in this new poll... If Pew is truly randomly dialing people and not just dropping +15 numbers to demoralize Republicans, the recent upswing in Republican respondents indicates... growing interest/enthusiasm among the GOP. Something the Fox poll picked up on as well...

I am also surprised Pew would leave the field with 2 days to go when their numbers have already moved 8 points in just a matter of days...

[rwilliam]

With regards to adjustments to party affiliation, here you go...

Current poll results:

Obama - 49%
McCain - 42%
Undecided - 9%

Party affiliation in current poll:

Democrat - 38%
Independent - 28%
Republican - 28%

Previous poll results:

Obama - 53%
McCain - 39%
Undecided - 8%

Party affiliation in previous poll:

Democrat - 40%
Independent - 30%
Republican - 25%

So, the gap in party affiliation changed from +15 for the Democrats to +10 for the Democrats.

However, that change of -5 actually led to a -7 drop for Obama’s overall lead (14 points to 7 points)! So, based on this, Obama must have actually lost support beyond what can just be attributed to the change in gap in party affiliation.

In other words... If they dropped their gap in party affiliation to +5 for the Democrats, would their poll show a TIED result? If so, that’s not good news for Obama - especially since +5 may still be unrealistic!

-Bob

[kesg]

Pew provides no explanation on how McCain could gain 8% in just a matter of days from its last poll.

I know. And with the same amount of undecideds, no less. McCain simply went from down 53-38 to down 49-42 -- essentially claiming that almost 5 million Obama supporters changed their mind and decided to vote for McCain. I'm not buying it, or, more precisely, I'm not buying the polling methodology that would lead inevitably to such a conclusion.

Well, wait a minute. Didn't Pew do the last poll before the Obamammercial the other day? Let me reconsider what I just said....:)

[Chet 99]

Where do you come up with Dem +10 in the latest poll? It looked like 39 to 34 from what I could tell.

[tatown]

I knew it, the old city slicker trick...lol.

[Bmaloney]

P - ewe

That’s what I think of this.

[kesg]

I couldn’t figure it out either, so I simply ended up doing what I do for TIPP polls.

[kesg]

I couldn’t figure it out either, so I simply ended up doing what I do for TIPP polls.