UPDATE More Zogby data available now. Obama loses 8 points among women? More below the fold.
The new LA Times poll shows Obama leading McCain 45% to 43%. This lead is down from the June LA Times poll that showed Obama leading McCain 49% to 37%. The LA Times and the Republicans would have you believe that the race is tightening and that McCain's negative attacks on Obama are working.
What may be the real story here is that independents are breaking for Obama. In June, independents favored McCain 44% to 36%. In the new August poll, independents favor Obama 46% to 35%.
Now if you've been paying attention you are asking yourself, "If Obama went from -8 to +11 with independents, then how did the race get closer?"
The answer is distribution of party affiliation, my dear reader.
If you follow me below the fold, I will deconstruct the math, hit the Republicans over the head with their own numbers, show that the race hasn't tightened at all, and see how much McCain's slime has hurt McCain's favorables..
TOP-LINE RESULTS
Let's recall back in June the hue and cry from the Republicans that the Times poll over-sampled Democrats. The Times reported a sample of 39% Democrats, 22% Republican, and 27% independent.
The LA Times poll tends to bring to the fore a dispute in the pollster community. The LA Times does not believe in weighting its sample by party identification. Campaign pollsters swear by it - as do Gallup and Rasmussen. I happen to agree with campaign pollsters that party identification is a stable enough variable to be used in compiling survey results.
Republican pollster Public Opinion Strategies publicly complained about the party sampling in the June survey. The Times noted:
The memo cites several other recent polls in which the party ID gap ranged as low as plus 6 percentage points for the Democrats to as high as plus 14.
It then asserts: "McCain’s double-digit deficit is not a reflection of reality, simply a result of an unusual party identification result in this survey.... If party identification on the L.A. Times survey is recalculated to ... 29% GOP / 39% Dem / 27% Ind / 5% Don’t Know/Refused, the ballot would be 40% McCain – 47% Obama."
(minor quibble, using these round numbers gave me 46%-40% Obama)
So I went looking for independent confirmation of the cited party distribution. Rasmussen surveys 15,000 voters each month to track party identification. The average monthly distribution in 2008 is 41% Democrat, 32% Republican, and 27% independent. This tracks pretty closely with the numbers asserted by POS (yes, I know, an appropriate acronym for Republican pollsters, but they picked it, not me) if you split the 5% Don't Know/Refused.
Using the POS distribution in their (legitimate) whining memo of June with the Times August survey (pg 13, Q2) yields an Obama advantage of... 46% to 40%.
Yes dear readers, the overall poll results are UNCHANGED when using the POS party distribution.
To paraphrase our friends at POS, McCain's surge is not a reflection of reality, simply a result of an unusual party identification result in this survey.
Don't let Republicans cherry-pick the results. As far as the LA Times poll is concerned, the race hasn't changed - we proved it with THEIR numbers.
::
FAVORABILITY
Re-calibrating the favorability ratings significantly changes the narrative as well.
June favorability re-calibrates from 59% Obama and 47% McCain to 55% Obama and 49% McCain. August favorability re-calibrates from 48% Obama and 46% McCain to 48% Obama and 43% McCain. Instead of Obama losing 11 points and McCain losing 1 point, the re-calibration shows Obama losing 7 points and McCain losing 6 points.
The original narrative defied history. History shows negative advertising brings both candidates' favorables down, not just the target. This historical perspective validates the results of the re-calibrated numbers. McCain's negative ads are driving down Obama's numbers, but at an equal expense to McCain. And more significantly, McCain's negative ads are not moving the top-line results.
::
Zogby
The latest Zogby poll shows McCain up by 5. The previous Zogby poll showed Obama up by 7. Let's look at this quote from Zogby:
"There were no wild swings, there isn't one group that is radically different than last month or even two months ago. It was just a steady decline for Obama across the board," Zogby said.
Without any wild swings in these groups, a 12 point swing is impossible unless Zogby is also changing the mix of party affiliation.
If anyone can find the internals for this poll, I will try to deconstruct them as well.
UPDATE 1 The damage is already done on the media narrative front for Obama. But I did find some more numbers from Zogby on his site (still no internals to be found).
Obama Support | July | August | Difference |
Democrats | 83% | 74% | -9 |
Women | 50% | 42% | -8 |
Catholics | 47% | 36% | -11 |
Ages <35 | 59% | 47% | -12 |
College Grads | 51% | 40% | -11 |
Live in Cities | 54% | 43% | -11 |
Income <$50,000 | 53% | 46% | -7 |
Southerners | 46% | 35% | -11 |
If Obama truly has lost 8 points among women, then I will EAT THIS DIARY. There is something rotten in the mix here.
This is also another LIKELY voter poll, not registered voters. For an explanation of poll skewing when moving from registered to likely voters, see Alan Abramowitz over at HuffPo deconstruct the recent USA Today/Gallup poll. In it he shows that the voters knocked out when moving from registered to likely preferred Obama 61% to 7%.
I believe it is irresponsible of Reuters to publish this survey without offering the internals for readers to review. At least the LA Times gave us the data to analyze.
::
Remember, it's easier to fill 24 hours of cable news when you report on a race that's "too close to call." But we know that the fundamentals just don't support it.