When we put out our Virginia poll yesterday with a party breakdown of 37% Democrats, 29% Republicans, and 34% independents people who wanted to discredit its results were quick to jump to the 'the sample is biased!' cry.
First off, for the umpteenth time, we don't weight for party. So we didn't 'create' those figures, the likely voters in Virginia did.
Let's dig a little deeper.
Our final Virginia Presidential poll last year was right on the mark. We said Barack Obama would take the state 52-46 and that's what happened. The party breakdown of the poll was 40% Democrats, 35% Republicans, and 25% independents.
Let's compare that to this week's poll: a 9% increase in independents since then with a 3% decline in Democrats and a 6% decline in Republicans.
Now let's take a bit of a deeper look at those independents. We found Barack Obama winning them 49/47 last year. Now his approval with them is a dreadful 35/56 and Creigh Deeds trails Bob McDonnell 53-37 with them as well.
A significant decrease in Republicans and a sudden significant rightward turn in the independent ranks? Pretty clear those two things are correlated. Those Republican voters from last year are still there. They're just not identifying as Republicans any more. But whether they identify themselves as Republicans or independents they're certainly still conservatives, and that's reflected in their appraisals of the President and who they plan to vote for this fall.
The spread between Democrats and Republicans in the party id breakdown becomes a lot less consequential when the independent ranks are so skewed in one direction.
Sorry, I don't buy it. In just about any poll, McDonnell is winning Independents by double-digit margins.
ReplyDeleteTo write that in the past two months, Republicans have dropped from 35% to 29% of the sample due to them now describing themselves as Independents doesn't pass the mustard.
And nobody is screaming bias, so please stop the whining. The SurveyUSA poll released yesterday is also suspect due to PARTY ID.
Republicans identifying more as independents is a long term trend dating back to last year.
ReplyDeleteThe reason the Republican id has dropped in the last two months is because Democrats had NO interest in this election over the summer, but are now becoming more engaged. Each poll is a snapshot in time and the early August poll reflected the fact that if the election had been held that day Democratic turnout was going to be completely pathetic. Now that's reverting to a more normal level.
Good reasoning, Tom. You're definitely doing a much better job on this race than SUSA or Rasmussen.
ReplyDeleteI figured that would be your response - enthusiasm.
ReplyDeleteBut I'll still treat this poll suspiciously as I have with the SurveyUSA poll released yesterday.
Also in the other thread, I view any poll where Blacks are more favorable to an unknown like Deeds than Wilder as an outlier.
TPM released some interesting data on PARTY ID:
ReplyDeleteSurveyUSA: McDonnell 55%, Deeds 41%, with a ±4% margin of error. The likely electorate is Republicans 32%, Democrats 32%, Independents 29%, and the respondents voted for John McCain in 2008 by 51%-44%.
• Rasmussen: McDonnell 51%, Deeds 42%, with a ±4.5% margin of error. The makeup of the likely electorate here is Democrats 43%, Republicans 38%, Independents 19%. Respondents were not asked about their 2008 presidential vote.
• Public Policy Polling (D): McDonnell 48%, Deeds 43%. The likely electorate is Democrats 37%, Republicans 29%, Independents 34%, and respondents voted for Barack Obama by 48%-45%.
http://tpmdc.talkingpointsmemo.com/2009/09/va-gov-race-all-about-likely-voters----and-gotv.php
Interesting that your numbers are going to be worse for Obama than Rasmussen even though Rasmussen shows a greater spread for McDonnell than you do.
ReplyDeleteSomewhat counterintuitive.
But SurveyUSA will almost certainly show horrific numbers for Obama.
You think Obama would win Virginia today if his disapproval is at 48-50%? He wasn't winning Arizona despite his approval being 47/47 according to your poll (the undecided were also disproportionately represented by McCain voters in the various matchups).
I began watching the party ID splits with your Sept. 1 poll, to figure why a 14-point McDonnell lead fell to 7-points. If the survey was more Democratic than the August pool, then it makes sense that Deeds was getting closer.
ReplyDeleteMaybe it wasn't the thesis, but just having more Democrats in the pool.
Looking forward to your next Virginia poll and the party ID split then.
Tom, you were supposed to release Obama's Virginia approval rating today. Is it still coming out?
ReplyDeleteI'm wondering since Rasmussen has it at 52% (the same poll showing Deeds trailing 51-42).