Nov 05 2006
The USA Today/Gallup poll just out gives the Reps a perfect trifecta in the polls as, again, a huge lead is cut. Today the Dems only enjoy a meager 7 percentage point lead over the Reps. A month ago it was a 23 point lead, and two weeks ago it was 13 points. Updating this in a few minutes.
A Democratic advantage of 23 percentage points a month ago and 13 points two weeks ago is now down to 7.
Gallup and USA Today are spinning like whirling dervishes to claim this still indicated dems will take the house. We know the senate is out of their grasp. Truth is, their supporters are very fragile. They feel like this election is slipping away they will stay home in a temper tantrum fit. In fact, Gallup admits as well they have had to adjust their turnout models and dems appear less likely to vote than before and the GOP supposeldy surges (or is finally talking to pollsters)
What’s shifted is the determination of Republicans to vote. The Democratic advantage among registered voters was 11 points, but Republican voters were more likely to be judged as sure to go to the polls, making the edge among likely voters smaller.
A month ago, the Democratic margin among registered and likely voters was identical.
Granted, in 2004 the margin had shrunk to where Dems where only up by 1 point by election day. In 2002 Gallup had Dems up by 3 points. In both years the Dems lost seats. Clearly, though, there is no wave. The average performance in the 6th year of a President’s term is for his party to lose 6 Senate seats and something like 24 House seats. The Dems will be performing below average this year – again.
Update: One last observation about the Gallup sample pool. In the PEW poll out today we saw GOTV efforts hitting 54%-64% of their sample voters. In the Gallup poll their sample is only reporting GOTV contacts by 15% by one party, and 27% by both. That is a GOTV contact rate of 42%, well below the Pew numbers. Seems these folks are not getting the same random samples! The GOTV contact numbers should be very close if the samples were both random measurements of the same population.