Here’s an interesting article I found from August about party affiliation and polling:
Pew’s latest poll to that point which showed a 10 point Obama lead, the party affiliation was D+13. The problem is that actual 2008 turnout was only D+7. Pew defends this result well, IMO, but it is interesting.
But Democrats haven’t had a double digit advantage in a very long time. Is support for the Democratic Party really rising that much, or are a lot of pollsters not named Rasmussen or Gallup making a mistake?
It says that Pew does adjust for race. Could the higher Democrat numbers be a function of assuming the same racial makeup of the electorate as 2008? If so, that’s quite an assumption.
The question becomes, should pollsters control for party ID, the way they control for things like race, sex, and ethnicity?
The problem, of course, is that party ID is a bit more flexible than race and sex. And there’s no indication that doing so systematically would get you more accurate poll results.
I remember back in late 2004, a lot of us in lefty blogland would look at the party breakdown in polls and say, ‘if these polls reflected what we knew about party ID more correctly, they’d show that Kerry was ahead.’ Kerry lost the election, of course, and there’s really no indication that he was ‘really’ ahead at some point in the final weeks before the election, and then slipped behind: he was clearly behind from the GOP convention on.
So I feel like I’ve had this conversation before, and I think that at best, if you’ve got a poll that’s a clear outlier, party ID is one of the things you look at to see why it might’ve come up with a different result than everyone else is getting.
I don’t think so. The electorate has been steadily getting less white for a long time now; I think the most you can say about the effect of Obama’s 2008 Presidential run on turnout is that he might’ve achieved 2012’s expected race/ethnic breakdowns one cycle earlier that would have otherwise happened.
So I think you’d expect roughly the same racial breakdowns as in 2008; you just wouldn’t expect the electorate to be less white than in 2008 because you got pretty much all of the expected 2004-2012 change between 2004 and 2008.
And I don’t see any signs at all that blacks will turn out in noticeably reduced numbers this time. They’re still quite proud of their guy.
Well, we do have a discrepancy in the polling. The major tracking polls are showing a very different race from everyone else. You figure either the major trackers have to converge with the other polls by November or else someone is going to have egg on their face.
Now that’s a fair statement. The Gallup and Rasmussen trackers really are showing a different race from what the other polls are showing. One group is right, and the other is wrong, and we can’t be sure which.
Nate Silver talks a bit about the discrepancy between polls with live interviewers that include cell phones in their sample, and robo-polls (which aren’t allowed to call cell phones). I don’t know which side of that divide the trackers are on.
Is it too early to say Romney has lost? Yes, it ain’t over til it’s over, and a major scandal could blow almost any president out of the water. Who knows what Romney’s superPAC might have in store? And one professor, to some critical acclaim, is predicting that Obama will win the popular vote, but since Romney’s strategy caters to the electoral game, he’ll win the electoral vote (his book keeps popping up in my Amazon searches).
But I think it’s pretty much over, the cat is essentially in the bag. Allan Lichtman made his prediction back in 2010, and Romney’s only hurt himself since then…and boy, has he hurt himself. You have Rich Lowry of National Review slicing up the candidate for his worldview near the end of his general campaign? The GOP’s nominee is in a tight spot.
A tracking poll is a continuing, ongoing poll - they do approximately the same number of interviews every day until the day before the election, and today’s poll result is the average of the last N days (I think Rasmussen’s N=3, and Gallup’s N=7) of interviews.
I’d put it this way: it’s as close to over as it can be, given the small absolute size of the lead.
I personally think the fact that there’s only 4% difference between the candidates in the RCP average understates the gap, given that ~46% of the electorate is pretty much locked in on each side. A 4% lead is pretty big, and a 6% lead would be dominating.
You know who was the last Presidential candidate to win the popular vote by more than 8.5%? Ronald Reagan, back in 1984.
I am just surprised at what a stinker the Romney campaign is turning out to be. I thought the GOP was supposed to be this lock-step Borg kind of entity with more money than God. What happened? They have been getting pummeled all week and seemingly have nothing to fight back with. They just fall farther behind every time Romney opens his mouth. I guess there is just no hiding that GOP ideas blow chunks.
I wasn’t rooting for him anyway, but honestly I am a little surprised that this is the best they can do.
538 gives this particular scenario a 2.5% chance. The reverse scenario (Romney wins the popular vote, but Obama wins the electoral vote) is at 2.2%.
Also, why give your book a title that will make it unsellable and an object of ridicule the day after the election? Apparently it’s a book about the Electoral College – at least make it usable as a college text.
Tracking polls both say tie. As a Romney supporter, I’ll put my hope in that. IT’s not a totally unreasonable belief. They poll every day, they are two of the largest polling organizations and they use the largest samples. I’m crossing my fingers that they are right and the other pollsters are wrong.
Both polls have been analyzed to within an inch of their lives by people who make a living in the scientific discipline of statistics, and both have been found to have either systematic sampling biases or other methodological faults that they refuse to correct for reasons they will not publicly explain. The means of the other major polls are well-centered on actual results, while theirs are consistently off, and in the same direction. That’s how.
There are some persons, naturally, who prefer to believe whoever tells them what they most *want *to believe. The rest of us are tragically condemned to reside forever in the reality-based community instead.
You miss the point being made. Team Romney has failed to deliver a vision that inspires … anyone. Maybe Obama will blow it and lose, but a Romney electoral victory now depends on an Obama fumble. And Obama has learned the hard way that the tape is always running and is unlikely to speak without filter fully engaged, is unlikely to stray from the game plan. He’ll run down the clock being Presidential. Team Obama has succeeded in more than neutralizing the only angles Team Romney has. Obama is trusted more with the economy and Romney mentioning Bain has been, rightly or wrongly, associated with negative features not positive.
Rasmussen is pretty dismissable as unreliable other than for having a GOP bias.
Gallup is a bigger question. My thread asking why they are different than the other reliable houses has not borne any fruit from the tree of knowledge. Maybe they do have it right and every other reliable house has it wrong, repetitively. Could be. Not impossible. Elvis if you could link me to some analysis of Gallup’s flaws I’d appreciate it.
If so, if the national numbers actually are tied, then you still have to contend with increasingly grim (from Team Romney POV) battleground state numbers. No not out of reach, but the path to winning enough of them to get the needed EVs gets harder and harder to see.
Gallup also puts Romney close in the battleground states. They have a clearly different take on the election than everyone else but Rasmussen.
And Rasmussen, whatever their methodological problems, is not dismissable. they tend to be fairly reliable. They aren’t in Zogby territory, that’s for sure.