When unfavorable polls for Romney come out the conservative blogs often note the party breakdown of the sample. It’s happening often enough that it’s possible some pollsters are oversampling Democrats. Unintentionally. I believe they don’t adjust their data for that, and it would make sense that they wouldn’t. Which is why although I want to believe the non-tracking polls are wrong, I’m not hopeful.
The only way this ends good for Romney is if he does well in the debates or Europe implodes or he stumbles upon an amazing message in the last two months. I wouldn’t say he’s toast yet, but he’s at a low point right now.
Update: Gallup reports that Democrats are now more enthusiastic than Republicans, which would explain why Democrats appear to be oversampled in the polls. There’s just more of them:
This may be a methodological error which’d be difficult to account for. For instance, it could be possible that democrats are more politically engaged or willing to talk to a stranger, or that Republicans work longer hours (polls show Republicans work ~3 hours longer on average, cite available on request).
If the first then it may be an accurate correlate for more likely voter.
If the second then we’d need some reason to believe that such is more so this election cycle than in other ones.
Interestingly a recent 538 discussed how Democratic leaners may be being undersampled in polls that restrict samples to landlines.
Pew’s take has been that the increased Democratic party ID is real and should not be corrected for, even though adjusting for likely voter pushes it back the other way a bit.
There’s something weird about Nate’s model. He’s got two forecasts, one for Nov. 6 (naturally), and a ‘now-cast’ that is basically, ‘if the election were held today rather than Nov. 6.’
Obama’s chances keep going up in the ‘now-cast’ and currently stand at 94.4%. (They were at 91.8% a week ago.) But while they’ve been going up over the past week in the now-cast, they’ve been dropping from 80.8% to 75.2% in the Nov. 6 forecast during the same week.
There are two reasons this doesn’t make sense to me.
Over the past week, Obama’s convention bounce should have been fading somewhat. If his now-cast changes are even holding steady, let alone going up, his Nov. 6 chances should be going up.
Each week, there is less time for Romney or circumstances to shake the Etch-A-Sketch. If Obama’s now-cast chances are holding steady, let alone going up, then his Nov. 6 chances should be going up.
For the most part, I think Nate’s got a much better handle on things than any other forecaster. But I’m running into a problem with the logic here.
As I understand it - and someone can correct me if I’m wrong - Nate thinks the positive convention numbers for Obama are still in play. He believes the bouces lasts longer than most people do, so his model corrects this by being skeptical about polls that give a boost for the president. As time goes on this skeptical stance will fade and the numbers will look more like the Now-cast.
The problem here is that being skeptical about Obama’s numbers in general means that he’s being skeptical about numbers that might be pro-Obama for reasons that have nothing to do with the convention bounce right now.
That is correct. The convention bounce correction actually holds pretty steady at something like 2% until it falls off in early October, as it is used in the Nov 6 forecast. The “Now-Cast” does not use those correction factors.
There is one other factor as well in the divergence between the “Now-Cast” and the Nov 6 forecast. Nate has historical data that shows regression towards the mean for the leading candidate until pretty late in the campaign. That is used in regressing the Nov 6 forecast down a bit - so a 3-point lead now might be 2-points or less in November just via regression.
True enough. And he’s always been a bit skeptical of his own bounce compensation scheme. He turned it off in 2008, but it actually would have helped smooth things out quite a bit. So he turned it back on this cycle. In particular he has some concerns because of the compressed nature of the conventions - normally they are a bit more spaced out.
With all of the secrets (Bain, his tax returns) and recent gaffes (Olympics comments to Brits, comments on Middle East, the 47%, etc.) plus his policies against women’s rights still not having all that much affect on the core of Romney voters - I think it is safe to say there is still only the slimmest margin of voters who are in flux at this time. Republicans might not love, or even like, Romney, but they hate Obama more. I can understand that - back in the day, I too would have voted for anybody but Bush.
I don’t even think the upcoming debates are going to matter - Romney could claim Israel is in South America, admit he has $9 trillion in off-shore accounts he has not paid tax on, and give us a peek at his magic underwear and he still wouldn’t lose a single Republican vote.
Thus I think the polls are going to stay relatively the same, with little bumps up and down right up to and including election day. The good news is that the polls continually show Obama ahead (albeit, sometimes by a narrow margin) in all the places he needs to win. The only real variable is voter turnout - and again, I think the Dems have learned to get out the vote better in recent years and should not have a problem doing so again.
I agree with the principle that there should be some ‘bounce skepticism’ factored in, and trust Nate’s judgment on how big the bounce was this time, and long it typically lasts before fading out. But that doesn’t change the fact that the ‘bounce skepticism’ that applied last week, should apply just a little bit less this week.
So let’s say that a week ago, his model gave Obama a 3% residual bounce from the convention, and now that’s down to 2.9%. If his standing in the polls has increased, so that he’d have a better chance of winning the election if it were held today than if it had been held a week ago, despite a slight fading of his convention bounce, then his chances for November 6 ought to be increasing rather than decreasing. So I still don’t get it, despite buying into Nate’s explicit assumptions.
Nate’s got a new post up today, talking about how the polls yesterday were kind of insane but on a second glance, things might be starting to look pretty grim for Mr. Romney.
He thinks if the election were held today, Obama would have a 95% chance of winning. He also thinks it’s possible that Obama’s bounce might still be in effect.
I agree that it’s too early to say that Romney’s forked, but it’s maybe not too early to start looking for the silver polish.
There’s a HufPo piece about Romney’s SuperPAC running low on money. Apparently they came out of August with only $6 million on hand and their takings in August were lower than they have been.
Could this mean an end of the commercials and telephone push-polls?
Wonder if all the big business whizzes will look at their ROI and decide that they’ve lost enough money on this one and cut him loose? It sure would be nice to go into October with those stupid “waterproof your basement” ads back on TV.
Romney is not going to be elected president. He’s a disaster of a candidate.
Seriously, you’re not in the dugout during a no-hitter. You’re not going to jinx it if you say what’s plainly obvious. In fact, I think the Democrats need to seriously consider shifting money away from the presidential race and start concentrating on the Congressional campaigns.
