ELECTIONATE NOVEMBER 4, 2012
-
Read Later
READ LATERAvailable only to subscribers. SUBSCRIBE TODAY
-
Listen
ARTICLE AUDIO
- Font Size
After a day of surveys from an irregular set of pollsters, the big picture remains unchanged but there are plenty of details to talk about.
Let's start with the Susquehanna poll of Pennsylvania, which is sure to catch fire as supposed confirmation of a tight race in the Keystone State. Folks should perk up if other polls start showing a tight race, but this poll just doesn't justify the attention that it is sure to receive. When other pollsters showed Obama leading Pennsylvania by nearly 10 points in September, Susquehanna showed Obama ahead by just 2 points. And after the first presidential debate, Susquehanna showed Romney ahead by 4 points in a private survey provided to the Washington Times. As a result, a tied race in Susquehanna is not unusual, it is not evidence of movement in Romney's direction, and it is insufficient to prove that Romney is ahead.
Next, let's consider the Columbus Dispatch survey in Ohio, which shows Obama ahead by 2 points, 50-48. The Columbus Dispatch poll is consistent with the average of recent polls in the Buckeye State, but it is conducted entirely by mail. The poll has a decent but imperfect track-record, so adding a poll with a unique methodology to the available data on Ohio should add more confidence in an Obama lead than a similar result from another random pollster doing all the same things as the other polls. Although it's unclear whether this is significant, one consequence of a mail poll is that it stretches back a little earlier--to October 24. The other interesting poll of Ohio came from Pulse Opinion Research/Let Freedom Ring (R). After showing Romney ahead by 1 point in Ohio in a poll released on October 22nd, POR/LOR now shows Obama up 2. The only poll showing Romney ahead in Ohio is from Wenzel Strategies, a Republican firm.
The University of New Hampshire found a tied race in their own state, which represents a substantial decline from their last poll showing Obama up 9. The UNH/WMUR surveys have been a little volatile this fall (ranging from Obama+15 to a tie since the DNC), but the big picture is that New Hampshire is probably the closest Kerry state. On average, Obama leads by just 1.6 points in the Granite State, but the polls range from a Rasmussen poll showing Romney ahead by 2 to a New England College poll showing Obama ahead by 6. Given the recent polls, the state's demographics, Romney's regional advantage, and its large number of independent voters, New Hampshire seems like one of Romney's best pick-up opportunities. The problem for Boston is that New Hampshire isn't especially important to the electoral math, at least so long as Obama's easiest route to victory goes through Ohio. However, there are two somewhat plausible in which New Hampshire can make the difference: a Romney win in Wisconsin and New Hampshire would overcome a defeat in Ohio, provided that Romney swept Colorado, Virginia, and Florida; a Romney win in New Hampshire can block Obama's so-called "western" path to victory through Colorado, Nevada, and Iowa.
Whether New Hampshire or Colorado is the weakest link in the "western route" is tough to say. But there's not much question that Obama is better positioned in Iowa than those two states. The vaunted Seltzer/Des Moines Register poll showed Obama ahead by 5 points, 47-42. The Seltzer poll is an excellent survey, but Seltzer's work is treated with such high regard that you get the sense people would take 1:1 odds on Seltzer results versus the field. While Seltzer has excellent achievements (like the '08 Iowa Caucus and Indiana), it's worth recalling that Seltzer showed Kerry winning Iowa by 3 and Obama winning by 16 in the general election. Nonetheless, the Seltzer poll is consistent with the balance of polling in Iowa, which now shows Obama ahead 48.4 to 45.3.
7 comments
Are you getting into the media "Let's make this a horserace" mode too, Nate? If you look at 538, where you can see pretty much every poll result back to last year, there have been exactly two red results in that time. Both were Susquehanna. There is a sea of blue results with just these two red. Whatever Susquehanna is doing, it is not normal sampling. Given that history, the divergence of Susquehanna by something like 6-8 points, the fact that Susquehanna now shows a tie with its methodology has to be taken as the same as anyone else with 5 or more for Obama. Far from showing a tie, Susquehanna confirms the present state of opinion in Pennsylvania. The only thing left for Romney is a big enthusiasm gap. It would have to be pretty darned big given that all the polls have been LV polls for a while. As Nate Silver wrote yesterday, the only way Romney wins is if the polls have been systematically biased against them. As I suggested would happen in the piece here about Silver, he has now ticked up to 85% probability for Obama. My guess is that he will tick up to 90% by Tuesday, leaving 10% as the risk that the polls are biased or that the Obama ground game falls down. The only surprise to me is that Virginia appears more likely to go blue than Colorado. But AaronW, intrepid poster and a son of Virginia insists that VA is going for Obama and there is no similar evidence for Colorado. Given that NV and IA (and PA and WI) are solid, the race is exactly as I described it at least a month ago: Romney has to sweep VA, CO, and OH to win. The better bet at this moment is that he loses them all. Given Silver's odds for these three, treating them as statistically independent there is a 40% chance of an Obama sweep of the three. The chance of a Romney sweep computed the same way? 6/10 of a percent. This is why Wang at Princeton gives Obama's overall probability as 99%. Silver discounts for some sort of correlated movement to Romney in the critical states and even says out loud that the contingency he provides for is the risk that the polls have been systematically biased. I put Obama at 90%, in between Silver and Wang (eminently fair and balanced of me). But today, even that does not seem certain enough to quell my anxiety.
- roidubouloi
November 4, 2012 at 8:45am
The only problem I foresee in PA is the knock-on effects of the storm, like loss of electricity, that reduce turnout in the eastern parts relative to the rest of the state. No other reason suggests PA will be a tossup.
- chaitless
November 4, 2012 at 9:42am
"After showing Romney ahead by 1 point in Ohio in a poll released on October 22nd, POR/LOR now shows Romney up 2." I think you mean Obama, judging by the table.
- rmutt
November 4, 2012 at 10:01am
Roid: the polling numbers are fine, but they really do not tell you anything about the ground game. It is there that, I think, the Community Organizer will come through.
- icarus-r
November 4, 2012 at 10:54am
chaitless according to PPL There are currently 22,804 customers out of power in 29 counties. In the Philly area according to Peco there are around 85,000 customers still without. But stores have been open as well as gas stations, water was never a problem. People seem to be coping well enough. All the schools are going to be open on Monday so all polling stations will be open on Tuesday. As most of the people without power are in outlying areas I have no idea why Romney is coming here and not in western Pa.
- blackton
November 4, 2012 at 10:55am
I incline to think so too, icarus. And there is upside for Obama relative to the polls. Except that no one cares about that upside. It is the unknown unknowns as well as the turnout risk that keep Silver from joining Wang at 99% ("taking no prisoners" as Robert Powell aptly described it). One cannot completely discount relatively low probability events which is why 90% probability of Obama being re-elected suddenly doesn't feel so high to me.
- roidubouloi
November 4, 2012 at 11:33am
I hope you guys are right. I have so many more gray hairs already.
- Sophia
November 4, 2012 at 12:35pm