The Non-Partisan Polling and Election Prediction Thread

So, we've got already got other threads for arguing about the race, who should win, and which of the candidates, in fact, have horns. So let's have a thread that's just about polls and electoral analysis, without value judgments.

You can even predict the state outcomes and share the link to your map in here. It's kinda fun:

http://www.270towin.com/2012_election_predictions.php

I'll probably toss mine up in a week or two.

So, lots of polls lately, as we get closer. You're all probably aware of the shifts in the national polls after the DNC and then after the debate. National polls seem to have settled into maybe a slight Romney lead, but they're pretty much all within the margin of error, so they're far from dispositive.

More pertinent are the state polls. And there's more drama there, because state polls can be big news even when a state's lead doesn't "flip," because the margin of each lead or deficit drives the decisions about where money and time are spent. You don't even have to win them to justify the focus; sometimes it's enough to force your opponent to spend time and money there, playing "defense" in a state they were already counting on. So, the margins matter.

I bring this up because the latest shift in the race has brought a lot of changing state polls, and they present both campaigns with a challenge in figuring out which ones represent real opportunities, and which are mirages or outliers. For example:

Pennsylvania
Pennsylvania is my home state, and has the fifth-most electoral votes in the nation (20). It's a huge prize, and it's voted for the Democratic candidate in each Presidential election since 1988, when Bush Sr. won it narrowly. Bush the Younger almost flipped it in 2004, but fell just short. Because of its demographics, aging population, and huge electoral vote count, it's often tantalized Republican candidates, though they've found it difficult to get over the hump here.

Late last week, a Siena poll showed Obama up just 3 here. Huge deal, if true, but again, Pennsylvania has eluded Republicans for awhile, and the poll looked like an outlier. It looked a little less so on Monday, when a Susquehanna poll gave Obama a 2 point lead.

On the flip side, two other polls later in the week gave Obama leads of 8 and 5, respectively, which are numbers more in keeping with the conventional wisdom pre-debate. So which is correct? That's what both campaigns have to decide.

Mirage?
Me, I tend to think Obama's probably got a 4-5 point lead here right now. That would make it technically winnable for Romney, but it's one of those situations where it's hard to imagine him winning it and needing it to win. That is, in any situation where his standing is strong enough in general that he's genuinely duking it out here, he's probably doing well enough in states like Ohio to win without Pennsylvania. But it's hard to say.

If we actually get a few more low single digit margin polls, Romney would be crazy not to make a push here. But my guess is we'll get a few more Obama +5-ish polls here soon and neither campaign will expend much in the way of resources.

I attached my map

As far as my state (CO) we've been a long time swing state, and I think we'll repeat our 2008 trend. It'll be close though.

Attachments

 

__________________

Agree on Colorado being important and close. If Romney loses Ohio, Colorado factors very heavily into the other combinations available to him.

Be stunned if he loses North Carolina, though, without a huge shift nationally, too. He hasn't shown too many huge leads there, but he's shown very consistent ones.

By the by, here's an update on ad spending in swing states. Publicly candidates will say lots of states are "in play," but if you want to see what they're really worried or hopeful about, you just need to look at where their money and human resources are deployed.

Romney's ramping up his ad buy in some states that make obvious sense; Florida, Iowa, Virginia, and Colorado. Interestingly, a pro-Romney Super PAC is airing some ads in Michigan, too. But it's probably telling that the Romney campaign proper isn't increasing spending there. They (rightly, in my mind) probably don't see it as worthwhile. Maybe if he'd gotten this mix of polls a month earlier.

Anyway, it'll be interesting to see what effect these things have. The campaigns have taken very different approaches here. Obama spent a ton of money up-front trying to "define" Romney. Romney saved more of his money for a late blitz, which is starting to get underway now. So Romney should have more to spend, but he might be spending it on an electorate less willing to pay heed to it. Hard to say. If nothing else, the increase in spending should give us a pretty clear picture of what parts of the map Romney's going to contest, though. So far, it's about the states you'd expect.

I can tell you that the west side of the state, i.e. Grand Rapids, and Detroit are blue and everyone else is red. So, perhaps it isn't worthwhile.

Sure it helps with that but I hate hearing about this stuff. All that money, it's a staggering amount merely for campaigning, could be much better utilized in almost any way.

I dig that. But keep in mind it goes to the people who make the ads (who I imagine are often creative types), and the people at the stations who air them and all that. To put it to better use would mean some of the people working on the ads or at the stations would be working somewhere else. Which is not to say that wouldn't be more efficient; it would. If we could find a better way to pick our candidates, that is.

Oh, forgot to mention that Obama has been spending a fair bit in Florida. Very interesting, because most of the electoral projection models have Romney looking good there; solid (if small) leads, demographic advantages, etc. But it might be one of those make-him-defend-it feints, because he needs to win it. Still, odd place for a feint, as its a large state and ad buys there are very expensive.

I heard in early voting it looks ugly for Rommney in Ohio. I think Florida and Virginia will be extremely close, those two are really toss ups. Same with NH, but that one isn't all to important

Early voting is up in Ohio, but they don't have party registration, so we don't actually know what that means. We can look at the rural/urban split, I suppose, but both are up. The only other thing we can look at are absentee ballot requests, which are way up in counties that have leaned Republican in the past.

Either way, they changed their procedures this year and mailed a ballot to every registered voter, so we're on totally new ground.

If you find the article about the early voting in Ohio, though, please post it. I'm trying to find as much of this stuff as possible.

New slew of polls just came out this afternoon:

• Pennsylvania: Obama +3 (Siena)
• Wisconsin: Obama +2 (PPP)
• Ohio: Romney +1 (ARG)
• Michigan: Obama +3 (FMW)

I believe the Wisconsin poll. I think that's winnable (though Romney's still trailing). Find it hard to think PA's that close, though, as much as I might like it to be. And I find Michigan harder to believe still.

It'll be interesting to see what each campaign does. Safe to say that if Michigan and Pennsylvania have to be defended substantially, that's a huge deal. But we'd need to see several more polls like this before either campaign starts putting their efforts there, I think.

It all probably reflects Obama's performance in the first debate. If he does well in the second debate it will halt Romney's momentum.

If memory seves me, Gore won Wisconsin by a squeaker so it is not a surprise it is close there. Wasn't Pennsylvania close also?

I think the Medicare proposal will keep causing problems for Romney in attracting swing voters. His big problem is if Ohio stays firmly in Obama's camp, it is much harder for him to make up for the loss as he has to make up for it by winning blue states.

__________________
It reminds me of a toilet paper on the trees
- Paula

Your memory is correct. Gore won Winconsin by .22 in 2000, and Bush flipped it by just .38 in 2004. But Obama took it handily in 2008. Pennsylvania, too, though Bush lost it narrowly both times. Obama doesn't necessarily need Wisconsin if he wins Ohio (though Romney might not need Ohio if he takes Winconsin), but pretty much every plausible electoral map Obama has involves Pennsylvania.

Re: swing voters. That actually relates to a pretty huge story throughout this election. The conventional wisdom is that Independents are huge in an election like this, but there are some reasons to doubt that here, which is actually good news for Democrats. In virtually every poll Romney's winning Indys...however, there is reason to believe that more of those Independents are conservatives who don't self-identify that way any more. So winning Independents might not be the bellwether it's been in the past.

Yeah Man...

Not posting every poll I find (way too many), but here are some interesting ones from the last couple of days:

• Iowa: Tied, 48-48 (ARG)
• Virginia: Romney +1 (ARG)
• Colorado: Obama +2 (Gravis)
• North Carolina: Romney +2 (PPP)
• Pennsylvania: Obama +4 (Morning Call/Muhlenberg)

IA, VA, and NC are all about in line with the consensus of polls. The newsworthy numbers here are CO and PA. Most CO polls have shown Romney with a slight edge, so this would be significant if the number is right. Gravis, however, has shown some pretty wacky results (for example, Obama very close in Arizona, which diverges heavily from the other polls there), so who knows. But Colorado is a huge part of any non-Ohio Romney victory strategy, so this bears watching.

The PA poll is part of what I was talking about before. If that's really the margin, then expect to see some spending in PA and/or some campaign visits. It'd be a big deal for Romney if he could pull within the margin of error there and make Obama defend it. Not sure if I buy that it's winnable this late (again, without a broader shift that renders its EVs superfluous), but I do buy that it's clearly not a 7-8 deficit any more, and is probably worth a second look. If we get a few more sub-5-point margins, this is gonna become a thing.

Also, related to this notion: some Romney ads are going up in Michigan, and Jill Biden is visiting Minnesota. Nothing huge on either front yet, but little tricklings of a defensive posture in both. We'd need to see more to draw much in the way of conclusions from it, though.

I just read the newest polls are showing Obama getting back some of what he lost. Everthing is in flux right now so who really knows. But I think the second debate with its town hall setting and Obama more charged up should be a win for him and that should reverse any Romney momentum.

Which polls are those? Whole point of this thread is to get into the details of this stuff, so by all means, share.

Color me skeptical on the practice of predicting the winner of debates before they've even, ya' know, taken place. But I'll point out that most debates are a rough wash in the polls when all's said and done. You need a real thorough clock-cleaning and a clear consensus on the winner to move them much, as we've seen. And it takes a big shift to be certain that that's even what did it, since a point or two is basically just noise.

What it might do is increase party morale, or, as you imply, simply stop Romney's gains. But the polls seem to suggest that those gains have already settled in, so I don't know if there's any bleeding to stop heading into tomorrow night.

The gains are not a lead so Romney needs to build on them or he still loses.

He is roughly dead even in national polls but still clearly behind in state polls.

There's considerable debate about whether or not the state polls will come in line with the national polls, the other way around, or whether or not there's an inherent split. You can read Nate Silver talking about the likelihood of each here. In short: nobody actually knows.

If we assume that there's an inherent split in strength, however, that would leave Romney needing to flip Iowa and Nevada (that's sans Ohio, mind you), where he's down by about a point in each based on the average.

Nevada polls according to this in recent years under estimate Democratic support:

* Nevada (6) - The poll averages still give President Obama the edge here, but it has shrunk in half over the last week, and is now down to 1.6%. Nevada is a state that Republicans thought they could crack in 2012, but it hasn't turned out that way so far. Again, I will point out that the polls in the Silver State have not been good in recent major elections - consider these examples: In 2010, the final poll average forecast a defeat for Sen. Harry Reid by 2.7%. Instead, Reid won by 5.6%, a difference of over 8%. Back in 2008, the polls forecast a 6.5% average win for Barack Obama - instead his margin was 12.5%. In 2004, the average polling edge for President Bush was 6.3%, but he won by only 3.5%. I cannot ignore that kind of error in the polls, so, in my head, I am adding a few percent to Obama's total in each Nevada poll. We'll see if the polls have a better track record in 2012.

This was from someone objectively analyzing the results, not a partisan.

Aye, I'm familiar with it; the theory behind that, if you're at all interested, is that the polls may have underestimated Latino turnout. Silver's written about that, too. Steps have been taken to try to guard against this (the specific how differs from model to model), but we obviously don't know how well that works, or if some of this is just noise from what's ultimately just a few data points.