The big picture remains that Obama leads by around four points, with a similar edge across the critical battleground states.


The poll most likely to scream “headline” is Gallup, which is so far out of line from the other registered voter surveys that I don’t even know what to say about it. Obama leads by about 7 points among registered voters, and, no, the methodological criticisms you’ve heard don’t explain a gap of that magnitude. As Harry Enten of the Guardian (@ForecasterEnten) tweeted today, we would probably blow this off as a clear outlier if it was named something other than Gallup. But while they have a long history, they haven’t exactly been the most accurate survey in recent years (showing a tied race in 2004 and an 11 point Obama lead in 2008), so there’s just not much cause to assume they’re better than the consensus of other polls, at least that I'm aware of. And Gallup’s case for a tight race was undermined by their traditional partner, Rasmussen, which showed Obama building a 3 point lead once leaners were included. It’s hard to see how Obama could lead by 3 points in a poll weighted to a Republican electorate, but it’s clear that Gallup is completely alone in showing a tied race (at least until Rasmussen returns to showing Romney up 3, or whatever). At the moment, Gallup stands entirely alone. The other trackers, let alone the less regular state and national polls, all show a clear Obama lead at or above post-DNC levels.

You might notice that I’ve included the RAND American Life Panel for the first time. As I alluded to, it’s an interesting methodology that I’ve enjoyed following, but it’s not totally analogous to a traditional poll? Why? Because it asks voters to estimate the probability that they’ll vote for a given candidate as well as the probability that they’ll actually vote, while a traditional poll just asks who you’ll vote for. From a methodological/statistical perspective, the probabilistic model is far more satisfying to my ears, but it’s tough to compare directly with a traditional poll when it’s not clear whether any differences are due to the differences in the firmness of support, likelihood of turning out, or just sampling. Nonetheless, I’ve become convinced that the poll is too valuable to keep to my own eyes, so here it is so you can take a look at it too. For what it’s worth, they show Obama building a 5 point lead—his largest lead of the campaign.

While the national picture was largely unchanged, there were many polls released in relatively under-surveyed states in the West and Midwest that help patch-up a few holes in the post-convention map. On balance, the news was good for Obama.

Start with Iowa, a wildly under-polled state where three surveys provided our first hints at the post-DNC landscape. Unlike the actual landscape of Iowa, the polls provided pretty uneven results, with Rasmussen showing Romney down 3 and YouGov/Economist and NBC/Marist showing Obama up by larger 5 and 8 point margins. In the post-DNC period, Rasmussen’s state polls have consistently diverged from the other surveys, so it seems fair to assume that other surveys would be more likely to follow in NBC/WSJ/Marist or YouGov/Economists footsteps if they ever got around to polling the state.

For most of the year, Obama looked relatively weak in Iowa, which wasn’t surprising given Obama’s broader weakness with white working class voters. If Iowa has, in fact, moved clearly into Obama’s favor, that’s a big blow to Romney’s electoral map. In most Electoral College gaming, I’ve generally assumed Iowa would go to Romney in a close election, given the demographics, the GOP-registration advantage, and a higher percentage of polls showing Romney ahead than most other places. But if you start constructing electoral maps with Iowa in Obama’s column (and Wisconsin, for that matter, where today’s polls seem to confirm a modest Obama), then Romney’s done if he loses Ohio and only one extra state would allow Obama to win with Virginia. And it’s worth noting that when the Iowa polls are coupled with two new surveys showing Obama ahead by 5 points in Colorado, suddenly yesterday's post about Romney’s ability to block Obama’s so-called “Western” strategy begins to look a little premature.

The good news for Romney is that Wisconsin, Iowa, and Colorado don’t appear to be moving further toward Obama than the nation as a whole, which means they’d be in play in the event of a close election. The bad news for Romney is the number 50, as well as the fact that 338 electoral votes support Obama as much or more than the national average, which is either a sign that Obama will have good odds in a close race, or a sign that the national average is underestimating the president’s lead.

If there’s any state where the polls look pretty good for Romney, it’s Nevada; a state where demographics and electoral history would tend to portend well for the president, but where Obama appears to only hold a modest lead. That said,  if Romney had to choose a state where he could receive some good poll numbers, it probably wouldn’t be the state where polls have systemically underestimated Democratic Senate and Presidential candidates for the last eight years. And despite today's relatively good news for the Republican nominee, Romney has trailed in every Nevada poll conducted in the year 2012.

Battleground state polls in Michigan, Pennsylvania, Florida, Ohio, and Virginia were consistent with other post-convention polls. A poll from High Point University showed Obama leading by 3 points in North Carolina, joining PPP in finding a slight Obama lead in the Tar Heel state. That finding was contradicted by YouGov/Economist, which found Obama trailing by 2 points among likely voters, but tied among registered voters. On balance, North Carolina looks quite close but perhaps tilting slightly toward Romney, as it has for much of the year.

I’ve touched on a few of the YouGov/Economist polls, but it’s worth considering them collectively. Their national survey shows Obama’s up 5 nationally and today’s battleground polls look quite consistent with a 5 point Obama lead. It’s also worth considering that in the event of a tied YouGov/Economist election and a uniform swing across the electoral map, Obama would lose the Electoral College as Romney pulls out narrow wins in Virginia and Ohio, as well as something of an upset win in Wisconsin. I wouldn't take this exact scenario too seriously, given that it's all from one poll. But it still illustrates that the case for a big Obama advantage in the Electoral College isn't as clear as the breadth of Obama's electoral opportunities might seem to suggest.

At this point, it’s a crime that the YouGov/Economist polls aren’t included in the RCP average: plenty of polls rely on online surveys to supplement their live interviews; the YouGov/Economist poll had a good record in 2010; their numbers have been consistent so far in 2012; and they’re run by capable individuals.