Correction: This post was based, largely, on an inaccurate release from Gallup. I've added a correction and a reconsideration of these issues in a follow-up. Thanks to readers for pointing out my mistake.

The latest bit of hope bouncing around the conservative blogosphere this afternoon seizes upon a Gallup release to argue that, all indications to the contrary, the early voting hasn't actually been good news for Democrats. Through Wednesday's polling, Gallup found that 11% of "registered voters who plan to vote" had already cast their ballots, and that early voters as a whole are "about evenly split politically." Drudge and others on the right were quick to trumpet that headline; it offered a ray of hope, compared with other reports of Democrats dominating the early turnout. 

In fact, the Gallup data is a little more complicated than the headline suggests. Among the 11% who have actually voted early, Gallup found that "early voting generally reflects the same Obama lead evident in the overall sample." Since this is Gallup, it's worth asking which sample, but that would seem to indicate a lead consistent with the range of 4-6% on the dates provided. The headline about an 'even split' was generated by the larger sample of 30% of voters who have either already voted or plan to vote early; Gallup found them tilting just 31-29% to Obama. 

That's a puzzling result (30% of these folks are undecided?) particularly since the larger sample is far more demographically friendly to Obama than the 11% who have cast their ballots. Among that 11%, past voting data combine with Gallup's numbers to indicate that roughly 55% of those who have already voted are over the age of 55, and that just 14% of them are under 35 - proportions badly out of balance with the electorate as a whole. Gallup's own demographic breakdowns drive home the point that this should have produced a very tight result. The larger sample, on the other hand, which more accurately reflects the age breakdown of the electorate, should have reflected Obama's overall lead. It's possible that the relatively small sample sizes are producing whacky results. My own hunch is that Gallup isn't sharing enough numbers; it's looking at age, and should be reporting on race as well. This looks like an artifact of the surge among black voters, who tend to skew older, but still support Obama. 

The key fact to bear in mind, though, is that early voters in the last two cycles have voted for the GOP, 60-40%. There's a reason for that, and it's not mostly about the GOP's turnout operations or the enthusiasm of its base, although those factors certainly played some role. Half of voters in Western states and more than a third of those in Southern states plan to vote early this year - but just 9% of those in the Northeast and 20% of those in the Midwest. On the whole, early voting is more popular - and more widely available - in red states than in blue ones. So the Gallup numbers are shocking in their suggestion that Obama is currently enjoying a significant lead, and that even once the initial surge of black voters recedes he'll slip back only into a virtual tie. That's a 20-point swing from recent cycles. If Obama is actually running even or marginally ahead among the 30% of voters who will vote before November 4, he should win going away.

[ed. note: Gallup is the most favorable of the four surveys that have released information on early voting from the perspective of the GOP. Rasmussen reports numbers that suggest Obama is doing slightly better among early voters than among the electorate as a whole - perhaps a 9-10 point lead. Zogby has had Obama up 20-27% among these voters in recent days, a number that probably says more about Zogby's dubious methodology - a 120 voter sample? - than about the election. And the extremely reputable Pew poll puts the race among early voters at 58-34%. This is an unsettling range of numbers - from Obama up 2% to Obama up 24% - and hopefully as more results roll in over the coming days, we'll gain greater clarity. But the bottom line is that none of these numbers preserve any hope for John McCain - Gallup and the others are just arguing about the margin of his defeat.]

Update: Re-reading this post, I realized that I'd buried the lede. This race, as Nate Silver recently explained, has been marked by an epistemological debate among pollsters regarding likely voter screens. For the first time, we have meaningful data that can serve as a benchmark and help to resolve that debate. The ballot-request tallies from the states that release them clearly indicate a substantial Democratic edge - and that's in states that voted for Bush in 2004. We don't know whom these voters have chosen, but it's safe to assume that, on average, the early voting in these states is tilting strongly to Obama. If Gallup - and to a lesser extent, Rasmussen - are producing early voter numbers that don't reflect this, then that's a very good reason to question their overall demographic modeling, and in Gallup's case, their likely voter model. And though Pew's numbers are tough to swallow, they have the virtue of according with the external indicators. Again, it'll be interesting to see what else we can read from these tea leaves as more numbers come in.