Ted Frank: Is 538 Wrong?

There have been many silly posts in the right-leaning blogosphere attacking Nate Silver at the 538 blog for his perceived pro-Obama tendencies. In fact, Silver created his model earlier this year, and has stuck with it as the campaign has progressed. So unless he’s lying about the results of his model, there’s no reason to accuse him of bias.

This, of course, leads to the question of whether his model is correct. In a nuanced post, Ted Frank sees one potential flaw. Conventional wisdom suggests that undecideds tend to break against the incumbent party, and that Republicans tend to outperform the polling data. Silver thinks that’s largely incorrect, but a good part of the reason that he does so is that Al Gore both won the undecideds and vastly outperformed his polls in 2000. As Frank points out, however, 2000 may have been an anomaly because it was the only election in recent memory that featured a last-minute, important “November surprise”–the revelation of Bush’s drunk driving conviction. If you take 2000′s results out of the picture Obama’s outlook looks pretty bad; an incumbent who is consistently polling less than 50% close to the election, as Obama is, is likely to lose.

It’s an intriguing point, and Frank provides the necessary caveats. I’ll add one more: Blog 538′s predictions haven’t been much out of line with those of betting markets, professional and otherwise. (Of course, there is also a chicken and egg problem here, to the extent that bettors think that Silver has the best model).

Meanwhile, I’m still intrigued by the debate Monday night, which I watched despite my better judgment. I thought Obama “won” the debate by seeming more confident while Romney often stuttered and sputtered. But I also thought that Obama adopted the aggressive posture of an underdog trying to make up ground, while Romney seemed to be playing it safe, the posture of someone trying to sit on a lead and run out the clock. It made me wonder whether the campaigns, at least, perceive that the dynamics of the race favor Romney.

As an aside, I’m still suspicious of pundits using polls that are within the margin of error as showing that one or the other candidate is in the lead, either nationally or in a given state. After reviewing the responses I received to my previous post on margins of error, I’m inclined to think that it’s a mistake to aggregate say, three polls from three different pollsters showing Romney with 1 to 3 point lead in Florida, but well within the margin of error, and conclude that Romney is leading on that basis. Silver, among others, clearly disagrees.

UPDATE: Here’s a piece by Mark Halperin discussing the Obama campaign’s (at least public) confidence in victory. Assuming this confidence is sincere, I’d love to ask the officials Halperin interviewed how one squares this performance with Obama’s clearly intentional demeanor in the final debate. I’m not saying it was the demeanor of a loser, but it surely wasn’t the demeanor of a candidate whose aides are telling him that he has the election sewn up. Interestingly, the Obamaites case for confidence seems to be that the demographic makeup of the 2012 electorate will mirror the unusually Democratic-leaning electorate of 2008, especially in the swing states. That’s quite a counterintuitive notion to base one’s campaign on, which is not to say it’s wrong. But there is one significant risk–if Obama’s ground game in the swing states results in a very favorable electorate in those states, but Romney wins the overall popular vote by a percentage point or two, it could be that a state like Pennsylvania, where many fewer resources have been invested, unexpectedly swings Republican, as Michael Barone suggests is possible here, costing Obama his electoral college victory. It’s a very interesting horse race