For about five minutes it was really fun to see everyone taking statistical modeling of political outcomes seriously. Well, ostensibly seriously. It was troubling that this was only done by the side whose preferred outcome was the one being forecast, but I had some hope that this might stick. Silly me. Immediately after the outcome became clear the same folks that were praising statistical models ten minutes prior discarded them in favor of ad hoc narrative.
This is unfortunate on a number of levels, not least because there is more at stake here than Nate Silver's reputation on Twitter. Silver was important because he was the closest that political science permeated the mainstream coverage of the election this cycle, and his case became both a bellwether and a rallying call. This in itself is unfortunate, since the consensus of various political science models of electoral outcomes had been predicting an Obama victory pretty consistently since last winter at the latest. So it's unsettling to see the lessons on offer being cast aside so soon. There are far more egregious examples than the one I'm about to present, but reformation begins at home. So I'll pick on my own home -- that being the IR blogosphere, in this case -- first.
Dan Nexon wrote:
But the 2012 election makes clear how far the GOP has strayed into the epistemological wilderness.With respect, Dan, I just don't think this is true. What the 2012 election shows -- or, instead, what it reiterates -- is that incumbent presidents have a very large electoral advantage when the economy is growing. Almost every model that I've seen based on "fundamentals" -- even the generic-candidate ones, which zero out all the "binders full of women" and "Big Bird" and even "47%" and Super PACs... all of it -- showed Obama with a pretty high likelihood of winning. And I think those models likely dramatically understated the case, considering the depth of the trough the economy has been digging out of since Obama took office. The most similar historical precedent we have led to a presidential administration so popular that, absent mitigating factors like human mortality, it seemingly would have continued in perpetuity, thus necessitating a constitutional amendment which instituted presidential term limits. So there were very good reasons to think Obama was the strong favorite all along, no matter what the GOP did and even before they did it.
And that was in fact the case. Obama fluctuated in between 60-75% chance of victory for most of the past year, according to the best political science of which I'm aware. See this very simple one, which was so simple that even poli-sci sympathizer Ezra Klein doubted it despite commissioning it! It had Obama at 67% and incorporated no new information after June, and not much up until then. I know that a few models here and there deviated from that expectation, but they were often the models with the least amount of empirical support behind them.
As such, I think the story of this election is that there is no story. At least, there isn't a story if we're going to take political science seriously, other than we should take political science seriously. And the "we" in that sentence begins with political scientists. If our models expect Obama to win because the economy grew out of a deep recession, plus he had an incumbent advantage, then we should accept that as the best explanation for his victory until another story with better evidence comes along. Those factors have historically been very powerful forces in presidential elections. They remain very powerful. The only other possible takeaway that would be well backed up by prior theory would involve demographic dynamics, but even that isn't really needed. And if the GOP is in the "epistemological wilderness" then how the hell did they sweep the 2010 midterms? Are epistemological wildernesses created in two short years?
Maybe. This is not an analogous event in need of explanation. This is exactly what we should, and did, expect. So I'd go with the models until given a better reason to abandon them.
P.S. I agree with Nexon's general take on the desirability of the current GOP. (Although I am less sanguine than he about the Democrats, maybe.) But the best case evidence for that does not come from this presidential election. As someone at The Monkey Cage pointed out recently, the GOP has lost four lost Senate elections in the past two cycles -- Delaware ‘10, Indiana ‘12, Nevada ‘10, Missouri ‘12 -- which almost certainly would've gone to the GOP had the Tea Party not moved the GOP platform away from the median voter. Those four seats currently represent the difference between majority and minority in the Senate. So maybe Nexon's got something there, but this is also in line with the common models: if you move to the extreme, you will lose.
P.P.S. It is also not exactly regrettable for someone who identifies with the left generally, as I believe Nexon does. If the GOP wisens up in the ways that many are asking them to (including both me and Nexon) one by-product will likely be less Democratic success in future elections.
This touches on another habitual harangue of mine, which is that the Tea Party is the left's best friend right now but that the left is too myopic to realize it. The left still has no guiding ideology of its own that is capable of moving the median voter leftward. , The American voting public still skews center-right despite it all. The only thing giving the left a chance are extremists on the right pulling the GOP away from the middle. Without the extremists on the right, the Democratic Party might not have won a presidential election from 1976-2008. Think about it. Without Perot, perhaps Clinton doesn't win in 1992 (although some political science suggests that Perot hurt Clinton at least as much as HW Bush, and the fundamentals were weakly in Clinton's favor). Without Clinton in 1992 it is unlikely that a Democrat would have won in 1996. And... that's it until 2008.
[Ed. Light edits for clarity shortly after posting.]