Last week’s election was a big victory for President Obama and the Democrats. But aside from a win for the Democrats, the election was also a win for the pollsters — you know, the trained statisticians who make their living surveying public opinion? These guys were under attack this year by Fox News and the conservative media. According to the perpetually paranoid over at Fox News, the pollsters who were showing Obama leading in the Electoral College for the entire year were just as liberally biased as the overwhelming number of scientists who believe in global warming and the statisticians in the Bureau of Labor Statistics who showed a decline in the unemployment rate in the run up to the election.
Even to the bitter end, Fox contributor and Harvey Fierstein impersonator Dick Morris was predicting an electoral college landslide for Romney, and the conservative media bought it hook line and sinker. I always try to keep Mark Twain’s maxim about statistics in mind, but when you have different polls with varied methodology all telling you something that’s at odds with your view of the world, that’s a pretty good indicator that your assumptions might be incorrect. Morris was contrite this week, explaining that he assumed a turnout more in line with 2004, but it’s not clear that there was any evidence to suggest this except the personal opinions of him and others on Fox.
I was having this debate months ago with my conservative uncle whose comeback for “the polls are showing you behind” was always “not according to Scott Rasmussen.” Rasumssen was was the king of the 500 person automated poll which assumed a strong Republican turnout based on responses to questions regarding party identification. Rasumussen’s polls consistently showed a Republican bias of a few points, which can make a real difference in a close election. But a little knowledge can be dangerous and Rasmussen’s polling bred a cottage industry of bloggers contesting the polling in the presidential race by adjusting the party identification mix the pollsters were predicting based on their interviews. The website unskewedpolls.com was the most prominent of the naysayers and they “specialized” in taking other peoples polls and recasting the results by adding more Republicans to the mix.
Meanwhile, the conservative media shills needed to find a visible scapegoat and they found it in Nate Silver, a statistician who turned to election prediction in 2007. Silver had a great record in 2008, predicting every state except Indiana for Obama. In the wake of that election, he was hired by the New York Times as a blogger, where (in case you were wondering) he did well predicting the Republican Congressional landslide year of 2010 as well.
Silver’s model was projecting an Obama win for most of the year based on his narrow but steady lead in the Electoral College polls. Oftentimes, his percentage prediction of an Obama win seemed over-optimistic, so you could quibble with the confidence level, but it’s hard to look at a guy who leads for most of the year in enough electoral college states to win the presidency and argue that he’s not the favorite. Plus, this is a statistical model. One assumes that if Romney was showing the same swing state resiliency, then it would have shown the same result for him.
By the Monday before the election, Silver had Obama at an 85% chance of victory. Meanwhile, the folks at Fox were still telling their viewers that Romney had the momentum and was going to win this thing. Dick Morris, George Will and others predicted a Romney landslide. Perpetually smarmy Wall Street Journal columnist Peggy Noonan wrote a blog post that Monday in which she predicted Romney would win the election based almost completely on… a feeling she had. Business Insider called it “the most anti-Nate Silver column imaginable,” not because she spoke about or even alluded to Silver, but because her analysis was almost completely devoid of empirical facts. This seemed like bravado at the time — a way to embolden the troops before a big fight–but in the aftermath it looked like they spent so much time in their own bubble that they couldn’t imagine any other objective reality where a majority could vote for Obama. Last week James Fallows likened it to the dismay attributed to Pauline Kael in the wake of the 1968 election when she couldn’t imagine how Nixon could have won, since “no one I know voted for him.”
To a certain extent, this makes sense. If you spend all your time talking to white Republicans who think that Obama is leading this country on a dangerous slide to socialism, that’s going to color your analysis. To be sure, Romney did carry white voters by a big margin and if the electorate turned out to be as white as they all seemed to think it would be, then we would have been looking at President Romney. But with all of the evidence pointing the other way, these guys should have known better. I have to imagine that there’s more than a few Fox viewers this week who feel like they’ve been had.