Though the results have yet to be completely finalized, the race for the White House was less of a toss-up than it was a solid win by Barack Obama, thanks to his advantage in the battleground states – just as Nate Silver and other forecasters in the United States had predicted.
After coming under attack for concluding that the Democratic incumbent was always heavily favoured to retain the presidency, the FiveThirtyEight blogger for The New York Times was vindicated last night, as his forecasting model called all 50 states correctly.
However, despite all of the noise about how the polls were not accurately reflecting American public opinion, the polls did quite well throughout the country and Mr. Silver was joined by several other forecasters in making the correct call. The Huffington Post ‘s Pollster and Votamatic.org, operated by Drew Linzer, were also successful in calling all 50 states accurately.
The poorest performer out of the more well-known forecasting websites, Electoral-Vote.com, only missed the call in Florida and considered North Carolina a tie (it was won by Mitt Romney). RealClearPolitics, Talking Points Memo ‘s PollTracker, the Intrade betting market, and the Princeton Election Consortium run by Sam Wang all erroneously picked the Republicans to win Florida, but made the right call in the other 49 states.
Aside from Intrade, all of these models used relatively similar systems based on polling data. FiveThirtyEight and Votamatic did the most massaging to the numbers, while the others took them more at face value (RCP, for example, used a simple average). Had the polls been inaccurate in this campaign, all of the forecasters would have had great difficulty in making the right call.
For that reason, some of them calculated Mr. Obama’s chances of re-election throughout the race. Mr. Wang gave the President a 99.2 per cent chance of coming out on top before the polls closed, while Pollster’s 91.4 per cent and FiveThirtyEight’s 90.9 per cent were only slightly less certain. Nevertheless, the forecasters were nowhere near calling the race a ‘toss-up’, as many of Mr. Silver’s critics considered the contest.
With results still pouring in from the West Coast, a region won comfortably by the Democrats, it may be too early to determine who came closest to projecting the popular vote accurately. But as of writing, Mr. Obama was besting Mr. Romney by 50.2 to 48.2 per cent. That puts Pollster (50.1 to 48.4 per cent) and FiveThirtyEight (50.8 to 48.3 per cent) closest to the mark, though the numbers are more likely to move in FiveThirtyEight’s favour as all of California’s votes are tallied.
Just how well did Mr. Silver do? His forecasts were incredibly close in all of the battleground states. He considered Florida a virtual toss-up, with Mr. Obama and Mr. Romney tied at 49.8 per cent apiece, but he gave the Democrats a 50.3 per cent chance of taking the state. The current split is 49.8 per cent for Mr. Obama and 49.3 per cent for Mr. Romney. Virginia was forecast to go to the Democrats by a margin of 50.7 to 48.7 per cent, and is currently in Mr. Obama’s column by a margin of 50.8 to 47.8 per cent. Mr. Silver’s margin of error in the other battleground states (Colorado, New Hampshire, North Carolina, and Ohio) averaged only 0.7 percentage points per candidate.
But though Mr. Silver’s performance was excellent, it was not perfect. In the races for the Senate, his forecast made the correct call in 31 of 33 states. He picked the Republicans to take the Senate seats in Montana and North Dakota, rather than the Democrats (who were leading as of writing). These two states were the outliers last night, as RCP, PollTracker, and Electoral-Vote all made incorrect calls in one or both of these states. Mr. Wang’s model got the final Senate numbers correct and Pollster went 33 for 33.
Because of this perfect call in the Senate, Pollster had the best overall result last night. The forecasting model, operated by Stanford University’s Simon Jackman, made all of the right calls for both the presidency and the Senate, and as of writing was the closest to the mark on the popular vote as well.
Nevertheless, Mr. Silver’s forecasts received the most attention during this campaign, and he was accordingly the main target of attack. His track record remains impressive. In 2008, Mr. Silver made the right call in 49 of 50 states and in all 35 Senate races. That gives him an accuracy rating of 98.2 per cent in the last two presidential elections. With his focus on probabilities, Mr. Silver would likely say that he was just going with the odds. But undoubtedly he will be happy to silence his critics – at least until the midterms in 2014.
Éric Grenier writes about politics and polls at ThreeHundredEight.com.