Now that the dust has slowly started to settle on the 2012 U.S. election campaign season, we are left with a picture of what happened, who won and who lost. Aside from the politicians who came in on the wrong side of the vote, the biggest losers of this election cycle are the prognosticators of the right, who were not only almost universally wrong about the outcome, but in many cases wrong by several orders of magnitude. An independent ranking of the best and worst forecasters, based on the number of key states predicted correctly, awards the 'broken forecast' prize to the man who called Pennsylvania, Minnesota, Wisconsin, Iowa, New Hampshire, Ohio, Colorado, Virginia and Florida for Mr. Romney, wrong in each and every case. Step forward Fox News pundit, Dick Morris. Close runners up for most woeful political prescience include another Fox News pundit and writer for the Washington Examiner, Michael Barone, as well as Steve Forbes, of Forbes Magazine and George Will of the Washington Post.
Among the pollsters, my 'getting it hopelessly wrong' award goes jointly to Gallup and Rasmussen Reports, for a faintly ludicrous set of polls, skewed (presumably by accident) in favour of the Republicans, in the days and weeks running up to election day. This mirrors the huge skew to the Republicans recorded by Rasmussen in 2010, and further back, dating to his calling of the election for Mr. Bush by nine percent in 2000 (Bush lost the popular vote).
The 'hopelessly wrong' award for state voting goes to Susquehanna Polling of Pennsylvania, who got their polls woefully skewed (again presumably by accident) to the benefit of Mr. Romney, while the broad consensus of the out-of-state polls simultaneously got the result almost spot on. Another loser was RealClearPolitics, who managed to eliminate from consideration some of the most accurate pollsters while sticking with many of the worst.
On the plus side, it was a very good night for five sages who called the state tally perfectly, these being Nate Silver of FiveThirtyEight at the New York Times, Markos Moulitsas and Daily Kos Elections, Simon Jackman of The Huffington Post, Josh Putnam of Davidson College and Drew Linzer of Emory University. More broadly speaking, it was a great night for serious statistical analysis, for the 'quants' as they are sometimes labelled (including Sam Wang, of the Princeton Election Consortium).
The unsung hero of this election, however, is not a particular pundit, or statistician, or polling organization. It is not an expert at all. The unsung hero is the anonymous crowd. The reason is that while an expert may know more than anyone else in the room, he is unlikely to know more than the room as a whole, to be wiser or cleverer than the crowd. It is an idea picked up and developed in a fascinating recent paper by David Rothschild and Justin Wolfers, of the Wharton School, University of Pennsylvania, who argue that better forecasts can be made by asking people who they think will win than by asking their personal voting intention.
This is the idea that is driving the growing belief in the power of markets like the betting exchange, Betfair, to unveil the future. Despite the swings and roundabouts in the polls and among the pundits, these markets never wavered in their belief that the president would be re-elected. It all comes down to the idea that those who know the most will tend to bet the most, and accurate forecasts can thus be obtained by following the money. This is the basis of the science of 'prediction markets', which are essentially speculative markets created or employed for the purpose of making predictions.
In the weeks to come, data will be crunched to see exactly how well these markets did perform in Election 2012. It will be a while until we know the details, but already I am confident enough to make at least one prediction -- they will have performed very well indeed.