In Defence of Liberty

Driven by data; ridden with liberty.

Four More Years: The Pollster’s Election

Barack Obama has been re-elected as the President of the United States, whilst Congress remains divided, with Republicans controlling the House of Representatives and Democrats retaining a majority in the Senate. This last piece will examine the effect that the 2012 election will have on political commentary in the future.

Thanks to correctly predicting the political outcome of all fifty states in the 2012 Presidential election, and predicting the popular vote to within 0.3%, Nate Silver gained worldwide news coverage and notable praise. His FiveThirtyEight blog for the New York Times, named after the 538 Electoral College votes, is a poll aggregation website that focuses on American politics.

Poll aggregation is a method of collecting all publicly-available polls on a particular election, fusing them, and seeking to distil out any historical inaccuracies and long-term biases. These polls are weighted for their sample size, increasing precision, and newer polls update the running central estimate. The confidence interval surrounding this central estimate decreases and contracts as the election looms, giving a more accurate prediction for the vote. A vast number of simulations of the election are made, yielding a final probability for one candidate’s victory or defeat. This basic method can be further adjusted and tightened, particularly for local effects when attempting to predict the outcome by congressional precincts. Such aggregation can also strive to ameliorate renowned polling biases, such as the Bradley Effect. This method is neither new, novel nor complex.

There are many other poll aggregators out there, such as Sam Wang of the Princeton Election Consortium, Simon Jackman of the Huffington Post, and Real Clear Politics. Poll aggregation does differ from more traditional metrics of predicting Presidential winners, such as the state of the economy, incumbent approval ratings, and the positions of opposing candidates, which Silver argues are useful for election forecasts well in advance of polling day, but are not particularly exact.

Whilst Silver has certainly refined his method, resulting in some startling accurate predictions, the response to these predictions has been similarly startling. Dylan Byers wrote an article for Politico, questioning whether Nate Silver was a “one-term celebrity”. Joe Scarborough, an MSNBC host of the ‘Morning Joe’ programme and former Republican congressman, directly attacked Silver on his show:

Nate Silver says this is a 73.6% chance that the president is going to win? Nobody in the campaign thinks they have a 73% chance – they think they have a 50.1% chance of winning. And you talk to the Romney people, it’s the same thing. Both sides understand it is close, and it could go either way. Any anybody who thinks that this race is anything but a toss-up right now is such an ideologue, they should be kept away from typewriters, computers, laptops and microphones for the next ten days, because they’re jokes.

Silver succinctly responded to this criticism, of being a ‘joke’ and an ‘ideologue’: “All you have to do is count to 270. It’s a pretty simple set of facts. I’m sorry that Joe is math-challenged.”

Obama and Romney were often close during the race on national polls; but the President is not elected on the popular vote. The Electoral College system means that each state (and the district of Washington DC) is assigned a number of delegate votes, based on the population of that state, so that those votes sum up to 538, the total number of congressman and senators in the United States federal legislature. When a candidate gains the most votes in a state of any candidate, they are assigned all the Electoral College votes for that state. As returning officers at each election announce a state for a candidate, the opposing candidate has fewer and fewer combinations of states capable of achieving 270 Electoral College votes, which is often called the ‘pathway to victory’. There is a reason that Silver’s model predicted such a high chance for President Obama’s re-election was when Obama and Romney converged in a national poll: Romney either bolstered his backing in his electoral citadels, or swelled his scintilla of support in Obama’s safe-havens. Meanwhile, Obama gained more and more favourable polling in swing states such as Ohio. Put simply, a candidate receiving 50.1% of the vote in a national survey does not translate into a 50.1% chance that candidate will win the election.

There will always be political commentators who claim electoral victory, despite all prevailing evidence that their preferred candidate will lose. There will always be political commentators who claim to be reading the pulse of the nation. In the presence of such accurate poll aggregation, illusory proclamations of ‘momentum’ and ‘comebacks’ now look like divining tea leaves for the winner. This is undoubtedly the election for the polling analysts, who have demonstrated that calm scrutiny of the latest polling and survey figures can be compelling reading, even worldwide. Out of the rubble of the narrative-builders, the pollsters will rise.

%d bloggers like this: