The latest odds from FiveThirtyEight
Elections are unpredictable. That’s why we have them. That said, modern polling and meta-polling do give a pretty decent sense of how the US Presidential election is going to play out. Added to that, their electoral college system also means that relatively few states (and votes) will make the difference between win or lose. The gold standard prognostication comes from Nate Silver of FiveThirtyEight. Despite an incredibly tight national race, he’s predicting an 85% chance that Barack Obama will win.
But what does ‘chance’ mean? Nate Silver is a statistical genius who got into election polling after politicians started messing with his full-time online poker playing.
Silver’s backstory reads like a quirky screenplay. While working as a transfer pricing consultant who helped companies minimize tax exposure from overseas investments, he spent his evenings and weekends perfecting an ingenious computer system for evaluating baseball players’ stats, which outperformed the analyses of many experts. A company running an Internet site for Rotisserie League fans bought his system. Silver used the money to stake himself as a full-time online poker player, quickly earning $400,000. Shortly thereafter, congressional grandstanding before the 2006 midterm elections stifled the online poker business with regulations, leading the best professional players, deprived of the overconfident amateurs they had been feasting on, to go after players like Silver. He lost $130,000. So he turned to politics, attempting to predict the partisan composition of the next Congress to help him decide whether to cut his losses and move on from poker. Thus was born the sideline that became fivethirtyeight.com. (The Prospect
538 refers to the number of electoral votes one needs to win the Presidency, it’s a combination of the number of Senators and Representatives plus a little sauce. What this means is that for most states it’s winner take all. George W. Bush won Florida by (a prescient) 537 votes and got all of that states 25 electoral votes. This meant that he could lose the popular vote and win the Presidency, albeit with a lot of help from the Supreme Court. And then you got the collapse of American power and economy as it were. But never mind.
The point is that predicting who’s going to win isn’t just predicting who’s more popular. If it’s just a popular vote statistical quirks and anomalies get filtered out by volume and it’s probably easier to predict. With each state voting on it’s own, however, small swings can make a big difference.
Silver and others account for a lot of different state and national polls, add economic and other factors and then run lots of simulated elections. The image I get is the last scene from the movie WarGame (spoiler embedded above) which shows a rogue computer playing tic-tac-toe for a long time and then running a bunch of nuclear simulations to get the point that there are no scenarios for global thermo-nuclear war that work out.
For Presidential elections, however, there are situations that do push one candidate or another over the required 270 electoral votes. In this current election, those simulations have Obama winning 85.5% of the time, with the other variance being the possibility that all the polling is somehow biased against Romney. This numbers bounce a lot, especially, after Obama’s disastrous first debate, but the overall trend for months has been towards an Obama win.
Now, getting away from the numbers, you could say this is because Mitt Romney is a lying and unlikable douche, pretending to believe toxic social positions he kinda doesn’t, pandering to the very wealthy and generally saying whatever he thinks is necessary to win. You could also say it’s because Barack Obama remains personally popular and did a lot with a bad hand, preventing a depression, passing healthcare and killing Osama Bin Laden. I think there’s also something to be said with the Republican party basically going insane and turning into a weird pyramid scheme that sells a perverted American dream to working class people getting them to vote against their interests and transfer money to the 1%. But those are all just possible factors across a thousand polls that go into statistical models.
Right now Barack Obama has the chances of a US football team leading by 3 points at the end of a game, a bit better actually. Elections are still unpredictable, but the most solid bet is that Obama will win. I sincerely hope so.