Nate Silver’s New York Times blog, FiveThirtyEight, was responsible for 20 per cent of visits to the NYT site this week – that is, one in five visits to the sixth most trafficked US news site.
He’s pretty popular.
And he currently predicts a nine-in-ten chance of an Obama victory.
A former sports statistician, Silver accurately predicted the outcome of 49 of the 50 states and all 35 senate races in the 2008 election – using pure numbers as his tools.
In his final estimate, he gave Obama a 91 percent chance of winning, estimating the Electoral College tally at 313-225 and the popular vote at 51-48 Obama.
Silver argued on The Colbert Report recently that his methodology is really quite simple. ‘People treat it like it’s Galileo, something heretical,’ he said. ‘There are many things that are much more complicated than looking at the polls and taking an average and counting to 270, right?’
Can numbers lie?
Silver’s huge hit rate isn’t explained just by his followers, though. (And by the way, some of those followers have admitted that they’re not simply seduced by his way with numbers, but are Obama supporters, comforted by his consistent predictions that their man will win.) Silver’s critics read his blog religiously too. Some claim that his model is wrong, while other believe that you can’t predict an election by math alone.
‘I do understand that math can be ironclad,’ said Jonah Goldberg at the LA Times. ‘But I like to think that people are different, more open to reason, and that the soul – particularly when multiplied into the complexity of a society – is not so easily number-crunched.’
Math versus ‘gut-level’ instincts
It’s a clash between those who believe that numbers don’t lie – and can tell the full story – and others who believe that numbers don’t allow for the twists, turns and inconsistencies of human nature. And that’s why some are calling this ‘the Moneyball election’.
‘In a way this is a perfect test case of the Michael Lewis Moneyball hypothesis,’ said James Fallows at the Atlantic. ‘Apart from Silver’s own background as a sports-stats analyst, we have an exceptionally clear instance of people judging from their experience, their ‘bones’, their personal instinct, etc that things are going one way (like veteran scouts saying that a prospect ‘looks like a Big Leaguer’), while data (on-base efficiencies in one case, swing-state polls in another) point in the opposite direction.’
Today, Adam Gopnik continued on the theme for the New Yorker’s blog, clearly outlining the pro and anti Silver arguments.
‘What [Silver and co are] saying is that you can look at much, much less evidence, but be confident in what it tells you just because you are confident that the future will be like the past – that polling averages, properly adjusted, are a nearly infallible guide to the results of elections. The pro-pundit class is saying, echoing those scouts and pros, that there are just too many variables – too much uncertainty with tens of millions of individuals acting according to the whims and moods and strictures of the moment – to think that this is so. As with ballplayers, so with politicians, the pros say: only a seasoned and practiced eye can suss out, exactly, the tools of each campaign – who they reach, what they say, how they react – and get the right answer. It’s a gut-level thing, they say, and they point, not unreasonably, to the many cases where polls are wrong, missing the point that, on the whole, they’re right.
A one-election analyst?
There’s a lot riding on this election – for America and the world, yes, but also for the way that elections are reported on and predicted. If Nate Silver’s predictions prove accurate once again, that will surely have an impact on future elections, as well as minting his own career. But if they fail, he’ll become, as some pundits have predicted, a ‘one-election analyst’. If your career is built on accuracy, you’d better be right.
‘One thing you’d rather not have happen when you’re seeking to make a forecast is to influence the attitudes of the people whose behavior you are trying to predict,’ Silver told BuzzFeed. ‘It’s clearly a very close election and conservatives and liberals who are concerned about the outcome probably ought to channel their angst by making sure that they vote, rather than either demonizing or deifying a blogger with a statistical model.’