Tuesday, October 30, 2012

Why Nate Silver is Wrong....

The Telegraph has unique insight into the USA's politics, as this blog shows.
In the history of presidential elections, has there ever been such an effort by one side to poll their way to victory? While the Republicans have spoken this season about jobs and debt – willing themselves to a moral victory – the Democrats have talked constantly about how well their guy is polling in one or two states. The goal is to create a sense of inevitability, to convince the public to vote for Obama because he’s a winner and who wouldn’t want to vote for the winner? We’ve witnessed the evolution of polling from an objective gauge of the public mood to a propaganda tool: partisan and inaccurate. Step forward Nate Silver of the New York Times. Nate has been an open supporter of the President and his newspaper just endorsed Obama (although it also went for Dukakis, so it ain’t that good at picking winners). But context doesn’t matter because maths is maths and maths can’t lie – and Nate says that, according to his model, Obama has a 74.6 per cent chance of winning. You might find that figure a little odd given that on the same page you’ll see that Obama is ahead by less than 3 per cent nationally and his advantage lies in one state, Ohio. It’s even odder when you consider how it conflicts with other polls that emerged this weekend giving a virtual tie in Wisconsin and Minnesota. It’s damn near-surreal when you discover that Gallup puts Romney ahead by four points among (and this distinction is critical) likely voters. Meanwhile, Obama’s job approval rating is heading downwards. Does Nate know something that the rest of the world doesn’t? A former business and baseball analyst, Nate came to fame in 2008 when he correctly predicted the outcome in 49 of 50 states in the presidential election. Frankly, a headless chicken could have done that. It was a wave election and we all knew Virginia and North Carolina were in play. Plus Nate had access to internal Obama polls that gave him an advantage over his market rivals. Nevertheless, this success turned Nate into a star – despite his own admission that his analysis technique is not modelled exclusively for politics. In an interview about his life’s work, he wrote: But the other thing too is on the blog I mostly focus on politics. And I think elections are a really interesting thing to study and to try to predict. But I don't particularly like politics. I find some of the people involved in politics, I don't think they're the most well-rounded or pleasant people necessarily, right? So I want to broaden my focus a little bit and say, look, by being data-driven and looking at how predictions go, doing analysis from statistics and everything else, we can look at business or sports or a lot different fields or science. And there's nothing about politics in particular that my interest and skill sets are uniquely suited to. [Italics are mine.] That noise you hear is the sound of the cat being let out of the bag. Appreciating that Nate’s system is rather more generic, interpretive and partisan makes sense of its central paradox: that while the rest of us are talking about Romney’s post-debate momentum, Silver still gives the race to the President by a huge margin. Here are some of the problems with his stats. 1. Nate isn’t very good at calling close elections. In 2010, he correctly predicted the outcome of the senate elections with the greatest leads. But in the 5 genuinely close races, he got it wrong in 3. For the House elections, Nate ran this extraordinary headline: “House Forecast: G.O.P. Plus 54-55 Seats; Significantly Larger or Smaller Gains Possible.” So, this oracle predicted that the results could have been “larger” or “smaller” – how prescient. In fact, they were much larger. The Republicans took 63 seats. 2. People make their minds up at the last minute, which confuses the outcome of close elections. Historically, voters have tended to break towards challengers, and particularly Republican challengers, in the last week. 3. Nate weights polls, meaning that he picks and chooses which data sets to run through his model. He has shown particular affection for Democrat-leaning pollsters like PPP, and this bias is evident in his use of state-wide polls. Silver embraces polling organisations that other writers avoid like the plague. Apparently, the New York Times isn’t as discriminating. 4. Nate ignores polls that contradict him. So PPP is right and Gallup is wrong. 5. Politics is even riskier than baseball and “stuff happens.” As columnist David Brooks put it in a critique of Silver’s polls: “Obama turns in a bad debate performance. Romney makes offensive comments at a fund-raiser. These unquantifiable events change the trajectories of tight campaigns. You can’t tell what’s about to happen. You certainly can’t tell how 100 million people are going to process what’s about to happen. You can’t calculate odds that capture unknown reactions to unknown events.” Nor can we determine turnout – and a lot of the polling in 2012 has presumed that as many Democrats will vote today as they did back in 2008. If that’s wrong, many predictions will be confounded. Brooks’ point is really the most powerful of all. Politics is not a science and it doesn’t lend itself well to predictions. Voters lie, natural disasters happen, scandals rock the White House. No one could have predicted – including Silver – that the debates would radically transform Romney’s image in the eyes of the voter. In some cases, that transformation happened in spite of Romney losing one or two of those debates. Silver’s stardom tells us two things about the Democratic Party. The first is that its obsession with numbers is part of a cold, mechanical way of looking at politics that divides the electorate up into blocks of voters who can be cobbled together into a winning coalition. Team Obama went out of its way in the 2000s to recruit professors and statisticians who would turn politics into something like baseball: hire the best players, master all the tricks, bet and gamble your way to victory. Grand narrative is gone. In its place are talking points designed to achieve a 51/49 per cent advantage: war on women, 99 per cent etc. Second, Nate’s success shows that Democrats are panicking. Losing the war of ideas, they’re resorting to bad maths. Last night a friend posted this on Facebook: “I want Nate Silver's data made into a blanket I can cuddle up with.” Sorry, but weighted polling served up by a partisan analyst is a very false comfort.

No comments :