Nate Silver, statistician and blogger, held an Ask Me Anything on reddit Tuesday. The topics discussed ranged from sports to the end of the world, but here were some of his thoughts on the 2012 election cycle:
1) Election forecasts can be reliable. While Silver admits people tend to overrate predictions within complex systems, he says, “There are certainly exceptions, and presidential elections are almost certainly one of them, but it’s a bit weird/ironic that I’m known for one of the exceptional cases.”
2) Bad process led to bad polling on the Romney campaign. “Groupthink and perverse incentives were the causes; to the extent their polling or analysis was bad, it flowed from that,” Silver says.
3) An Obama win was predictable early on. Silver feels a culture clash occurred when news outlets fishing for a story about a contested election were faced with the numbers, which told a different story. “Part of what [we] were saying for much of the campaign—both at different stages of the general election and perhaps even more emphatically in the end-stage of the primary when Romney pretty much had things wrapped up—is that the outcome had become fairly certain,” he says.
4) Silver downplayed his election model’s accuracy in the beginning. “Stupid poker analogy: part of playing well is in maximizing the amount of value you get from a hand in the event that things go well, in addition to mitigating your losses if they don’t,” he says. So he emphasized the model’s uncertainty early on, and only began rebuffing pundits in the final weeks of the election. His model hadn’t changed.
5) Politics is more frustrating than sports. Silver finds politics harder to analyze because between the partisans and pundits there are fewer reality checks. A win’s a win in sports. “In politics, you can go on being delusional for years at a time,” he says.
Oh, and he’s surprised there’s no election equivalent to baseball-reference.com.