Here’s a pretty good piece in Slate:
The election guru said Trump had no shot. Where did he go wrong?
By Leon Neyfakh
A few comments in defense of Silver:
- Donald Trump has not yet quite gone through the formality of winning the GOP nomination.
- Silver has been pretty explicit that nomination campaigns are harder to predict than general elections.
- Silver’s method of aggregating and weighing poll results in the fall of 2012 was about as good as could be done. The danger facing him was always that the polls would turn out to be wrong due to some kind of technological or social change in polling response behavior. For example, in 1936, the Literary Digest’s telephone poll turned out to be too early: people without phones were significantly more pro-FDR than people with phones. In 1996, all the pollsters except Zogby missed that the race was going to be moderately closer than they expected (for reasons that I’ve never seen well-explained, but the polling industry seemed to fix the problem by 2000, whatever the problem was).
It’s quite possible that in some future election, it will turn out that current polling techniques are disastrously outdated. But that didn’t happen in 2012.
In fact, Silver’s prediction of a 2 point Obama win turned out to be wrong by 2 points. It just happened to be on the
conservative liberal side: Obama won by 4 points. But nobody much cared.