Sort of.
He was wrong in the primary. Really wrong. And he’s actually written a column about it since…
In the general election, I don’t quite think I would say that he was wrong, because for statisticians “right” and “wrong” are not such binary concepts. I wrote a little about this in one of the other election threads…
For reference, here is the article Nate posted on election day:
Most people read this headline and say “Well, he said most of the outcomes come up Clinton, and Trump won, ergo Nate Silver was wrong” - which is not really accurate. The article acknowledges that there is a wide range of outcomes and gave Trump a 28.6% chance of winning, and as explained here…
“The goal of a probabilistic model is not to provide deterministic predictions (“Clinton will win Wisconsin”) but instead to provide an assessment of probabilities and risks.”
It’s fun to take potshots at the nerds who were “wrong” about Trump, but the 538 models weren’t as “wrong” as everyone is making them out to be. At one end of the possible range of outcomes was a landslide for Hillary; in the middle was a close win for Hillary; and at the other end was a narrow win for Trump. We ended up at that “other end” considering that it’s basically a dead-heat in the popular vote, and a matter of pulling out all of the key states (Wisconsin, Michigan, Pennsylvania) that were extremely close.
Trump’s win (or any Electoral-popular-vote split) like a football team that gets to the playoffs by going 11-5 with four blowout losses and eight wins by a TD or less. Trump didn’t get landslide wins in many states and he lost a couple blowouts, but he won all of the close ones.