No! I have an extremely negative association with 538 (not their fault). I don't want to see that logo for at least another year.
back
2 comments
I simply don't get this. Do you view fivethirtyeight as a kind of pagan god, who gives you what you want? On the contrary, to the best of my knowledge the crew over there try to describe reality. Sometimes that reality is good and sometimes it's bad, but it's real.
Care to elaborate?
Their model of the 2016 US Presidential election reflected the same polling error that all the other models showed. They gave Trump a 30% chance of winning, along with some commentary about ways that systemic error could easily swing things in his direction. Trump then won the election with results well-within their error margins, but that wasn't enough for some people.
It's hard for some members of our society to acknowledge that 30%-chance events happen on occasion, so 538 takes the fall.
It's not even clear if there were systemic polling errors: there were two last-minute changes that affected the election. There was the unexpected last-minute letter from Comey on October 28, and 538 did in fact report Clinton's chances to be in freefall from then until days before the election. There was also the unexpected competence of Kushner's data-science work right at the end to "micro-target" specific undecided voters and likely Hillary voters (unexpected in part because Trump had expressed skepticism at those strategies, and that organization was underfunded). It certainly seems possible that 538 had the data exactly right all along.
I suppose it's neither here nor there now but I strongly suspect that both candidates' internal polling showed much closer numbers than the nationally published polls.
In the last week of the campaign, Hillary was still visiting states that should have been in the bag for her and Trump was still visiting states that he should have written off as lost.
Maybe they were just tactical anomalies but I think it makes more sense if they both knew that the race was closer than everyone else thought.
> It's hard for some members of our society to acknowledge that 30%-chance events happen on occasion, so 538 takes the fall.
indeed, with a well-calibrated predictor, they should happen 30% of the time.
Their election night coverage was one of better ones, but in the lead up to the election, Nate Silver got a lot wrong about Trump. He even had to write a mea culpa about it. Perhaps that is what OP is talking about?
>but in the lead up to the election, Nate Silver got a lot wrong about Trump. He even had to write a mea culpa about it.
That was for the Republican primary.