Given the ass-kicking 2016 gave most political prognosticators, which ones will you be paying attention to in 2018?
Is Sam Wang still your guy, even with bugs in his teeth? Nate Silver all the way, for being less certain than the others? Charlie Cook? Drew Linzer? Someone else? Are you going to ignore all the talking heads and just read polls?
Can you explain why you think Nate Silver wasn’t accurate. He put the odds of Clinton winning at around 70%. Things that have a 30% probability occur with some frequency. In fact, that frequency can be determined as around 30% of the time. I will continue to go by 538, which has been unbelievably accurate.
FWIW, I don’t believe I made the claim that he wasn’t accurate here. I said he was “less certain then the others”, which he was. For an overview, from the NYT link above, these organizations all predicted Hillary would win, with varying degrees of likelihood / certitude:
99% = PEC
98% = HuffPo
92% = DK
89% = PW
85% = NYT
71% = 538
If I might gently suggest: the problem was never Nate Silver.
It was Nate Silver’s eager readers, who read the words, “Clinton has a 70% chance of winning,” then turned to their friends and said, “Clinton’s gonna win.”
Which prognosticators? None.
What they think has no bearing on how I’ll vote, so I will digest the issues and the candidates’ stances on the issues with as little spin as possible from the prognosticators and pundits. And then I’ll mark my ballot giving not one damn if someone else thinks my preferred candidate has a 2% or 98 % chance of winning.
I wasn’t asking if they’d affect your vote, I was asking where you’d turn to get campaign- and election-related news and predictions in the run-up to the 2018 election.
Yep. No one would celebrate a doctor telling them they had a 70% change of being alive next week.
Who’ll YOU be paying attention to, Ditka?
(Me, 538/Silver.)
Probably mostly the RCP polling averages.
I’ve learned my lesson. I’m going to pay closer attention to what thoughtful people like J.D.Vance have to say.
Remind me, what did the RCP rolling averages say about the probability of Trump winning?
I think you probably know that RCP doesn’t give percentages for probability of winning, but instead focusing on polling averages. Here was their final no-toss-up map, which gave Trump 266 ECV vs 272 for Clinton. Nate Silver’s modeling gave Trump an average of 235 ECV, and Sam Wang famously said “If Trump wins more than 240 electoral votes, I will eat a bug”. They were all “wrong”, but I think it’s pretty obvious which one was least “wrong” and which one was most “wrong”.
Is it really obvious? Because it looks to me as though both 538 and RCP got the same number of states right. They only differed on 2 states, Nevada and Florida; 538 got Nevada right (and RCP got it wrong} and RCP got Florida right (and 538 got it wrong).
And so that we’re both clear, RCP projected Trump to lose. That’s not surprising, given that 538 and RCP were both using the same polling data.
And to reiterate what someone said earlier, things with a 30% chance of happening happen approximately 30% of the time, to America’s great misfortune.
I think you (and Nate Silver ;)) missed North Carolina.
ETA: I get your point about 30%, that’s why I was putting “wrong” in quotes.
Nate Silver. His track record in 2012 was perfect state-for-state, IIRC, and in 2008 it was near-perfect. And in 2016 he stuck by his model, which was the most accurate of the various polling-aggregate models that I’m aware of.
Shoot, according to all those articles in the Times and the Post, Trump supporters still support Trump! Hardly seems worth holding an election at this point.
What about just the RCP no-toss-up map? It got FL and NC (and ME2) right where Nate got them “wrong”, whereas he got NV right while RCP got it wrong. Are you discounting RCP because you don’t consider it a “polling-aggregate model”, or because you think they were less accurate than 538?
I wouldn’t discount them – I’ll probably look at them too, but in my understanding they were less accurate than Nate in both '08 and '12, and so I still think Nate’s record is the best. As I understand it, RCP doesn’t weight polls differently, which means they allow total crap polls to influence their results – Nate weighs polls based on their track records, and I think that results in a more accurate product, in general, even if in a few states RCP happened to be better in '16.
Thanks for the clarification. I haven’t actually compared their '12 or '08 performances. You may be right.
I’m a Nate Silver fanboy.