That's just a journalist using the wrong vocabulary. If you ever listen to Silver, he's always really clear he's not making a prediction. It's like his mantra.Ok, I am aware of that but this years race was characterized by a lot of uncertainty due to undecideds and high third party support so they were bound to do worse. Also, again doing worse or getting it wrong does not mean that it's not real science or even inherently flawed nor does getting it right mean that the history professors method is anything other than pseudoscience. I can make a one question model which would have accurately predicted this race while also characterising every president ever, though it would give a 50:50 prediction for all previous race. Is the candidate a man?
Well I'd say hopefully, but then Pence
His probabilistic models from previous elections are still online if you don't believe me and want to check them out.
The polls should factor in early voting directly in the questioning of voters. They should also try to tease out shy or hard to contact voters. These are what polls are there to do. Silver isn't a pollster.
Just have to suck it up and take it and hope you guys vote well in 2 years time. Hope that the Supreme Court doesn't lose anymore judges in that time either, and then hope that the Dems get their shit together and a decent candidate is put forward in 4 years time, all the while hoping the Supreme Leader doesn't cause WWIII in the meantime. How can you possibly know that though? Do you think any model which didn't show Trump as the favourite has to have been wrong? That just doesn't make any sense, at all.
Once early voting is open, most polls ask "Have you already voted, and if so who for?" as one of the first questions.
I agree mate.
Early voting is a bit of an incomplete picture as far as I know. What gets released varies state by state with Nevada being really the only one that releases raw data of who has how many votes. I think your second point is more where the problem lies, as the early voting picture is very incomplete and didn't help the people who looked at it this year, with the difficulty of predicting what demographic is going to surge in what state and who you are not reaching with your polling. Though those are not really Silver's problems to fix.
Yep, they're factored in to most polls.
You're just ignoring everything people are telling you now.
No. Early voting data is incomplete and only gives you part of the picture (http://www.realclearpolitics.com/ar...voting_a_poor_predictor_of_final_results.html) so there is no real reason poll based models should bend over backwards to shoehorn it in. Plenty of people did look at it this year and came to the conclusion Clinton had it in the bag. It's incomplete nature means you might as well be reading tea leaves, you'll just see what you want to. Also, your preferred model from the history professor does not include early voting nor any other data.
And it would have been wrong for the odds of a Clinton win in Nevada to soar given that in the end she only won the state by 2%. It was marginal, and could have swung on election day.
http://projects.fivethirtyeight.com/2016-election-forecast/nevada/ http://www.nytimes.com/elections/results/nevada
Nevada's a little different as a heavy majority of people there vote early, and registration rates are predictive of results historically. Colorado is similar, though to a lesser extent. If 538 had been showing Trump as favourite there on election day then there would be cause to argue, as there was strong evidence that wasn't the case. But in the end, as they showed her narrowly up (which wasn't the case on the RCP average), as you say, you can't really pick fault.
I get what you're saying, but it really was close in Nevada. Put it this way, if just 1 in 75 voters had switched from Clinton to Trump, she would have lost. You can't go much beyond 60/40 odds in a situation like that I think, even if a big chunk are voting early.
I agree, I think Ralston mentioned that Trump would have to win election day by 10 points or something which was unheard of, but as we saw elsewhere, unheard of was happening quite a lot in rural areas. She was up in the EV by similar levels as Obama '12 but the final result was 4% closer, higher than the national swing, so something along those lines must've happened.
I don't see how it can be argued. Of cos plenty of people don't vote anyway, but the EC is still a big factor why many people don't vote.
I hope not, President Pence would be even worse than Trump with his archaic beliefs.
