Calling it a battleground state doesn’t mean it’s in play, as the polling doesn’t show that it’s close at all. The closest poll in NH was two weeks ago, showing Hillary up 3, but everything else recent is +7 or more for her.
Nate’s own forecast for NH gives Hillary an 86.4% chance of winning it. He gives Trump a better chance of winning the presidency than he does of winning NH. NH isn’t in play.
She COULD lose every state that is still in play-- AZ, NV, IA, OH, NC and FL-- and still win the election.
To your point, a lot of it is from the assumption that the states are correlated, so if the polls are off in one state, they are likely off in other states too. With no correlation at all, the odds would be 99.8% in favor of Clinton. So yeah, missing the prediction in Florida or North Carolina alone would still be OK for Clinton. But if that happens, it also might point to missing it in PA, WI, NH, etc.
Still, even if the polls are off, or Trump swings, or whatever, I just don’t see a path for him. Let’s say the polls are off, or large swaths of the population move in Trumps direction-- how about 5 percentage points between now and election day. If we shift every state’s current RCP average 5 points in Trump’s direction, this is the map we end up with.
Minnesota is the only state that doesn’t put one candidate clearly on top if the polls shift 5 points toward Trump. But considering Minnesota’s electoral history, plus the fact that Wisconsin and Michigan still both go for Clinton, I’m comfortable flicking MN into Clinton’s column. So she STILL ends up at 272. Even spotting Trump 5 percentage points in each state!
I just don’t see how Nate’s numbers could give Trump a 15.5% chance of winning this thing (and since I last posted, his chances ticked up slightly again). I just think 538 is starting to look a little ridiculous with their forecast numbers, when every other major forecaster (NYT Upshot, Princeton, Predictwise) has her at 91% or better and climbing daily.
Going beyond the raw polling numbers and looking at early voting (edge: Clinton), voter reg #s (edge: Clinton in FL anyway), party support for the candidate (edge: Clinton), cash on hand (edge: Clinton), GOTV ground game (edge: Clinton), TV ads (edge: Clinton), and media narrative (“look at this interesting and close horse race” has shifted to “look at the disaster the Trump campaign has become”), Trump’s chances are moving quickly from “slim” territory to “none” territory, Nate’s numbers notwithstanding.
Based just on polling numbers alone, I give Hillary a 94% chance. Add all of those other things into the mix, I kick it up to Wang’s forecast number.
There’s one advantage that Trump has. “We’re gonna lose ‘our countreh’ if we don’t vote for Donald Trump.” That’s what keeps him alive, and it’s a powerful force to be reckoned with.
Hillary has the same advantage. Surely you’ve noticed Clinton hammering on Trump’s “might not accept the results” as an attack on American democracy and the general Trump=Hitler vibe from her supporters.
It’s hard to imagine that I took a lot of statistics in college. My only defense is that it was a lot of years ago. So if this is a stupid question, I apologize.
Is there a point at which 538’s three forecasts (polls plus, polls only, and now cast) converge? OK, if Polls Plus is attempting to account for factors not considered in just the polls, maybe that forecast never matches the other two. But shouldn’t Polls Only and Now Cast eventually be identical?
PA has narrowed to a 4.4% Hillary lead on RCP so in your hypothetical Trump +5 scenario he would win. Obviously a very long shot. I believe that 538 puts some weight on the fact that there are still a lot of undecided voters, 7% in the RCP four-way. This may account for their comparatively high Trump percentage.
If you wanted to create a semi-plausible scenario for Trump winning it would be:
a) There is some bad news for Hillary perhaps based on Wikileaks.
b) That causes the undecided to break in his direction
c) Polls are underestimating his support by a couple of points perhaps because he has succeeded in mobilizing some white working class voters who don’t usually vote and don’t answer polls.
I don’t think this is at all likely but it doesn’t seem completely crazy.
Polls-Only and Now-Cast have been within a few tenths of a percent for the last week. Most of the difference is likely random, due to the percentages being calculated using Monte Carlo simulations.
Don’t forget, Nate puts more probability in the positive possibilities as well as the negative. Both Sam and Nate have the same number of median expected votes right now (330 vs. 332, Nate is actually the higher one). Nate would tell you that all of those positives that you list might put Clinton into landslide territory (19% chance of landslide, higher than the chance of a Trump win), while Sam would say all those positives don’t mean much, and landslide’s not that likely either.
Some of these are impossible to model, but I wonder if Nate includes things that theoretically could be, like early voting patterns and late registrations. Since the trend toward significant early voting is relatively recent, I’d guess there’s too little historical data to analyze it effective. Maybe in 8-12 years.
That would be the very definition of “picking polls”.
If you look at the polling history for New Hampshire, it’s been an extremely uniform lead for Clinton since early August. There’s no reason so suggest that Trump is suddenly going to close the gap and overtake Clinton there.
@TroutMan:
Well, it’s a little longer than 8-12 years.
For many Americans, Election Day is already here
Pew’s own research put it at 36% of all votes cast were mail in or early ballot box in 2012. That’s why some new wikileak/October surprise next week would make a lot less of difference than some might think.
It’s not the same advantage. There is no single factor in this race that is as powerful as the notion that the people are voting to ‘save’ ‘their’ society and their tradition of being at the apex of power in American society. What higher purpose could there be for civic action?
This country is still overwhelmingly white, and there are many whites who fear that ‘their’ country is slipping away from them and that this might be their last chance to save it. Not everyone who feels that way is necessarily overtly racist either. There are many Trump supporters or people who are open to voting for him who don’t really think of themselves as being racist. There are different levels of racism, and it operates on subconscious levels.
But this is what is driving his campaign and it is powerful stuff - so powerful that people can vote for him knowing that he has likely committed a series of sexual assaults that other people would be thrown in jail for, not paid income taxes over a 20 year period, and has basically debased women and numerous ethic groups with his rhetoric. What other candidate could withstand public scrutiny with such a checkered past and sordid campaign?
And this is the problem for Trump - any rise in the polls would have to be sharp and decidedly in his favor. I might be wrong on this one but I have the feeling that early voting is going to be significantly higher than in previous years - like possibly 40-50%. People are literally stressing out about this election and just want it to be over.
I simply disagree. For some people, yes, but for a good number of people "putting America’s Hitler in the White House " is just as much of a threat that requires civic action to stop. I mean look at yourself, you’re worried that even a Clinton win isn’t good enough - it has to be extra ‘real’ and be big enough to be an obvious public rebuke of Trump.
And once again, Dopers ignore Pennsylvania, where the latest poll is rather close: it shows Clinton/Trump/Johnson at 45/42/5. PA is not locked up.
Pennsylvania has a whopping 20 electoral votes. If Trump takes Pennsylvania, he probably wins the White House even without NC. (If both PA and NC go for Trump it’s game over.)
Of course we all hope for Hillary landslide. But Trump victory is still a real possibility.