Fairleigh-Dickinson, rated A at 538 came with a poll that goes to October 2 showing Clinton 10 points ahead of Trump.
http://elections.huffingtonpost.com/pollster/polls/fairleigh-dickinson-ssrs-25961
Agreed! I decided to see what would happen if non-college-educated whites had 25% greater than expected turnout, AND all minority groups had a near 25% depressed turnout (presumably due to voter ID laws and Trump’s poll goons), AND a 10% third party vote. Clinton still barely wins with 272 EVs. Granted, the result depends a lot on what states in which the demographics are altered. Still, it’s encouraging to know that the 538 model still has Clinton winning even the turnout is historically bumped in Trump’s favor.
Big bumps on 538 today:
Polls-plus ↑ 0.9% (73.5%-26.5%)
Polls-only ↑ 1.3% (77.5%-22.5%)
Now-cast ↑ 1.1% (83.3%-16.7%)
From several Emerson polls, including some slightly odd results (NV tie, FL Trump +1, AZ Clinton +2) and a UNF Florida poll (Clinton +3).
Also of note – as of now, in the “polls only”, Clinton’s ‘weakest’ state for her path to 270 EVs is NH at 77.5%. Next weakest is CO at 79.1%. All others are over 80%. When they’re all over 80 I’ll feel pretty good.
What the heck? Now-cast is now 85.1%-14.9%, and Iowa is blue in Polls-only at 51%-48.9%.
Latest, from a HRC +11 MI poll:
Polls-plus ↑ 0.8% (74.7%-25.3%)
Polls-only ↑ 0.9% (78.3%-21.7%)
Now-cast ↑ 1.8% (85.1%-14.9%)
IA is blue in polls-only, and AZ is blue in the Now-cast.
Are these new numbers reflective of the 1st debate’s results, or is something else going on?
Just more polls from the post-debate period. Maybe also reactions from Trump’s craziness afterwards.
Your guess is as good as mine.
My WAG is that every day now brings new bad news for Trump, often in multiple doses, and it’s having a cumulative effect.
And finally we seem to have had a hiatus from the once-steady drumbeat of all those ‘raises questions’ and ‘bad optics’ Clinton Foundation and Clinton emails stories, so it’s a drumbeat of bad Trump news with nothing to counterbalance it.
I think 538’s forecast might just be catching up to reality. Every other forecast has been much higher than Silver’s for a while.
This poll isn’t on 538, though.
I think it is – check the 5 Oct update at 10:23 AM (ET).
I don’t think the polls are even taking into account the whole tax revelation yet, are they?
At the rapid rate new polls are being added to 538, and the way the PO forecast keeps climbing for Hillary, I fully expect the PO to be higher than 80% by the end of the day tomorrow. That would put Silver’s forecast in line with NYT Upshot, Daily Kos, HuffPost, PredictWise and within 10 points or so of Wang (who is currently at 92% Hillary).
The 538 PO has gone up 12 points since Sunday.
ETA: The latest from NH (Suffolk) has Clinton +2, and IN (Public Opinion Strategies) Trump +5. This moves the PO up to 78.7 for Hills.
Of course the town hall debate is Sunday, which could change the direction of things. Or speed it up. So I guess change the velocity of things is more scientifically accurate.
There’s a 17.6% swing between Polls-plus and the Now-cast for Hillary in Arizona, which seems like an awfully large disparity. Even with the Now-cast showing Arizona blue, I still believe Trump will take it.
It’s probably only useful to look at the Now-cast when the election is happening now-ish.
I look at the Now-cast for trend only, but the swing between it and Polls plus is still striking to me.
If you look at Arizona’s internals on 538, the first 2 post-debate Arizona polls were just added, at (adjusted) Clinton+4 and Clinton+6. The series of Trump-up polls were from before the debate and the new data is suddenly overriding the national-poll based extrapolations from the pro-Trump polls. So it’s legit given the data, but the question is whether those new polls are outliers or will hold.
The difference nationally between the Now-Cast and Polls-Plus is still hovering around 10 percentage points, with Polls-Only in between.
I’d have expected the numbers to start converging more by now. The difference between now-cast and polls-only is that they lie to the computer and tell it the election is today. The differences between polls-only and polls-plus, according to 538:
- Polls-plus combines polls with an economic index; polls-only does not.
- Polls-plus will include a convention bounce adjustment; polls-only will not.
- Polls-plus starts by assuming that likely voter polls are better for Republicans; polls-only makes no such assumption. Both models revise this assumption as more data becomes available.
- Polls-plus subtracts points from third-party candidates early in the race, while polls-only does not.
- Both models employ a regression that is based on demographics and past voting history. But polls-only weights the regression less and places less emphasis on past voting history.
- Polls-only accounts for more uncertainty than polls-plus.
- Polls-plus and polls-only will tend to converge as the election approaches.
The state of the economy is pretty much baked in by now, and the convention bounce is long gone. So that takes care of (1) and (2). Pretty much everyone’s using likely voters now (only 2 of 538’s 25 most highly weighted national polls are RV rather than LV) so (3) seems to be inconsequential at this point. It’s no longer early in the race, so kiss (4) goodbye.
I’m not sure what to say about (5) or (6), but if 4 out of 6 differing factors are now out of play, (7) should be happening: there should be convergence.
And of course, with just 33 days left until Election Day, and early voting already having started in some states, the uncertainty due to the remaining time between now and Election Day should be shrinking. It should still be a real thing, but a bit smaller than it was a month ago.
Latest from 538:
Polls-plus ↑ 0.4% (75.7%-24.3%)
Polls-only ↑ 0.5% (79.3%-20.7%)
Now-cast ↑ 0.3% (86.5%-13.5%)
Due to some trendlines towards HRC for some Republican leaning national polls (Rasmussen and LA Times).