538: Voters really did switch to Trump at the last minute

[pnwmom]

http://fivethirtyeight.com/features/voters-really-did-switch-to-trump-at-the-last-minute/?ex_cid=story-facebook

Donald Trump’s somewhat surprising win has forced many political analysts to wonder: Were we wrong all along in thinking Hillary Clinton had the upper hand, or was late-breaking movement to Trump part of the reason why polling averages missed his upset Electoral College victory? There’s certainly evidence that the polls underestimated Trump’s support in crucial Midwest states. But the latest wave of the Institute for the Study of Citizens and Politics panel survey that my University of Pennsylvania colleague Diana Mutz and I have been overseeing is now complete, and it provides new evidence that voters did shift to Trump in the final weeks of the campaign, too.

SNIP

At first glance, it might seem as if Clinton in October 2016 was in roughly the same position as Obama was in October 2012, at least with respect to the distribution of votes nationally: Both enjoyed margins of 7 percentage points among exactly the same group of people. But there were critical differences, even beyond the fact that the geographic distribution of support is crucial in making one candidate president. First, the number of undecided respondents in 2016 was 21 percent, significantly outpacing the 15 percent we saw in 2012. Second, our 2016 survey ended on Oct. 24, leaving two full weeks before the Nov. 8 election for people’s minds to change. There was still a lot of time on the clock.

And while most people’s support remained the same, the changes we did observe were consequential. Consider the table below, showing panelists’ support in the October 2016 poll versus their support in the post-election poll, which took place from Nov. 28 to Dec. 7. Eighty-nine percent of the 1,075 American adults reported the same preference in both waves, whether it was for Clinton (38.0 percent), Trump (35.2 percent) or neither (15.8 percent). But among those who did move, Trump had the advantage. While no one moved from Trump to Clinton, 0.9 percent of our respondents moved from Clinton to Trump. Although that 0.9 percent isn’t a lot, those changes are especially influential, since they simultaneously reduce Clinton’s tally and add to Trump’s. If there were a comparable swing in the national electorate, 1.2 million votes would move to Trump.

Trump also outpaced Clinton among people who were previously undecided or third-party backers, with 3.1 percent of respondents moving from those categories to Trump while just 2.3 percent did the same for Clinton. Clinton also saw 3.1 percent of her October supporters defecting to third-party candidates or becoming undecided. Trump lost just 1.7 percent.

In all, Trump picked up 4.0 percentage points among people who hadn’t been with him in mid-October, and shed just 1.7 percentage points for a net gain of 2.3 points. Clinton picked up a smaller fraction — 2.3 points — and shed 4.0 points for a net loss of 1.7 points. . . .

SNIP

[Kashkakat v.2.0]

1. I dont understand all the verbiosity - how does that explain exit polls being so far off in those

"crucial swing states" run by Repub governments... more so then in Dem states.

If these polls referenced above were done so far after the election, how does it rule out that certain small segment of people who want to align with a winner? If they lie one way (too embarrassed to say their voting for DT) who's to say they wont lie the other way?

[pnwmom]

2. A few days before the election, there were still large numbers of undecideds.

On the day of the election, those undecideds mostly split for DT. Also, she lost some voters whose support had been weak.

[triron]

4. That doesn't explain the exit poll 'phenomena'.

Only shenanigans does.

[LonePirate]

14. I think we need to accept that exit polls are now fundamentally flawed.

We need to stop believing they are scientifically representative or accurate.

[triron]

19. Please read Ron Baiman's paper.

Even if they are 'inaccurate' (which I do not believe), to be skewed so improbably in one direction is not explainable except as voter disenfranchisement/election fraud by the party toward which they are skewed.
Any other explanation seems grasping at straw.

[LonePirate]

26. It's certainly possible (likely?) that both election fraud occurs and exit polls are flawed.

I think cases can be made for both of those occurring. Election fraud is probably the most under-reported and uninvestigated threats/crimes in our country. We just need to be very careful about using problematic exit polls to prove election fraud.

[triron]

30. The assumptions made

about turnout models causing exit poll inaccuracy are at least as suspect as the exit polls results. It is an ad hoc explanation.

[Yupster]

39. The problem is self selection

If the exit poll table is 100 feet from the polling place there is plenty of space to avoid speaking to them if you want to. Exit pollsters are also often younger women and there are often banners on the table identifying the exit poll site by organization sponsoring it. The exit poll questionnaires also have as many as 40 questions to answer.

So who would tend to self select toward that table and who would tend to walk in the other direction? It's just not a random sample. Angry older white men don't go near an exit pollster and a young woman is unlikely to chase them down when there's a much friendlier face right behind them. They are supposed to interview every fifth voter. Turns out that's a lot different than interviewing one of every five voters.

Back in 2004 there was an article in one of the newspaper magazines (NYT?) by an exit pollster trying to explain why she thought the exit polls were so wrong. She said that all through the day she felt she was not getting a good sample of respondents. She said she made an honest effort to question everyone she was supposed to, but some people just blew past her with a wave and she felt during the day it was skewing her results. When the exit polls turned out so wrong, she said she wasn't surprised and she in a tiny way blamed herself for it.

The fact is that angry older white male voters are not going to stop and talk to a young woman working for a media conglomerate at the same rate as other people. So who are the Republican's most reliable voters?

[mythology]

43. Baiman's paper is worthless because he doesn't understand the flaws of exit polling

Exit polls have many factors to overcome, and anybody selling some idiotic theory based on the "raw data" of an exit poll should be considered the same level of charlatan as those selling theories on how climate change isn't real or vaccines result in autism.

https://www.thenation.com/article/reminder-exit-poll-conspiracy-theories-are-totally-baseless/

http://fivethirtyeight.com/features/ten-reasons-why-you-should-ignore-exit/

[Kashkakat v.2.0]

32. Uh, no. You got it bass ackwards. First we get the secure, verifiable voting system

in place, THEN we see how accurate (or not) the exit polling is.

IMHO we don't need to wait around for "proof" that electronic election fraud has been committed - its alarming enough that proof exists that it COULD BE.... Isnt that enough??? You didn't see bankers wait around to see if anyone would steal money from electronic banking systems - no they immediately instituted safeguards, security, audit procedures capable of detecting fraud.

Then of course theres that whole mess with Crosscheck - how many showed up to vote and were either turned away or cast a provisional ballot that would not be counted - people thinking they voted but their ballots not counted would account for osme of the discrepancy.

[LonePirate]

34. Yes, your sequence is how it should be but people are performing it in reverse now.

[triron]

40. I think you got it right

[forthemiddle]

35. When States like Florida

Had close to 70% early voting, how can exit polling be correct?

I asked this question before the election and never heard a good explanation.

[BzaDem]

38. Exit polls included a weighted component of early voters calculated through a pre-election poll.

Which ironically can make them more accurate (since pre-election polls are typically much more accurate than exit polls).

[triron]

41. Wonder about this as well.

If exit polls only for same day voting no way Trump won Florida!
Hillary had huge early vote advantage.

[jmg257]

17. It doesn't. They weren't. nt

[mythology]

22. The exit polls weren't off.

People insist on trying to use the "raw data" without understanding that is useless.

[Yo_Mama]

23. Exit polls aren't true samples. They do not contact (much less get a response)

from enough people to extrapolate the results from their raw data. What they do, instead, is go through their sample and adjust their sample to make it look demographically like the sample that they believe will vote, and that is substantially based on the previous election cycle exit polls, plus voting record, for Likely Voter polls.

There are several sources of error contained in this method, but the most obvious (and most lethal) source of error is a change in turnout between elections. In essence, the method they use would be very accurate if they already had the exit polling for the election they are trying to predict, but of course they don't have that yet.

Turnout changes from election to election are hard to detect in advance. When both candidates have high unfavorables, as they did this year, they can be a bear. No matter how one tries to adjust the Likely Voter screen, it will be wrong.

Checking against exit polling shows that turnout changes, high third party candidate involvement and most specifically, high turnout in some areas shifted the results. But it also confirms that the voting tallies were accurate.

[realmirage]

3. Fuck 538. They got it wrong, this is an attempt to save face

I'll never trust that site again

[pnwmom]

5. Polls cannot measure last minute changes in mind because the polling

has to stop a few days before the election.

[Ace Rothstein]

31. And there were a lot of undecideds this year.

[zipplewrath]

6. Actually, they were "getting" it right

They were one of the few that was showing her losing ground all the way up to the election. She was down to about 55% chance of victory on election day. That was down from upwards of 70% in the late summer and early fall. Nate even mentioned that her "lead" was soft going in because of all of the "undecideds". Remember, she only lost the EC because of about 85,000 votes spread over about 4 states. That's REAL tight. Basically, Nate got it right.

[HoneyBadger]

11. Nate does not just poll

He adds secret sauce to the actual results in an effort to be predictive. To what extent being predictive actually cook the books, who knows. I consider it more art than science. I.e. him getting it right is akin to flipping a coin and getting it right.

[Guy Whitey Corngood]

20. He doesn't do polling. He bases his predictions on other people's polls and uses

statistics.

[HoneyBadger]

13. Getting it right at the end of Election Day is meaningless

Getting it right a month before is the real trick.

[zipplewrath]

28. Statistical flaw

The classic problem in the use of statistics for single events is that it is a logical fallacy that statistics have anything to do with the outcome of a single event. We know the statistical probability of flipping a coin, but just because it comes up heads once in a while doesn't mean the predictions for tails was wrong.

In other words, the problem is with the reader. Everyone should have been concerned when her chances were falling. When they fell below 60%, there was a real reason to worry. More importantly, when she "only" had 44% of nominal support, that left a lot of room for things to "break" his way. When you have 49%, it's hard to lose. I got concerned when she fell to 55% but never thought he could win that many Midwestern states. And he barely did. The flip side is that the basic reason he won one of them, is why he won more.

[Renew Deal]

8. Who got it right?

[Hassin Bin Sober]

16. Hmm. That's the same response I got when I posted that Silver was warning...

... the Clinton margin could disappear with a slight polling mistake. Silver warned it wasn't in the bag.

'Fuck Nate Silver, he just wants a horse race!!! Sam Wang has it at 99%'

Or words to that effect.

I think too many people got caught up with the big 65/35 number on Clinton's side and confused it with a polling number when it was really "odds" or chances of winning.

I think he even said something about the numbers being great odds for betting on a game but not so much when the fate of our country is at stake.

What really hit home was he gave Trump the same Chance as the Cubs winning the series. *gulp*

Like everybody else, I got caught up with the "in the bag" feelings. But when I read Silver's warnings and I started looking at the margins in the battleground states, I got that sinking feeling I had in 2004. About a week before the 2004 election, I read about all the anti gay marriage legislation the repigs put on the ballots around the country and I thought "uh oh"

[Buckeye_Democrat]

42. Nate Silver was one of the few prognisticators who warned about...

... the much larger percentage of "undecided" voters this time compared to 2012. He wrote articles about why he gave Obama a much stronger probability of victory over Romney compared to Clinton over Trump, despite pretty similar polling margins, and undecided voters were the common theme of the articles.

He also had a disadvantage because there weren't many polls done in the final days before the election... after Comey pulled his BS stunt.

Compared to other sites, like the Princeton Consortium (Wang), Silver was far less optimistic... and many people on DU hated him for it back then!

Silver might have been wrong too, but he was "less wrong" than the others.

[ucrdem]

7. Some did, no doubt, just not enough to mean anything.

Comey found nothing and said so. His tittle tattle might have inspired a clueless independent or two but not enough to put him over so Team T relied on their now-familiar vote rigging to produce that miraculous 6-point leap in WI for instance.

Comey would have loved to sink Hillary the way he's sunk many another Dem -- the Democratic mayoral candidate of my own town, for instance -- but he didn't have the stuff to do it. So he gave a kayfabe performance and now the RW noise machine is pretending he took her out. He didn't.

[pnwmom]

10. The evidence says otherwise. DT's win in key swing states was so narrow --

less than 1% -- that this small shift accounts for it.

[ucrdem]

15. Not really. In WI he won by 1 point, but he leapt from 40 to 48 points to do it:

Comey is a poster boy but the real thuggery went bump in the night. As usual.

[pnwmom]

36. So? If the Comey bombs hadn't been dropped his leap wouldn't have been that big. n/t

[mtnsnake]

9. If anyone switched to Trump at the last minute

it was probably a large number of Gary Johnson's supporters and also some of the supporters of that other 3rd party dipshit, Jill Stein.

[HoneyBadger]

12. Yes, but supposedly Nate's polls predict this

3rd party ALWAYS pivots at the end. What date defines the end is the $100,000 question. Traditionally you get an October surprise. I.e. campaigns attempt to come up with a big win in early October, the idea being later is simply too late.

[Tiggeroshii]

18. Polling indicated how quickly the electorate was to switch at any miniscule piece of non-information

In September I think, the state by state polling on rcp looked like election day.

[yardwork]

21. So Comey's unethical behavior handed the election to Trump.

[oasis]

24. Comey is responsible for any disaster Trump brings on our nation/planet.

His interference in the election makes him America's number one deplorable.

[Dem2]

25. Because they got the permission slip from Comey

People, they wouldn't have gone out for him if they weren't encouraged at the last minute.

Uggh, nobody is ever going to understand what happened and how knowing he suddenly had a chance was all he needed to secure a very narrow victory. Comey. Jail. Comey. Jail. Jesus, we are so stupid.

[hollowdweller]

27. Private server was her achilles heel.

After the whole Whitewater thing which was totally bogus, you would have thought that Clinton would have done ANYTHING to avoid the appearance she was hiding anything after her SECOND run and her history.

I know in some way she thought having the private server would PROTECT her from the very thing she had to endure, but I'm not exactly sure why she thought that.

Then she made it worse by deciding herself which emails to delete, which played right into their hands and created suspicion.

People were tired of the Benghazi witch hunt, and without the emails this would have played out early in the campaign.

Now we dems have to suffer the consequences.

[yallerdawg]

33. We can't just believe her?

She said it was just easier to do it this way.

If she could do it over, she wouldn't do it this way.

That worked for me Day One!

[Justice]

29. COMEY did it.

[Generator]

37. This defies all common sense and human nature

It's more of the equally hated bullshit. If you had to wait until the last minute with a clear danger to sanity and all human decency to decide that oh yes her e mail server is the same as you know rounding up people and deporting them and sexually molesting women and disrespecting military dead and the disabled and on and on. We no longer can believe exit polls. So we have no check on anything. That's how they knew it was stolen in the Ukraine. We ain't winning shit again unless we actually demand fair elections. Nothing to see here. Move along. Democrats make me despair as much as anyone. Always making nice with Nazi's.

https://www.thenation.com/article/reminder-exit-poll-conspiracy-theories-are-totally-baseless/

Oh look the Nation! I hear they LOVE Putin. Oh well.