2018 exit polls getting defanged

[Marionumber1]

Ever since electronic voting machines surged to prominence around the year 2000, polls and election results have suddenly ended up at wide divergence from each other. And this has generally gone in the same direction: a "red shift", with vote counts coming out to the right of pre-election and exit polls. Exit poll discrepancies have been good forensic indicators of fraud in many past elections. For example, in 2004 (http://cavdef.org/w/index.php?title=2004_general_election) they pointed to a Kerry victory that was stolen for Bush, in 2006 (http://cavdef.org/w/index.php?title=2006_general_election) they indicated that Democrats won a larger victory than reported, and in 2016 (http://www.mintpressnews.com/donald-trump-warned-of-a-rigged-election-was-he-right/224326/) they pointed to fraud favoring Clinton in the primaries and then Trump in the general election.

Now it turns out that the exit polls might be getting modified so that the red shift goes away: https://www.msn.com/en-us/news/politics/exit-pollsters-make-changes-after-2016-breakdown/ar-BBPeybL Because the 2016 exit polls were so "inaccurate", and MSM groupthink dictates that this can only be due to polling error rather than election fraud, the exit pollsters are revising how they weight their surveys. That appears to be a euphemism for pre-adjusting their polls to the right, so that future red shifts will be covered up. The news outlet most upset by the persistent red shift is Fox:

Their plans were officially unveiled back in May. Arnon Mishkin, a consultant on Fox News’ decision desk, delivered a presentation at a pollsters’ convention that month, explaining why Fox had left the network consortium. Mishkin cited a “pro-Democratic skew” in the results of the traditional exit poll going back decades.

“This is a longstanding problem,” he said. “It’s been exacerbated to a certain extent in recent years, but it’s been around forever. And it’s something that we’ve collectively addressed.”

In fact, Fox and the Associated Press have both left the exit polling consortium to do their own post-election telephone/Web survey.

We'll see on Tuesday night what effect these changes to the exit poll will have. But I have a bad feeling that its goal is to help obscure a pro-GOP election theft operation.

[JackRiddler]

I know. I know. So hard to take this on, even if you have no denial about it.

[Marionumber1]

The Florida exit poll (https://drive.google.com/open?id=1QncXa6i-LAf_Xkp1oaAphRXK_BhL5Z58) says that 18% of black women supported DeSantis. Seemed pretty weird to me (and many others who observed the same thing) given DeSantis's support from overt racists. And the same poll has only 8% of black men voting for DeSantis, which breaks the general pattern of women voting more Democratic. What do people think of this? Is there likely some legitimate demographic explanation, or does it point to polls being skewed against the Democrats in an implausible way, perhaps consistent with trying to cover up a red shift?

[Elvis]

I for one am flummoxed. The PDF is CNN's, but who conducted/compiled the exit polls?

The Florida exit poll (https://drive.google.com/open?id=1QncXa6i-LAf_Xkp1oaAphRXK_BhL5Z58) says that 18% of black women supported DeSantis. Seemed pretty weird to me (and many others who observed the same thing) given DeSantis's support from overt racists. And the same poll has only 8% of black men voting for DeSantis, which breaks the general pattern of women voting more Democratic. What do people think of this? Is there likely some legitimate demographic explanation, or does it point to polls being skewed against the Democrats in an implausible way, perhaps consistent with trying to cover up a red shift?

[JackRiddler]

.

I noticed that too. Such a big difference between AA men and women cannot be correct. It can only be polling failure or fixing.

.

[Marionumber1]

I for one am flummoxed. The PDF is CNN's, but who conducted/compiled the exit polls?

The exit polls are arranged for by a media consortium that includes ABC, CBS, CNN, and NBC (formerly also AP and Fox until this year), and conducted by polling company Edison Research. It's the same group that's done the exit polls since 2004. CNN just happens to be a member of the media consortium that posted the exit polls on their website.

[Marionumber1]

.

I noticed that too. Such a big difference between AA men and women cannot be correct. It can only be polling failure or fixing.

.

In my opinion, there definitely seems to be something wrong with the poll. But the key question is whether it's just a one-off mistake or the result of trying to red-shift the polls in advance so that the disparity with the official results won't be as large. I'm inclined to believe it's the latter, and it suggests that more authentic exit poll results would show Gillum multiple percentage points ahead of what the official results say (rather than the virtual tie shown by the exit polls, which is in line with the official results).

[JackRiddler]

.

Can't tell without the raw data. Of course, can't necessarily tell with it, either. This seems to me a clumsy fix. A less noticeable and consistent shift in the white-people numbers would be better camouflaged. -- And how do we know they didn't do that anyway?! -- I lean to artifact, poor sampling, or sample low enough that this is random, outside MoE.

.

[Marionumber1]

.

Can't tell without the raw data. Of course, can't necessarily tell with it, either. This seems to me a clumsy fix. A less noticeable and consistent shift in the white-people numbers would be better camouflaged. -- And how do we know they didn't do that anyway?! -- I lean to artifact, poor sampling, or sample low enough that this is random, outside MoE.

.

That is a fair point. However, I imagine that a pre-adjustment of the exit polls to be more Republican was not a malicious "fix", but rather a way to make their poll more in line with official results. Methodological purity doesn't matter in a business like polling. If properly-done polls are consistently coming out to the left of vote counts, it looks embarrassing, and the pollsters will usually try to compensate by improperly altering their methodology for the next election. Basically, we have a feedback loop where corrupted vote counts become the baseline for polls. So it's possible that this weird result for black women voters was a manifestation of that, and it appears super clumsy because they're applying all sorts of bizarre tricks to meet their (red-shifted) baseline. But yes, we'll probably never know for sure. We know for a fact that they did red-shift the exit polls in advance (and that likely did include doing something to white voters), but the official explanation is that it's to make the exit polls more accurate, not less. Alas, exit polls were already difficult tools to discern election integrity evidence from, and it's only gotten harder.

[Elvis]

Oh, okay, Edison still; I haven;'t kept up and figured they'd have gone through some name changes or something. I harped to anyone who'd listen about glaringly obvious problems with Edison's 2004 exit polls. I'll just assume Edison is dirty and it's a fix.

The exit polls are arranged for by a media consortium that includes ABC, CBS, CNN, and NBC (formerly also AP and Fox until this year), and conducted by polling company Edison Research. It's the same group that's done the exit polls since 2004.

[Marionumber1]

Edison has always had a weird dynamic when it comes to their exit polls. At least from 2004 until 2016, they released exit polls that were more or less accurate, skewed slightly Republican if anything. Yet whenever those exit polls didn't match official results, they were quick to find fault within their own polls, and denounce any theory that the polls indicated fraud. In 2004, this became a lengthy academic debate between election integrity group US Count Votes and lead pollster Warren Mitofsky + his parade of "experts": http://cavdef.org/w/index.php?title=2004_general_election#Exit_polls

On a possibly sinister note, Mitofsky died in 2006 shortly after the debate essentially ended in favor of USCV, and Jonathan Simon hints in his book Code Red that it might be suspicious due to the timing. The book Votescam details Jim and Ken Collier's attempt to pierce the media consortium responsible for election predicting and reporting, which they came to believe had CIA connections, and they were constantly told by people including Mitofsky that their questions were "not a proper area of inquiry". Following Mitofsky's death, Edison stayed out of the limelight and just let the exit polls stand as they were, not bothering to comment on any red shifts. It was only in 2016, when the polls showed Bernie ahead in the primaries and Hillary ahead in the general, that they were finally forced to address the issue again. And that led to where we are now with the intentional pre-skewing of the exit polls in 2018.

[Elvis]

Edison has always had a weird dynamic when it comes to their exit polls. At least from 2004 until 2016, they released exit polls that were more or less accurate, skewed slightly Republican if anything.

Woops, you're right — I should have said that it was the election results that had "glaringly obvious problems." It was Edison's exit polls that proved it (at least mathematically, odds-wise).

Wasn't 2004 the year of the "shy Republican" who, it was claimed, accounted for the exit poll/election results discrepancies? That was rich.

[Marionumber1]

Edison has always had a weird dynamic when it comes to their exit polls. At least from 2004 until 2016, they released exit polls that were more or less accurate, skewed slightly Republican if anything.

Woops, you're right — I should have said that it was the election results that had "glaringly obvious problems." It was Edison's exit polls that proved it (at least mathematically, odds-wise).

Wasn't 2004 the year of the "shy Republican" who, it was claimed, accounted for the exit poll/election results discrepancies? That was rich.

Yeah, their explanation came to be known as the "reluctant Bush responder" (rBr) hypothesis: that Bush voters were less inclined to talk to exit pollsters than Kerry voters. Aside from the general implausibility of that idea, US Count Votes pointed out (http://www.electionmathematics.org/em-exitpolls/Exit_Polls_2004_Edison-Mitofsky.pdf) that Edison's own data contradicted this explanation in numerous ways:

• Response rates were highest in overwhelmingly Bush precincts, not what we'd expect if the Bush voters were the shy ones
• Maybe the Bush voters were only shy in the company of Kerry voters? The exit poll discrepancies went up as the percentage of Bush voters in a precinct did, which is inconsistent with that explanation too (but is consistent with the fact that more Republican precincts are often riper for fraud in favor of Bush).
• Supposing that a difference in Bush and Kerry response rates is the issue, matching Edison's data would require nonsensical response rates. Bush voters would need slightly higher response rates in overwhelmingly Kerry precincts and Kerry voters would need significantly higher response rates (sometimes over 100%) in the precincts with more Bush voters
• Although there was a high exit poll discrepancy in the presidential race, it did not show up for the Senate races. If there was a pro-Democratic sample bias, one would expect all exit-polled races to be skewed in that same direction.

The fact that the debate somehow continued for several months even after this report was released is pretty incredible.

[Elvis]

^^^^ Yes, it's coming back to me. At one time I had all that data saved, along with analyses. A few years ago in a computer crash, I lost my big 'electronic voting' folder containing all that and much more. From 2004 I tried to get conversations going about it, but almost no one seemed to think it was important, and I was sometimes called a "conspiracy theorist" for talking about it. I gave up.


I've heard people say in the same breath, "There's always been vote fraud, get over it!!"—then, "Besides, there's no way that e-vote fraud would ever be allowed to happen, and in the first place no American [cue America the Beautiful] would actually tamper with our votes!!"

[Grizzly]

Stand up fuckers!