2016 Postmortem
Related: About this forumWhat would the US do if we saw this type of exit poll/election result discrepency in Russia?
tonyt53
(5,737 posts)RKP5637
(67,112 posts)Txbluedog
(1,128 posts)People lie in them all the time
NWCorona
(8,541 posts)Igel
(35,359 posts)There were some serious miscalls nobody remembers because it mutes the meme.
They went fuzzy when voting stopped being one-day in-person. Because all exit polls require a set of assumptions about what the electorate is like. Then you take the results from a few polling places and extrapolate out.
They went fuzzy because the groups doing the exit polling went to fewer and fewer polling places. That made the extrapolation riskier--if those polling places were slightly anomalous, then the pollster's screwed.
If the early-voting/mail-voting electorate isn't like the in-person electorate, if the in-person electorate is sampled in a bias way or doesn't tell the truth, then the assumptions fall flat.
The fix is to see how a number of sampled polling places *did* vote. Then you can do an analysis to sort out how each of the demographics actually voted, and go back and fix your assumptions. You assume that 18% of the electorate is going to vote for a person, that the Af-Am electorate in your polling place reflects general Af-Am views but they don't, you assume that when old white male (R) turn out at a 48% rate in precinct XYZ that it means all old white male (R) turn out at that rate but they don't because in that precinct the dog catcher was 'one of them' and boosted the turnout, and you're seriously in statistical "there be dragons here" territory.
If the assumptions suit one side in the election, then the unadjusted, erroneous results are assumed to be true. Why? Because confirmation bias rules. Numeracy would help, but only a little. Numeracy doesn't overwhelm confirmation bias.
And there's also random error. You can't account for it, you just try to calculate it and say that you're 95% sure (if you're assumptions are correct) that your predictions are accurate to within some margin of error. Perhaps +/- 2.5% (depends on the sample size, among other things).
Yes, exit polls work elsewhere. Where 90% or more of the polling stations are sampled. Where most people tell the truth. Where the sample rate isn't a really small number. Where the exit polls aren't updated, so the initial results are inherently and insanely biased but taken as gospel by those the bias favors.
Then there's weirdness from having poll workers who don't follow instructions.
LoverOfLiberty
(1,438 posts)with facts.
RKP5637
(67,112 posts)J_J_
(1,213 posts)Over the past decades, exit polling has evolved into an exact science. Indeed, among pollsters and statisticians, such surveys are thought to be the most reliable Exit polls in Germany, for example, have never missed the mark by more than three-tenths of one percent. Exit polls are almost never wrong, Dick Morris, a political consultant who has worked for both Republicans and Democrats, noted after the 2004 vote. Such surveys are so reliable, he added, that they are used as guides to the relative honesty of elections in Third World countries. In 2003, vote tampering revealed by exit polling in the Republic of Georgia forced Eduard Shevardnadze to step down. And in November 2004, exit polling in the Ukraine paid for by the Bush administration exposed election fraud that denied Viktor Yushchenko the presidency. Robert F. Kennedy, Jr.
leftinportland
(247 posts)Exit polls are used worldwide for the specific reason that they are accurate within a percentage point or two. They are used worldwide to validate the integrity of an election. Exit polls were used extensively in the US before electronic voting. I remember in the 60's, this is how elections were called because it took too much time waiting for the votes to be counted.
stonecutter357
(12,697 posts)NCTraveler
(30,481 posts)TalkingDog
(9,001 posts)NCTraveler
(30,481 posts)TalkingDog
(9,001 posts)The exit polling in the states with no paper trail is divergent. (and the votes are skewed)
The exit polling in the states with a paper trail is consistent. (and the votes are consistent)
It's called critical thinking.
NCTraveler
(30,481 posts)intheflow
(28,504 posts)They're not interested in facts, they're interested in believing what they're told to believe. They've been told that Clinton won. What more proof do they need? Critical thinkers need not apply.
TwilightZone
(25,485 posts)When one of the sources involved is "My Dad's View of Election Fraud Study", one might want to take the whole thing with a grain of salt.
LexVegas
(6,101 posts)geek tragedy
(68,868 posts)Smarmie Doofus
(14,498 posts)aren't exactly proud of their primary vote.
Hmmm.... I wonder why that would be.
rjsquirrel
(4,762 posts)as ours usually are?
This is absurd. Exit polls are not votes. We don't poll to choose election winners. No one cheated Bernie out of anything. His people just didn't show up or there aren't enough of them or both.
Hundreds of people would have to be in on any scheme to pull off voter fraud on the level crazed Bernie Bots are suggesting. Maybe thousands of people. They'd all have to keep a secret that could ruin the party and its nominee to the grave with them. Why would a candidate who locked up the nomination mathematically in March even take such a risk, even if you believe all the bulshit about how corrupt she is? (And it is misogynist and right wing bullshit.)
Prove it in court or stop whining.
Absurd. The delusional level of denial among the last Bernie soldiers makes me think you're all Susan Sarandons. You never stood with the party. You're not my allies as a progressive. And I don't care whom you support in the general, except to note how very privileged you must be to think Trump would be alright and/or the same as Clinton. It's a sad spectacle watching the marginal hard left campus elite spin these webs of conspiracy theory.
Clinton is our nominee. I was proud to support her and I remain so. I'm just as progressive as any Bernie supporter. But I am not a naive 20 year old voting for the first time or a campus radical spinning tales of "revolution" around a seminar table and mocking the ignorance of millions more actual voters who came out for Clinton, including sizable majorities of poor and minority voters.
DemocratSinceBirth
(99,714 posts)fleabiscuit
(4,542 posts)rjsquirrel
(4,762 posts)squirrels eat nuts!
DemocratSinceBirth
(99,714 posts)fleabiscuit
(4,542 posts)J_J_
(1,213 posts)Around the world, exit polls have
been used to verify the integrity of elections. The United States has funded exit polls in Eastern Europe to detect fraud. Discrepancies between exit polls and the official vote count have been used to successfully overturn election results in Ukraine, Serbia, and Georgia.
http://electiondefensealliance.org/frequently_asked_questions_about_exit_polls
Response to J_J_ (Reply #50)
rjsquirrel This message was self-deleted by its author.
k8conant
(3,030 posts)rjsquirrel
(4,762 posts)Any scheme the could produce 3.5 million vote difference would involve hundreds of people keeping a secret that would destroy the party for a generation.
That's why this conspiracy theory bullshit is so laughable. It's the same as birtherisn or 9/11 trutherism. It's a joke except you're apparently being serious.
The Democratic Party should never have allowed Sanders to run in our primary. Leftist nonsense and he never was a committed democrat..
I don't want your alliance or your vote. We will be fine without CT in our party.
k8conant
(3,030 posts)I live near Charles Town so I'll be glad to let everyone there know how you feel!
rjsquirrel
(4,762 posts)And no one from Connecticut either.
J_J_
(1,213 posts)nt
Response to J_J_ (Reply #79)
rjsquirrel This message was self-deleted by its author.
djean111
(14,255 posts)"Democracy" and assert that it is obvious who the Russian establishment meant to "win". And maybe talk about sending election observers. That is exactly what would happen.
fleabiscuit
(4,542 posts)djean111
(14,255 posts)snooper2
(30,151 posts)fleabiscuit
(4,542 posts)djean111
(14,255 posts)And just calling something "ridiculous" in order to deflect/minimize/shut down conversation does not work very well at DU - the most it gets is a Full Ignore. It is about as effective as the ROFL guy. As in not effective.
rjsquirrel
(4,762 posts)You mean like respecting candidates who win massively larger numbers of actual votes than your guy?
Didn't think so. Hence the excuses for your undemocratic desire for a coronation and/or coup d'etat in the name of a "revolution" that doesn't exist and never did. Revolutionaries don't run for office under a major party banner.
Godhumor
(6,437 posts)We do not recommend using exit polls to look for fraud in other countries. In fact, we devote an entire chapter to exit poll weaknesses in our USAID manual on election fraud.
Initial exit polls are often very wrong before further waves are polled and compiled. Again, because they're not intended to get the result of the election right, but to, after the election is complete, work out how different groups of people voted.
rjsquirrel
(4,762 posts)JTFrog
(14,274 posts)Skinner has made it clear that RT is not an acceptable source on DU. Guess folks will just have to try to cram it all in before the 20th. I had really been looking forward to the end of this ridiculousness today. *sigh*
Oh well, just a couple more days.
brooklynite
(94,745 posts)frazzled
(18,402 posts)I decided to take just one state to check: Mississippi. Averaging out the demographic results, Hillary Clinton looked to be preferred by voters in the 80-85 percent range For example:
Are you:
Male (36%) Clinton 79% Sanders 19%
Female (64%) Clinton 85% Sanders 15%
Which age group do you belong?
17-44 (40%) Clinton 72% Sanders 27%
45+ (60%) Clinton 90% Sanders 9%
http://www.cbsnews.com/elections/2016/primaries/democrat/mississippi/exit/
Clinton won that primary 82.6% to 16.5%, right in line with the exit poll (completely within the margin of error). Yet this chart claims a 10.4% discrepancy in Clinton's favor. Completely false.
I don't have the time or interest to investigate each of these state numbers. But I call bullshit right now. Get a grip: big discrepancies would have been commented upon both in the press and by the campaigns. They were not.
NurseJackie
(42,862 posts)MyNameGoesHere
(7,638 posts)I am going to ask Jakov to finish it for me.
TwilightZone
(25,485 posts)we have to look at the sources and wonder why you would believe that RT and "Turn-Trout" are terribly reliable.
Igel
(35,359 posts)Depending which "US" you're looking at.
If you like Russia and the exit polls favored the anti-Russian candidate, you'd think they were flawed. If they were biased towards the candidate you liked, you say the elections were rigged. This is just confirmation bias talking.
If you're a politician and want to say something negative about Russia, you'd say they were rigged. If you were a political ally of Russia, you'd say that the exit polls were funded by USAID and were intended to embarrass Russia. This is less confirmation bias and more playing political games, thus speaks political expediency.
If you're a statistician, you'd want to figure out where the problem was. You might assume that the polls were rigged. Or not. But perhaps they sampled a very small, biased group, in which case you'd chuckle and send them a copy of a 1st-year stats textbook. You might assume that voters weren't trustworthy (meaning, in other words, that the voters didn't find the pollsters trustworthy). There are contexts in which poll questions need to be very carefully worded.
Otherwise you get results like having a set of people think bin Ladin is still alive. But that a large portion of those thinking bin Ladin is alive also thinks the US raid in Pakistan was a fraud, because bin Ladin was dead long before that. That poll measured not belief in bin Ladin's alivedness, but trust in the media and government. You have to know what the poll data means.
randome
(34,845 posts)People who don't have an understanding of statistics or computers are just positive that things are being done by shadowy forces out to destroy their nominee! It's the only explanation!
I'm still refraining from criticizing Sanders for the remainder of the week but his supporters have always been his worst enemy and I have no problem calling this the bullshit it is.
ChisolmTrailDem
(9,463 posts)but not when the charge is being leveled at one of ours.
I remember the Bartcop support here when he was all over bush's ass about voting fraud but apparently it's impossible for this to happen with a Democrat.
vintx
(1,748 posts)It makes it very clear how it was possible to shift this party to where the republicans used to be, as that party shifted off the charts into the far right fringe
It's still very sad that so many here seem perfectly ok with that.
ChisolmTrailDem
(9,463 posts)...when it's Clinton.
randome
(34,845 posts)But since exit polling has always been unreliable, most don't see the saucer full of aliens come to probe us that you want us to believe in.
ChisolmTrailDem
(9,463 posts)...in them as do you, apparently.
LanternWaste
(37,748 posts)And here I thought Literalism was applicable to Fundy's Only.
(Metaphor, allegory and satire are literary constructs you may wish to look into)
CorkySt.Clair
(1,507 posts)Sad.
randome
(34,845 posts)It's always the things we can't see or understand or prove that prove there is a conspiracy.
"Here we go 'round the mobius strip, the mobius strip, the mobius strip..."
CorkySt.Clair
(1,507 posts)Our resident crank with their walls of texts and links to fever swamp sites. I won't name names but...
NorthCarolina
(11,197 posts)the candidates amenable to their goals, to massaging the poll results and controlling the media message. There will be no revolution as long as we cling to a two party system wherein both parties are manipulated by the same forces.
djean111
(14,255 posts)NorthCarolina
(11,197 posts)is a NEW political party that follows the mold set by the Sanders campaign, that relies solely on small donations, and that shuns corrupting PAC donations, and Corporate donations. The pool of those willing to run for public office for the sake of doing the peoples work and not self aggrandizement would be much smaller of course, but IMO that is what will be necessary. The two party system has outlived it's usefulness for far too many in this country...time for change.
djean111
(14,255 posts)The Democratic Party has moved to the right, BIG time, and I cannot go with it. I will not.
NorthCarolina
(11,197 posts)I have been a registered Democrat since 1976, but I no longer recognize this party at all. It's sad, but it's not the end of the world. History shows that political parties come and go with the times. Because the deep level of control over the process by moneyed interests, the two parties we have now have managed to survive longer than they probably would have otherwise, and certainly longer than they should have for any party in a truly "representative" democracy.
ChisolmTrailDem
(9,463 posts)Tarc
(10,476 posts)The exact same trash was hidden 6-1 a few minutes ago, yet this one stands, is the reason most won't be sorry to see the backside of the jury system.
CorkySt.Clair
(1,507 posts)Despite their whining to the contrary. I've seen vile attacks on Democrats and Hillary allowed to stand in recent days and the votes aren't even close. It's their last stand before the board is returned to people who are interested in seeing Trump defeated.
onenote
(42,768 posts)I picked one example at random: Mississippi, where the chart claims the ballot box results are 10 percent higher for Clinton than the exit poll results.
First I looked at the actual voting data: There were 220,790 ballots cast in the Mississippi Democratic primary (including a few minor candidates).
Then I looked up the exit poll results. http://www.cnn.com/election/primaries/polls/ms/Dem
According to the exit poll data, 64 percent of the voters were women, 36 percent men. And women favored Clinton over Sanders by 85% to 15% and men preferred Clinton over Sanders by 79% to 19%.
Thus, according to the exit polls, 141,306 votes were cast by women of which 120,110 were for Clinton and 21,196 for Sanders.
It also means 79,484 votes were cast by men and 62,793 were for Clinton and 15,102 for Sanders.
Adding up the numbers, the result of the exit poll suggests a final tally of 182,903 for Clinton and 36,298 for Sanders.
That's 82.8 percent for and 16.4 percent for Sanders.
Then I went back to the actual results:
182,447 for Clinton (82.6%)
36,348 for Sanders (16.5%)
In other words, the exit poll results were actually quite accurate and the claim in the OP's chart that the exit polls showed Sanders doing ten percent better than the actual result is bullshit.
I didn't bother to check any others, just that one. But if the creators of the chart will lie about Mississippi, there is no reason to think they haven't lied about other states.
On edit: I'm aware that the claim will be made that the final exit poll results are adjusted to match the final actual results. But that puts the onus on those making the claims to show the unadjusted numbers and give a source for them.
whatthehey
(3,660 posts)AL exit polls
http://www.cbsnews.com/elections/2016/primaries/democrat/alabama/exit/
If you take 60% female at 80% Clinton and 40% male at 73% Clinton you get 77.2% overall Clinton
And we got....
http://www.nytimes.com/elections/results/alabama
77.8% Clinton
I'm pretty sure their sample size had a greater than 0.6% MOE so in other words an exact match, not a 15.7% skew....
Stop believing bullshit lies that confirm your bias. Check the fucking data dammit. Took me less than 2 minutes.
stillwaiting
(3,795 posts)For some reason we have to adjust our exit polls now to match reported results.
That is some strong democracy for us.
It is the original unadjusted exit polls that show the large discrepancies.
whatthehey
(3,660 posts)Florencenj2point0
(435 posts)pdsimdars
(6,007 posts)Our state department uses them to measure election integrity in other countries. In some countries when the exit polls deviate by more than 2%, people flip out. Apparently here in the US, no one cares. 12% deviation in NY?
felix_numinous
(5,198 posts)are ignoring these stats.
pdsimdars
(6,007 posts)that they decided to cancel all the exit polls in CA. I guess they didn't want any more evidence than was already out there.
Nye Bevan
(25,406 posts)Is this being investigated?
Rex
(65,616 posts)That like never happens.
charlyvi
(6,537 posts)as much as ours do.
JoePhilly
(27,787 posts)Lord Magus
(1,999 posts)They're made-up conspiracy theory nonsense. Exactly what Russia Today likes to peddle in.
http://www.thenation.com/article/reminder-exit-poll-conspiracy-theories-are-totally-baseless/