I’ve seen this article several times, mainly linked by Sanders supporters on Facebook. It claims that a combination of exit polls and actual vote counts (similar to what is seen by some as suspicious in 2004 in Ohio) proves beyond all doubt that Hillary only beat Bernie due to outright election fraud.
Does anyone know anything about this? Is it reputable? Has it been debunked?
Thanks.
Patent nonsense without looking at it. She won, fair and square.
The fucking idiot in that article doesn’t know ‘versus’ vs ‘verse’
This is the legacy of 2000. We can’t have a presidential election anymore without the losing side’s supporters claiming it was rigged.
If (and when!) Trump loses, it’s gonna be a lawsuit shit show.
Totally absurd. There were only two or maybe three candidates, all of very roughly equal standing. There is nothing about Sanders to indicate that he had a 99.9999% chance of winning: he didn’t have the funding, the name-recognition, the supporters, or the machine. Frankly, he did a lot better than he “should have” and that’s a triumph he and his supporters can brag on.
Meanwhile, even the hottest hater of Clinton or Schultz can’t point to any examples of “voter fraud.” Sanders did well in caucuses, less well in elections. It is actually possible to argue that he benefited from the bias inherent in the caucus system, and did “better than he should have.”
In any case, numbers like “one in 77 billion” are just made-up horse-shit, without any conceivable validity. If you want to say “one in four,” we’ll listen. But if you go nuts, you lose our interest.
(It’s like the Indian swami or guru who could lift incredible weights, even into his 80s. Okay, show me a photograph of him lifting a 200 pound bag of rice, and I’m impressed. But show me a photograph of him lifting an 80 ton steel beam…and I can only laugh.)
What data was reviewed? The article is pretty vague.
Okay, I understand what reported vote counts are. But where did they get the data on “discrepancies”? And what the heck are “Various subsets of demographic polling data”?
The article also says that several different statistical models were applied to the data. Did the results obtained by different models give consistent outcomes? If not, what were the different outcomes? If nine tests were conducted and nine of them say Clinton won then a tenth one that says Sanders won becomes a lot less credible.
So, no one has a link to “a team of statisticians pointed out the following errors…” or anything of that sort? I’ll google around some more I guess…
Here it’s discussed on Snopes, seems to require additional peer review, the Snopes people seem skeptical but don’t go so far as to dismiss it out of hand.
The way I read the article was that there was a 1 in 77 billion probability that ALL of the reported anomalies were due to chance. So basically there was some intentional or unintentional cheating* but that doesn’t mean there was ever a hope in hell Sanders would get the nomination.
*For example, in Colorado Sanders was cheated out of a delegate until someone noticed the miscount the next day.
Garbage in, garbage out statistics. The same statisticians who proved the moon landings were fake no doubt.
It seems they have two arguments:
Clinton did consistently worse in exit polls than in reported polling results.
Clinton did far better in states which use electronic voting machines with no paper record; IOW, states where election fraud would be relatively easy to pull off.
The first, if true (I suspect cherry-picking), is interesting, but exit polls are of limited reliability, as the snopes article discusses in detail. In general, the election results matched up pretty well with the pre-election polling data, with the exception of Sanders’ big upset in Michigan. The authors don’t seem to dwell much on this inconvenient truth.
And when you search through the appendices to see which states use those theoretically hackable voting machines…they are almost all in the deep South, where people are stupid, so naturally Clinton did well there…just as the pre-election polls had suggested she would.Why no, I’m not bitter. Why do you ask?
Did the article also talk about the statistical probability that Clinton would be able to pull this off without any evidence?
That number also raises the question of how do you assign relative probabilities to all the different possibilities at each variation point, in order for it to compound exponentially into “1 in 77 billion”. And whether you are only counting discrepancies that would have had an actual effect in the delegate numbers. All that so that it’s not just a statement of a 1 in 77 billion chance that given the universe in June of 2015, in the next 365 days all things would happen in the exact fashion that they did … which means that there could be a whole lot of other scenarios where all this does not happen and Clinton still wins but the big difference is she picks Hickenlooper, or wins Oregon but loses Tennessee, or suchlike things.
She pulls off all her other nefarious deeds without evidence, so why not this too?
I wrote about exit polls in another post a while back–I’l see if I can dig it up. In the meantime, a summary:
Basically, exit polls are not in the least predictive, and are not meant to be. Despite assertions from Sanders supporters and statements from people like Robert Kennedy Jr., there is no reason to think that exit polls ought to accurately reflect the actual vote count. This is the case for several reasons, including the fact that very few polling places in any given state are actually surveyed.
In this particular case, there are a number of reasons to believe that exit polls significantly overstated Sanders’s actual level of support, among them:
–Younger people are known to be much more likely to respond to requests from exit pollsters than older people, and left-wingers are more likely to respond than more conservative people. The typical Sanders voter (very leftish and younger) is thus more likely to be included in an exit poll than the typical Clinton voter.
–It is really difficult to follow protocol at a crowded polling place (ask every tenth person…), leading to potential unconscious bias on the part of the people hired to carry out the polls–who may be more inclined to seek out people who are most like them. I don’t have a breakdown, but I have seen indications that many of the interviewers are young and white–the profile of the typical Sanders voter.
–In states where early voting is possible, indicators are that Clinton did especially well among people who took advantage of this opportunity. People who voted on the day of were disproportionately more likely to be voting for Sanders. Again, exit polls would inflate the actual share of Sanders’s votes as it would not include the people who voted earlier in the polling period.
–Exit polls do change as the day goes on as different waves of voters show up at the polls. There is some evidence that Clinton voters may have been more likely to vote later in the day, which would make an early Sanders edge in exit polling seem to disappear as the day goes on–which might look suspicious to someone primed to look for irregularities.
The other important thing to keep in mind is that exit polls do not truly represent independent events. The 1 in 77 million or whatever figure is based on the notion that the exit polls indicated fraud in New York, and in a completely separate event that the Illinois exit polls indicated fraud there, and so on and forth, and so we can multiply the odds in each state and arrive at some frankly overwhelming figure with an enormous number in the denominator. Of course, you can’t, because the biases everywhere all point in the same direction: the direction of exaggerating Sanders’s support. If we walk down the street and see 17 Chinese restaurants and no others, perhaps it’s a conspiracy of some kind, or perhaps we’re simply in Chinatown–or in China.
That’s a part of the answer. I’ll see if I can find that other post and link to it. Suffice it to say for now that there are very good reasons in general not to trust the exit polls, and very good reasons in particular not to accept that they demonstrate any kind of fraud designed to screw Sanders out of the nomination. Hope this helps.
We have less reason to be embarrassed.
There’s an approximately 1 in 77 billion chance that the article authors did not pull numbers out of their ass.
You’re probably thinking of Sri Chinmoy:
So what’s happening here is some statisticians have prepared a report that uses statistics to show a certain thing. Keep in mind, the correctly crafted statistics can show damn near anything you want. But since it was done with some degree of statistical professionalism, some group of statisticians is going to have to take the time to investigate and point out flaws.
Is it possible they will find no flaws? I guess, sure. But until we’ve seen this survive the rigors of that process it’s the equivalent of a scientific article that has yet to undergo peer review–and those can often show things pretty fantastical, until peer review shows not so much.
So I wouldn’t necessarily say we have easy, statistical “bullshit” cards to play on this, we have some rhetorical/logical explorations of why it might be bullshit, but give it time–the people that spend time refuting things in this field will eventually address this.
I just noticed that the website refers to a “Berkeley study” (as well as a “Stanford study”) pointing to election fraud by Clinton. The thing is, that the “study” is apparently authored by two people living in Berkeley, and not, as implied, a study done by the University of California in Berkeley.
Chances of that website being misleading: 1:1.