The Devil Is in the Digits - Statistical analysis of the Iranian election results

The proposition in the book Blink by Malcolm Gladwell, if I understood it, might add a viewpoint to this discussion about mathematical analysis. Sometimes our unconscious spots fraud before any analysis does.

0 Replies

@dlowan,

dlowan wrote:

Let's keep my goddam back off any but one thread!!!

Okayokayokay ... I'll back off.

0 Replies

@fbaezer,

fbaezer wrote:

Only that the probability is so low, it's really unthinkable that it wasn't a fraud.

"Unthinkable"? Arguments from lack of imagination tend to be unsound. "I can't think of any way this result could have come from a clean election, therefore it has to be fraud" has the same logical standing as "I can't think of any way this critter could have evolved, therefore God created it."

1 Reply

As i pointed out in another thread, it no longer matters if there were fraud or not. A sufficient number of people are convinced that electoral fraud took place that it has become the functional truth, not matter what the ultimate truth is.

0 Replies

@Thomas,

Except that is clearly not a supporting argument but a mere description of the low likelihood of the result being natural. How it could be natural isn't something hard to conceive I imagine that anyone who can read and understand the thread can conceive of it being coincidence.

But like you said, we bet our lives on less than that sometimes.

2 Replies

@Robert Gentel,

Well, evidently fbaezer is having trouble conceiving of it being coincidence. And the trouble with publication bias is that you don't know what the relevant likelihood is.

1 Reply

@Thomas,

I don't understand the argument about publication bias changing the likelihoods. It certainly makes the argument seem less powerful if they cherry pick the tests with deviation from the models but unless the other tests can rule out fraud by having been passed I don't get how it would impeach the mathematical conclusions of one test.

0 Replies

OK, I'll concede to Thomas: there's 99.99% chance it was a fraudulent election.

He reminds me of the joke about the actuarian, the physicist and the mathematician in Scotland. They see black sheep while on the train.
-Sheep in Scotland are black -says the actuarian.
-Some sheep in Scotland are black -corrects the physicist.
-Some sheep in Scotland have a visible black half -says Thomas, the mathematician.

1 Reply

@Robert Gentel,

Robert Gentel wrote:

.......I have not examined their evidence in detail, so I can't vouch for their conclusions..........

Neither can the 2 authors of the study you cite: neither their statistical analysis nor their psephology (they are political scientists, not mathematicians) bears much scrutiny. In fact they appear to be suffering from a common delusion known as the "gambler's fallacy":

Quote:

....Subjects act as if every segment of the random sequence must reflect the true proportion; if the sequence has strayed from the population proportion, a corrective bias in the other direction is expected. This has been called the gambler’s fallacy.

The heart of the gambler's fallacy is a misconception of the fairness of the laws of chance. The gambler feels that the fairness of the coin entitles him to expect that any devi¬ation in one direction will soon be cancelled by a corresponding deviation in the other. Even the fairest of coins, however, given the limitations of its memory and moral sense, cannot be as fair as the gambler expects it to be. This fallacy is not unique to gamblers. ...

There's certainly a preponderance of evidence that at least some rigging took place with those election results, but quantification (like the laughably "precise" 99.99% posted on this page) of the probabilities requires far more rigorous analysis than undertaken in either the original article or by anyone here, though only Gentel so far admits it.

0 Replies

@Robert Gentel,

Robert Gentel wrote:

Except that is clearly not a supporting argument but a mere description of the low likelihood of the result being natural. How it could be natural isn't something hard to conceive I imagine that anyone who can read and understand the thread can conceive of it being coincidence.

But like you said, we bet our lives on less than that sometimes.

Exactly! Said clearly and more succintly than the excerpt I just posted - and keep in mind that the author got a Nobel prize in economics for his discovery.

0 Replies

@fbaezer,

fbaezer wrote:

OK, I'll concede to Thomas: there's 99.99% chance it was a fraudulent election.

As I explained earlier, even that is not necessarily true: Your statement here depends on implicit assumptions you made about any tests the authors might have run on the election results, but whose results they did not publish in the Washington Post. These assumptions, however, may be wrong.

Granted, if the authors took the election data and only ran the two tests they described in the Post, their results would support your conclusion. But if they ran two hundred tests on them (which is quick and easy to do with modern statistics software), if two out of those two hundred tests turned up suspicious-looking results, and if the authors then mentioned only those two in the Washington Post article, your reasoning becomes dead wrong. Given enough tests to run, a few of them are bound to come up with odd results -- just by mere chance, and even if the election was clean in the first place.

You don't have the information that you need to distinguish between the first and the second scenario you just described., Therefore, you don't know what the actual likelihood of a fraudulent election is, given the data described in the Washington Post and elsewhere.

1 Reply

@Thomas,

Here's where empirical experience meets logic.

Doubtless, there is doubt and all your logic is correct.

I'd take bloody serious notice of Fbaezer's experience in this area, though.

Sounds to me like there's damn good reason to think it fraudulent.

Makes me think of the beginning of Chaucer's Wife of Bath's Tale.

I strongly agree with whoever said that a passionate telling off of Iran by the US or Britain, especially, would be highly counter-productive.

1 Reply

@dlowan,

Grrrrr -- I'm not saying the Iranian was clean. I suspect it wasn't, too. I'm only taking exception to this "99.9%" stuff, and against the argument that 'if it's unthinkable that this is a clean election, it probably wasn't.'

dlowan wrote:

Here's where empirical experience meets logic.

As a trained experimental physicist, I, too, have experience with the crunching of empirical data. Part of this experience is with an all-too-human habit so common we even coined a term for it: 'to torture the data until they speak.' If you think you know what the data ought to be showing, and the straightforward interpretation of the data doesn't support it, you will stubbornly try yet another statistical model to fit the data to, and yet another test to run on the data, until the data says what you expect it to say. It is astoundingly easy, even for people as brilliant as myself, to utterly fool themselves with statistics.

It is precisely this experience that sets off my warning lights when I hear this "it is unthinkable" line of reasoning. Everyone, including myself, expects the Iranian elections to be rigged. Everyone running those statistical tests on the distribution of digits expects a positive outcome. This raises a probability that amateur statisticians will torture the data until they speak, and repress any thoughts that that's what they're doing. It's a probability, not a certainty, but the probability is much greater than 0.1 percent.

3 Replies

@Robert Gentel,

the number of identical digits could a random rare event

probability indicates likelihood not intent

the invigilators may have used random number generators not wmd's (whoppers of mass digits)

the dealing is done notwithstanding the veracity of the report's observations about number patterns

0 Replies

@Thomas,

Maybe it's your manner that sets me off, then.

1 Reply

@Thomas,

Have you considered the possibility he was pulling your leg, and was astonished to have it come off in his hand?

1 Reply

@Thomas,

I'm sure that no one in this discussion is willing to give much heed to what i post here--oh well, so sad, too bad. However, earlier in the thread, i posted a link to some people claiming that the election were rigged, and they knew that from the statistical data--but they were offering a theory of statistical error completely at variance with the one presented in the Washington Post article. Nobody appears to have gotten the joke, which is because, i suspect, no one bothered to look at the link. Torturing the data, indeed. This sort of exercise expects to find the evidence it is looking for, because it assumes, a priori that the results were fudged. That makes this exercise no different than the assumption of fraud which RG charges the media with displaying.

Once again, it no longer matters if the election were fraudulent, because so many people believe that to be the truth, that it functionally has become the truth.

1 Reply

@dlowan,

Easily possible. My manner does that to people.

1 Reply

@roger,

He being Fbaezer being a tad ironic?

That's what I thought.

0 Replies

I still think most of the election ballots were not examined, counted and tallied. Their claim that the small sample recount produced three million illegal ballots. How could they examine that many ballots, determine that they are illegal, from a small sample in such a short time? I don't think they counted any ballots in the so-called recount. I think they just pick mistic numbers.

BBB

1 Reply

The Devil Is in the Digits - Statistical analysis of the Iranian election results

Related Topics

Quick Links

My Account

able2know