Comments on FiveThirtyEight: Politics Done Right: If He Did It

酒店經紀人, 菲梵酒店經紀, 酒店經紀, 禮服酒店上班, 酒店小姐兼職, 便服酒店經紀, ...

2009-10-10T08:32:51.304-04:00

酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店經紀,
酒店打工經紀,
制服酒店工作,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
專業酒店經紀,
合法酒店經紀,
酒店暑假打工,
酒店寒假打工,
酒店經紀人,
菲梵酒店經紀,
酒店經紀,
禮服酒店上班,
酒店小姐兼職,
便服酒店工作,
酒店打工經紀,
制服酒店經紀,
酒店經紀,
菲
梵,

Hey Nate, Have you seen this photo essay (w/audio...

2009-06-18T05:03:02.169-04:00

Hey Nate,

Have you seen this photo essay (w/audio commentary) from the New Yorker covering one of Aahmadinejad's visits to Tabriz (Mousavi's "home town") in April.
It's about 2 minutes long. Well worth the time to give a bit of color to the rural discussion.

http://www.newyorker.com/online/multimedia/2009/04/13/090413_audioslideshow_ahmadinejad

It's only one data point, but it reminds me how little real information we're working with.

Cheers,
Diesel

Yes! This makes perfect sense! It's quite reasonab...

2009-06-18T02:28:39.999-04:00

Yes! This makes perfect sense! It's quite reasonable to wonder why Karroubi's votes went to the incumbent...until you consider ECONOMIC factors. NOBODY is talking about the fact that Mousavi is a total right-wing privatizer. He's basically the libertarian party candidate, if you will. Karroubi, on the other hand, is more of a traditional left-liberal.

Put it this way: if Dennis Kucinich ran for president in 2004 as an independent and got 10% of the vote, who do you think would get his votes in 2008? Obama, right? But Iran's politics aren't like that.

Imagine if you will that GW Bush had been an economic liberal...he passed single payer health care and expanded the welfare state while being hard right on foreign policy. Now imagine that the dems had nominated someone to the right of Obama on economics...like say, they convinced McCain to run as a democrat...so, Bush the right-wing socialist vs. McCain the middle-of the road reformer with right-wing economic policy. Wouldn't the Kucinich voters vote for this Bush?

This is a horribly tortured analogy. The fact is that the Iranian election is another fake "velvet revolution" to install an economic liberal. And NOBODY is reporting it. Be the one to stand up for truth!

EARTH TO NATE!!!! This is your worst analysis ever...

2009-06-17T23:08:24.278-04:00

EARTH TO NATE!!!! This is your worst analysis ever!

How many times do people have to tell you that the 2005 elections were boycotted by a large swath of moderates and thus not comparable to 2009...

How many times do people have to tell you the votes boxes were stuffed in 2005 and thus not comparable to any year prior or since...

How many times do people have to tell you that you like the base "common sense" knowledge to do a numbers comparison on the ground in Iran..

Nate, I love you man, but STFU on Iran or find a clue!

These results look even stranger once you take int...

2009-06-17T20:14:16.946-04:00

These results look even stranger once you take into account the fact that in 2005 many people boycotted the election, while in 2009 about 85% voted, many of which were those who boycotted the previous election. And as you might know, the boycotters are those who think that reformists are not liberal enough, we need more radical changes. So, as you can guess, when boycotters decide to vote, they definitely won't vote for Ahmadinejad.

Besided, 85% turnout means people are pissed off with the current president.

OK, fresh cup of coffee, I finished the chi-square...

2009-06-17T15:43:27.881-04:00

OK, fresh cup of coffee, I finished the chi-squares:

Ahmedinejad: 15.3
Razaee: 3.2
Korroubi: 28.1
Moussavi: 10.0

In a 9-row column, expected value (50/50 chance for over/under) for chi-square is 8 (the "degrees of freedom" number).
Moussavi's number is perfectly normal (p >25% of being that large).
Razaee's number is low but not freakishly so (p between 5% and 10% of being that low; low chi-squares mean data that is "too good", deliberately tailored to match the Benford percentages, but I doubt the Iranians have that kind of sophistication).
Ahmedinejad's number is on the borderline of statistical significance (5% chance of chi-square > 15.5, so his number has a p-value just over that .05 threshold of significance).
Karroubi's number is just horribly improbable.

jdk: been there, done that. The number of 7's in...

2009-06-17T14:45:43.168-04:00

jdk: been there, done that. The number of 7's in the Karroubi column using busiestday's data is 4.4 standard deviations from the expected, for well under a .0001 probability; now that probability should be multiplied by 36 since we have four candidates and nine first-digit-totals for each, and we have picked out the worst-looking one (that is, the probability that the worst of 36 comes out as a 1 in 10,000 chance is 36 in 10,000), but still we have over 99% confidence these numbers are bogus.
Above, I computed the chi-square which is a more appropriate approach to a table of data like this; the figure for the Karroubi column, 28.1, is way off the charts, negligible probability of this happening by chance. The other columns are not so bad. My impression of the way these vote totals were obtained is: the votes were switched between Ahmedinejad and Moussavi in most precincts where Moussavi was ahead; votes for the minor candidates were just reassigned to Ahmedinejad and low numbers for the actual candidate invented out of whole cloth.
But as has been expressed above, figuring out the how is not so terribly important. We do need to satisfy ourselves that yes, these numbers are nakedly fraudulent.

wv: fomanti, could this foment an anti-regime movement?

Benford's Law is really the only analysis that mak...

2009-06-17T13:01:14.739-04:00

Benford's Law is really the only analysis that makes any sense in this circumstance. (Others have already posted. As to "significance", the way to handle this is to see if any of the data falls outside 3 std of the expected percentage (hence a sign that it makes sense to look for a special cause (cf. Deming and Shewhart).

p-bar = expected percentage
N = number of obvervations (i.e., precints or voting districts)

Upper and Lower Control Limits =

p-bar +/- 3 times
[
[p-bar times (p-bar minus 1)]
divide by N
]

If there was fraud and the frauders were not sophisticated, the fraud will out itself in the numbers.

Joseph: "Is it really possible to reduce Iranian p...

2009-06-17T12:41:36.341-04:00

Joseph: "Is it really possible to reduce Iranian politics to this left-right continuum? I'm not saying it isn't but I'm skeptical that politics in such a different system and culture would necessarily be so easily reducable to such a Western formulation of politics.

I would argue that it's not really possible to reduce Western politics to a simple left-right continuum. There are at least four dimensions in play in Western politics. (social vs. fiscal, personal vs. private freedom, authoritarian vs. libertarian, liberal vs. conservative, etc.)

That said, the "conservative" vs. "reform" paradigm is how the Iranians appear to be classifying themselves. Therefore, for the sake of this analysis, we can probably accept that paradigm as having moderate legitimacy with how the voters are viewing the candidates.

Of course, to our western perspective, their "reform" candidates are still very conservative...

Opus 132 at 5:04 a.m. I would not suggest it's im...

2009-06-17T11:49:05.115-04:00

Opus 132 at 5:04 a.m.

I would not suggest it's impossible those who devised the reported numbers resorted to a convenient premise in case they needed defending. Your suggestion of the Ballen/Doherty study serving that purpose seems plausible; indeed, at seems to fit the notion of 'steal from one column - add to another' Nate is posting about here.

Moreover, since you and I agree, and all posting comments here agree with Nate's Iran Position 2.0, that, whatever the rationale(s) behind the reported numbers, it's wildly improbable they accurate report the faithfully recorded of the votes of several tens of millions of diverse Iranians, then the exercise we're engaged in here, led by Nate, is whether the thinking behind the numbers chosen to be reported are rooted in something more solid than the effluent to which, I take it, you interpreted I was referring.

Even effluent comes to its condition from some other. Indeed, were one to credit the confabulators with pure novelty, one would also expect of anyone capable of such imagination to come up with a product that's also more plausible at some level. MY point is that, since the device itself (the invention, the product, the fiction) is so implausible, it's MORE worthwhile to consider the motivations for reporting such a transparent fiction, than it is to root through the scat of the invention in pursuit of its origins.

But I have to concede that the choice as to which pursuit is more worthwhile is a matter of personal preference. I'm comfortable with your hypothesis until its credibly shot down; but it still doesn't go to motivation for the fiction, and I'm saying we have quite a bit of evidence, including the dimensions of the implausibility, to enable us to hypothesize as well about motivation. I assume you'd not be offended by my notion that there's possible value in attempting to extend the findings of mathematical exercise in an effort to derive meaning.

Is it really possible to reduce Iranian politics t...

2009-06-17T10:36:04.239-04:00

Is it really possible to reduce Iranian politics to this left-right continuum? I'm not saying it isn't but I'm skeptical that politics in such a different system and culture would necessarily be so easily reducable to such a Western formulation of politics.

@Opus 132 "Regarding the low votes figures for th...

2009-06-17T09:00:17.686-04:00

@Opus 132

"Regarding the low votes figures for the two minor candidates, Rezaee and Karroubi, which are being cited repeatedly as "proof" of cheating..."

The TFT data looked at the nationwide vote and doesn't seem to capture the veyry high votes Mousavi and Karroubi would have gotten in their hometown/province/ethnic enclaves. The poll simply doesn't take into account the density of votes in places like Lorestan.

What's the regional/provincial spread on those polled by TFT?

Actually, are the second round provincial results ...

2009-06-17T05:18:36.652-04:00

Actually, are the second round provincial results for the 2005 elections available anywhere? The only results table I can find seem to only have provisional national vote totals.

Regarding the low votes figures for the two minor ...

2009-06-17T05:09:36.691-04:00

Regarding the low votes figures for the two minor candidates, Rezaee and Karroubi, which are being cited repeatedly as "proof" of cheating: the Ballen/Dougherty "Terror Free Tomorrow" Western poll PREDICTED this almost exactly... they would each be in the 1% - 2% range. So q.e.d, you can't CITE this as proof of cheating. Maybe they DID in fact have votes removed from them... but there is a logical, simpler explanation.

Well,how about this for an explanation of the reported numbers:

Instead of the "pulling numbers out of his ass" as Joan phrases it,the designer of this fraudulent vote used the totals of an existing poll. That's right,why not use the figures from that American poll that showed that Ahmadinejad would win big,make up supporting detail,and eliminate counting the actual ballots.The resultant phony numbers would look legitimate thanks to the work of those blessed Americans Ballen and Doherty.

Voila,the (almost) perfect crime!

bruce b does have one point, beyond the rhetoric a...

2009-06-17T04:42:54.900-04:00

bruce b does have one point, beyond the rhetoric and the failure to read the previous posts on the site. It would be interesting to look at a regression analysis of the 2005 first round -> second round results.

@busiestday: on the other hand, the Moussavi colu...

2009-06-17T04:37:46.004-04:00

@busiestday: on the other hand, the Moussavi column has a chi-square of 10.0, meaningless (about a 25% chance of a chi-square that large).

Leave aside that several commenters here seem so d...

2009-06-17T04:22:50.889-04:00

Leave aside that several commenters here seem so deeply mired in their own self-geekitude, they're missing the barely-concealed sarcasm in Nate's post, has it not occurred to anyone that the [not an ayotollah] Khameni picked the number he did had nothing whatsoever to do with an attempt to steal the election fairy away from Iranian voters, but instead was deeply personal?

While the aparatchick Ahmadinejad was still in school and working waaaHAAAY-behind-the-lines of the Iran-Iraq war, all of NAA K, Mousavi, Rafsanjani, Khatami and every single other one of the various semi-ayotollahs who has actually had his name mentioned in the dispatches of this uprising, were all in the upper reaches of the Khomeni Kabinet. It seems clear that Mousavi is far from a 'natural' reformer; he worked very well as P.M. through the toughest time in post-revolution Iran, and with NAA K as President, with reports being consistent that NAA K hated how unrelentingly disputatious Mousavi was - which implies that since Mousavi could not have survived so long there, indeed THRIVED, without the support of Ayatollah Khomeni, he of all the surviving founders is Mister Most Likely To Be Considered Chief Rival to NAA K - and was and is among if not the biggest sophist of the bunch.

Again: the reports are that NAA K utterly despises Mousavi, and so despises the idea of having to deal with him, plus very likely fears his standing with the mullah mafia.

So: it makes sense that the 63% 'award' was very likely, MOST likely a transparent slap in Mousavi's face.

And no unarguable facility with multiple regressions analysis is going to compete with someone pulling a number out of his ass to shove in his opponent's face.

(To those who would counter that there's no way kindly looking Mousavi would ever sacrifice the lives of kids to win a personal grudge match, consider just the reported incidents of literally hundreds of thousands of citizens including kids - PARTICULARLY kids - sacrificed by Iranian Republican government on the watches of Khomeni and Khameni - including mass murders of helpless prisoners of the mullahtocracy and long-term "chain murder" binges.]

As ugly as are the street assaults, the real battle is taking place in Qom -- and no matter who wins, the post-uprising murders will go on for weeks if not months.

@busiestday: I calculated the chi-square for the ...

2009-06-17T04:13:03.002-04:00

@busiestday: I calculated the chi-square for the Korroubi column (after making a horrible algorithmic error, hence my deleted post) at 28.1, and at df 8 there is only a .005 chance of chi-square greater than 21.955; even applying Benferroni's rule of thumb (there are four columns, and we are picking out the worst-looking, so there is a 2% chance that the worst of four comes out with chi-square > 21.955) we have over 98% confidence that the Null Hypothesis is false (Null Hypothesis meaning the preconditions for Benford's Law, numbers resulting from a random count starting at zero with no preferred stopping points).

What do you think about stratfor's analysis that w...

2009-06-17T03:16:12.483-04:00

What do you think about stratfor's analysis that western focus on sources which spoke english and lived in the urban centers is driving a misunderstanding of internal iran dynamics?

http://www.stratfor.com/weekly/20090615_western_misconceptions_meet_iranian_reality

Yes, Iran is poorer than Canada. But tabulating el...

2009-06-17T02:52:23.034-04:00

Yes, Iran is poorer than Canada. But tabulating elections with paper ballots is a labour-intensive activity, and cheap labour is abundant in Iran. So even if they have less money with which to hire poll workers than Canada, those workers are much cheaper. With unemployment at 12.5% officially (and surely higher unofficially), there is a much larger reservoir of poll workers to draw on.

As for Canada having an advantage in its smallness... err... Canada is the world's largest country, and has one of the lowest population densities.

Secondly, Iran did not count all of the votes in 3 hours. According to IRNA 66% of the ballots had been counted by early Saturday. That story is dated June 12 7:10 EST, or 4 am June 13th in Iran.

http://www.reuters.com/article/topNews/idUSEVA14340720090612?feedType=RSS&feedName=topNews

Most of Iran's polls closed at 9:30 am EST, or 6:30 PM Iranian time. Where does that 3 hour number come from? In a few cases where there were long lines, a 6 hour extension was granted. However, in the majority of cases, there were 9 hours to count, not 3.

Source: http://www.npr.org/templates/story/story.php?storyId=105287388&ft=1&f=1004

By the end of Saturday, 80% of the vote had been counted.

http://english.aljazeera.net/news/middleeast/2009/06/2009612195749149733.html

Are these numbers "too fast"? You claim this is not typical for Iranian elections. I think you are pulling this claim out of your rear end.

In the 2005 Iranian election, the polls closed at 22:22 June 17th in Ireland (see below for source), but that was partly because of a 4 hour extension. So polls closed in Iran at 2 am on June 18th, though officially at 10 pm on June 17th.

http://www.rte.ie/news/2005/0617/iran.html

60% of ballots were counted at some point on Saturday, when this article was released. Of course a result took longer to announce because it was a multi-candidate race, and a small difference in votes would have a large impact on the outcome (since only the top two guys go to the runoff).

http://www.washingtonpost.com/wp-dyn/content/article/2005/06/17/AR2005061701344.html

According to the Seattle Times, 66% of the vote was counted in 2005 by June 18th 12:47 am Seattle time, or 12:47 PM Iran time (10 hours after polls closed, and 14 hours after they closed in most of the country).

So yes, the vote count was a little slower in 2005. However, the difference was not radical and could easily be explained by different allocation of resources, and the added difficulties of a multi-candidate race.

The 3 hour figure (which should be 3 and a half hours) assumes that vote counting didn't start right away, but forgets that there was a 6 hour extension. It also ignores that only 66% of the vote was counted in this time, not all.

If you want an example of fast, Canada usually has close to 100% of votes counted by 3 am.

Regarding the low votes figures for the two minor ...

2009-06-17T02:07:19.639-04:00

@hosertohoosier: "There is simply no way to count ...

2009-06-17T00:59:41.272-04:00

@hosertohoosier: "There is simply no way to count 45 million something ballots that fast in a country like Iran. They can not afford to hire that many people to make it that fast." I find statements like the frustrating because I agree with the point the author is trying to make, even though their statements are actually false. There most certainly are methods that *might* work to count 45 million ballots in three hours in Iran, and many of those methods are things the fairly wealthy country of Iran could afford to do. Iran almost certainly hired enough people to do it, because they needed that many poll workers in the first place, and one fast method is to count in the polling place as soon as polls close.

I doubt Iran did implement a plan to really count the vote in three hours, and I doubt they would have succeeded in producing announced semi-official results in that time had they tried. I therefore doubt Iran can demonstrate that the vote count is credible, and the onus should be on the Iranian government to demonstrate that they did as they claimed. But if you base your argument on the false idea that such things are impossible, you are easily defeated, because then all you are asking Iran to do is show that there is a method they could have used to get a count in three hours. It is, at the outside margin *possible* to count 45 million ballots and more or less report the totals in 3 hours, so you aren't asking for enough.

@juvanya; "Also, in every previous election, it took days to get results." That's a valid point. What actions other than fraud can the Iranian government claim to have taken that speeded up the count. If none, then why were previous counts slower?

(If the Iranian power structure weathers this storm, I am confident they will never again make the mistake of reporting results so quickly.)

@hosertohoosier: That's because they have reliable...

2009-06-17T00:01:43.504-04:00

@hosertohoosier:
That's because they have reliable opinion polls and small areas to go with. There is simply no way to count 45 million something ballots that fast in a country like Iran. They can not afford to hire that many people to make it that fast.

Also, in every previous election, it took days to get results.

Nate an interesting analysis. I believe however t...

2009-06-16T23:46:41.353-04:00

Nate an interesting analysis. I believe however that the mechanics were slightly different.

I think that the reformist voters you have identified, those who previously supported Karroubi and Rafsanjani, did in fact change their votes, but not to Ahmadinejad. I believe that seeing that Mousavi had a legitimate chance of getting elected these voters moved to him precipitating a groundswell and an actual margin of victory for Mousavi of close to the 60% he claims.

Faced with this fact, Ahmadinejad then stole the election by the transparent but effective device of simply moving some 20% of the total vote from Mousavi's column to his own. When Ahmadinejad did this there was probably some preference for changing the results in rural provinces, that were harder to monitor, rather than in Tehran. This would account for the massive, and entirely unbelievable shift of Karroubi support to Mousavi.

Faced with this fact, Ahmadinejad then stole the election by the transparent but effective device of simply moving some 20% of the total vote from Mousavi's column to his own. When Ahmadinejad did this there was probably some preference in changing the results in rural provinces that were harder to monitor rather than in Tehran. This would account for what the massive, and entirely unbelievable shift of Karroubi support to Mousavi.

Regarding the hand-counting in 3 hours thing, a fe...

2009-06-16T23:17:15.125-04:00

Regarding the hand-counting in 3 hours thing, a few points. Firstly, you silly Americans, there are lots of countries that count paper ballots (Canada, for instance), where the result of elections are known within a few hours. You don't need to count every vote in order to call an election (from a media standpoint).

Iran may also have had advance polls, such most of the results had been counted already by election day.

As for the comment "but counting paper ballots would require hundreds of thousands of workers", err... yes, you are right. That is how you do it. Again, using my Canadian comparison, we had 170,000 ballot counters in 2006. That was for 15 million votes (I don't know how long they worked, but they got paid $185 a pop).

Iran's elections had 40 million votes, so assuming they worked at Canadian speeds, they probably needed about 450,000 vote counters (probably paid less than $185).