Statistical Suspicions In Iran's Election 512
hoytak writes "An expert in electoral fraud, professor Walter Melbane, has released a detailed analysis (PDF) of available data in Iran's controversial election (summary here). While he did not find significant indications of fraud, he does note that all the deviations from the predicted model are in Ahmadinejad's favor: 'In general, combining the 2005 and 2009 data conveys the impression that a substantial core of the 2009 results reflected natural political process... [These] stand in contrast to the unusual pattern in which all of the notable discrepancies between the support Ahmadinejad actually received and the support the model predicts are always negative. This pattern needs to be explained before one can have confidence that natural election processes were not supplemented with artificial manipulations.'" In related news, EsonLinji notes reports in the Seattle PI and other sources that the US State Department has asked Twitter to delay system maintenance to prevent cutting off Iranians who have been relying on the service during the post-election crisis. And if you would like to help ease the communication crunch, reader RCulpepper tips a blog post detailing how to set up a proxy server for users with Iranian IP addresses.
Re:Statistical nothing (Score:5, Informative)
We hand count around 10M votes in Canada in a few hours each federal election (which is around once a year these days....) You can say "well, that's Canada and this is Iran", but Iranians have the same hands Canadians do. (Well, minus those cut off due to Sharia, if Iran practices that.)
There's a good chance the election was manipulated but that's no indication at all.
Re:Statistical nothing (Score:5, Informative)
They somehow managed to hand-count ~40M votes in a couple hours. It doesn't take a brain surgeon (or a statistician, in this case) to realize there's something fishy going on.
How so? I believe the way it works in Canada is that ballots are counted at each polling station and parties are free to have a representative oversee the election process. This ensures that we have an unofficial count a couple of hours after the polling stations close. (See The Electoral System of Canada [elections.ca], on page 34 of the PDF)
The official count comes, by law, up to seven days later, but it usually doesn't differ from the unofficial count.
Re:Come on, It's Iran already (Score:5, Informative)
While he did not find significant indications of fraud
QED. The null hypothesis was not rejected, therefore your study determined nothing. Speculation is not science.
If you know anything about statistics... (Score:3, Informative)
Re:Statistical nothing (Score:5, Informative)
AFAIK the official line was that the boxes were sealed and were brought to a central location for counting.
This was after the elections observers from the opposition parties were kicked out of the polling places, of course.
ProxyBox Virtual Appliance (Score:5, Informative)
Mirror 1 [exstatic.org]
Mirror 2 [128.210.109.29]
Proxies:
Squid installed and listening on ports: 7, 13, 53, 993, 995, 3128
Polipo installed and listening on port: 8123. Polipo is routed through Tor.
Tor: port 9050 (a socks5 proxy)
Ziproxy: Port 8080 (good for low bandwidth connections. It recompress images & text.
Socat: Must be run manually, but listens on port 443 and routes through Squid.
SSH enabled, listening on ports 22,80,2222,22222
2 Users: root:#iran and iran:election. If you enable ssh to the world, change the root password (passwd). This should enable ssh tunneling.
-
I created this for people on Fark who were having problems with squid. Everyone here shouldn't have a problem. It's a bare bones (netinst) debian install with all the above installed and setup.
I did NOT put ACLs in because there are reports here: http://iran.sharearchy.com/ [sharearchy.com] that the ACL list is actually blocking some people in Iran.
And could one of the mods please change to the coral cache of Austin's website? He's already getting DDoS'd by Iran all this morning. Slashdot isn't going to help anything.
If any /.ers would like to help make it smaller, better, faster (VPN?), jjarvis98 at gmail.com
And you're free to inspect it to make sure I'm not trying to r00t you.
Re:Proxy (Score:3, Informative)
Re:If you know anything about statistics... (Score:2, Informative)
...you know that a small random sample of the population tells you what the general population is like to a very high degree of certainty.
Statistics allows you to extrapolate results from a small sample set if, and only if, the the entire sample set follows a known statistical model.
Re:Proxy (Score:4, Informative)
I did email him twice but got no response. I also tailed my squid logs all night and nobody used it. I would like to help out here but am not much use if no one can find my proxy. Oh well.
Re:Slashdot (Score:2, Informative)
http://slashdot.org/submission/1021265/Grassroots-PetitionOnlineorg-taken-down-by-DDos [slashdot.org]
It seems peitiononline.com is under ddos attack. What I thought was interesting is that their number two most popular petition is
Investigation into crimes committed by Ali Khamenei
Re:The results match pre-election poll (Score:5, Informative)
The poll was done by Terror Free Tomorrow: The Center for Public Opinion and New American Foundation. The Washington Post merely did an article on the findings from the poll.
From the survey [terrorfreetomorrow.org] linked to in the article:
TFT and KA use telephone interviewing instead of face-to-face research in Iran because of the political and social constraints inside Iran. Face-to-face interviewing in Iran can be difficult for interviewers who risk possible prosecution and imprisonment. Face-to-face interviewing also poses issues related to access to households and respondents due to social considerations. Access to female respondents across the Middle East can be challenging.
I'm not sure how much better over-the-phone polling is in Iran. Many in Iran are leery of being called by random strangers over the telephone asking them political questions. Whenever we call our relatives in Iran, we are extremely careful with what we say over the phone. More to the point, when you have a brutal regime and some random person calls and asks: "Who will you vote for in Presidential Elections?", I wouldn't be surprised if they answer in one way and vote in another.
I won't dismiss the findings of this survey outright - they did conduct a scientific polling, something that I haven't done. It's just difficult taking the survey very seriously when what you see happening in real life - thousands and thousands of bloodied protesters taking the streets and demanding change - and compare it with a polling sample of 1001 Iranians, as stated in their Methodology section on page 25 of the pdf document. I'm also thinking back to both the entrance and exit polls in the 2004 U.S. elections, where John Kerry was said to have won by a large margin, only to find that the opposite had happened.
I think it is evident that I am quite anti-Ahmadinejad and anti-Mullah and especially anti-Arab when it comes to my ancestral country. But I will concede that he won if more information is released and it points in favor of his victory.
Re:This reads like electoral interference to me (Score:5, Informative)
Just to clarify, Iran is a mountainous and largely forested country inhabited neither by Arabs nor Arabic speakers.
Re:Proxy (Score:1, Informative)
http://www.boingboing.net/2009/06/16/cyberwar-guide-for-i.html has information about passing proxy data. Cory Doctorow, as obviously shown in his teenager targeted book "Little Brother" is a master of cyber protesting and fighting the man. Proxies just aren't that useful if you simply cut off Internet access with a shovel. The real world works in the same way it did a thousand years ago, twitter or no.
Remember Remember, the Naieveity of Novemember, the way people are dragged about and shot. I see no reason your ignorance should not be point'd out. http://imgs.xkcd.com/comics/security.png
Re:The results match pre-election poll (Score:3, Informative)
Seriously, that's an incredulous omission to make, nevermind the fact that the poll itself was conducted a month ago. It is in these past two weeks that voter's opinion would better reflect their voting preferences, you know, after the actual presidential debates.
Also, in Iran you get only 30 days to actually campaign. This poll was taken right at the start of campaigning. Of course the current Pres will fare better in the polls then, more people are familiar with his platform.
Re:If you know anything about statistics... (Score:3, Informative)
CNN repeatably has stated that they are going on second and third hand information. That they are watching German and british news sites for information.
It is in all their articles.
Re:If you know anything about statistics... (Score:2, Informative)
In many parts of the world voting is not a two horse race or a highest vote wins system (I don't know which applies to Iran). Some of us get to express a series of preferences with lower order preferences being significant if no-one get 50% of the votes on first preferences (often the case here in Australia). Counting a voter's first preference is indeed fairly quick, but tallying a preferential ballot can take quite some time, especially if the result is close or the electoral area is large and contains many polling stations.
A casual glance at the last Federal election results for polling booths in my electorate shows that none handled more than 5000 voters (excluding absent/postal votes) and they are able to return a provisional first preference count for the House of Representatives within hours. Usually the "certain winners" in provisional counts are sufficient to identify the new Government. Senate votes and close races typically took days to be settled fully.
Re:Statistical nothing (Score:2, Informative)
Iran was, by the very definition of the term, a third world country. Not that the term has any meaning, any more, as the Soviet Bloc, the second world, no longer exists.
Mod parent up. (Score:3, Informative)
The study did not reject the null hypothesis ( that the results are not rigged.)
The strongest pattern observed was that in 2009 Amadi-nejad (sp?) did the best in districts he did the best in 2005. (no surprise). The authors note that to the extent that prediction for 2009 deviated from a model based on 2005, then his deviations were always above the extrapolated line. The thought here is that if the model were perfect then the deivations should be unbiased and thus depending on their distributions nature have about as much mass or events above and below the line-- more or less. But to have them all above the line is, assuming the model predicts well, surprising.
But this assumes the model predicts well.
Other people have noted that with an 85% turnout, common sense suggests this should favor the challenger. that is, angry people are more motivated to turn out. But while perhaps compelling it's not a hard rule. Iceland had a recent election where something like 70% turned out and the incumbent won. Likewise, even obama did not win by a margin anywhere close to the level of excess turnout. So it's quite clear that excessive turnout is not all favoring the challenger. Perhaps an enhanced fraction but by no means all.
It's also worth noting that 3 weeks prior to the election A was leading by double digits in some unscientific polls. In those (non rigorous) polls about 50-60% where undecided to declined to state. so there was a large latent swing vote. But again it's not reasonable to assume that all of the swing vote would go to the challenger. Hence A's early lead of committed voters would give him a suggested advantage. Admittedly the poll is not scientific, probably did not poll women as much, and I'd assume there's intimidation as well for people to respond honestly.
This is not to say that A won. Not at all. It is to say that proving that A lost is a hard sell and should not be based on statistics. What should be learned here is that in order for the winner to govern he has to convince the losers they lost. And you can't do that by denying poll observers to the challenger, having the incumbent's office too tightly coupled to the voting authorities, and then doing some jackass stunt like reporting results before the polls close (assuming that allegation is true-- there's some doubt on that.)
SO now they losers are not convinced they lost. They might be right, they mught be wrong, but the important thing is they are not convinced.
These same reasons are why electronic voting is a bad idea as well I note.
Re:The problem of time (Score:3, Informative)
Canada, with paper ballots, 1/2 the population and 7x size has achieved this for at least 40 years. Does Iran lack Canadas 1970 technology? I doubt it.
This isn't a red flag, at best a pale beige. They might have a little better communications infrastructure, for an obvious explanation.
No offence to OP's analysis guy, but CNN-style instant analysis has a very odd smell, a bit like napalm in the morning.
Re:Makes me feel good on the inside. (Score:3, Informative)
You're so right. Look at all the democracies we promote. Saudi Arabia, Pakistan, Indonesia, Syria, Egypt... USA! USA!
No one believes that load of crap you just spewed, at least no one who's taken a cursory look at the list of countries we've overthrown and vandalized [wikipedia.org]. Especially Iran, where we not only overthrew their democratically elected government in 1953, but may have even fomented the Iran-Iraq War in 1980.
I'm sorry, comrade! I did not mean to be unpatriotic and dare criticize our motherland, which is the true force of Good in this Evil world.
Re:If you know anything about statistics... (Score:3, Informative)
In many parts of the world voting is not a two horse race or a highest vote wins system (I don't know which applies to Iran).
Iran requires a true majority of votes (50% of all votes, not just more than any other candidate). The other two candidates only got about 1% each, so the election was effectively whichever of Ahmadinejad and Mousavi got more votes.
Re:ProxyBox Virtual Appliance (Score:1, Informative)
Re:Come on, It's Iran already (Score:5, Informative)
Juan Cole politely tells you that you're full of shit here [juancole.com]. He knows more about this than you do. Go read him.
Re:The results match pre-election poll (Score:1, Informative)
Re:What if they are? (Score:3, Informative)
Here's the thing that most slashdotters don't get about Iran, they are not just Muslims, previous to the Islamic revolution in 79 a large portion of the population was Zoroastrian and Baha'i. The US/British backed Shah was so bad that when the Islamic revolutionaries offered them another choice they jumped at it. Unfortunately the Ayatollah set about suppressing Zoroastrians and exiled all Baha'i who didn't convert. This is why the main Baha'i complex is in Israel (as a side note, every Israeli Jew I've met has said the most oppressed people in the middle east are the Persians).
Persian Iranians are amongst the most secular and accepting (of other religions) people I've met, even the Muslims. Granted I've only ever met the ones that have moved to Australia.
From what I've been told this has ignited a lot of Racial tensions in Iran, the Ayatollah and ruling council have set up a separate police force made almost exclusively from non-Persians (mostly Palestinian and Lebanese migrants) for the purposes of guaranteeing that the government has a force to use against its own people. The government will keep the Iranian army from being involved as they consist of mostly ethnic Persians and will not fire on their own people. Protesters have taken to killing non-Persians in some parts of Tehran in reprisal for suppressing the protests. Disclaimer: This last paragraph is third hand info (given to me by my Persian mate).
Re:Come on, It's Iran already (Score:3, Informative)
Juan Cole politely tells you that you're full of shit here. ... Go read him.
He has a different opinion, yes. Personally I find his analysis far from persuasive.
IMHO he overemphasises the "culture wars" of a decade ago (if they ever existed as he imagines) and completely ignores appeal of A'jad jingositic nationalism, ineed he completely ignores the last 4 years of A'jad. Moreover Cole's assumption that support for Khatami in 2000 is indicative of a the success of moderates against religionists seems a little odd if we remember that Khatami was himself a cleric and indeed a sayad. Nor do can I accept that issues of class (at least if we use that terms in a more expansive sense than an old-school Marxist might) and "culture wars" can neatly be separated.
Cole's greatest failing, however, is that he simply ignores that actual data. He fails to present any statistical evidence of election fraud? He fails to account for the fact that nationwide polling [washingtonpost.com] showed A'jad receiving a level of support more or less in the order of the eventual result? Even intelligent and reasonably well informed people can fall victim to believing what they want to believe, rather than what the evidence suggests.
He knows more about this than you do.
I would not dispute that, but nothing in the particular posting you linked to indicates so. I suggest you go back to that article, and read some of the responses, especially those from people with 1st hand knowledge and those citing empirical evidence. These comments pretty much dispose of Cole's analysis.
NOTE: Nothing I have written should be taken to imply that I'm making any assertion that voting fraud did not take place. Rather my point is that in the absence of solid evidence of fraud, and in the light of previous polling we ought not to presume fraud.
Re:Mod parent up. (Score:3, Informative)
Other people have noted that with an 85% turnout, common sense suggests this should favor the challenger. that is, angry people are more motivated to turn out. But while perhaps compelling it's not a hard rule. Iceland had a recent election where something like 70% turned out and the incumbent won.
Pardon? Iceland recently had an election, yes. But turnout was 85.1% compared to last election's 83.6% - not a big change (and lower than the turnout in the election before 87.5% were there was no change in government). Yes, the winner was the incumbent - on a technicality, because the coalition only came in power a few weeks before after the previous government coalition, (close) winners of the last election (by 1 seat) had resigned after massive public protests.
Re:The results match pre-election poll (Score:3, Informative)
With 400 million US "invested" beforehand in support for the "Iranian opposition"
The $400 million went to supporting groups like the Jundallah, not for reformist politicians like Mousavi and Tehrani students who like rock music and hate the moral police.
http://en.wikipedia.org/wiki/Jundallah [wikipedia.org]
No amount of money can buy a popular uprising. Sorry.
Re:If you know anything about statistics... (Score:1, Informative)
Wrong angle. It's better understood as "4chan being a force for underground communications." They're doing what they're best at. Albeit, for once, for a good cause.
Re:What if they are? (Score:3, Informative)
Here's the thing that most slashdotters don't get about Iran, they are not just Muslims, previous to the Islamic revolution in 79 a large portion of the population was Zoroastrian and Baha'i.
I do in fact get that. Iran is one of the most ethnically and religiously diverse countries in the Middle East.
From what I've been told this has ignited a lot of Racial tensions in Iran, the Ayatollah and ruling council have set up a separate police force made almost exclusively from non-Persians (mostly Palestinian and Lebanese migrants) for the purposes of guaranteeing that the government has a force to use against its own people.
I can confirm this. Those aren't "migrants", though, they're members of Lebanese Hizballah and Palestinian Hamas. Yet another reason I want to see this revolution succeed.
I'm going to go listen to "Yallah Yah Nasrallah" now.