Did Mahmoud Ahmadinejad Steal the 2009 Iran Election?

Charges that the Iranian government brutally mistreated protesters after the 2009 presidential election must be taken very seriously. A protester’s human rights should not depend on the merits of his position, just as our respect for a soldier should not depend on the merits of the war he is sent to fight. The question considered here, however, is not whether the government mistreated those who protested the election result, nor whether Iran’s government ought to be run by different people with different policies. Nor is the question whether more candidates ought to have been declared eligible to run – a complaint not made by Mir-Hossein Mousavi until after the election. Obviously he made the list, and the exclusion of other candidates probably improved his chances. The question here is simply whether Mahmoud Ahmadinejad won the election, fair and square.

Here is the officially reported outcome:

Iran Presidential Election – June 12, 2009

Mahmoud Ahmadinejad – 24,525,491 (62.6%)

Mir-Hossein Mousavi – 13,258,464 (33.8%)

Mohsen Rezaei – 656,150 (1.7%)

Mehdi Karroubi – 330,183 (0.8%)

Valid votes – 38,770,288 (98.9%)

Blank or invalid ballots – 421,005 (1.1%)

Total votes – 39,191,293 (100.0%)

Were these figures correct? They were reported by Iran’s Interior Ministry, an institution that has been vilified and ridiculed in the Western press and in which most Iranians themselves do not express a great deal of confidence – much as they reportedly feel about Iran’s Guardian Council, which monitors the Interior Ministry’s election-related activities. But that does not matter because we have no need to rely on the figures reported by the Interior Ministry, nor on the Guardian Council. What does matter is that these reported vote totals match the sum of local vote counts reported by the 45,692 polling stations at which Iranians voted in this election. At local polling stations all across Iran, tens of thousands of observers selected by Mir-Houssein Mousavi personally monitored the voting all day long and closely watched the vote counting after the polls closed. Not one of Mousavi’s 40,676 registered observers complained that day that he had been turned away or prevented from observing any activity at his polling station. Not one has claimed that the Interior Ministry reported a vote count for his polling station that did not match the vote count he had personally witnessed, or that he was deceived or lacked an adequate basis for approving the count. These facts – disputed loudly and often by Mousavi supporters after the election results were reported, but not by his actual election-day observers – are difficult to ignore. Few outside observers appear to understand this.

This is not entirely sufficient even so, since several thousand polling stations were not observed by Mousavi representatives, and he has alleged other wrongdoing. Nonetheless, whether Ahmadinejad won Iran’s 2009 presidential election, fairly, is a question that easily can be answered. The tests proposed below are straightforward. The necessary data have long been available. The results ought to convince any fair-minded skeptic.

Analysis of Mousavi Complaints

By the end of Election Day, the three opposition candidates had filed 646 complaints with the Guardian Council, which soon claimed to have investigated them even though nearly all involved local irregularities that could have little effect on the lopsided outcome. In addition, 10% of the ballots were recounted eight days after the election, with video cameras and hundreds of opposition observers looking on. No significant discrepancy was found, and the candidate whose representatives had observed the recount (Rezai) withdrew his complaint three days later.

Mousavi ignored all of this, however, having shifted his attack to sweeping allegations such as “the way the results were pre-planned,” and to a more extreme remedy: nullification of the election. Most of Mousavi’s new allegations – involving subjects such as “the role of shadow institutions” and “abuse of power” – were phrased too generally to permit an investigation. Instead of supplying requested details, Mousavi encouraged his supporters to stage protest rallies, which led to harsh government crackdowns. A few complaints nevertheless were developed sufficiently to be assessed here.

Complaint: There Were More Votes Than Eligible Voters In Some Areas

Early reports indicated that the votes in two of Iran’s thirty provinces slightly exceeded the number of eligible voters, which Mousavi cited as evidence of fraud. Similar “excess voting” had occurred in earlier high-turnout elections, such as the 1997 election won by the reform candidate, Mohammad Khatami.

Voters do not register in Iran, though few Western observers appear to have known this. In a presidential election, any Iranian age 18 or over may vote at any polling station in Iran – even outside Iran: hundreds of thousands of traveling and expatriate Iranians voted in the 2009 election (overwhelmingly for Mousavi, as it happened) at polling stations set up in 95 countries outside Iran.

Because no voter registration records exist, measuring turnout depends considerably on how one counts eligible voters (the denominator of the voter-turnout fraction) in the area measured. The less accurate the count of eligible voters, the more likely that “excess voting” will be found. Some independent calculations of 2009 voter turnout were based on 2006 census figures; others used residential data supplied by an independent Iranian news agency. The Interior Ministry said its own turnout calculations were based on birth certificate registries, while other government agencies used voting data from the 2005 presidential election. Mousavi did not disclose what data he had used. The Guardian Council claimed that all such measures had flaws, and that only one test of voter turnout was meaningful in a presidential election: Nationwide, did the votes exceed the eligible voters? Although the 2009 turnout was the highest ever for an election (85%), it was well under 100% and far short of the 98% turnout for the 1979 referendum held to ratify the creation of the Islamic Republic.

The measure of voter turnout also depends on the size of the measured area. “Excess voting” appears far more frequently when smaller areas are measured. An influx of students, soldiers, vacationers or commuters into a small city, for example, will affect turnout figures much more for the city than for its province. Although the American Enterprise Institute later concluded that no province-wide “excess voting” had occurred after all, Iran’s Interior Minister announced that voter turnout had exceeded 100% in 48 small cities. The proportion of “excess votes” had been extremely high in some areas. For the affluent north Tehran suburb of Shemiran (the most pro-Mousavi area in all of Iran), the Interior Minister reported that the number of votes was 13 times the number of eligible voters – up from 8 times in 2005.

In short, “excess voting” has long been common in Iran. It occurred more often in the 2009 election because voter turnout was higher than ever. It does not mean that fraud occurred. Nor, of course, does “excess voting” exclude the possibility of fraud. It is not easy in Iran, however, to stuff ballot boxes or vote in more than one place. Responsibility for ensuring fair elections is entrusted to Iran’s Guardian Council, which monitors the election-related activities carried out by the Interior Ministry. Whether or not one respects either institution, a candidate may post an observer at every polling station to monitor compliance with elaborate procedures designed to ensure that elections are fair.

Each voter is required on election day to present an identification card, called a “shenasnameh,” which bears the voter’s photograph, thumb print and unique identifying number. The voter’s name and number are entered into a computer and recorded in writing at the polling station, and are written again on the stub of his ballot. Before voting, the voter must press a purple-ink thumb print onto his ballot stub, which is then separated from the ballot and dropped into a “stub box.” Once the ballot and stub have been separated, it is impossible to determine how the voter voted. Each voter’s identification card is stamped to prevent him from voting more than once. A unique stamp is created for each election so that poll workers can easily spot it when they check a voter’s identification card.

All of this occurs in full view of candidates’ observers at each polling station where they are present. Representatives of the Guardian Council, the Interior Ministry, the local judiciary, the local police, and members of the public also serve as observers. Many polling stations are located in schools, where local teachers often act as observers. Typically, 14 or more observers monitor all election-day activities at each polling station. Observers verify that the stub box and ballot box are empty and then sealed before voting begins. They watch all day as each voter’s credentials are examined, he receives a blank ballot and presses his thumb-print onto the stub, the stub is separated from the ballot and dropped into the stub box, the voter enters a private voting booth, and finally he emerges and drops his completed ballot into the ballot box. The observers watch the ballot box closely to make sure no one except a voter drops anything into it.

When the polls close, the observers watch as the “stub box” is opened and the stubs are counted, and then as the ballot box is opened and the ballots are counted to ensure there are as many ballots as stubs. If the stubs outnumber the ballots, all ballots will be counted and the discrepancy will be noted in the election report. If the ballots outnumber the stubs, the discrepancy will be noted and the number of “extra” ballots will be randomly removed from the ballot box before the vote count begins.

The observers continue watching as the actual vote count takes place. Election officials examine each ballot to confirm that the voter’s choice was clearly indicated. Challenges are discussed and resolved among the election officials and observers. The final count for each candidate is written on a government form – Form 22 – which also states how many blank ballots were supplied to the polling station and how many are left. Five originals of Form 22 are signed by election officials and each observer. If a candidate’s observer disagrees with the count, he will refuse to sign (and presumably will notify the candidate). One signed original of the Form 22 is placed inside the ballot box, which is then re-sealed in the observers’ presence and handed over to a local election official to hold for a legally prescribed period of time.

The ballot box is not delivered to the Interior Ministry, even if a recount occurs. Many analysts mistakenly believed that the 45,692 ballot boxes in the 2009 election were to be physically transported to Tehran for counting – under “police escort” in some accounts, sometimes with stop-overs at “local wards” and “provincial committees,” and even with multiple observers along for the ride. Some analysts even considered it evidence of fraud that Mousavi observers had been barred from riding along on these imaginary journeys to Tehran.

Signed originals of the Form 22 are delivered to the Interior Ministry in Tehran and three other officials. A copy is given to each observer. The Form 22 information is also transmitted electronically on election night to district or county government offices, where candidate observers also are present. Form 22 information from numerous polling stations is summed up there to yield district-level and county-level vote totals, which are then transmitted electronically (and later physically) to the Interior Ministry in Tehran. To expedite the national vote tabulation in the 2009 election, Form 22 information was also transmitted directly from each polling station to the Interior Ministry. Observers are present when any electronic transmission occurs.

The government’s explanation for “excess voting” in the 2009 presidential election was the same as in previous elections: Regardless of where he lives, any eligible Iranian voter may vote at any polling station anywhere in the world. This explanation was widely ignored or distorted in post-election press coverage. Many stories reported that the Iranian government had admitted “voting errors.” According to Dr. Ali Ansari, author of the frequently cited Chatham House Preliminary Analysis, the government had even conceded that “possibly 3 million votes were missing.” He was referring to the Interior Ministry’s announcement that “excess voting” had occurred in 48 small cities. The Ministry spokesman had explained that local turnouts exceeding 100% had been more common this time because turnout had been extremely high; this did not mean fraud had occurred. Moreover, the spokesman had added, the outcome would have been the same even if fraud had occurred – in fact, even if all 3 million votes cast in those 48 small cities had been fraudulent.

The spokesman’s last remark might have struck most listeners as harmless. But some took a dimmer view – several dimmer views, in fact, linked only by the phrase “3 million votes” and a shared suspicion that Ahmadinejad’s vote-riggers were to blame for whatever foul play had occurred. Some commentators merely expressed concern about “irregularities” and “discrepancies” that “could affect 3 million votes.” Others, such as Dr. Ansari, suspected that the 3 million votes were more than just “affected” – they were “missing.” Still others reached precisely the opposite conclusion. There were not 3 million too few votes, but rather 3 million too many: “[T]he number of votes recorded in 50 cities exceeded the number of eligible voters there by 3 million.” The “too few” and “too many” interpretations of the spokesman’s remark soon were harmonized in an explanation that appealed to many Mousavi supporters: the three million votes were neither missing nor excessive – they had simply been stolen from Mousavi and given to Ahmadinejad.

If the candidates’ totals were adjusted to reflect this vote theft, Ahmadinejad’s 24.5 million vote total would drop to 21.5 million, within striking distance of a run-off election (19.6 million votes, 50% of the total), and Mousavi’s total would jump to well over 16 million. In light of this shocking revelation, who could doubt that more misconduct would soon be uncovered? As if all of this were not bad enough, the Iranian government appeared to find nothing wrong with it – it was perfectly legal and “normal.” A Guardian Council spokesman had shrugged off the government’s massive fraud at another press conference, in an astonishing remark promptly reported in hundreds of stories around the world:

[T]he [opposition] candidates, who claim more than 100 percent of those eligible have cast their ballot in 80 to 170 cities, are not accurate – the incident has happened in only 50 cities…

In short, 3 million votes had been stolen from Mousavi and given to Ahmadinejad, and the Iranian government’s reaction essentially had been “So what? Mousavi still falls short.” Nearly as upsetting was the government’s brazenness: holding press conferences to announce its own fraud.

In fact, the Iranian government had never conceded any voting errors as a result of “excess voting,” nor that a single vote was missing – much less that Ahmadinejad had stolen 3 million votes from Mousavi. Nor was evidence of fraud reported for any of these 48 small cities. Vote-tossing and ballot-box stuffing had been rampant, according to Mousavi supporters, but apparently no one could remember who had done it, or where, or how. Many people had voted multiple times, but not a single example was cited. Not one of Mousavi’s thousands of polling-station observers stepped forward to claim that misconduct had occurred at his polling station. These claims of vote theft, ballot-box stuffing and multiple-voting appear to have had nothing at their base but fertile imagination and ignorance (or ignoring) of Iran’s “vote-anywhere” rule.

Complaint: Results Reported By Local Polling Stations Were Altered By Election Officials In Tehran

Mousavi contended that vote counts reported by polling stations were altered by the Interior Ministry in Tehran. The vote tabulation allegedly took place in locked rooms from which opposition representatives were barred. The Guardian Council denied this, and claimed that “many of [the candidates’ observers] left their desks [at election headquarters] at 6 AM on [the morning after the election].” As will become clear, it is not necessary to resolve this disagreement.

Most of these “locked-room” allegations were made by persons who appeared to believe, mistakenly, that the Interior Ministry counts ballots in Tehran. Ballots are counted only at the polling stations, by local election officials, with many observers looking on. The Interior Ministry’s task is only to tabulate these field counts and generate county-level, province-level and national-level election reports. Even so, a risk of fraud exists. If one assumes these Interior Ministry officials were mere cat’s paws of Ahmadinejad, as many Mousavi supporters insisted, they might have altered the field reports, producing “official” results that showed Ahmadinejad with a majority of the votes (or even 62.6%, as was reported). This is precisely what Mousavi and many others alleged. Some Mousavi aides asserted that no votes at all were counted. Dr. Ali Ansari agreed: “I don’t think they actually counted the votes, though that’s hard to prove.” Hundreds of other commentators made similar statements.

Ironically, any such mischief would be a blessing in disguise for Mousavi – an opportunity to prove his case. For the first time ever, Interior Ministry officials in the 2009 election reported a per-candidate vote count for every ballot box (see note 1). This simplified Mousavi’s task. He needed merely to show that a ballot-box count reported by the Interior Ministry did not match the Form 22 ballot-box count witnessed by a Mousavi observer. The following passage, and the two paragraphs that follow it, make clear how the ballot-box-level reports issued in the 2009 election made it easy to detect vote-counting fraud:

Counting process. The two-stage counting process presents perhaps the most troubling aspect of [Iranian] elections. At each polling station, after the end of voting hours, the votes are counted and recorded on Form 22 in the presence of representatives from the candidates, the Interior Ministry, and the Guardian Council. These forms are secret, however; the results are not announced to the press or released to the candidates. Instead, in the second stage of the counting process, the forms are sent to the Interior Ministry, where the votes are tallied and published on Form 28, which reports the votes by province or county. But because there is no supervision of the preparation, there is no way to compare Form 28 to Form 22. In other words, it is possible for agents from the Guardian Council or the Interior Ministry to change the vote totals before announcing them.

This possibility had existed for every Iranian presidential election before 2009. Once the Form 22 information from a particular polling station had been reported to the Interior Ministry, it would become a small component of regional totals later reported on a Form 28. There would be no way to verify a Form 28 because the hundreds of Form 22’s that had been summed up to yield its reported totals would not have been published and candidates’ representatives would not have monitored the Interior Ministry’s tabulation process. A candidate’s election-day observer would know only the vote count reported by his own polling station.

This changed in the 2009 presidential election. The Interior Ministry added a crucial detail to its report. Instead of reporting only county-level and province-level totals, it also reported the vote count for each ballot-box – the very same vote-count number reported on a Form 22. For the first time, it was possible and quite easy, to challenge any ballot-box count: just compare the Form 22 field count with the Interior Ministry’s official count.

The Guardian Council claims that it asked Mousavi “time and time again to provide the council with any evidence or examples about the discrepancy” in ballot-box counts, but that “no documents or evidence were received.” Mousavi has not disputed this, nor has he ever cited a discrepancy for any of the 45,692 ballot boxes. Even if thousands of his would-be observers were improperly turned away, as Mousavi insists (see next section), tens of thousands of them observed election-day activities at polling stations all across Iran and indicated their approval of the reported result – either by signing the Form 22 or, at least, by failing to dispute the vote counts reported by the Interior Ministry for their polling stations. The Guardian Council claims to have “written evidence” of approvals which “if necessary can be given to the media to inform the public,” though it has not specified the nature or extent of its “written evidence.” Mousavi has not asked that any written evidence be released.

Since the necessary data have long been available to compare ballot-box counts, only two explanations for Mousavi’s silence come to mind: either no such discrepancy exists, or no one has bothered to check. To exclude the second possibility, someone should make these comparisons now – ideally for every ballot box, but at least for several thousand chosen to yield a valid statistical sample. If all tested ballot-box counts match, it will follow that vote-count fraud was not committed either by the Interior Ministry or at any polling station where a Mousavi representative observed the vote count and did not dispute the reported result. This would leave only the possibility that vote counts were falsified at “unobserved” polling stations, which can be determined by comparing “unobserved” with similar “observed” ballot boxes as discussed in the next section.

All of this begs a question, of course: Haven’t these ballot-box comparisons already been made? When the Interior Ministry released its official ballot-box-level reports, was there a single election-day observer in all of Iran who did not immediately compare his personal notes of the vote count with the Interior Ministry’s reported count for his polling station? And if the two counts did not match, is there a single Mousavi observer who would not have reported the discrepancy immediately? One is tempted to answer “no” to both questions.

Finally on this point, might some of Mousavi’s polling station observers have been deceived by Ahmadinejad’s vote-riggers? Though this possibility cannot be dismissed entirely, it seems unlikely and Mousavi has not identified any polling station where this allegedly occurred. A typical polling station observer is smart, zealous, alert and well-trained to spot signs of polling-station fraud – that is his only reason for being there, after all. Nor can it be said that some types of polling-station fraud are undetectable even by the best of observers. The activities carried on at a polling station are not complicated or difficult to monitor. Absent at least an allegation that an observer was deceived, his approval of a field count should be considered sufficient evidence that the count was correct.