A Rejoinder to the Chatham House report on Iran’s 2009 presidential election offering a new analysis on the results
The recent Iranian presidential election, held on June 12 of 2009, has become embroiled in considerable controversy, not least of all because it is the first of 25 elections that have been held in the last 30 years of the history of the Islamic Republic where the final outcome of the count has been rejected outright and regarded as invalid by many observers. “A Preliminary analysis of the voting figures in Iran’s 2009 presidential election” was released on the 21st June by the British think tank, Chatham House, and authored by Dr Ansari, an historian and professor of Iranians studies at St Andrews, who also serves as a consultant to the UK Government. Involved in the analysis, and contributing to the research, were two graduate students, Tom Rintoul and Daniel Berman. The study attempted to highlight aspects of the Iranian presidential election, which appeared to be implausible, problematic at best, and indeed more conducive to the notion that they have been artificially manipulated, although this was not explicitly stated. This followed Dr Ansari’s initial reaction to the election results on the 13th June which he described as “nonsense” and speculated that he didn’t think that the votes in the election had even been counted1. In its executive summary, the report raised the following four key points:i) Firstly, that the number of votes exceeded the number of registered voters in two provinces. ii) Secondly, that there is no correlation between the increase in turnout and swing to Ahmadinejad. iii) Thirdly that the incumbent would have to have won over many reformist voters and iv) Finally, that the claim that Ahmadinejad is popular in the countryside is a myth. A response to all these points is contained within the main body of the discussion here.
IRREGULARITIES AND ANOMALIES IN THE VOTER TURNOUT
One of the key observations mentioned in the executive summary of the Chatham House report is that the turnout in two provinces, namely Mazandaran and Yazd, surpassed the number of eligible voters registered in the 2006 census. This alone may appear to be the case of a major anomaly and reason to suspect an artificial amplification of the vote tally. In fact, it turns out that the turnout levels for both provinces were actually only slightly above 100% nor does this not take into consideration natural population growth or indeed migration of people between the provinces over the course of the last three years. When compared with the previous one taken in 1996, the 2006 census does show that the population in Mazandaran has increased by a year-on-year rate of about 1.1%2. Iran’s electoral watchdog - the Guardians council - has stated, voters are free to cast ballots anywhere they choose so long as they have a proper ID card. In the case of Mazandaran, we should not be surprised to observe that the turnout has exceeded the number of eligible voters because in the summertime many Iranians shelter from the heat of the big cities in the cool of the Caspian littoral region2 where they either own or rent holiday villas and apartments, or indeed stay with friends and relatives3. In addition, as with many statistics reported for Iran, the accuracy of the 2006 census may be wanting as could the estimate of the number of eligible voters this time round which may have been nearer 50 million, if birth certificates are any reliable measure, according to Iran’s national statistics centre. In any case, the national turnout, around 85%, was the highest on record and this in itself means that exceptionally high figures over 90% should not necessarily be viewed with scepticism. The situation in Yazd province is harder to determine, but Mr Mousavi did in fact win three out of the ten districts there, including the provincial capital. This was expected given that his principle supporter, former president Mohammad Khatami, who appeared side by side with Mr Mousavi on campaign flyers, is a native of Ardakan that Mr Mousavi carried. It would seem odd to call into question results that showed stronger support for Mousavi here than what he received nationwide. Moreover, the population growth rate based on the difference between the 1996 and 2006 census shows an annual increase of about 2.8% which is significantly higher than the national average for the same period4.
The mobility of the electorate, due to reasons that include pilgrimage, vacationing, seasonal migrations (as in the case of nomadic tribes), people working or studying away from their home area, expatriates visiting, as well as administrative redistricting since the taking of the census, can help explain and account for many, if not all, of the perceived discrepancies that the Guardians council acknowledged as occurring in as many as 50 townships throughout the country. Even so, some of the districts where the excess votes were recorded are small, remote places frequented by few outsiders. But is also worth noting that this has also happened in previous elections where there too was a very high turnout, such as in 1997 presidential election which none would dispute as being fraudulent. In a debate presented by PressTV5, Dr Ansari defended his report and referred to the Guardians Council’s admission as one that had conceded “up to 3 million votes had gone missing”. However, one of the other guests present in the debate, Professor Marandi of the University of Tehran, responded by saying that this had been in fact misreported and that the total number of votes in the 50 districts in question amounted to 3 million, not that such a huge number had simply disappeared or been added. In order to explain the near 100% turnouts in Mazandaran and Yazd, the authors make reference to allegations of “tombstone voting” which is also mentioned by dissident scholar Mehdi Khalaji of the Washington institute for Near East policy6. As well as the registering of the dead, other possibilities of fraud include polling station volunteers cheating illiterate voters by not writing in their chosen candidate or certain ballot boxes being pre-stuffed. It is worth realising that, as the authors acknowledge, none of these allegations of electoral fraud and irregularities are actually new in Iran and have been reported in previous elections where the overall outcome has not been called into serious question by any of the competing sides – this includes three national elections held under Mr Ahmadinejad’s tenure as president which were deemed to be fair in as much as the actual result of the count was regarded as being accurate. The past 30 years have served to produce a system in place that seeks to minimise any instances of electoral fraud to allow for competitive contests. The voting procedure is described by one ministry worker: “Each voter presents identification, and his or her name and information is entered into a computer, and also recorded in writing. The voter's thumbs are inked with purple dye and these are printed on the stub of the ballot. The voter's identification is stamped to prevent multiple voting at different polling stations, and there is also a computer and written record of everyone who voted at each polling place.7” If this was indeed upheld in this way, then it seems unlikely that any form of tombstone or multiple voting would have been possible. The election itself was monitored not just be officials from the Interior Ministry and Guardians council but also by the local authorities, who include teachers and police, the judiciary’s general inspectorate8, as well as thousands of observers from all four candidates who signed off the results for each ballot box. The report claims that there was no relationship between the increase in turnout and a swing to any one particular candidate. This is incorrect. What the results indicate is that there is no uniform swing to either candidate across all 30 provinces and 366 districts; instead, there are regional and local swings, some of which appear to be almost independent of any national trend. Many districts, notably the capital Tehran and its affluent northern suburb of Shemiranat, showed a huge swing from the 2005 election when a substantial increase in the turnout (22%) resulted in Mousavi comfortably winning with a 2:1 margin (63-32) in the case of the latter. But in the working class suburbs to the south of the city, namely Pakdasht, Islamishahr and Robat Karim, Mr Ahmadinejad won by a similarly large margin. Overall, Mousavi won districts in 12 of the 30 provinces. The CH report finds that “the massive increase in voter participation results in substantially less variation in turnout between provinces, with the standard deviation amongst provincial turnouts falling by just over 23% since 2005 and that regional variations in electoral participation have disappeared.” But it is difficult to appreciate why the authors see any significance in this. In 1997, too, there was also little variation in turnout simply because it was very high as Professor Salehi-Isfahani succinctly points out: “ The CH report claims that the fact that the variation in participation across provinces has dropped is evidence of fraud. But anyone familiar with elementary statistics knows that the standard deviation of any variable limited to 100% from above would drop as its mean increases. So, because the participation rate increased by about 35%, it is hardly surprising that the standard deviation fell by 23%9.”If we turn to Figure 8 of the CH report, the turnout levels reported for each province do still show that provinces with the lowest turnout levels in 2005, such as Kordestan, remained so in the case of the 2009 election. The results, thus, show that there was not one uniform nationwide swing to Ahmadinejad and that there were distinct differences at the local levels with districts even adjacent to one another reporting different swings to either candidate. In general, according to Walter Mebane, “natural processes in 2009 have Ahmadinejad tending to do best in towns where his support in 2005 was highest and tending to do worst in towns where the turnout surged the most.” It was also widely believed that the higher the turnout the likelier the support for reformist candidates. This, however, was simply an assumption – those who did not vote in 2005 may not have had reason for voting then irrespective of their political leaning.
POLITICAL AFFILIATION OF THE ELECTORATE
The CH report’s categorization and identification of the electorate with respect to perceived support for political factions, which really matters only to the elite and not to the general populace, is also misleading as is suggesting that a “decade of conflict” would have created a deeply partisan voting population. According to Abbas Barzgar, writing in the Guardian newspaper11, reformist president Mohammad Khatami was not swept to office in 1997 on a tide of liberalism or commitment to any ideological stance, but rather because he appeared to be an honest, charismatic anti-establishment figure and one untainted by official corruption. The fact that he was a black-turbaned seyed, a descendant of the Prophet Muhammad, and a disciple of the late father of the Islamic revolution, Ayatollah Khomeini, likely also played well with the religious masses. The personality and style of the candidate himself, and not merely his policy agenda, was the crucial factor in propelling Khatami to his landslide victory. This is a classic case, argues Barzgar, of western analysts misinterpreting certain political developments in Iran as signs of “liberal inevitability”. Moreover, the distinction between both factions appears somewhat blurred since both Mousavi in 2009 and also Qalibaf in 2005 described themselves as “principle-ist reformists12” to the derision and consternation of many. There is indeed little evidence to suggest any solid political allegiance to any one party or group amongst the general public in Iran that is so evident in many western states where an almost tribal loyalty exists. The conservative-reformist divide that emerged in the mid 1990s was essentially an internal development within the ruling elite similar to that which occurred in the ancient Roman senate, involving the split between the optimates and the populares, or that of the Tories and Whigs of eighteenth century Britain – namely, a power struggle within the establishment that was played out in the public domain. Initially, the conservatives were seen as being tied to the traditional alliance of the clergy, the bazaar and the governmental security and intelligence apparatus, while the reformists were viewed as having popular appeal amongst the masses, particularly women and youth. In 2003, this situation changed when a new order of conservatives arrived on the scene winning nationwide municipal elections that were set against a low turnout: Rather than coming from the traditional elements, the so-called “neo-conservatives” of Iran, led by the Abadgaran group, were mostly affiliated with, and had served in, the Revolutionary Guard and thus represented another sector of the Islamic establishment. This resulted in a decisive shift in the political dynamics whereby an otherwise unenthused conservative base was revitalized and remodeled by a new political class of revolutionary ideologues, many of whom were former guardsmen and war veterans.As the mayor of Tehran from 2003-2005, Ahmadinejad frequently criticized the then Khatami administration for not responding to the bread and butter issues and needs of the general public and accused the administration for being part of an entrenched and corrupt managerial system. In his 2005 election, he presented himself as “mardomyar” or “friend of the people” set against the vested interests of the reformists who had used any popular support for greater freedom only to further their own political ambitions. In this respect, this development represented a distinct departure from the rivalry of the last decade which the CH report refers to: the reformists were cast as no longer being a democratic, anti-establishment movement but one that had become all too familiar with power and privilege. The stage was set for hardliners to make a comeback by winning support among the people on a platform of social conservatism (focusing on family and community values), economic populism and a resurgent nationalism fuelled by Iran’s nuclear program. It is against this backdrop and development that the results of the 2009 should be judged and evaluated and not during the period that marked Khatami’s first term that the authors seem to be referring to.
WHERE DID AHMADINEJAD SOURCE HIS VOTES?
Another major concern raised in the report is that, according to the official data, President Ahmadinejad received approximately 13m more votes in this election than the combined conservative vote in the first round of the 2005 election. Naturally, this begs the question as to where all these extra votes came from. However, this is actually a somewhat spurious statement because the report deliberately omits any reference to the 2005 election’s second round. This may have been because the authors suppose that those people who voted for Ahmadinejad in the second round did so purely on a tactical basis, namely an anti-Rafsanjani one, and that what transpired in the second round is not indicative of their true political inclinations. While this may be correct to some extent, it could also have been because many voters had been won over to Mr Ahmadinejad’s populist and nationalist platform. Indeed, in 2005, Karroubi campaigned on a promise to share the nation’s oil wealth amongst the people, a move regarded as having no economic sense, but which nonetheless resounded in poorer parts of the country. His credentials and commitment to constitutional and social reform was not what he actually enticed voters with, and therefore there is some doubt as to whether those who voted for him could be described as being reformist-inclined. If, however, we take the result of the second round of the 2005 election and compare it to that of the 2009 election we see very little difference (61.69 to 62.46%). The Terror Free Tomorrow Poll, conducted with the support of the BBC 3 weeks prior to the election12, gave Mr Ahmadinejad a similar margin, as did the local Aleph poll13 conducted a few days before June 12th. Moreover, both surveys gave insignificant levels of support to the two minor candidates standing in the election, Mohsen Rezai and Mehdi Karroubi.It is assumed that people who did not vote for Ahmadinejad in 2005 would vote for Mousavi this time round and this is a reasonable enough hypothesis although it can only be regarded as a general rule.
Much has been made of Lorestan, the home province of Karroubi, since Mr Ahmadinejad won with an impressive 70% of the vote in this province even though he only took about 7% of the vote in the first round of the 2005 election – indeed, the votes of all three conservative candidates combined, totaled only about 18%. But when he faced off with Rafsanjani in the second round, he managed to secure as much as 50%. So, in the subsequent round of voting Mr Ahmadinejad must have succeeded in winning over a sizeable percentage of Karroubi and reformist voters assuming that all those who voted for Rafsanjani in the first round did so in the
second. Figure 2 of the report, “Did Karroubi voters defect to Ahmadinejad?” should be seen in the context of whether Karroubi voters switched to Ahmadinejad in 2005 and did not revert back to Karroubi, rather than that they defected in any such way in the recent election. It is thus only an assumption of the authors that those who voted for Karroubi in the first round of 2005 were likely to do so in 2009 when there is no real reason why they should do so. Lorestan, a deeply traditional and socially conservative region of Iran, did in fact vote for the conservative candidate Nateq-Nouri in the 1997 election14 against a national trend of support for Khatami.The observation is thus made that Karroubi’s share of the vote in Lorestan has collapsed entirely from about 55% to about 5%. Even so, Karroubi did poll more than 5 times better in Lorestan than he did nationwide and nearly 25 times better in his native township of Aligodarz. The collapse of Mr Karroubi’s vote from the 2005 election is not really that surprising and, as stated above, had been predicted by several voter surveys. Back in 2005, he was a reasonably well-known figure who had served for four years as Speaker of the Iranian parliament and was very much in the public light. This time, he was not campaigning from any position of authority within the government, although he did have the backing of his own political party and newspaper. The platform of economic populism that had epitomized his 2005 bid and which he repeated again in 2009 had been largely undercut and usurped by Mr Ahmadinejad this time round. The report even admits the possibility of this: “Ahmadinejad’s supporters claim that people voted for Ahmadinejad in 2009 for precisely the same reasons that they voted for Karroubi in 2005”. In 1997, there were two frontrunners and two well-known but minor candidates. The outcome of the result was that they minors received very little support at all since the election was regarded as being a two-horse race. It is also questionable as to whether Mr Karroubi can be seen as a “man of the people” since he is, after all, suspected of being linked to the same cleptocratic clerical mafia as Hashemi Rafsanjani. Indeed, Karroubi’s claim to be a populist was seriously called into question in one of the presidential debates where Ahmadinejad accused him of accepting bribes and for enjoying a wealthy and privileged lifestyle in his somewhat palatial home15. Overall, if we look at the numbers involved, we can claim that support for Mr Rafsanjani and Mr Moin (who was the principle reformist candidate) in the first round translated into support for the former in the second round, whereas Ahmadinejad was able to draw support not only from his two other conservative rivals but also from that of Mr Karroubi and Mr Mehralizadeh. Mostafa Moin, and not Mehdi Karroubi, was regarded as being the refomist movement’s leading candidate and was backed most of the 2nd Khordad alliance’s parties. Thus, on top of the 11.5 million total for the conservative candidates, Ahmadinejad secured the support of 5.8 million supporters of both Karroubi and Mehralizadeh in the second round to reach the decisive figure of 17.3 million.
Ahmadinejad’s biggest swing was in Kerman province (35%) that went from having the second lowest vote proportion for him in 2005 to the highest in 2009. However, this is not totally unexpected since it is the home of Ahmadinejad’s rival in the 2005 election, Hashemi Rafsanjani. The home advantage that Rafsanjani enjoyed at the time meant that it is difficult to compare support for Mr Ahmadinejad then and now. The greatest swing against Mr Ahmadinejad was in West Azerbaijan which he lost to Mr Mousavi.He also nearly lost in the provinces of Tehran and Ardabil with a large decline in his share of the vote reported in both. Even so, there are still 7 million out of 11.5 million new votes for Ahmadinejad that would have come from those who did not participate in 2005 or were too young at the time to do so at the time. Professor Farideh Farhi of the university of Hawaii has stated that she "simply, simply cannot believe" that Ahmadinejad could have won so many of these voters. Again, it is easy to think of the 40% of the electorate as having boycotted the 2005 election out of protest with the system but it was more likely to be a case of disinterest and apathy, certainly outside of the big cities. The election was widely expected to mean the return of Hashemi Rafsanjani and was a fairly dull event unlike in the 2009 contest that galvanized the electorate. It was in the big cities that the Mousavi campaign was so vibrant and confident in the closing days before the election and should represent changes compared to 2005. Yet many show swings of various degrees against the incumbent with the most dramatic being in Yazd, Tehran and Tabriz. Had such a sizeable swing been uniform and across all districts, Mr Ahmadinejad would have been forced into a second round or could even have lost outright in the first round. Of particular interest to the following section is the result from the city of Qazvin that, despite the significant swing against Ahmadinejad in the urban area itself of around 7%, heavy support in the rural districts of the small province ensured an overall swing in favor of him in the province of Qazvin.
THE POPULARITY OF AHMADINEJAD AMONG RURAL VOTERS
The votes of rural communities appeared to be decisive in the presidential election and it was reported that up to 75%16 of voters in the countryside had supported the incumbent this time round, as indeed they had done in the 2001 re-election of Khatami. The CH report cites trends shows that rural voters, who make up about a third of the population, tend not to support conservative candidates in past presidential elections. In the case of the first round of the 2005 elections, as well as those of 2001 and 1997, rural provinces did indeed, by and large, vote for reformist candidates. That said, rural voters do not tend to be natural supporters of the reform movement and its political goals. Illiteracy in villages in Iran is believed to be much higher than the national average of around 20%17, and so distinctly liberal issues, such as freedom of the press, are of lesser priority than those concerned with relieving poverty and infrastructure improvement. As well as this, government involvement in the lives of ordinary people is generally seen as being less pervasive and intrusive than in the major cities due to the fact that a traditional system of political and social organization remains in place within these communities to this day. As such, people in rural areas of the country are, on the whole, less concerned with the political and social agenda that typifies the urban and middle-class reformist movement, and are more concerned about economic issues as well as those relating to government regulation and red tape. If Iranians in rural areas have had a track record of voting for reformists, it is probably out of expectation of promises for social justice that Mr Khatami declared he would deliver on as part of his 1997 presidential election bid.
It is also a mistake to associate any level of support for the conservative faction in rural Iran with Mr Ahmadinejad himself who has attempted to avoid being seen as linked to any one particular political group. As Professor Salehi-Isfahani points out, the electorate in rural communities is more prone and responsive to being swayed with populist promises, be they from reformists or indeed conservatives, such as the sharing of the national oil wealth among the people as well as fighting corruption and discrimination. The simple lifestyle that Ahmadinejad likes to project as leading also plays well with people who may see the “son of a blacksmith from Aradan” as being of their own social class and thus deserving of their support. In the course of the last four years, Ahmadinejad, indeed, has focused much of his government’s attention on rural development and meeting the needs of those who live in the more than 50,000 villages across the country and, crucially, has visited nearly every district in the country, which is an unprecedented achievement by any ruler in the nation’s history. Critics have argued that this is just recourse to demagogy, tantamount to campaigning at the expense of the national treasury, but there is no doubt that simply by visiting these areas he would have received exposure and attracted support. Indeed, an article in the Washington Post18, claiming that support for him was falling in the run up to election day, nonetheless reported that many residents of the town had appreciated the simple fact that he was the first head of government to visit their small town and were prepared to vote for him merely on account of this. Gauging public opinion in rural areas of the country is not easy, but a rare journey and insight into rural Iran was reported 2 years ago by the Christian Science Monitor19 entitled “Ahmadinejad: rock star in rural Iran” which indicated that Ahmadinejad enjoyed in the countryside the kind of support which Khatami received at the height of his popularity and was thronged with well-wishers. It is unlikely that such sentiment would have abated in the two years since.
Just a month before the election, there was heavy criticism of the Government’s decision to dispense cash and food to as many as 5.5 million people living in rural areas20. Whether one agrees with the rationale and political ethics involved in such a decision, which could be seen as a blatant act of voter bribery, it clearly would have influenced the political inclinations of many of the beneficiaries of such handouts. Therefore, while the Chatham House report attempts to suggest that Ahmadinejad and other conservative candidates were “markedly unpopular” in rural districts, it is presumptive to portray the countryside as being unfavorable towards Ahmadinejad on account of what transpired in the first round or previous elections where he did not stand. The hitherto unknown candidate of 2005 did not have the chance or the time to present himself to those who live outside of the big cities. Besides Tehran where he was mayor, he was little known throughout the country at the time of his election except in his native Semnan province as well in Ardabil where he had served as a governor. It also neglects the fact that Ahmadinejad has had four years with which to build his base by supporting this constituency. Were we to look at Ronald Reagan’s re-election in 1984, we would see that he swept the board taking with him many Democratic strongholds with the support of so-called “Reagan Democrats”. Had the election gone down party lines, Walter Mondale may have had a decent chance of ousting the incumbent president.
A reference is also made to the comments of professor Eric Hooglund21 who is a respected scholar of rural Iran and who has been conducting fieldwork in Iran for over 30 years, although his research over the last 4 years has been relatively limited. In a somewhat anecdotal and subjective account of events in one village, Baghe Iman, which is apparently 15 miles from Shiraz, the provincial capital of Fars, Hooglund described how contacts he knew there had reported a sense of outrage in the village that Ahmadinejad has won the election – the problem is attempting to verify the claim since the name appears to be fictitious and there is no such place on the map. It is also deliberately misleading to suppose that rural political sentiment is indicative of the views of a single village. Interestingly, Hooglund describes what appears to be intimidation against Ahmadinejad supporters, one of the very things cited by Mr Mousavi as happening to his followers and a reason why the election ought to be annulled. A close analysis of the individual ballot boxes should shed some light on sharp differences should Hooglund deign to provide the real name for the village in question and its district. Finally, the incumbency factor, especially in rural Iran, should not be dismissed. While most of the Iranian peasantry has access to radio or television sets, their source of information beyond the state-run media is limited. Due to Mr Mousavi’s 20-year absence from politics, it was always going to be an uphill struggle even making himself known among residents of the villages and small towns. These places were far removed from the colorful and raucous scenes associated with the Mousavi campaign in Tehran and the big cities. Compounded to this, is the fact that Iranians were not given a choice of candidates whose box they had to tick, rather they had to write in the name of their chosen representative. In 2001, Mr Khatami won with nearly 78% of the vote both in urban and rural areas alike. The simple fact that he was better known and traveled compared to his rivals was a major reason for his overwhelming reelection victory. Ahmadinejad, therefore, had a clear advantage as a sitting president among this group. Incumbents in previous presidential contests – 1985,1993,2001 – had all won by a landslide.
ETHNIC AND REGIONAL VARIATIONS IN THE RESULTS
The CH report claims that “regional variations in participation have disappeared” and that much of Iran’s rural population is comprised of ethnic minorities - Lors, Baluch, Kurds, Azeris and Arabs amongst others. These ethnic and religious minorities are seen as having a history both of voting reformist and also of voting for members of their own ethnic group. For example, they were an important segment of Khatami’s vote in 1997 and voted largely for Karroubi and Moin in 2005. It is true that many ethnic minorities have had a tendency to vote for reformists in both parliamentary and presidential elections in Iran, however, as we have seen in the case of Lorestan and also in Mazandaran, – both of them home to ethnic minorities- these were the only two provinces that voted against the national trend in 1997 and voted for the conservative candidate, Mr Nateq-Nouri, instead. In the 2005 election, home advantage was apparent but it was not the case that candidates won a majority of support in their native regions. Indeed, Mostafa Moin did quite poorly in Esfahan province where he hails from. Only Karroubi was able to garner a majority in Lorestan.It is also worth remembering that many of the individuals within Iran’s civilian and military establishment are themselves of ethnic origin, including Ayatollah Khamenei whose father comes from the same town in East Azerbaijan as Mr Mousavi does. It has even been alleged that ethnic Azeris have a disproportionate influence in the government compared to all other groups, including Persians22. This may not be surprising when one considers that the Shiite Safavid and Qajari empires had their roots in Iranian Azerbaijan. In 2009, as it turns out, ethnic minority support for Mousavi was high in many parts of the country. Of the two provinces that Mousavi won, namely West Azerbaijan and Sistan va Baluchestan, both are ethnic regions as well as homes to large numbers of Sunni Muslims. Many commentators have since remarked that Mr Mousavi should have done much better in his home province of East Azerbaijan but it should be noted that Mr Mehralizadeh, also a native of Azerbaijan, received only about 29% of the Azeri vote in 2005. Nonetheless, Mr Mousavi did win in his home district of Shabestar and did very well in all three Azeri regions compared to elsewhere in the country. This somewhat makes a mockery of Karim Sadjadpour of the Carnegie Endowment’s comment that Azeris should have voted for Mousavi as blacks in America voted for Obama.In the 2005 election there was indeed more regional variation, but this was also because there were no clear frontrunners and there were more candidates in the race, all coming from different parts of the country. In the presidential election of 2001, when the incumbent president Khatami was running for reelection, these regional variations were not apparent at all despite the presence of ten candidates – Khatami won right across the board by a large margin.In the 2009 election, ethnic voting does indeed play a significant part. Outside of his native Lorestan, Mr Karroubi also did relatively well in parts of Ilam and northern Khuzestan: towns inhabited by ethnic Lors. The total Lori-speaking population in Iran only numbers some 2.6 million or about 3.7% of the total Iranian population. Next, turning to Rezai, we find that he also appears to have done well in his home province of Khuzestan winning a plurality in one town, Lali in Khuzestan, and a sizeable share of the vote in others. Lali is a Bakhtiyari town where Mr Rezai’s clan lives. Bakhtiyaris also live in Chaharmal Bakhtiyari, as the name implies. Here, in Koohrang and in the town of Izeh, another Bakhtiyari town in Khuzestan, Mr Rezai attained around 30% of the vote way above his national average of 1.7%.Mr Mousavi appears to have done best in one region but it is not his home province, rather the border region of Sistan va Baluchestan, a hotbed of ethnic separatist activity and which supported the reformist Moin in the 2005 election. Clearly, Baluchi Sunnis were the most pro-Mousavi ethnic and religious minority and the result from Khash County, where he received as much as 81% of the vote, is far in excess of Mousavi’s national average have about 34%. In the case of Ahmadinejad some districts, all of them from Kerman province, are exceptionally high (>90%) and are all from rural districts as well. While naturally suspect, they are not necessarily unbelievable. Even so, these are higher than even in Ahmadinejad’s native town of Aradan in Semnan province where he received 87.40%. It is results like this that ought to be the subject of any independent probe to determine if any artificial manipulation had taken place. At this point not enough is known about the circumstances pertinent to these districts. However, even if these results were annulled the outcome of the election would not in any way be changed.
In relation to Mr Mousavi, 37 out of the 46 districts he won (i.e. 80%) were in ethnic regions. Mousavi carried many Azeri Turk, Kurdish and Baluchi districts. These were expected to vote in the way that they did and so there is nothing surprising about this. Few Persian-speaking areas outside of Tehran province were supportive of the challenger. Indeed, Mr. Ahmadinejad appears to have won the Persian heartland convincingly, which may, possibly, suggest a backlash against the fact that Mr Mousavi was an ethnic Azeri. The New York Times reported the views of Iranian travelers just over the border in Basmagh in Iraq23. From the interviews conducted, Ahmadinejad supporters were all from the distinctly Persian areas of Esfahan, Qom and Shiraz while Mousavi followers were from Tehran and Kurdish regions. One isolated result which appears to be interesting is that in the remote rural, Persian-speaking district of Darmiyan in South Khorasan, Mr Mousavi received as much as 47.23% of the vote. This is significant because South Khorasan is regarded as being an Ahmadinejad stronghold. Indeed, in neighboring districts Ahmadinejad was winning 70-80% of the vote. It serves to illustrate just how complex and how localized voting preferences across the country actually were. In short, it is hard to make out why the Chatham House report portrayed the election results as showing no diversity and variety across the 366 districts and 30 provinces. In 2005, Mr Rafsanjani won 38 districts, many also in ethnic areas.
STATISTICAL AND NUMERICAL ANALYSIS
The Chatham House report refrains from an in depth numerical analysis of the election data deferring to that to others. It does, however, at least mention the work of Polish astronomer Boudwijn Roukema who found that too many leading 2’s and not enough leading 1’s in Ahmadinejad’s numbers. Roukema also reports another anomaly: too many leading 7’s for Karroubi’s totals. A first digit rule, known as Benford’s law, states that in lists of numbers from many (but not all) real-life sources of data, the leading digit is distributed in a specific, non-uniform way. However, it is also acknowledged that first digit distributions to not always conform with Benford’s expectation and that this was noticed by Benford himself. The reason for this is because were we to have a scenario where all towns reported 1000 voters, all subsequently voted, then the frequency of the numbers would broadly match of the vote proportions. In the case of election returns, one big reason Benford’s Law can fail is that voting areas aren’t necessarily randomly sized .For example, as Sam Wang of Princeton university comments, if voting areas are set up to be uniformly sized to contain 100,000 people each, then close races will produce lots of 4’s and 5’s as leading digits24. However, ignoring the first digit gives lists of numbers that are not subject to the size constraint and should therefore follow Benford's law. Mebane himself has conducted an exhaustive mathematical analysis into the results and has demonstrated that they display a large number of outliers. So, while the overall share of the vote for Ahmadinejad was 62-63%, there were results that strayed far from this general trend. In addition, Mebane found that when looking at vote totals for whole towns, the second digits appeared to follow the expectation of Benford’s law. “A statistically sharp approach to statistical testing—taking the multiple testing into account—fails to provide evidence against the hypothesis that the second digits are distributed according to Benford’s Law.” Mebane acknowledges that the “model is complicated, but it is easy to see that several aspects of the results seem natural. Places that strongly supported Ahmadinejad in the first stage of the 2005 election tended to support him in 2009…places that strongly supported Karroubi in 2005 tended strongly to support him in 2009, especially so as 2009 turnout surged above 2005 levels.” However, Mebane finds that an inspection of the second digits at the ballot box level reveal that “the tests are insignificant for Mousavi and highly significant—well beyond 99 percent confidence—for the other three candidates The results for both Karroubi and Rezaei reflect the very small proportions of the votes each received.” Mebane concludes that in the case of the two minor candidates:” this might be caused by either (a) inherently low levels of support, (b) voters strategically abandoning the candidates, or (c) fraudulent.” Based on opinion poll surveys in the run up to this election and the results of the 1997 election where the two minor candidates each received about 2% of the vote, it is plausible to infer both (a) and (b) as likely explanations. Of particular interest is that Mebane found “No significant distortions are apparent for Mousavi’s vote counts.”
One indication of tampering is when a ballot box contains an unusually low or high proportion of spoilt ballot papers — for example, adding fake votes, or accepting invalid votes in favour of your preferred candidate that should have been discarded will decrease the proportion of invalid votes, while invalidating legitimate votes for other candidates will increase it. Also, when people commit fraud by adding extra votes, they often forget to add invalid ones. Mebane remarks “to support the benign interpretation, the additional evidence needs to explain how the strong support for Ahmadinejad happens to line up so strongly with the proportion of invalid votes in the ballot-box vote counts. The increase in Ahmadinejad’s average vote proportion as the invalid vote proportion decreases in this interval is very steep. As the invalid vote proportion falls from 0.00385 down to zero, the average Ahmadinejad vote proportion increases steadily up to a value of about 0.78 (at about invalid= 0.0018). Such a steep relationship makes it implausible to argue that the relationship between invalid vote proportions and the respective shifts in votes for Ahmadinejad or Mousavi reflects changes in protest votes, i.e., in blank or spoiled ballots cast by people who liked none of the candidates.” However the distribution of invalid votes in the Iranian election follows a general trend whereby ethnic areas, such as Kordestan and Khuzestan, tend to include a greater proportion of spoilt ballots that are interpreted as being a sign of protest against the system as a whole. Moreover, in some ethnic towns where Ahmadinejad did very poorly, such as Zaboli in Sistan va Baluchestan, the invalid proportion was as low as 0.19%, far below the national average of over 1%. In the Persian speaking areas, Tehran and Qom provinces reflected the largest proportion of invalid ballots that suggests that this too may have been an urban protest vote by the middle class there. Therefore, as with previous elections, ethnic and urban areas, as opposed to Persian rural areas, attracted more spoilt ballots and this helps explain there is a strong correlation between support for Ahmadinejad and the proportion of invalid votes in as many as 10,000 ballot boxes. Separate to this research, two graduate students at Colombia University, Beber and Sacco, have looked at the last digits from the provinces for all four candidates and noticed that there was a distinct anomaly – there were too few 5’s and too many 7’s, have also conducted a study. They concluded there was a 1 in 500 chance at best that this was the result of a natural process having compared results with a “clean” Swedish election. They deduced that they were likely to have been manually fabricated. However, had the results been computer generated, the last digits we might expect to find a uniform distribution. When the district results, which were released at the same as those from the provinces, were analysed there was no sign of any fraudulent manipulation: “After we wrote our op-ed using the province-level data, we've now also done some preliminary tests with the county-level data. In the latter dataset, the last digits don't appear fraudulent.25” Moreover if we examine the penultimate digits we see that they do conform to a more even distribution.
For the last digits, the mean is 12, the variance is 19.2 and the standard deviation is 4.381. A chi square test produces a ÷2 of 16.0 that gives us a p value of 0.06688 to 9 degrees of freedom.
The percentage of numbers 0-4: 40.0%
The percentage of numbers 5-9: 60.0%
For the penultimate digits, the mean is 12, the variance is 11.2 and the standard deviation is 3.346. A chi square test produces a ÷2 of 9.3333 that gives us a p value of 0.407 to 9 degrees of freedom.
The percentage of numbers 0-4: 44.2%
The percentage of numbers 5-9: 55.8%
The number 5, whose frequency was well below average for the last digits, is now dead on the mean. It is clear that we should expect to find a greater frequency of higher value numbers as we approach the last digit together with an increase in the variance. This is essentially Benford's expectation but in reverse. And this is what is occurring here.
A more careful study of the individual ballot boxes should confirm that natural processes were involved although the possibility of fraud and ballot stuffing in remote parts of the country is a distinct possibility. However, it is clear that even in cities like Khorramabad in Lorestan, for example, Mousavi still won the tallies of 19 ballot boxes as Kaveh Afrasiabi notes26. This suggests that he was popular in some areas within the city district and is, of course, to be expected. An examination of ballot boxes from five cities, below, where Ahmadinejad won decisively, still shows that Mousavi won in certain areas. This indicates that much of the vote was being split along class and residential lines with central areas generally more supportive of Mousavi than those in the outlying areas. Clearly, the absence of ballot box data for previous elections means that a comparison is not available which could prove very insightful.
A final world must be said for the results of the voting of over 200,000 Iranians living or travelling outside of the country who also cast ballots in the presidential election. Mr Mousavi indeed won the overall share of the overseas vote. However this belies the great variation evident in different parts of the world. It is manifestly clear that Iranian expatriates in the West overwhelmingly backed Mousavi (by about 80%) -which may explain a sense of outrage at the result of the election amongst such voters- while those in the Arab world were essentially split between the two main candidates. These results have not been disputed as being fraudulent since they tend to reflect what was predicted. However, what is of interest to us, are the results from Iraq, Syria and Saudi Arabia because votes cast here were by travelling Iranians who were resident in Iran and not expatriates. The share of the vote in these three countries essentially matched that of the overall vote (63.5 – 32.7) that is indicative of the fact that Iranians inside of Iran were voting by a margin of nearly 2:1 for the incumbent president.
The Chatham House report, though a preliminary one, clearly set out to cast doubt on the Iranian election without offering anything other than a superficial analysis which Dr Afrasiabi has called a “a poor substitute for any hard evidence.” The distribution of votes across the provinces and districts conforms to general trends and comports to a natural outcome. Statistical studies have proved inexact and inconclusive as far as detecting any real evidence of fraudulent manipulation. If cheating did occur, it must have been localized and generally restricted to remote parts of the country where the population levels would not have been significant enough to sway the final outcome –We thus conclude that the 10th Iranian presidential election is a genuine reflection of the will of the Iranian people and that Dr Mahmoud Ahmadinejad is the duly elected president of the Islamic Republic of Iran.
[For more information on the report of Reza Esfandiari and Yousef Bozorgmehr, contact firstname.lastname@example.org. ]