The devastating statistic comes from the third bullet above. According to the official tallies, there were over 10.6 million new voters in 2009, which is usually a sign of a reformist upsurge. To have achieved his reported victory, the conservative Ahmadinijad would have had to have captured all former conservative voters, all former centrist (Rafsanjani) voters, all new voters, and up to 44% of former reformist voters. This could not occur except in someone's wildest fantasies, hence, sadly, there was massive rigging of the count. The interesting question is why the rigging was done so crudely, and the most likely conjecture is that it was done by the Khamenei government under pressure as early results showed Mousavi to be the winner and "corrective action" had to be done on the spur of the moment. If so, this gives credence to the comment from the senior author of the Chatham report, quoted in The New York Times article cited above, despite its pouring gasoline on the fire rather than oil on the roily waters.
I don't think they actually counted the votes, though that's hard to prove," said Ali Ansari, a professor at the Institute of Iranian Studies at the University of St. Andrews in Scotland and one of the authors of a study of the election results issued by Chatham House, a London-based research group.
The tension between the Ballen & Doherty (Rockefeller Brothers Foundation-funded) research article and the Chatham results should, on current evidence, be resolved in favor of the latter, especially since Chatham uses the data set forth by the Iranian government. There are numerous rebuttals of both the liberties taken by Ballen & Doherty in their Washington Post op-ed article and of their polling itself. Their op-ed article refers to the study as being "three weeks" before the election, but in fact it was conducted from May 11-20, a range of 24-33 days prior to the poll. Although they report finding more than a 2-1 margin favoring Ahmadi-Nejad, that was amongst those who revealed which candidate they would vote for (34% vs.17% of those polled). A full 27% were undecided and 22% did not support anybody (i.e., likely: none of your business). Critiques may be read here:
http://www.openleft.com/diary/13774/on-that-iranian-poll
http://voices.washingtonpost.com/behind-the-numbers/2009/06/about_those_iran_polls.html
http://www.juancole.com/2009/06/iran-election-fraud-moaddel-on-ballen.html.
By contrast the Chatham report looks stronger upon examination and reflection. The dramatic impossible requirement of capturing all former conservative voters, all former centrist (Rafsanjani) voters, all new voters, and at least some of the reform voters occurs in only 10 of Iran's 30 provinces. Superficially, this suggests that 20 of the 30 results were legitimate and perhaps enough to hold on to the Ahmadi-Nejad victory. But that is a superficial analysis. The 10 provinces where the fraud could be explicitly proven are those where, not surprisingly, there were relatively few votes for Ahmadi-Nejad in 2005 and relatively few non-voters in 2005 - that is why the statistical ballot stuffing could be easily detected and reported. But anyone rigging the vote count, -- i.e., the Guardian Council -- would not, in its wisdom, decide to rig the vote in just that third of the provinces where the 2005 electoral background would allow such detection. The rigging would occur as needed, although the 2005 distribution in the other 20 provinces (where there were more non-voters and more who voted for Ahmadi-Nejad) would not allow the same demonstration of rigging. Thus, the idea that the rigging occurred only these 10 provinces and would not alter the overall outcome is not supportable. Thus, on current evidence, the Ballen and Doherty results should be scanted in favor of the later and much better grounded Chatham analysis.
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).