FiveThirtyEight provides a considered post showing that the graph alleging that a strong linear relationship in voting patterns is evidence of electoral fraud is not a smoking gun.
To be clear, these results certainly do not prove that Iran’s election was clean. I have no particular reason to believe the results reported by the Interior Ministry. But I also don’t have any particular reason to disbelieve them, at least based on the statistical evidence. If Moosavi truly did have the support of a majority of Iran’s citizenry, the best evidence we will have of that is what happens in the streets of Tehran over the next days and weeks.
He also provides a link to Juan Cole’s blog which gives top reasons for thinking the election was stolen.
this post-election situation looks to me like a crime scene. And here is how I would reconstruct the crime.
As the real numbers started coming into the Interior Ministry late on Friday, it became clear that Mousavi was winning. Mousavi’s spokesman abroad, filmmaker Mohsen Makhbalbaf, alleges that the ministry even contacted Mousavi’s camp and said it would begin preparing the population for this victory.
The ministry must have informed Supreme Leader Ali Khamenei, who has had a feud with Mousavi for over 30 years, who found this outcome unsupportable. And, apparently, he and other top leaders had been so confident of an Ahmadinejad win that they had made no contingency plans for what to do if he looked as though he would lose.
They therefore sent blanket instructions to the Electoral Commission to falsify the vote counts.
This clumsy cover-up then produced the incredible result of an Ahmadinejad landlside in Tabriz and Isfahan and Tehran.
The reason for which Rezaie and Karoubi had to be assigned such implausibly low totals was to make sure Ahmadinejad got over 51% of the vote and thus avoid a run-off between him and Mousavi next Friday, which would have given the Mousavi camp a chance to attempt to rally the public and forestall further tampering with the election.
This scenario accounts for all known anomalies and is consistent with what we know of the major players.