Today's Washington Post op-ed by Ken Ballen and Patrick Doherty concerning a public opinion survey they conducted in Iran on that nation's presidential election is both worth a read, and highly disturbing for what it omits. Here is the first paragraph:
The election results in Iran may reflect the will of the Iranian people. Many experts are claiming that the margin of victory of incumbent President Mahmoud Ahmadinejad was the result of fraud or manipulation, but our nationwide public opinion survey of Iranians three weeks before the vote showed Ahmadinejad leading by a more than 2 to 1 margin -- greater than his actual apparent margin of victory in Friday's election.
Anyone regularly involved in poll analysis would notice two red flags in this paragraph:
Because of these two rhetorical slight of hands that were employed instead of simply listing the poll results and actual dates the poll was completed, my sense of the integrity of this poll drops immediately. It wouldn't even have taken any longer to just list the numbers, but for some reason the pollsters immediately employed rhetorical moves instead.
- The actual results of the poll were 34%-14%, with 27% undecided and 22% falling into some strange category of not supporting anyone nor being undecided. To instead report the results as "more than a 2 to 1 margin" is to use a rhetorical trick that creates more similarity between the final result and the poll than the numbers actually suggest.
- The poll was conducted from May 11-20, which is actually a range of 24-33 days before the election, not "three weeks." To call it three weeks instead of providing the actual dates make it sound as though the poll was taken closer to the election than it actually was.
Undecideds break heavily for lesser known challengers within the context of American elections. It is difficult to imagine why that would be any different in other countries, given that the same phenomenon of the electorate making up its collective mind about an incumbent before the election season would still be in effect. Further, check out the following two paragraphs from the poll that they failed to mention entirely in their op-ed (hat-tip Juan Cole):
' A close examination of our survey results reveals that the race may actually be closer than a first look at the numbers would indicate. More than 60 percent of those who state they don't know who they will vote for in the Presidential elections reflect individuals who favor political reform and change in the current system.'(...)
The current mood indicates that none of the candidates will likely pass the 50 percent threshold needed to automatically win; meaning that a second round runoff between the two highest finishers, as things stand, Mr. Ahmadinejad and Mr. Moussavi, is likely.
So, due to the large number of undecideds and the mood of the undecideds, in writing about their own poll, Ballen and Doherty actually predicted a second-round run-off with no one reaching 50%+1. However, when official results come out showing the incumbent winning with 63%, they publish an op-ed in the Washington Post suggesting that the election was clean, based on their own polling?
I don't know anything about these pollsters. However, it is pretty bizarre to argue, at different times, that a single poll both shows the incumbent likely to not reach 50%, and also that a 63% result for the incumbent was probably legitimate. I really can't wrap my head around that one. Even though I absolutely hate making implications like this, such a strange turnabout in analysis is enough to make one wonder if there is another agenda at play.
Anyway, this isn't going to be settled by statistical analysis. Right now, the real numbers to be asking about are how many people are protesting, and how long they will continue to do so. On that front, the number seems to be rising from thousands, to hundreds of thousands, to millions.