Nate Silver

Ken Ballen and Patrick Doherty, who work for a nonprofit group called Terror Free Tomorrow: The Center for Public Opinion, are out with a comment in today's Washington Post which claims that their poll of 1,001 Iranians conducted last month predicted Mahmoud Ahmadinejad's victory.

Ballen and Doherty are doing admirable and important work. Regular readers will know how difficult it is to conduct a good poll in the United States. Take that difficulty to the fifth power, and you'll have some sense for how difficult it is to conduct a good poll in Iran.

Unfortunately, while the poll itself may be valid, Ballen and Doherty's characterization of it is misleading. Rather than giving one more confidence in the official results, the poll raises more questions than it resolves.

Ballen and Doherty wrote in the Post that their poll showed "Ahmadinejad leading by a more than 2 to 1 margin -- greater than his actual apparent margin of victory in Friday's election." But let's look at what their poll (.pdf) actually said:

Well, indeed, Ahmadinejad has more than twice as much of the vote as his next-closest rival, Mir Hossein Mousavi. But he also only has 33.8 percent of the total vote. Between them, indeed, Ahmadinejad and Mousavi only have 47.4 percent of the vote. Where does the rest of the vote go?

Not to the other candidates, Karroubi and Rezai. The poll -- correctly, apparently -- predicted that they would only account for a small fraction of the vote.

Some 7.6 percent of the respondents said they weren't planning to vote for anybody. There's nothing particularly suspicious about this; turnout in Iran, while high by American standards, is certianly not 100 percent and this poll was not screened for likelihood of voting, as most American-based polls are. An earlier question, Q25, asked people directly whether they intended to vote in the June elections and 6.8 percent said no, closely matching this figure.

It's the other two categories, however, which give one pause.

Firstly, some 27.4 percent of Iranians told TFT they were undecided. By comparison, a month before the U.S. presidential election, about 5-9 percent of respondents generally claimed to be undecided. Perhaps it is folly to try and extrapolate the Western experience to Iran -- but for 27 percent of the voters to claim to be undecided one month before a high-profile, high-turnout election strikes me as unlikely. Iran is a relatively sophisticated and dare I say stubborn country where people debate politics regularly and vigorously. They might not have told TFT whom they were planning to support -- but that doesn't mean they were truly undecided.

And indeed, the next category speaks directly to this. Some 15.1 percent of respondents refused to disclose who they were voting for. This is not mere modesty; indeed, the Iranians in TFT's survey were very forthcoming about a whole host of controversial issues, ranging from their perception on Iran's governance structure to their feelings about the United States and Israel to their opinions on Iran's nuclear program. On these questions, just a couple percent of the respondents refused to answer -- but the number shot up to 15 percent for the Presidential tally.

Ballen and Doherty claim that the forthcoming and surprisingly liberal responses to these other sorts of questions give their poll more credibility:

Ballen and Doherty's point would be worthwhile -- if support for Ahmadinejad was as robust as they were claiming. But instead, the presidential preferences of almost half of their survey cohort are unaccounted for -- 42.5 percent either refused to answer the question or said they were undecided, and another 7.6 percent said they weren't planning to vote at all.

While it is dangerous to make inferences about the preferences of undecided voters, the fact that the Iranians in their survey did tend to favor reformist positions on most issues, and had generally tepid reviews of Mr. Ahmadinejad performance, would seem to provide a few hints. For example:

And on Mr. Ahmadinejad's performance:

The Iranians were more conservative on other issues -- they overwhelmingly favored the development of a nuclear energy program, for instance, although were more evenly split about nuclear weapons. They had a quite negative perception about the United States government, although they were largely sympathetic toward the American people themselves.

Overall, however, the poll revealed that Iranians gave Ahmadinejad tepid reviews on the performance of the economy, and favored a much less bellicose foreign policy than he has pursued. One would think that under those circumstances, the incumbent would be in a fight for his political future.

Let's accept the poll's contention that about one-third of Iranians -- or about 36 percent of those who were planning to vote -- are hard-core supporters of Mr. Ahmadinejad. There are certainly conservative elements in Iran, and I have no particular reason to doubt this figure. That still leaves Ahmadinejad far short of the margin he would need to carry the election, however.

But here's the catch. If you have 15 percent of the electorate refusing to say whom they'll vote for, and if probably about half of the 27 percent in the "don't know" category are in fact "soft" refusals who are similarly reluctant to reveal their preferences, those votes won't necessarily have wound up in Mr. Mousavi's column. If these Iranians were too intimidated to reveal their preferences to a pollster, they may also have been too intimidated to vote as they really pleased on Friday.

The swing votes in Iran are not those blue-haired ladies who take 40 minutes in the ballot booth and call the election clerk over every few minutes. They are rather the perhaps 30 percent of the population who were trying weigh the potential risk to their persons or their standing in the community in voting against Mr. Ahmadinejad, against what might be a relatively small benefit in voting for Mr. Mousavi, whose reforms could be easily vetoed by the Ayatollah. These swing voters may also have been worried that their votes wouldn't have been counted anyway: about one-third of Iranians in the survey didn't believe, didn't say or didn't know whether they expected to have a free and fair election.

If you take that 30 percent swing vote and add it to Ahmadinejad's 33 percent base, he could have won the election with 63 percent of the vote, as he ostensibly did on Friday. If you take it and add it to Mousavi's column, Ahmadinejad would have gone down to a solid defeat.

The point that few commentators are realizing -- Al Giordano is an exception -- is that this story really isn't about the way that the votes were counted. It's about whether Iran is capable at this point of having an election in which the democratic will of its electorate is properly reflected. If Ahmadinejad hired a bunch of thugs to hold every Iranian at gunpoint while they were casting their ballots, it would not have been difficult for him to get 63 percent of the vote -- indeed, he'd probably have wound up with very close to 100 percent. This would be an election -- and there would be no need at all to tamper with the results. But it wouldn't be an expression of democracy. We need to separate out those two concepts. Ahmadinejad, as far as we know, did not go so far as to hold anyone at gunpoint. But the tentacles of fear in Iran run deep.

¤ ¤ ¤ ¤ ¤

Source: http://www.fivethirtyeight.com/2009/06/did-polling-predict-ahmadinejad-victory.html
Photo: http://farm4.static.flickr.com/3328/3629447955_6dba6caa69.jpg?v=0