Sunday, June 14, 2009

Put a little effort into your fraud

Since a) I'm really not an expert on Iranian politics and b) I really didn't know how to pick a side in the recent elections, I've refrained from weighing in on the allegations of voter fraud there. But Andrew Sullivan's graph (via LGM) is a dead giveaway:

In other words, as the results of the election came in, they steadily reported the exact same vote share for the two candidates. This, of course, means that the state has zero credibility on the matter of elections.

3 comments:

Josh Putnam said...

I found Nate Silver's exercise in applying the same type of "wave returns" to the 2008 US presidential election interesting.

Seth Masket said...

With due respect to Nate, I think he's wrong on this one. The example he uses divides up the American states alphabetically. That's basically the same as saying that election returns will come in randomly. So of course it will basically follow a linear path, as each wave is a random sample of the national vote.

But we know that the speed at which districts can count and report votes is not distributed randomly. At least in the U.S., it often takes longer in poorer districts, more rural districts, etc. Some areas use different voting technologies, which can dramatically affect reporting speed. These differences are somewhat correlated with vote choice. For an example, just think back to Florida during the 2000 presidential election. Early returns were so favorable to Gore that most networks felt comfortable calling it for him. Then they un-called it, and a few called it for Bush when returns started overwhelmingly swinging his way. The final vote, of course, was almost exactly 50-50.

I don't know how Iran conducts its elections, whether there are substantial differences in voting technology across different regions, etc. But the idea that each wave of reported votes would be a near-perfect random sample of the entire national electorate really, really strains my credulity.

Milan said...

Apparently 65-70% of Iran's population is 30 years old or younger. Also, according to Joe Biden in a television interview 70%of the Iranian vote comes from the cities. These two demographies highly supported Mousavi going into the election, while the older/rural, more "religiously conservative" voters supported Ahmadinejad. The demography of the increased turnout would also benefit Mousavi more than Ahmadinejad.

I'm not too familiar with this field, but with seeing the majority of Iranian voters belonging to a dominant demography, I wouldn't be surprised that in reality, the exact same vote share for the two candidates were steadily reported. I only believe that Mousavi would be leading Ahmadinejad 2-1 rather than the "real results". Many experts believed that the real results were "flip-flopped" for Ahmadinejad.

This would be interesting to see!