I’m obsessing over the aftermath of the Iranian election today, and thought our readership might be interested in this analysis by Walter Mebane, from UMich, of Iranian election results.

He’s taken the Ministry of Interior‘s posted results (or a copy in Google docs), and done a quick check against Benford’s law, which predicts the statistical distribution of initial digits of numbers. With the available data (returns for ~350 districts), he reports that nothing looks particularly wrong, but would love to have more detailed (polling station level) data, as you shouldn’t expect to see much anyway.

Nate Silver also has a critique up of some earlier statistical complaints made about the announced election results. If anyone sees something else along this lines, I’d love to hear.

### Like this:

Like Loading...

*Related*

By the way, there’s a newer post of 538 pointing out that flashy statistical tricks are unecessary: the official numbers are completely preposterous when compared to previous elections.

That’s extremely interesting.

If he wants to read the numbers in Eastern Arabic Numerals, see http://en.wikipedia.org/wiki/Eastern_Arabic_numerals.

This is a meta-comment really, but it’s funny how low the correlation is between news and reality.

See also http://arxiv.org/abs/0906.2789, a paper by an astronomer which is a bit less cautious.