The Iranian elections appear to be a mess. But were they fixed?

One of the most frequently cited pieces of evidence for those attempting to answer that question in the affirmative is a chart which appears at Andrew Sullivan's site. It shows 6 different vote announcements made by the Iranian Interior Ministry over the course of the night. After all the vote announcements, incumbent President Ahmadinejad maintained roughly the same vote share:

The gist of the argument that this is proof of an illegitimate election is quoted by Sullivan:

Statistically and mathematically, it is impossible to maintain such perfect linear relations between the votes of any two candidates in any election — and at all stages of vote counting. This is particularly true about Iran, a large country with a variety of ethnic groups who usually vote for a candidate who is ethnically one of their own.

Before I examine this any further, let me say that I hold no brief for Ahmadinejad. And I would love nothing more than for the entire regime in Tehran to come tumbling down.

But as someone who has watched a lot of election returns come in from a lot of places, the chart above doesn't seem that odd. For whatever reason, elections sometimes hew to their final results early on, and stay there.

Consider Kyle's and my recent liveblog of the Virginia Democratic Gubernatorial Primary. There aren't many more diverse electorates in America than Virginia Democrats. You have both urban and rural black voters, farmers, coal miners, large student populations and affluent suburbanites, all voting in the same primary.

Yet the results were steady as they came in last Tuesday. With 4% of the precincts in, Deeds was up 54%-23%-23%. This is awfully close to the 50%-26%-24% final outcome. With only a quarter of the precincts in, 30 minutes after the polls closed, the results were 51%-25%-24%. They changed little as the remaining 70% of the votes came in.

This isn't to say that results are always consistent. Just that they sometimes are, and that a consistency doesn't mean fraud (either that, or we have a much bigger problem with Diebold in this country that many believe!)

To further test this, I decided to run a simulation of sorts using U.S. Presidential Elections. I wanted to see if results trickling in for our Presidential elections have a different result than what we saw in Iran. Do we see substantial deviations from the trendline? After all, the US is a large, diverse country, like Iran. It is broken into various regions, all with different electoral loyalties. Some areas of the country are extremely Democratic (many Congressional districts gave Obama more than 80% of the vote in 2008), while some are extremely Republican.

Unfortunately, I can't find a list of different popular vote results reported at different times over the course of election night (this is unsurprising, since our focus is on the electoral vote). What I can do, however, is look at how the popular vote changed as state polls closed.

State poll closing times are here. We have nine poll closing times for the states, which gives us three more observations than we have in Iran. Also, our first poll closings are in Kentucky and Indiana, which we would expect to naturally skew things toward Republicans. So our system is set up to see deviations of the type we didn't see in Iran.

Although in reality it takes time to count state polls, here we'll assume for simplicity's sake that all returns for a state come in at once. In other words, we'll assume that at 7pm we had all the results reported from GA, SC, etc.

With the Obama-McCain matchup, we can plot the number of votes for McCain and for Obama at different poll closing times. We can then plot a trendline over them, and come up with something that looks an awful lot like the Iran chart. We see the following:

Looking at the chart, the deviations from the trendlines don't look that dissimilar to the slight deviations from the trendline we see in the Iran chart. And remember, the first states to close are IN, GA, SC, KY, etc – which means the election is front-loaded toward Republicans – while the last few pols to close are CA, OR, WA, HI and AK – a selection that is loaded for the Democrats. Yet we still see little deviation from the trend.

As Jay has demonstrated, 2008 was one of the most polarized elections in recent history (if not the most polarized). In other words, there was a lot of diversity in how states voted, which in turn makes it more likely that there would be substantial movement in the returns.

What if we look at an election with low polarization, 1988:

By the 8pm closing time, we have a 55%-45% Bush lead, which is very close to the final call. And looking at the chart we see only very minor deviations from the overall trendline; the largest deviation is from the first closing time; we would expect this for IN and KY in 1988.

Anyway, the Iranian election may well have been fixed. There are other strange events that occurred that are difficult to explain. But this chart doesn't seem to be particularly good proof that the election was fixed. Even with the vast diversity of our country, our Presidential returns look at awful lot like the Iranian ones. Proponents of regime change from within in Iran should focus on other, stronger talking points.