Stats Check!

Some more "partial results" from the Iraqi elections are out, covering the 10 southernmost provinces including Baghdad (with the proviso that only 45% of Baghdad polling stations have reported). The main line of reporting on this has been the very strong indeed showing of the UIA list - they could get as much as 66% of the seats in parliament on current trends, although that will fall as the Shia-minority provinces report. Juan Cole reckons they might finish up with 50% of parliament, which would put them in the position of choosing between a small-coalition solution with Allawi's lot and minor parties or a big coalition with the Kurds. A headline figure of 3.3 million votes counted is given.

What does this tell us about turnout?

It's hard to make any predictions, because it's not entirely easy to get population stats by province. That's why the election was based on the food ration cards, but I've yet to find figures for those. This site has historical tables and an estimate for 2002. Now, the current figure for Iraq's population is 21,722,287. The estimate given for 2002 is 27,022,200, or 24% greater. If we assume a constant error across Iraq (and frankly, who wouldn't assume constant errors in Iraq?), we arrive at a figure of 12,457,844 for the provinces in question (just the total adjusted down by 24%). The Iraqi government gives a (round!) figure of 14,000,000 electors in Iraq, or an eligibility rate of 64%. That in turn gives us an electorate of 7,975,020 for the areas reporting already. As the other headline figure is of some 3.3 million votes counted from those areas, we arrive at a 41% turnout.

Naturally, this is a very rough estimate: if the forecast error in the province figures for the south is greater than the north, obviously our adjustments will be out. If the figure of 14,000,000 electors is dodgy, then the percentages will be wrong. And if the other 55% of Baghdad polling stations report a widely different profile, then the whole thing could turn out (ha!) to be utter shite.

It certainly throws light on the claims of turnouts as high as 72% nationally, though. The currently-standard figure of 57% ("8 million voters"), though, certainly seems high, especially when it is taken into consideration that the low-turnout areas are still to report. If turnout there is on the optimistic side, 41% might well be achieved with the help of a good Kurdish showing. If not, though, it could well be wildly optimistic. If - say - the north's overall turnout is half that in the south, 20.5%, then an overall figure would be around 32%.

I'll buy a pint on that one.