Afghanistan’s elections: let’s talk turnout

The IEC has released a figure for the indicative turnout of yesterday’s poll (40 percent). It is now being widely repeated and compared to other figures, including previous elections and turnout percentages in our home countries. It happens so often. For some reason nobody finds it necessary to understand where these random figures come from and what they mean. For some reason nobody finds it necessary to ask: 40 percent of what?

The IEC is, to their credit, quite transparent about where the "indicative turnout" of 40 percent came from:

IEC Indicative Turnout: Polling Centers reported 4,632 turnout by 16h. The maximum number of voters allocated to these 4,632 PCs corresponds to 9,203,586 voters. 3,642,444 votes were cast by the end of polling representing 40% of the maximum number of voters (9,203,586).

So they seem to have counted the total number of ballots that were given out (and only those that were given to polling stations that reported turnout on polling day) and divided them by the total number of votes cast at these polling stations. That is not a turnout figure. It just tells you that 40 percent of the distributed ballots were used.*

For anyone who wonders how this tallies with past elections, here is a short list to refresh our memories (and don’t forget these totals represent the number of ballots cast; there is no way to know how many voters these figures represent):

2004: 7.4 million (first presidential election)

2005: 6.4 million (first parliamentary and provincial council election)

2009: 4.8 million (presidential, post-audit) / 6.0 million (provincial council)

2010: 3.6 million (parliamentary election)

The declining trend signifies several things, most prominently a growing disillusionment and disengagement with the process, and the impact of a worsening security situation; but also factors such as a decrease in the number of ballots distributed (during the first presidential election 18 million ballots were printed) and a decrease in the number of polling centres. Basically there were fewer opportunities to cast ballots – either through a genuine vote, or through the various forms of electoral fraud.

Turnout figures have always been a problem in Afghan elections (see also last year’s pre-polling blog on the same subject). Nobody knows how many voters there are in Afghanistan, mainly because nobody knows how many people live in Afghanistan (although CSO estimates have improved and now stand at 23.6 million in mid-2004). Over the years 17.5 million voter cards have been distributed and it is now widely accepted that several million of them are either duplicates or not linked to actual voters. Recently the IEC has been using the figure of 12.5 million estimated voters, while UN SRSG Steffan di Mistura floated the figure of 10.5 million estimated voters, just days before the election.(**)

It is reasonable to adjust the total number of estimated voters and to stop using the inflated voter card figure, but to announce a turnout figure of 40 percent, that is based on an unusual calculation, and that has not been corrected or caveated for fraud-induced inflated numbers is a misrepresentation. 

* Otherwise, the less ballot papers you distribute, the higher your turnout would be. Actually, in several areas polling centres ran out of ballot papers, with voters waiting for more to come. Does this mean that turnout in those areas was 100 percent? Or more than 100 percent? 

CORRECTION / ADDITION: The IEC released a fact sheet on their voter turnout calculation after the posting of this blog. It states that "turnout figures are expressed as a percentage of voters who could have voted and did". The baseline figure is the 11.4 million voters that the IEC prepared for, minus the polling stations that did not open on polling day, and – in a later stage – the polling stations excluded by the IEC or ECC. This means that turnout is calculated without counting (1) the eligible voters living in areas where polling centres could not be secured, and (2) the eligible voters whose votes have been used in electoral manipulation.

** He meant to reassure by arguing that additional fake voter cards would not change the trajectory of the election: "There are already 17.5 million cards, legal cards, issued since 2003. But the previous 2005 elections, or the largest number of voters, in 2004, was not more than seven million. And this year, based on the statistics, based on previous elections and national statistics, the voters who are eligible for voting are likely to be around 10.5 million, not the figure of 17.5 – 10.5 million. In these last elections there were around five to six million who actually voted, so let’s put things into perspective."

Martine van Bijlert is the co-director of the AfghanistanAnalysts Network, where this was originally published.