Only hours after polling centres had closed on the evening of the second round of voting in the Afghan presidential election, the country’s Independent Election Commission (IEC) was already giving the first turnout figures. They were high. Some media broadcast them without question; other outlets were more sceptical, citing anecdotal evidence suggesting turnout might actually have been lower. Either way, doubt was cast on the IEC and Dr Abdullah’s team has since held the figure up as evidence of organised fraud by the Ghani team and the IEC. Thomas Ruttig, a senior analyst at AAN, has had a close look at the various statistics available, collected and compared additional data and done some maths. He comes to the conclusion that the IEC’s optimistic announcement stood on shaky ground, as all of Afghanistan’s post-2001 election were conducted on the basis of outdated statistics. That this has not been cleaned up after any of the previous elections, has made it easier to make complaints – whether justified or not – about the actual process of ballot and count. (With contributions by Kate Clark.) 1, 2, 3, 4 .... 7.1 million? How many Afghans have really voted in the second round? Hard to say, writes Thomas Ruttig, looking at the contested turnout figures, as all post-2001 Afghan elections have been conducted on the basis of inconsistent data that can be interpreted to fit any political agenda. Photo: Borhan Osman
On the evening of the second election day, 14 June 2014, the chairman of the Independent Election Commission (IEC), Yusef Nuristani, informed a press conference that, as in the first round on 5 April, there had again been a high voter turnout: about seven million people, and that this represented a turnout of about 58 per cent (62 per cent men and 38 per cent women). However, in its extremely meagre press release published later on its website, the IEC restricted itself much more, saying only that “while final figures are yet to be ascertained, preliminary indicators suggest a turnout of more than 7 million voters.”
It seems that even the IEC concluded that Nuristani – who may have felt obliged to say something early in the process – had said more than available data allowed. After all, counting had only just started. IEC member Abdul Rahman Hotaki explained in a TV interview broadcast two days later, on 16 June, that Nuristani’s figures had been based “on sampling we did on the distribution of ballots papers in the polling centres.” Indeed, even the IEC’s provincial offices did not have full figures at that time, as we double-checked for example with the one in Bamyan.
Figures not adding up
As Hotaki stated, the turnout percentage is, in general, calculated on the basis of the total number of ‘potential votes’, meaning ballot papers available. In most countries, this kind of calculation would be done on the basis of the number of registered voters. But the series of post-2001 voter registrations has been so botched that no reliable, consolidated voter registry exists. The number of distributed voter cards has reached 21 million which means that between 7.5 and ten million additional voter cards are floating around that are not linked to real voters. (1) However, the IEC does not explain anywhere on its website, including in any of its “fact sheets”’ – neither in the one on voter registration nor in the one with figures on polling stations and IEC staff – which data it bases turnout and other calculations on. It does not even give the figures of how many people have registered as voters.
Moreover, Nuristani’s figures do not add up in the first place. His absolute figure (number of voters who turned out) does not match the relative figure (percentage). Seven million is not 58 per cent of the distributed voter registration cards (which should be equivalent to the number of registered voters but is not, due to massive over-registration). Nor is seven million 58 per cent of the 13.5 million maximum number of eligible voters (of 18 years of age or older) or of the number of ‘potential votes’.
To be on the safe side, let us do the electoral maths using both ways of calculating outcomes, looking at the ‘potential votes’ first. Given that the IEC had planned 23,136 polling stations to be open on election day (see our analysis here) and that each polling station was provided with 600 ballot papers, there would be 13,881,600 potential votes, ie ballot papers. However, 58 per cent of this would have produced a turnout of just over eight, rather than seven million. It would have been even higher if we had taken as a basis the number of 15 million ballot papers ‘prepared’ for the second round, the figure provided by IEC member Abdul Rahman Hotaki in his 16 June TV interview. (Many of the extra ballot papers were for contingency, which is a standard procedure.)
Hotaki also said in the interview that the IEC had prepared one million more ballot papers in the second round, ie 15 million, than in the first round (14 million) – “which means that almost 1.5 million more citizens found the opportunity to vote.” This is nothing if not cryptic and really raises the question if Hotaki fully understands the figures on which the electoral procedures are based. (Also, when giving the number of polling stations prepared for the second round he used a figure from an outdated list. (2))
The lack of a census and a voter registry
As we have shown earlier, the maximum number of eligible voters – calculated on latest official Afghan population figures – is 13.5 million. Calculating the turnout based on this figure, 58 per cent would be 7.83 million, and a seven million turnout would be 51.8 per cent. The difference to the figure of seven million is substantial, almost as large as the difference between the official figures on Dr Abdullah’s and Dr Ghani’s share of votes in the first round (2,973,707 and 2,082,417, see here).
Another, and maybe the mother of all problems with numbers in Afghanistan, is the lack of reliable population data. The Afghan government and its electoral institutions rely on the latest figures of the Central Statistics Office (CSO) which puts Afghanistan’s total population at around 27 million in 2012 (25.5m settled and 1.5m nomadic; here). But all figures used today, including those of the CSO hail from a census that was carried out in 1979. It was never completed and consisted of samples and earlier household listings only – it was actually never supposed to be a full census in the first place. The new census envisioned in the final document of the 2001 Afghanistan conference in Bonn has never been carried out, stuck in technical and security problems and blocked by the controversial question whether the potentially polarising ‘ethnicity question’ should be asked.
Furthermore, there are discrepancies between the existing CSO figures about the population in certain provinces and the reported high turnout. The Abdullah campaign, for example, as well as complaining about the seven million turnout figure, also specifically questioned the turnout in Khost province. It says a turnout of around 440,000 in the second round (compared to 113,000 in the first round) was a sign for massive fraud. (3) It further points out that the CSO figures indicate a total population of 525,800 for Khost, and it does indeed seem unlikely that the province would then turn out 400,000 voters given that more than half of the Afghan population is considered to be under the voting age of 18. However the population figures are themselves estimations and their reliability is questionable.
Moreover, as the Ghani team has pointed out, the CSO figures are not the only basis for estimating population. Its spokesman, Faizullah Zaki, pointed to statistics from the National Solidarity Programme which he said suggested the population of Khost province is about 1.3 million. (4) He also said that Dr Abdullah had agreed and approved, as Dr Ghani had done, the number and location of ballot papers to be sent to the provinces. The 420,000 ballot papers, plus five per cent reserve, for Khost, he said, had been agreed by both candidates. “If the Abdullah team had thought the actual number was less, why didn’t they complain at the time?”
Window-dressing tops solutions
Summing up, it appears as if we do not simply face a maths problem, but a political one. We have discussed transparency, or the lack of it, in the country’s electoral institutions several times (here and here). We have also found how superficial transparency seems to be motivated mainly by attempts to ‘pacify’ complainants. This attitude seems to re-appear regarding election data. There is unwillingness to provide exact figures, as, for example, in the case of the number of election observers, witnesses and candidates’ agents. AAN asked for them, got ‘round’ figures and the promise that exact figures would follow; we are still waiting. The Afghan media is constantly facing similar problems and has repeatedly complained about the lack of access to information and the unwillingness of Afghan institutions to inform the public. The IEC has also managed to bury the gaps in its knowledge under a big amount of other data on its website. In other words: window-dressing continues to top real transparency.
On the back of complaints about the IEC’s turnout figure, Dr Abdullah decided on 18 June to withdraw observers from the IEC, effectively boycotting the count. The situation is now very grave indeed. But in order to solve it, it needs to be understood that the problems are just not a product of the 2014 election. All post-2001 Afghan elections have been conducted on the basis of out-dated statistics and inconsistent data that can be interpreted to fit any political agenda. This and other fundamental problems that need reform have been known and put on record since the first election cycle in 2004/05; in 2010, Democracy International even produced a summary of “consensus recommendations”, based on the a “hundreds of documents” available. (5) It is the joint failure of the international community and the Afghan government – of which both current candidates were members for a number of years – not to have done this homework, particularly in creating a consolidated voter registry.
(1) Apparently, this was also still used in the first round, see our reporting then.
(2) Hotaki said that the IEC had added more than 2540 polling stations to the number used in the first round on 5 April, as a reaction of the surprisingly high turnout then and the resulting ballot paper shortages. According to him, this brought the number of polling stations to 23,313. This, however, is a figure from an outdated list; the final number published by the IEC on 10 June was rather 23,136 (see here). Hotaki also admitted that he did not know how many of the 15 million ballot papers prepared had actually been distributed to the polling stations; he said this question came “too early” and might have to do with the fact that the IEC has not full picture about the contingency polling stations and additional ballot papers distributed to ‘regular’ polling stations where ballots run out early.
(3) The figures of turnout and CSO comparison in Khost originated with the Abdullah team and were cited in a number of newspaper articles after Dr Abdullah gave reporters an off-the-record briefing (which several shared with AAN) on 16 June. Asking his spokesman about this, however, AAN was told, “The [turnout] figure was based on rumours… We need some time to figure out the exact figures.” And when asked again, he said, “We don’t know about the source, but rumours are flying all over the city.” He continued to cite the 400,000 turnout figure for Khost during the interview, however.
(4) We looked for the statistics on the National Solidarity Programme website but could not find them.
(5) There was also a ‘Declaration of Principles for Electoral Reform’ published on 17 October 2012 by FEFA and 16 other Afghan civil society organisations, including most of the important networks, and 34 political parties from across the political spectrum, both opposition and pro-government (see an earlier AAN analysis of the document here). Unfortunately, the full document is not available under the link we gave then anymore.
This article was last updated on 9 Mar 2020
Central Statistics Organisation