Political Landscape

Afghanistan’s 2019 Election (12): Scrutinising the IEC’s partial turnout figures

IEC workers unload ballot boxes to be taken to a counting centre in Kabul the day after the elections. Photo by WAKIL KOHSAR / AFP

IEC workers unload ballot boxes to be taken to a counting centre in Kabul the day after the elections. Photo by WAKIL KOHSAR / AFP

The Independent Election Commission (IEC) has provided partial turnout figures for Saturday’s presidential election, but the numbers and the manner in which they have been released are somewhat baffling. Two days after the vote, it is also still not clear how many polling centres opened on polling day. The fluidity of the figures and a sudden jump in reported turnout, from one to two million voters, raise important questions that the IEC will need to answer swiftly to avoid confusion and allegations that they may not be in control of their own process. The AAN team takes a closer look at the figures that have been released so far and explains why they may be troubling.

 Number of polling centres

Two days after Afghanistan’s fourth presidential election, it is still not clear how many polling centres opened. As reported previously (see this AAN election primer), the IEC had planned to open 5,373 polling centres across the country (the rest of the 7,366 had already been deemed too insecure). Ahead of the election, the Ministry of Interior added 451 more which it said it could not secure (see here). It was always expected that more would fail to open on E-Day. The question is: how many and how long would it take for this number to become clear?

On Saturday 28 September 2019, in the afternoon, Abdul Moqim Abdul Rahimzai, Director-General for Operations at the Ministry of Interior, which is responsible for election security, said that 4,905 polling centres had been “active across the country” and that 468 polling centres were “inactive due to different reasons” (see also this report). That number of 4,905 active polling centres was also given by the IEC in its Saturday press release and at their Saturday evening press conference.

The figure was, however, revised by IEC commissioner Mawlana Abdullah on Sunday. He was quoted in this Tolonews tweet, as saying:

5,373 polling centers were supposed to be operational, which makes 29,586 polling sites. From the original number, 617 centers remained closed & another 234 centers aren’t reliable and it’s difficult to say whether elections were held there or not.

According to this calculation, the IEC was working with a figure of 4,522 active polling centres with a further 234 centres whose status remains unclear. If those 234 all opened, the maximum possible number of open polling centres would be 4,756. Figures shared by Commissioner Abdullah at the IEC’s Sunday press conference also added up to 4,522. They comprised a list of 3,736 polling centres that already had reported turnout figures (published on his Facebook page and 786 more polling centres whose results, he said, were still on their way.

Emerging turnout figures

During the two days since the election, the IEC also provided partial turnout figures. It did so during press conferences held on Saturday evening and Sunday evening. During the first press conference, several hours after the closure of polling centres on Saturday at 5 pm, the IEC said that 1,051,998 people had voted in 2,597 polling centres (for a recording of the IEC press conference, see Tolonews’s 10pm bulletin, starting from the 48th minute). That number of polling centres would represent 48 per cent of all planned centres and 57 per cent of the 4,522 centres that commissioner Abdullah later said had opened.

During the Saturday press conference, the IEC also read out a provincial breakdown of the partial turnout, citing the number of polling centres and the number of votes cast per province. The numbers are shown in the first column in the table below. The IEC said these figures had been received “via mobile phones and reports by the provincial polling centres and station coordinators.”

In Sunday’s press conference, the IEC provided a set of updated partial turnout figures. According to those figures, the IEC had received reports of 2,196,463 votes cast in 3,736 polling centres (that would represent 70 per cent of all planned centres and 83 per cent of the 4,522 active PCs) and 17,689 polling stations (which would represent around 60 per cent of those planned. There are no figures yet for the number of polling stations that were active). The IEC did not mention how it received these later figures. It also said they did not include newly-received figures from Takhar, Faryab, Farah and Kunduz. A provincial breakdown by number of PCs and votes cast is shown in the second column in the table below, based on a photo of the results’ sheet that commissioner Mawlana Abdullah published on Facebook.

The IEC has so far failed to provide any information in writing, either on its largely idle website, or in the form of a media hand-out. Instead, it has so far exclusively used Twitter and Facebook, channels from which information can easily disappear, possibly depriving the country of an archive of its election data.

In the absence of a more detailed explanation of the partial turnout figures, it is difficult to know how to read them. However, the figures as presented raise questions that need to be answered swiftly in order to avoid confusion and to counter any loss of transparency.

Most importantly, there has been a huge jump in turnout between the first partial update, which was based on the numbers provided by 2,597 polling centres, and the second partial update based on the figures provided by 3,736 polling centres. Based on these figures, it appears that the additional reporting of 1,166 polling centres is supposed to be responsible for an additional 1,144,465 votes. This represents a near doubling of the total turnout, while the number of polling centres only rose by about 50 per cent. This seems particularly unlikely given that polling centres in secure areas are the ones that are likely to have reported back first and without problems. These are also the polling centres where most votes are expected to have been cast, as opposed to those in more remote and insecure areas, which tend to come in later.

It is, of course, possible that the figure given at the end of the first day, on Saturday, consisted of incomplete polling centre data, ie that a large number of polling centres reported the results of some, but not all of their polling stations, and only did so in the day after. However, it is unclear why this would be the case and why it would have happened on such a large scale – and why the IEC had not mentioned that this was the situation.

The figures as given suggest an average turnout of 124 voters per polling station (2,196,463 votes cast in 17,689 polling stations). Although this is not impossible, it does not seem to match most observations, even in urban and secure areas. Whether this average turnout is, indeed, credible should get clearer as observation reports are collated and shared.

Zooming in on the provincial level data

The provinces that reported the greatest increase in votes between the first and the second update have also tended to show the most implausible results. For example, Nangrahar, which reported a turnout of 22,813 votes from 309 polling centres on Saturday, a day later, suddenly reported 254,871 votes from 390 polling centres – more than ten times the number reported on E-Day. Helmand showed a similarly implausible increase:  on Saturday it reported a turnout of 18,641 votes from 68 polling centres, while on Sunday, this had jumped three and a half times to 70,218 votes from 61 polling centres, ie fewer centres than had been reported earlier (more on this below).

In the table below, similarly implausible instances of reporting have been marked in red.

Jumps in turnout that look suspicious are marked in red; jumps that are theoretically possible but require additional data to verify that they are plausible are marked in purple; a seemingly reasonable increase in turnout, or no changes, are not marked. Table: AAN

Jumps in turnout that look suspicious are marked in red; jumps that are theoretically possible but require additional data to verify that they are plausible are marked in purple; a seemingly reasonable increase in turnout, or no changes, are not marked. Table: AAN

Jumps in turnout are also noticeable in several other provinces. In Baghlan, turnout increased threefold between the two reports: on Saturday, 86 polling centres reported 13,695 votes, while on Sunday, 104 polling centres (18 more than on the first day) reported a turnout of 44,777 votes. Baghlan did experience a phone network lockdown in parts of the province, so it is plausible there was a delay in some of the reporting, but it is unclear why the IEC would then have reported so many polling centres already on the first day.

Maidan Wardak was a curious case, too, similar to Helmand. On Saturday, the IEC reported a turnout of only 4,763 from 67 polling centres. Sunday’s figure was four times that, 20,560, but from fewer polling centres (64). No explanation for this drop in centres was provided.

Maybe the most interesting cases are the provinces that had already reported votes from all polling centres that were allocated to them. For example, Panjshir reported a turnout of 10,144 from all 96 polling centres on Saturday, while on Sunday almost a doubling in turnout from the same number of centres was reported – suddenly there were 19,205 votes. (The day before, local authorities had already reported a figure of 19,000 – see here) In Kunar province, the same thing happened. On Saturday, the turnout from all 110 Kunar’s polling centres was 39,998; on Sunday, with the same number of polling centres, it had doubled to 67,383 votes.

Again, it is possible that these figure represent a lagging of the different systems – that the IEC, for instance, had already started reporting the number of polling centres that had called in their results, while the actual votes had not yet been centrally tallied. However, to clear up the confusion, the IEC should release the details of their tally process, as well as the lists of polling centres that did and did not open, and of the polling centres that called in their reports late. It is, in particular, important to understand where the high increases in reported turnout in certain provinces came from.


Some form of confusion is inevitable in a country as geographically and infrastructurally diverse as Afghanistan. This is already reflected in the electoral calendar, that foresees that the announcements of the preliminary results will take place more than three weeks after election day. Not all polling centres are able to report back on the evening of the election, or even in the immediate days after, if only because of a lack of internet and/or phone coverage in their areas. Experience from previous elections, however, shows that communication is sometimes intentionally cut to create confusion and opportunities for manipulation. The seeming sudden leaps in voter numbers do seem reminiscent of previous elections.

For this reason, the IEC needs to explain the basis of its figures and convince voters and observers of their solidity, if it wants to reassure its audience that it understands its own data and is still in control of it. This is particularly relevant in the face of building fraud allegations. For now, it is unclear whether the IEC’s lack of transparency is because it is simply struggling to respond to the situation, or whether it is intentionally providing only the bare minimum of data. Whichever is the case, it creates the impression of an organisation not wanting outsiders to look into the workings of the election or follow what it is doing.

Tagged with: , , ,
Thematic Category: Political Landscape