What to Watch in the Elections (4): The count, the complaints and the result
Elections in Afghanistan can be overwhelming, surreal and utterly confusing, even for those who have followed the politics of the country for a long time. To bring those who were not here during previous elections up to speed, and to remind those who may have forgotten the details of what it was like, AAN‚Äôs Martine van Bijlert answers some crucial questions. She discusses voter registration, the securing of polling stations, election day fraud, and the laborious process to settle on an election outcome ‚Äď in four installments. In this one, the fourth, she looks at¬†what happens after the polling stations close: the count, the complaints and the results.
What is the most important thing we need to know about the period after election day?
The election is not won or lost on election day, but in the processes after: during the audits and disqualifications by the Independent Electoral Commission (IEC) and the complaints adjudication by the Independent Electoral Complaints Commission (IECC).
These are messy, contested and confusing processes that can take a long time to sort out. The IEC will be faced with the hugely unattractive task of deciding which votes to count and which to discard, while the IECC will be sifting through hundreds of complaints ‚Äď some serious and with far-reaching consequences, others frivolous, off-point or fabricated ‚Äď hampered by the fact that its provincial offices were set up very late. Well over a million votes were disqualified in both 2009 and 2010, representing respectively 20 per cent and 25 per cent of the total vote, in processes that were often opaque. (1) Quite likely, this time will be the same.
A complicating factor is that there is no agreed mechanism to solve disputes, should they arise; this has been improvised every single time. In 2009, the UN designed a sampled audit in response to a dispute between the IEC and the IECC, while in 2010 the president called in the Supreme Court and established a highly controversial special tribunal (see AAN reporting¬†here and also this recent report by AREU).¬†With the waning role of the international community, both directly in the elections but also more widely, it is not quite clear how a possible dispute may be resolved this time. There may be a role for President Karzai, or for whatever institution he calls in, if an impasse should arise.
When it comes to fraud mitigation, one of the biggest problems the IEC faces is that a strict implementation of its own anti-fraud guidelines will most probably lead to massive and geographically uneven disqualification (more in insecure areas, less in secure). In the run-up to the 2014 elections, discussions around the idea of ‚Äėinclusivity‚Äô suggest that the IEC might seek, or be encouraged to seek, a balance between strict fraud correction and making sure that all areas are represented in the results. This introduces an enormous discretion into the process, as the IEC decides which suspicious polling stations to count and which to discard.
So let‚Äôs start at the beginning: What happens on election day once the polling stations close?
The count and tally procedures are largely the same as during the 2009 and 2010 elections, with the initial count taking place at the polling stations immediately after they have closed for voting. (In 2004 and 2005, the count was done at regional and provincial count centres, respectively. The idea, at the time, was that removing the count from the area would better protect it against intimidation and interference, but many candidates and local observers felt that that it only complicated the observation and many were suspicious of tampering during transport.)
At the end of the day, the ballot boxes are opened, the numbers on the seals recorded at the beginning of the day are noted again, and the ballots are counted, twice, in the presence of observers ‚Äď if there are any. The results of the count are recorded on the Reconciliation and Results Form, both in letters and numbers. The recorded results on both the form and its copies are covered with transparent tape to avoid tampering. One copy of the form is posted at the polling station; the others are handed to observers. The ballots are then returned to the boxes, which are resealed with new numbered seals. The forms are placed in tamper-evident bags (these are bags that cannot be re-opened without it showing) to be sent to Kabul.
In areas where observers can move freely, the count at the polling stations can be quite a crowded affair with many onlookers. In other areas, however, the count will be largely unmonitored, sometimes after observers have been forcibly removed or told that their presence is no longer needed.
After the count, all material is handed over to the District Field Coordinator (DFC); this may be on the same evening or much later, depending on logistics, how long the count takes and whether the staff wants to part with the ballots (or still may want to tamper with them). The material is then transported to the provincial office or warehouse, which could take days or longer, where the sealed boxes are logged and stored. The tamper-evident bag with the forms is sent to the tally centre in Kabul.
The IEC has recognized the importance of control and has sought to improve the tracking of all sensitive election material as it moves through the system and throughout the country, using sign-off forms and barcode readers. However, in practice this will not prevent local strongmen or electoral staff from seizing whole ballot packs (boxes, ballots, seals), fabricating a vote and trying to cover up after the fact.
The tally process in Kabul, according to the IEC, is a ‚Äúseven step-process, broken into three separate phases,‚ÄĚ¬†which include (a) intake of the tamper-evident bags and verification of the documents, (b) auditing of suspicious shipments, and (c) tabulation of the results.
There will probably be an explosion of rumours and reports of fraud in the days after the vote, as all candidates, supporters and observers are acutely aware of its importance in the competition. Some reports will be accurate, detailed and based on facts; some will be vague; and some will be exaggerated or even fabricated. Social media will likely play a big role, with various campaign teams posting photographs and videos showing evidence ‚Äď real or fabricated ‚Äď of fraud perpetrated in favour of their rivals. There will likely to be a lot of noise, with candidates and their supporters clamouring for audits and disqualifications, as well as for the re-insertion of quarantined polling stations.
What is the official timeline?
The election is on 5 April 2014. According to its own timeline,¬†the IEC hopes to finalise the count (including all the audits) by 20 April 2014 and to announce its preliminary result on 24 April 2014, less than three weeks after the vote. This seems optimistic.
The IECC, which will start receiving and adjudicating complaints almost immediately after the polls close, hopes to complete its work by 7 May 2014, handing over its decisions to the IEC on 8 May 2014. The IEC is then scheduled to announce its final results on 14 May 2014.
The provincial council election, in the meantime, is scheduled to have preliminary results announced on 17 May 2014, with the IEC finalising and submitting its decisions to the IEC on 31 May, so that the IEC can announce the final provincial council results on 7 June 2014.
In 2009 and 2010, the final results ‚Äď of, respectively, the presidential first round and the parliamentary election ‚Äď were indeed announced roughly within two months, but not without considerable upheaval. In 2009, it took the design of a complicated audit to do the trick, while in 2010 the aftermath of the election, after the final results, took almost a whole year. (2)
But if we have such an organized count immediately after the vote, why does it take weeks to announce the preliminary results? And why is there so much confusion?
Unfortunately, the process is not as organised as it appears on paper. First, the count at the polling stations will not be as quick and clear-cut as one would hope. In 2009 and 2010, in many places the count ‚Äď or part of it ‚Äď was postponed until the next morning, and in several cases in more remote areas it continued over several consecutive days, usually alongside considerable ballot stuffing. Polling staff was often unreachable by phone or used stalling techniques. And even after all forms had arrived at the tally centre in Kabul and were being processed, confusion remained over how many polling stations had returned results or how many votes they represented. As a result, the basic data ‚Äď number of votes cast and the number ultimately counted ‚Äď was highly fluid;¬†up to¬†and even after¬†the announcement of the preliminary results.
Confusion in the past was mainly caused by the messiness of what was returned to the IEC as a result of ballot stuffing (top-up or wholesale), multiple voting, crude tampering with forms or ballots, ‚Äėlost‚Äô ballot boxes, etc., with varying levels of sophistication and scale.
As a result, the information that the IEC has to work with ‚Äď the reports that the IEC in Kabul receives from the district and provincial offices, the ballot boxes that end up at the provincial centres, the forms and tamper-evident bags that are sent to Kabul ‚Äď show a wide and sometimes astonishing range of disarray. Boxes are not properly sealed, ballots have not been properly handled, tamper-evident bags and results sheets have been clearly tampered with, reports are incomplete or inaccurate, and the irregularities are often not properly logged. The IEC must then flag the problems and sort out whether they are due to misunderstandings, accidents, neglect, or crude manipulation ‚Äď and in each case¬†whether this is grounds for disqualification.
Moreover, the IEC‚Äôs fraud mitigation policy (which is not publicly available and is discussed in more detail below) is based on several crosschecks of data that, in principle, should correspond and add up. In practice, they rarely do. The problem is therefore not that the IEC has insufficient information, but rather that the information is messy and conflicting, as witnessed in the confusion during past elections over turnout, polling stations and the final number of valid votes cast (see¬†also the first part of this series on voter registration¬†and the second part of this series on everything around polling stations).
Is it clear how the IEC will decide which votes to disqualify?
It is not clear now¬†and may not be later either. We know when the IEC intends to check the data for irregularities, but not how it will treat its findings or how much transparency there will be when it decides to quarantine, disqualify or re-include votes into the count. In past elections, there was very little.
The instances at which data is checked during the tally process include the following:
- at the tally centre check-in, result forms from all over the country are checked for physical irregularities and tampering;
- during data entry, results are subject to mathematical and quality control checks embedded in the database software;
- incoming suspicious results may be subject to audits involving checking electoral material stored at the provincial offices/warehouses or recounting¬† polling stations‚Äô ballots;
- before consolidated results are presented to the IEC chief electoral office and the IEC commissioners, they are checked by an Operations Group consisting of senior IEC staff (Operations Group also advises commissioners on the need to carry out investigations with regard to any of the results sheet received); and
- the IEC says, according to its strategy, it will transparently follow up on issues brought to its attention by observers or media.
The fraud mitigation strategy is based on identifying apparent irregularities and inconsistencies, such as obvious tampering, but also the use of more ballots than provided or from batches other than the ones allocated (as all batches are now individually numbered). No mention is made of other fraud triggers that were used (and sometimes discarded) in the past, such as finding more than 95 per cent of all votes cast for a single candidate. This could however fall under the unspecified ‚Äúsuspicious results‚ÄĚ that warrant further investigation, referred to in the strategy.
The strategy is somewhat circumspect on how the IEC may react to evidence of fraud: ‚ÄúThe results of any audits are presented to the IEC Commissioners. If evidence of fraud is detected, the IEC Commissioners may disqualify specific results, or quarantine results pending further investigation, including by possible referral to the ECC.‚ÄĚ
In the past, lack of transparency has been a huge issue. Large numbers of quarantined polling stations were disqualified or re-included in the count without explanation (see previous reporting here and here).
Particularly towards the end of the process, as time was running out and the IEC was sorting through the remains of the mess, votes were added and subtracted without it being clear whether these were intentional reshuffles to rearrange the list of winners ‚Äď particularly in the parliamentary and provincial council elections ‚Äď or desperate attempts to rectify previous oversights (as may have been the case with the 511 ‚Äúmissing polling stations‚ÄĚ in 2010 that turned up in the count after the announcement of the preliminary results; see here for more detail).
Is that why AAN has spent so much time poring over the polling station results in the past?
Yes, that‚Äôs one of the reasons. Apart from announcing regular provisional count figures, the IEC will be posting several sets of results as they come in. These include scans of the original polling station results forms (which may or may not be legible), lists of quarantined and/or disqualified polling stations, and polling-station results as they are being entered into the database.
The polling-station results as entered into the database are interesting, as they should reflect the cleared and cleaned-up results, i.e. after all the checks and possible audits. Analysing these results for suspicious patterns thus gives a fair idea of how effective fraud-correction measures have been.
Indications of manipulation that can be found when going through the polling-station results include the following
- implausibly high numbers of ballots per polling station, particularly when exceeding the maximum number of ballots provided;
- implausibly large occurrences of round numbers (100, 200, etc.);
- all or almost all votes cast for one or a limited number of candidates, particularly if in combination with no or little sprinkling of votes for other candidates;
- patterns of ‚Äėcandidate partnering‚Äô (that is, repeated instances of the same candidate combinations, particularly when receiving implausibly high or round numbers of votes);
- and, finally, implausibly high numbers of female votes and low numbers of invalid or blank votes. (3)
Examples of analysis based on results patterns can be found in many of AAN‚Äôs 2009 and 2010 post-election dispatches, as well as in this report.¬†Comparisons of AAN‚Äôs analyses, based on results patterns, with audit findings and disqualifications by the IEC and IECC, have shown that they are indeed fairly good indicators of fraud.
Another reason to study polling-station results is that it brings irregularities to light that can go unnoticed in the chaos of simultaneous and often unexplained interventions. An after-the-fact analysis of the 2009 data, for instance, showed that in three provinces (Kandahar, Ghazni and Paktika) tens of thousands of votes were added to the provincial council count after the announcement of the preliminary results ‚Äď ostensibly in an attempt to offset intrusive ECC disqualifications (for details see pp. 12‚Äď15 in this AAN report).
How much might the results change after the IEC has announced preliminary results?
This depends on several factors, in particular how thoroughly the IEC has dealt with indications of fraud and the extent to which the IEC and IECC agree on how the fraud needs to be dealt with. Moreover, the way the IEC and IECC operate differs: the IEC seeks to clean up its data so that it can properly count it, while the IECC first responds to complaints that it receives (although it can launch independent investigations). This can lead to differing results.
In the last two elections, the IECC (at the time, it was still called ECC) significantly changed the outcome of the election ‚Äď although each in very different ways, largely in response to the way the IEC handled its responsibilities. In 2009, the IECC tried to make up for the IEC‚Äôs refusal to engage in mass disqualifications based on its own fraud triggers, while in 2010 the IECC mainly complemented the IEC‚Äôs intervention after the IEC had already disqualified 14 per cent of all polling stations (representing 23 per cent of the vote). And although the IECC added little to that, its main intervention was to exclude 27 parliamentary candidates, incidentally without much clarity on the exact basis for the decisions. (4)
The 2010 results were fiercely contested, including by the executive and the judiciary, culminating in a protracted process that ‚Äď at different points ‚Äď included the Attorney General‚Äôs Office¬†an ad hoc special election tribunal,¬†the Supreme Court,¬†and the president,¬†as well as extensive protests.¬†The drawn-out dispute finally resulted in the removal of nine parliamentarians¬†almost a year later, illustrating that even after the final results the outcome of an election may still change. (All links above AAN reporting.)
It is anyone‚Äôs guess how the IEC and IECC will function this time. According to the new law on the structure, duties and powers of the IEC and the IECC, the IEC is responsible for the announcement of the final results, based on the decisions of the IECC ‚Äď which are final and unalterable (incidentally, according to this law, both the decisions of the IECC (art. 26) as the announcement of the IECC (art. 14.5) are final and unalterable, which may still leave room for dispute). The Provincial Complaints Commissions (the PIECCs) are the primary authority to deal with complaints, although the central commission can review PIECC decisions based on renewed complaints or at its own discretion. Like in past elections, the IECC has struggled to recruit its staff on time, and it will be faced with ‚Äď sometimes justified ‚Äď allegations of partiality of its PIECCs.
There have also been irregularities and misgivings surrounding the appointment of both organs (see here and here),¬†leading to suspicions that the commissioners may be beholden to Karzai ‚Äď and to whoever is his candidate of choice. But the commissioners will be under a whole range of pressures and they may have loyalties of their own, while within the IEC, the Secretariat and the Operations Group play important roles.
Interestingly, when the IEC and IECC were called to brief parliament on 31 March on the preparations for the elections, they both declined. The IEC simply said it was too busy, but it would be prepared to come after election day, while the IECC responded that it did not answer to parliament or any other institute.
How will the presidential election end?
In the end, the election will be determined by a complicated mix of voter decisions, mass voter mobilisation, observer vigilance, security incidents, logistical setbacks, fraud and, finally, the disqualification process. With three strong contenders, no obvious favourite and a messy process, it is unlikely that the vote and the count will result in an obvious and uncontested outcome. So whether the election gets decided in a first or second round (or something in between), the result will probably be rooted in extensive political negotiation. The process will be tense;¬†there will be posturing, shows of force and¬†threats of violence ‚Äď¬†in an attempt to infuse the¬†political negotiations with a sense of urgency and to force a sharing of power ‚Äď¬†but if the normal pattern of Afghan politics is followed, the candidates will probably stand down in time.
In the first three elections, the international community, led by the UN, played a significant role. In the 2010 election, in a drawn-out and very messy process, it was the president and several Afghan institutions that shaped the outcome.¬†In the current situation, with frayed relations and a president highly suspicious of international interference (but who may also be leveraging this to silence possible criticism), the international community is walking a tightrope. It will need to balance a cautious approach ‚Äď withholding early judgment on how much better or worse than expected the elections were ‚Äď with a clear-eyed view of the problems that emerge. There will be a need for careful behind-the-scenes mediation, as the various players try to settle on an outcome, which will be complicated by the need not to be seen as too much in a hurry or partial.
(1) In 2009, at least 1.2 million ballots were disqualified in the presidential vote. The IEC and (then) ECC ran out of time to properly process the 2009 provincial council vote, so disqualifications were more limited here.¬†Around 1.4 million were disqualified in the 2010 parliamentary elections.
During the 2004 and 2005 elections, disqualification processes were not very developed. In 2004, no significant attempt was made to audit the vote, even though fraud did occur. In the clamorous 2005 election, candidates created much commotion over the widespread irregularities, but fraud corrections by the IEC (called JEMB, Joint Electoral Management Board, at the time) and ECC were relatively limited, resulting in the exclusion of 672 polling stations and 74 individual boxes.
(2) In 2009 election day was on 20 August; the preliminary results were announced less than a month later, on 17 September 2009, but only after IEC dropped important anti-fraud triggers to meet the deadline.¬†The ECC released the findings of a joint audit,¬†which had been purposely designed to do a quick job, another month later, on 20 October 2009. Based on these findings the IEC started planning for a second round, but on 2 November 2009 Dr Abdullah pulled out of the race and Karzai was declared the winner.¬†On 24 December 2009, we were still waiting for the provincial council results of eleven provinces.
In 2010 (election day was on 18 September), the preliminary results were announced (after having been delayed three times) on 20 October 2010,¬†but there were some awkward loose ends when the IEC found out it had forgotten to count a few hundred polling stations.¬†On 24 November, the IEC released the final results for almost all provinces (except Ghazni) based on the decisions of the ECC and its own further investigations.¬†The aftermath of the election, however, lasted a whole year and included interventions by the president, the Supreme Court, the Attorney General‚Äôs Office and an ad hoc special election tribunal. (All links AAN reporting.)
(3) Given the low level of education, one would expect a certain proportion of invalid votes due to misunderstanding. All provinces that showed very low numbers (2 per cent or less) also had been problematic in terms of irregularities (the lowest percentages were found in Paktika, Nuristan, Paktia and Khost). In comparison, Herat province reported 5.5 per cent invalid votes, Balkh 6.8 per cent and Kabul 7.3 per cent. See p. 10 in AAN‚Äôs report Who Controls the Vote? for details.
(4) The results of the 2009 presidential election were shaped by a complicated tussle in which the IEC discarded its own anti-fraud measures,¬†disregarded an IECC order to undo the decision¬†and announced preliminary results that included a huge number of votes from clearly suspect polling stations.¬†The IECC then ran out of time to properly investigate, after which the UN designed a complicated and improvised sampled audit¬†so that an outcome could be reached. At least 1.2 million votes were disqualified. The simultaneous provincial council vote, just as messy, was however left largely alone. The disqualifications in the presidential election were unevenly spread over the country; in the five provinces hit hardest, over 60 per cent of the preliminary results as announced by the IEC were thrown out (Paktika 88.5 per cent, Nuristan 82.5 per cent, Kandahar 71.6 per cent , Paktia 67 per cent and Khost 63 per cent). For more detail, see here. (All links AAN reporting.)
After the investigation into the 2009 presidential vote was completed, the ECC continued to examine the ‚Ä®more¬†than 500 so‚Äźcalled ‚ÄėPriority A‚Äô complaints in the provincial council poll (that is, the complaints that, if found valid, could potentially alter the outcome of the election). The ECC issued 34 separate decisions, one for each province, among others ordering the IEC to invalidate ballots where it had found clear and convincing evidence of fraud. This resulted in the exclusion of 722 polling stations (out of almost 24,000 ‚Äď the exact number open on election day will never be known). The greatest number of invalidations occurred in Kandahar, Nangarhar, Paktika, Nuristan, Faryab and Paktia (more than 60 polling stations invalidated in each). The ECC acknowledged in its final report that after the protracted audit process in the presidential vote, it did not have the time or resources left to conduct the provincial council investigations.
In 2010, the situation was very different. The IEC disqualified a large number of votes based on its own investigations: it audited and/or recounted 6,813 polling stations and excluded 2,543, resulting in the disqualification of 1,330,782 votes (an average of 523 ballots per station). The IECC did not engage much in disqualification of polling stations, although it did some, and instead removed 27 Wolesi Jirga candidates, most of whom had expected to win (see here for more detail).