Five questions about the Kenyan electoral system

Once again, Kenya has witnessed an intensely competitive election campaign.

Whether you support President Uhuru Kenyatta or his main challenger Raila Odinga, one thing is clear: the presidential contest is close.

In turn, a particularly tight race has inevitably placed the electoral system and those who manage it under greater strain.

In the last few days the Independent Electoral and Boundaries Commission (IEBC) has moved to clarify its position on some pressing issues and to deal with others through the introduction of new guidelines.

This is an admirable effort, but as we head towards polling day five important questions remain regarding the strength of the country’s electoral system.

DEMANDS

1) How effective was the IEBC’s response to the audit?

Following opposition and civil society demands, the Register of Voters was subjected to an independent audit by KMPG.

This generated valuable insights into the strengths and weaknesses of the roll. 

From the resulting report, we know that KMPG identified a number of problems including duplicate registrations, multiple registrations associated with the same ID numbers, and the inclusion of a possible 1,037,260 dead voters.

The question is therefore how many of these issues were effectively resolved.

DEAD VOTERS

The IEBC says that it has worked hard to clean the list, but we also know that in some cases their hands were tied.

For example, the IEBC could not remove all of the dead voters because it did not have some of the information that would have been required to do this.

Unfortunately, it also seems that some of the dead voters that were reported to the Commission have not been removed. According to InformAction (IFA), of the 29 dead voters whose information the organisation shared with the IEBC, 96.6% remained on the registrar after it was certified as being final.

A further question is raised about the quality of the roll by AFRICOG’s finding that in 14 counties the number of voters actually increased between the audit and the final register when one would have expected it to fall due to the removal of duplicates.

REGISTER OF VOTERS

While these counties are split fairly evenly between those likely to vote for the government and the opposition, greater clarity from the IEBC on what explains these increases would help to assuage the concerns of election monitoring groups.

2) What proportion of the electorate will be biometrically verified to vote?

The limitations of the Register of Voters means that it is particularly important that the new Kenya Integrated Elections Management System (KIEM) kits are used effectively.

By using voter’s fingerprints to verify their identity, the technology makes sure that the dead can’t vote.

However having the machines and making them work are two different things.

In the 2013 polls, the Elections Observation Group (ELOG) found that the electronic verification kits failed in more than half of all polling stations.

NEW EQUIPMENT

This time round, the IEBC has purchased new equipment – the KIEMS – which have performed well so far.

However, the opposition remains concerned that some kits may malfunction, or not be used correctly, leading officials to fall back on manual processes that could facilitate multiple voting or ballot box stuffing.

One effective check against this would be for electoral officials to make a record of the number of voters processed through the kits – information they are set up to collect – and to include this information on form 34A.

This would significantly increase the transparency of the process by allowing monitors and observers to check what proportion of voters were verified through the KIEMS, and in which polling stations the technology was least effectively used.

3) How will the electoral data be scrutinised and corrected?

DISCREPANCIES

Of course, recording the number of people processed electronically only helps if this information is then scrutinised to identify discrepancies.

One of the important improvements of the KIEM kits over their predecessors is that electoral officials will submit both a set of election results and an image of form 34A, which details the results and is signed by the presiding office and party agents.

This allows for the data and the forms to be compared to identify any errors – accidental or otherwise – that have been introduced into the system.

However, one thing that seems less clear is at what level these results will be reviewed and what will happen if they are found to contain mistakes.

We know these can creep in as a result of basic human error, because this is what the IEBC says happened during its test of the vote transmission system on Wednesday, when the wrong figures were entered for some candidates.

MISTAKES

Given this, there needs to be a clear system for alerting parties and voters to any mistakes, and a clear process for rectifying them.

It seems that polling station results will be checked and aggregated at the constituency level.

However, mistakes cannot be fixed at this level, because results transmitted from the polling station level “cannot be changed by anyone” and the transmission of constituency level results to the national level will not use the KIEMS.

These stipulations are clearly important to protect the integrity of the system but raise a tricky issue: errors detected at the constituency level will not be corrected in the live stream of results that will be broadcast based on the polling station submissions – this will have to be done later.

4) Will the vote transmission system work?

In 2013, the vote transmission system failed horribly. The main problem with the vote transmission in 2013, Kenyans were told, was a “server issue”.

SERVERS

The reason that the servers went down, depending on who you believe, is either that they were hacked from outside, or that they simply lacked the capacity to deal with the flow of information.

Given this, you might have thought that conducting a full test of all of the kits working at the same time would have been a priority ahead of the 2017 election.

Only in this way can you ensure that the system will be able to stand up to the rigours of polling day.

Instead, the IEBC opted to only test the equipment in certain locations, and then only briefly.

This is worrying, because while the equipment may well work fine on the day, we don’t know that it will – and any failure of the system, even if it is purely technical, will be interpreted by NASA as evidence of manipulation.

5) How many party agents will be allowed?

Party agents of both sides will play an important role in safeguarding against electoral fraud, and a great deal of time and money has been invested by the main parties in developing stronger and more technically savvy monitoring systems.

REGULATIONS

However, there is a serious question about how many party agents will be allowed in each polling station.

Although the electoral regulations give candidates the right to post party agents, the high number of independent candidates in some areas has seen election officials place a limit on the number of agents that can be present at any one time.

However, this varies significantly across the country – while some presiding officers say that they will impose a limit of 10 agents, others have adopted a “the more the merrier” approach.

This inconsistency is likely to anger candidates, especially if some agents are turned away after they have been recruited and paid.

A clear directive from the IEBC about what candidates and parties should expect could help avoid unnecessary confusion and frustration.

One of the great challenges of Kenyan elections is managing popular perceptions.

CONFLICT

Early results in favour of one candidate can create the impression they are well ahead, even if in reality the race remains close.

It is therefore wise to explicitly set out which results are final and which are merely provisional.

According to the IEBC, where there is a conflict between the data entered into the KIEMS kits and the scanned images of form 34A it is the latter that will prevail.

This is consistent with 2013 and makes sense, as it is the form and not he data that party agents approve by adding their signature.

However, this means that the figures that will be broadcast at Bomas, and which come directly from the polling station level, are likely to be slightly different from the final result announced by the IEBC.

HUMAN ERROR

On the one hand, any human error that occurs when inputting the data into the KIEMS kits will only be picked up in the final result.

On the other, the votes from any polling stations in which KIEMS kits fail to transmit – and it would be incredible if everything worked everywhere – will not be added to the Bomas live stream at all.

This means that there is a good chance that the KIEMS total will be a little lower than the final tally announced by the IEBC.

The risk, of course, is that supporters of the losing side interpret any discrepancy as evidence that votes have been artificially added during the tallying process when this may not necessarily have been the case.

It is therefore important that the IEBC explains ahead of time that while large discrepancies will indicate rigging, small ones might not.

Where close elections are concerned, perceptions can be as important as reality.

Nic Cheeseman (@fromagehomme) is Professor of Democracy at the University of Birmingham