Sunday, March 17, 2013

The IEBC Has Some Explaining to Do

Yesterday, Raila Odinga launched his Supreme Court petition to challenge the results of the March 4th Kenyan general election. Topping Raila's list of grievances is his allegation that the voter registry, sometime between the December 18th final registration deadline and election day, was manipulated to favour his opponent, president-elect Uhuru Kenyatta.

He has a point.

From March 4-9, as I was tracking the progress of the election on Excel spreadsheets, inconsistencies between the Independent Electoral and Boundaries Commission's (IEBC) December 18th registered voters list and whatever the organization had used on election day were already apparent. Indeed, there was no publicly accessible voter registry on election day; whatever changes were made to the rolls from December 18th to March 4th (during the IEBC's "cleaning up" of the registry on February 23rd, for instance) weren't posted to the IEBC's website, as pointed out in a brilliant article in yesterday's Daily Nation.

The official post-election IEBC voter registry, finally published in last Friday's Nation, was significantly different than the December 18th rolls, both in absolute numbers of registered voters and in their distribution across the 47 counties. In 11 counties, voter registration numbers shifted by 3 per cent or more; in one case, the figure was as high as 12.13 per cent.

My latest Excel spreadsheet shows that the changes almost universally benefitted Uhuru, and probably denied Raila a chance at a run-off.

The Bullet Points

On February 23rd, the IEBC claimed to have struck 20,000 names off the voter registry due to duplicate registrations. If that's all they did, why did the number of registered voters rise by 12,497, from 14.337m to 14.350m, on election day?

In Raila's strongholds, there were 2,126 fewer registered voters on March 4th compared to December 18th. In Uhuru's there were almost 45,000 more.

In Raila's heartland (Siaya, Kisumu, and Homabay), registered voters fell by 9,380, or almost 1 per cent.

Had voting been conducted using the December 18th voter list, Raila would have gained 13,590 votes, and Uhuru would have lost 23,686 (assuming March 4th voting patterns). That's a net swing of over 37,000 votes, not even taking account of the share held by fringe candidates. Given that Uhuru squeaked by with a mere 8,000+ votes to capture 50.07% of total ballots cast, it takes no intricate Excel algorithms to conclude that, had the December 18th voter roll been used, the March 4th election would have resulted in a run-off. A run-off that Raila would have had a very good chance of winning.

Of course, none of this is proof of foul play. There could be perfectly plausible reasons for the pattern in the redistribution of registered voters. But so far, the IEBC has yet to provide any. The organization has some serious explaining to do, both to Raila Odinga, the CORD Alliance, and the Kenyan people.

The complete spreadsheet data is available on Google Docs.


  1. You do realize Raila needed more than 800K votes to offer any meaningful challenge to Uhuru.You also remember the BVR register did not capture some voters and the Final register released in February had the list of those voters...finaly the issue of Voter register was never asked prior to the election...CORD could have asked for the clarification once the final register was published...IGNORANCE IS NO DEFENCE

  2. Mudavadi's supporters and probably most of the fringe votes would have come over to Raila in a run-off. And with so much on the line, it's likely that Raila's turnouts in his strongholds (which hurt him on March 4th) would have improved. Would have been very close.

    As I wrote in the post, there could very well be a reasonable explanation for the registry alterations. Though I find it odd that the eligible voters not captured by the BVR registry all ended up in Jubilee strongholds.

  3. in the sparsely populated counties, like West Pokot n TURKANA close to 25K voters were not reflected in the Dec 18th Register. Also remember that a Court order asked the IEBC to register prisoner's

  4. Hi Jay
    This article goes into some of the analysis that has been done on the problems associated with the voter register. There was a register published after December 18 but even that one is problematic:

  5. Thanks for the clarification and for the link. Will take a look.

  6. An excellent article in The East African that raises many of the same points: