Eve of Poll Prediction

This has turned out to be one of the hardest I’ve done over the years.

Now based in Nairobi, the picture blurs rather than sharpens. Western correspondents, recent opinion polls and the large gentlemen in nice suits in Nairobi all say it is decided – for “Baba na Martha” (Raila Odinga and Martha Karua). But the same men acknowledge that Ruto might get more votes (if that were the only deciding factor). Recent polls clearly favour Raila by unexpected and somewhat suspiciously large margins, but an entirely unrepresentative personal sample of urban southern Kikuyu tonight confirms my suspicion – some will not answer honestly when asked, but will tell pollsters what they think they want to hear. They intend to vote Ruto. The election is still neck and neck and there is a path to victory for Ruto – though it is narrower than when I started reporting on this poll. But there is also an assumption among many that if early results show a Ruto win is coming, the incumbents will – with the support of IT experts or insiders in the IEBC – somehow find a way to ‘stop Ruto’. Its less clear that Ruto has the same ability if the result is so close the other way round (a narrow Raila win), but some believe he too has hidden access (Davis Chirchir is his "occult priest" of IT).

So, final predictions. I still predict William Ruto will win the presidential election narrowly – by 51% or 52% to 48 or 49% for Raila. 8.3 m for Ruto to 8.1 m votes for Raila with negligible votes to the other two minor candidates (<1% combined), based on a mid-range runout of 74-75%. This is a tiny majority, but over 50% and meets all other requirements. But it is far from certain. Small changes to my estimates (much lower turnout in Mount Kenya, a complete ANC/FORD-K collapse in Western) could flip this the other way, rising to a 51%-49% Raila win. But still, I predict a small Ruto win, to be followed by aggressive attempts in the courts to invalidate it.

Deputy President Ruto is the change candidate, ex-MP and opposition leader Raila Odinga now the establishment. Raila is elderly and some see him as a figurehead President, more concerned with the trappings of power than the substance (as in 2008-13 when he was Prime Minister). This would allow "business as usual" and maintain a place for current elites and their business allies close to the "money tree" of state jobs and contracts. Ruto is more praetorian, a ‘man on horseback’ populist multi-millionaire (whose constituency is now the poorest), who has bet everything on this one throw of the dice, promising impossible miracles. His victory would result in radical change to the insider community and he is therefore extremely unpopular there. Both candidates promise to respect the outcome if they lose (as long as they lose fairly). But neither will believe they have lost fairly if they do. Kenyans want peace, but they also want to be heard, and fear of rigging inside IEBC is deep-rooted.

There will be significant challenges in the IEBC no matter their calm demeanor. The KPMG audit of the voters register showed serious issues with identity management and change management in such a high integrity IT system. The use of multifactor authentication and named user accounts are critical and suggestions that hundreds of generic accounts with shared passwords are being used to manage even some aspects of the election are terrifying if true. Even if everyone was working with the best will in the world, this present the opportunity for abuse. And some will not be shy to take such opportunities.

Discrepancies and confusions will abound even without rigging. And electoral abuse can’t be excluded. Some warning signs of rigging at national or constituency level include:

On the Day:

• High use of manual registers in particular swing or homeland areas. Massive inexplicable failure of KIEMS kits in specific regions.
• Multi-constituency violence or other mass polling station disruption (eg too few ballot papers) in the homelands of one candidate.
• Multi-constituency protests at fake polling stations or transposed results.

As the results come in:

• Turnouts above 85% in specific constituencies. These indicate “top up” voting where officials (after polls close) use the manual register to mark names and vote for those voters who didn’t vote for themselves. Siaya, Kisumu and Homa Bay most likely for Azimio; Kericho, Bomet, Nandi for Ruto.
• Significant (more than 5%) discrepancies between total votes for each of the five trackable elections in a county (Presidency vs parliamentary vs Women Rep vs Governor vs Senator).
• In a constituency, spoilt votes exceeding the margin of victory.
• Unusually high or low turnouts compared to other constituencies and compared to the 2013 and 2017 elections.
• Most obvious, integrities between forms and electronic results and otherwise unexplained use of non-standard forms without IEBC security features.
• And of course Kenya’s own 1988 and 2007 special: results which simply don’t add up because someone clumsily modified one result on a form without modifying the others….

I’ll be providing real-time independent analysis of the results as they come in from early in the morning on the 10th, and trying to judge for whatever data we have its arithmetic plausibility. In what is likely to be a high-tech election in every sense, the five-way match will be my leading tool, but its only definitive once the majority of the results are in (which can be too late for the international observers' interim reports). Turnouts will be an early indicator (as will entirely inexplicable results and tiny majorities in contested seats). There are always typos and errors in media reports, so every number needs cross checking.

​IEBC has a portal again at: https://forms.iebc.or.ke/  (though right now it’s reporting that it’s a “simulation” – not reassuring). What will be on it on the day, we shall see. Also, we shall see whether this year the media televises actual results from counting centers round the country (the official results), rather than just the results announced from IEBC. Twitter and Facebook are likely to get there first for some.

​In the House, little has changed. I still see the Azimio coalition as narrowly the largest in the 349-member house with about 165 elected MPs and Women Representatives (+ 6 proportional nominees), giving circa 170. Kenya Kwanza will -  I predict - have around 160 plus proportional nominees (circa 165 in total). So, right now, no overall majority and roughly 10 independents and non-aligned small parties could hold the balance.  But this is nothing more than a well-informed guess - some seats are completely unpredictable on a 4-way contest and Azimio will lose seats unexpectedly to KK due to “friendly fire” between allied parties. If Jubilee collapses on the day in Mount Kenya (as I still suspect but cannot prove), these results would reverse. I’m confident though that UDA will be the largest party, ODM second, Jubilee third.  

​I have not attempted to call the Senate this time – just too many candidates for now.

The below diagram shows the 290 constituency MP seats  - most now blank - with their winners, which I will populate as we go. Pre-coloured seats indicate places where the other alliance didn't manage to register a candidate (with the most likely winner - but it could also be an aligned independent):

Among governors, its close again but I now give Kenya Kwanza 24-25 to Azimio 21-22 (with one independent). As the establishment party, Jubilee has influential senior individuals who have a chance at victory even in a county which votes Ruto and UDA – Tharaka-Nithi, possibly Nyandarua, but the UDA wave will I predict sweep most of Mount Kenya. I think ANC will lose the governorships of both Kakamega and Vihiga to ODM, but FORD-Kenya will win Bungoma and (less confidently) Trans-Nzoia. Samburu and Turkana are both contested but I’ve called them both for UDA, but narrowly. On the coast, ODM or Jubilee will win most governorships, but I think ANC might take Lamu and UDA Kwale. I’ve called both Kajiado and Narok governorships for UDA, though in Kajiado only because it’s a three-way race, and I expect Azimio to lead in the other contests. The Kalenjin Rift is safe for UDA and Nyanza for ODM. Any independent successes (such as Kidero in Homa Bay) are still part of the same family.

This has been taxing to do (entirely unfunded, to ensure independence). There are more and more candidates to track and media outlets to monitor. So if it turns out I’m entirely wrong, I will probably retire. You should know when to quit and I've called them roughly correctly since 1992 (apart from 2007, which I missed, but which I would have called for Raila). Still, let us see what the next four days hold ! Every Kenyan election holds something new.

Stop press:  news just in that Mombasa and Kakamega Governor polls have been deferred due to errors with the ballot papers, also two West Pokot seats. Doesn't seem any partisan logic and doesn't favour any one obviously (maybe it slightly disfavours Azimio as they should win them both, but KK should win at least one of the two Pokot seats). But they left it very late and administratively its not a good start.