Log into your member account to listen to this article. Not a member? Join the herd.

With matchmaking almost complete, the Kenyan political dance now takes on a more structured character as we head into election season proper between April and August 2022.

With the collapse of the One Kenya Alliance, we now know that, nationwide, it is a two-horse presidential race between William Ruto and Raila Odinga, each backed by supporting alliances, but with Raila’s Azimio la Umoja also backed by President Uhuru Kenyatta and the state apparatus. The contortions required for an incumbent president to support his strongest opponent to succeed him destroyed the ruling Jubilee party during 2020-21. This leaves a rump Jubilee allied with ODM and 20+ other “Azimio-friendly” parties facing the recently formed Kenya Kwanza alliance, centred on Deputy President Ruto’s UDA.

The two alliances are now neck and neck overall, with most models giving both candidates between 7.5 million and 8.5 million votes each, leaving the result open and subject to influence from other factors, including the financial resources of the two alliances, state support and simple luck.

With the onboarding of Luhya leaders Musalia Mudavadi and Moses Wetangula to his new coalition during February, Ruto has strengthened his national credentials and gained a strong base in the Western region, but benefited less than he had hoped nationwide, as a substantial element in FORD-Kenya and the ANC defected to Azimio. Raila’s Azimio alliance has lost most of the Kikuyu, Embu and Meru, but elsewhere has strengthened its position in the last six months, picking up support in northern Kenya and the coast and recently winning over a very reluctant Kalonzo Musyoka. He was the last big player left in the now collapsed OKA alliance.

Whether Raila can keep Musyoka—who now controls maybe 50 per cent of the Kamba vote—is less certain, as Musyoka is openly angry and prepared for trouble. But this last-minute shotgun marriage gives the ramshackle Jubilee-ODM-Wiper-KANU alliance a narrow majority (52-53 per cent) in the National Assembly once more (which the government had lost for a few weeks in February).

The diagram below shows the political allegiances of the elected MPs in the national assembly as of early March 2022, after Musyoka finally declared for Azimio. Orange indicates Azimio (for now, as they have not decided on a colour as a brand) and Yellow for Kenya Kwanza (the colour of the UDA, its largest component).

A square represents one elected MP. There are also 47 elected women representatives and 12 party-nominated MPs (not shown). They follow a similar pattern but with a slightly greater leaning towards Azimio as a result of the vulnerability of nominated MPs to party recall.

My calculations show that Azimio has the backing of 150 elected MPs while Kenya Kwanza has 134, with six having refused to declare their stand or flip-flopping so fast their position cannot be determined.

Both presidential candidates have failed to nominate a running mate as deputy president, holding back from making that choice for as long as possible, until defection from their alliances by spurned partners becomes impossible. Both have onboarded very senior allies (Luhya Mudavadi, Kamba Musyoka) without formally naming them as deputy, leaving the Kikuyu, Embu and Meru vote still potentially swingable by a strong choice of DP by one or both alliances.

This last-minute shotgun marriage gives the ramshackle Jubilee-ODM-Wiper-KANU alliance a narrow majority.

In truth though, there are few senior Kikuyu, Meru, Embu or Mbeere with the mettle to become Deputy President to either candidate. Martha Karua is, in my view, the best Kikuyu candidate for Raila, but they have a difficult history together and she commands respect more than she does votes. Peter Munya might work for the Meru but not elsewhere. Peter Kenneth is a safe pair of hands but he is more of technocrat than a rabble-rouser. For Ruto, none of the floated names (Wahome, Gachagua) makes real sense. They are not national-level players, and many have skeletons bursting out of their closets. Susan Kihika in Nakuru might be my pick (a female DP is a distinct possibility) but she is still new and the Nakuru governorship is hers for the taking so a 50-50 chance at DP might be less attractive than a definite governorship. Anne Waiguru would be another option, but again, she has a solid chance of re-election and she does not have the common touch. Justin Muturi would do, but he doesn’t set hearts alight. Mudavadi would motivate the Luhya more than Muturi would sustain the Kikuyu.

At the county level, the model below shows my prediction as to which alliance and presidential candidate will (based on multiple factors) win the presidential election in that county. Where opinion polls have been (properly) conducted, those numbers have been incorporated.


As we can see, the two alliances are again neck and neck, with a fractional advantage for Kenya Kwanza. In most counties, the leading alliance is now pretty clear to all but diehard supporters. The final result will be determined by the size of the winner’s majority, so turnout will play a huge role in the outcome.

Another huge influence on constituency and county elections (but not the presidential election) will be “friendly fire”, i.e. split votes between different candidates from the same alliance. Both alliances have decided not to hold cross-alliance primaries, leaving multiple candidates from the same alliance competing for the seat in different parties. This was the bane of the pro-Kibaki parties in 2007 and is likely to have the same effect on Azimio. Real three- and four-horse races will result across at least a fifth of the country (with Jubilee candidates standing against ODM and DAP-K candidates for example, or UDA, ANC and FORD-Kenya competing against each other), making the result in a first-past-the post system hard to predict. But assuming no upsets, we can predict that UDA and ODM will be the largest parties in the next parliament, and UDA will be the largest, as it is the dominant player in the Kenya Kwanza coalition everywhere except among the Luhya, while support for Azimio is more evenly spread between constituent parties.

Regional models do not tell the whole story, however, for in presidential elections some counties are more equal than others. At the extremes, Nairobi has 2.5 million voters, while tiny Lamu has 80,000. Presidentially, therefore, Nairobi’s dominance is worth 30 times the dominance of Lamu. Historically, Kenyan elections have been about many things, including real issues (corruption, economic reform, ethnic clashes, multi-party democracy), personal loyalties and alliances and loyalty to party (although the parties change every five years), but underlying those have been consistent ethnic community preferences. To use a pithy phrase coined in 1992, that election was about “bribe and tribe”, and while the situation has changed in the last 30 years, that characterisation is still familiar to many.

In truth though, there are few senior Kikuyu with the mettle to become Deputy President to either candidate.

In the last two general elections, an ethnically-based predictive model—looking at the population of each ethnic group, registration, predicted turnout and estimated voting intention for each community—proved extremely accurate in predicting overall results. Whether the turnout and voting preference predictions will be accurate this time cannot be confirmed until the day, and outcomes can change with events, such as unexpected selections of Deputy Presidents, wars, disease, sudden deaths, major defections, blatant election rigging or catastrophic faux pas by leading candidates. But it did well in 2013 and 2017.

Overall, my current national presidential model is based on the input that there are between 51 and 52 million Kenyans today, of which roughly 30 million are aged 18 and over. We know the ethnicity of these people in 2019. Of the 30 million, we know that 22 million have registered to vote, 13 per cent more than in 2017. But we also know that voting registration rates have historically differed between communities and have varied depending on whether they have ethnic “skin in the game”.

Looking at historical registration rates and turnout, and current registration by constituency, gives us an estimate of the number of actual voters in 2022—I predict between 16 and 17 million. We do not know the ethnicity of those registered voters, above all because the government has failed to release the census results regarding ethnicity at the district or county level since 1989. But we have 1989 and can make a good guess at the changes since. Adding to the mix the predicted voting preferences of each community, one can then estimate the votes for each candidate.

There are four important notes to this model though. First, for this election, William Ruto has run an effective grassroots, economy-based campaign, the first true national populist campaign since Matiba in 1992, openly critical of the wealthy elites and appealing to the poor and marginalised of every ethnicity (most Kenyans). While senior politicians and most elites have turned against him, the poor have not; his “hustler versus dynasties” campaign has given them hope. I therefore predict that in this election we shall see some cross-ethnic economic voting. I think that Ruto will poll 25 per cent and above in places like Ukambani (if Musyoka does not become Raila’s running mate), Kisii and at the (Christian) Coast even when much of the political establishment is with Azimio.

The two alliances are again neck and neck, with a fractional advantage for Kenya Kwanza.

Religion is another cross-ethnic political theme, stronger than in most previous polls, with Ruto explicitly branding himself as a Christian leader and receiving the support of grassroots religious groups in Christian areas but correspondingly finding it heavy going in Muslim areas of northern Kenya and the Coast.

Thirdly, this model does not take material election rigging into account. Top-up voting in the homelands of the two candidates is common (whether you like it or not) and is to some extent factored in. Material rigging (entirely faked results) is different—it can produce totally unpredictable results. But the Kenyan courts have become increasingly assertive on such matters and the risks of rigging were made evident with the annulment of the 2017 presidential result (even though no rigging was proven). Whoever wins, I’m sure the other will petition.

Finally, this assumes both alliances have the financial resources to finish the campaign. Both have big donors, and both are seeking funds from home and abroad. But Azimio has the dubious honour of having close connections to the state apparatus and its opportunities to free up funds. On the other hand, Ruto has personally bankrolled an insurgent campaign for two years, and even his deep pockets must one day run dry.

Putting all this together, the result of my current model (all other things staying equal) is a narrow Ruto victory of 8.3 million to Raila’s 8 million (51 per cent-49 per cent). But this is not cast in stone. A 5 per cent error in my prediction of the Kikuyu vote (estimated at 71 per cent Ruto to 29 per cent Raila in my model) changes that result to a narrow Azimio win. In one sense, it is all to play for. However, in another, given the country’s violent political history and the endless disputes over election rigging, such a narrow victory for either alliance is unlikely to be a comfortable outcome for Kenya.