OK, so as promised, here is my first attempt at looking at the numbers and what they are telling us about the outcome of the March 4th general election in Kenya.

14.3 million Kenyans registered to vote this year. Out of this (based on historical turnout rates) about 11 million will actually show up to vote. If the opinion polls are right, neither Uhuru Kenyatta nor Raila Odinga (the top two frontrunners) will get the requisite 50% plus one vote required to win the election. It is likely that there will be a runoff. About 4% of voters remain undecided. The polling trend (see below) suggests that the race will tighten over the next six weeks before the election.

The first opinion polls after the party nominations show Deputy Prime Minister Uhuru Kenyatta ahead of Prime Minister Raila Odinga in the raw vote, at least according to my analysis. The overall national head to head match up in the two polls released Monday show Mr. Odinga leading Mr. Kenyatta by (48-40, Infotrak) and (40-36, Ipsos).

The regional poll tallies, on the other hand, show a different story. In these Mr. Kenyatta emerges with a lead of between 490,000 and 630,000 of the accounted for votes depending on the turnout models used. The average of the tallies show that if elections were held over last weekend Mr. Kenyatta would garner 4.5 million votes to Mr. Odinga’s 3.9. This leaves about 24% of the (potential) votes cast either spread out among the other presidential contenders or undecided.

Here is how I arrived at the numbers:

The surveys by Infotrak and Ipsos (the two firms correctly predicted the outcome of the 2010 referendum) gave regional tallies of how the top two coalitions did among those surveyed. With a few modifications (like assigning the GEMA counties in Eastern region the Central region poll results), I assigned these tallies to the different counties within the regions. I then estimated voter turnout using the numbers from the three most recent national voting exercises – 2002 and 2007 elections and the 2010 referendum. Because of the anomalies in the presidential election in 2007, I used the constituency turnout figures (In these figures, for instance, Juja and Nithi did not have turnouts exceeding 100% as was the case in the presidential election in 2007). Of course there are counties in which the popularity of either Kenyatta or Odinga vary by constituency but this is the best we can do for now. I then used the estimated county turnout rate and the regional polling results to estimate the expected vote count for either candidate in each county using IEBC’s figures of registered voters.

It is important to note that among the two polls, Infotrak asked respondents about their preferred ticket (Kenyatta and Ruto vs. Odinga and Musyoka) while Ipsos asked about individual presidential candidates. The discrepancy in the national polling average and the raw numbers I show here might be because of incorrect weighting of the different regions by the polling companies. The fact that Kenyans vote along ethnic lines and voters are geographically concentrated means that the regional polling numbers might provide a better picture than the national numbers. National polls appear to be over-estimating Odinga’s support by about 3 percentage points on average.

Uhuru Kenyatta is ahead in the raw figures for the following reasons:

1. The first reason is that Mr. Kenyatta has the numbers. The combined GEMA registered voters number 3.9 million. That is 27.3% of the registered voters. Mr. Kenyatta obviously won’t bag 100% of these votes but it doesn’t hurt to have a vote rich base.
2. His stronghold of the wider Mt. Kenya region had the highest voter registration rate in the country. This, combined with the fact that his running mate brings in the populous Rift Valley region, gives Kenyatta a slight edge off the gates.
3. Kenyatta’s strongholds (Mt. Kenya) and Rift Valley have historically had higher turnout rates than the regions that Odinga will need to win on March 4th. In 2002 Kenyatta’s strongholds had a higher turnout rate by 5 percentage points. In 2007 it was 10%.
4. The combined high population, higher registration rates and expected higher turnout means that Mr. Kenyatta is presently the favorite to win the first round of the March 4th presidential poll.

How can Odinga win?

1. A lot of voters (24%) remain spread out among the smaller candidates or are undecided. Come election day these voters may break for Mr. Odinga for the reasons I gave in an earlier post.
2. Mr. Odinga’s other path to victory is by ensuring high turnout in his strongholds of Nyanza, Western and Coast regions. Just by matching the expected turnout in Mr. Kenyatta’s strongholds he would reduce the deficit to about 250,000 votes.
3. He must also eat into some of Mr. Kenyatta’s support in the Rift Valley and Central regions. If the election is a mere census then Mr. Kenyatta will win the first round (the second round is another story all together). For Mr. Odinga to win he must convince voters in Mr. Kenyatta’s strongholds that he is the better candidate.

Facing reality:

For a while it seemed like this election was Mr. Odinga’s to lose. I have since softened on this a little bit. Despite his many problems, Mr. Kenyatta can still win this election, at least the first round. In the second round everything will be contingent on who between Messrs Kenyatta and Odinga can bag the roughly 20% of votes that will go to various smaller candidates in the first round. As things stand Mr. Odinga is the likely beneficiary of these votes.

A lot will happen between now and March 4th. But key things to consider include:

1. If turnout is low on March 4th Mr. Kenyatta will emerge the winner. His (national) base is relatively wealthier and more urban (or more accurately, more politically engaged – if you doubt this see the voter registration numbers for Kiambu county alone) than Mr. Odinga’s and thus will have a higher turnout. Having failed to match Mr. Kenyatta’s voter registration rates, Mr. Odinga needs upwards of 80% of those registered in his strongholds to show up to vote, or else he will lose.
2. Mr. Kenyatta appeared to be the better organized candidate in getting his base to register to vote. And given the way in which his party handled the nominations exercise, it is likely that he will out-organize Mr. Odinga in getting his supporters to the polls. This spells more trouble for Mr. Odinga.
3. The nominations exercise gave Mr. Odinga’s coalition bad press for four consecutive days. His home base of Nyanza was the worst affected. Seemingly undemocratic nomination exercises – in which Odinga’s allies controversially won party primaries – in the region may depress turnout, something that Odinga should be worrying about A LOT. Watch out for how Mr. Odinga’s party handles the nominations fallout in his Nyanza backyard.
4. Musalia Mudavadi appears to have made gains in Western province – he is polling there at 26%. His gain is Odinga’s loss. If Mr. Mudavadi continues to gain in the next 40 days then we shall almost be assured of a run off, after Mr. Kenyatta wins the first round.