The key perhaps is in history. Certain things tend to remain the same irrespective of the time or surrounding events. Taking the polls and matching them against the 2002 voter turnout makes for very interesting reading expecially when you tweak the turnout for obvious changes on account of the different candidates this time around. 

Is it possible for a Kibaki victory and especially if you use the Steadman Polls released yesterday? It is not going to be an easy win, and the margin may very well be miniscule, even 1% but it can be done. Here is how.

Take the Steadman Group poll result on a per province basis. Next take the 2002 voter turnout - irrespective of who people voted for. And now comes the tricky bit loaded of course with my assumptions:

Raila Odinga

As Raila Odinga will be a candidate in this election, I expect that the turnout in Nyanza will rise by 11% to 65% from the 56% in 2002. Note that there was a huge turnout in the Gusii districts on account of Nyachae being a candidate. I am assuming the increase this time is driven by Raila's candidature.

In Central Province, I assume that there will only be a 9% rise in turnout increasing to 75% from 66% five years ago. This will be on account of the general feeling of isolation and peril in Central Kenya this year. The high turnout in 2002 was on account of the fact that there were 2 Central Kenyan candidates and now we only have one, so it is only the heat of the moment that will boost turnout.

The Kalonzo factor as well as Mt Kenya sentiment will increase the turnout to 65% from 61% but since Steadman has already taken care of Kalonzo in this poll, then the increase will be proportionately spreadout.

Nairobi province has seen a great increase in voter registration. My view is that there will be a lot more people who normally do not vote but will go out this time and vote Kibaki. So I am increasing voter turnout by 13% to 52% from 42 in 2002. I am however maintaining a Raila lead of 46% against Kibaki's 43% for Nairobi.

In the case of Rift Valley province, I think the turnout of 68% in 2002 was Moi inspired so I am factoring in that inspiration and estimating that the turn -out this time will be 65% as I do not think there is that much electoral excitement and incentive to go to the vote. Note that even if I retained the turnout at 68% Kibaki would still have a marginal lead.

In both Coast and Western provinces, I do not think there is a sufficient shift in fundamentals that would affect the overall voter turnout. In 2002, NARC had the overall lead in the election with a Wamalwa for VP candidate whilst in 2007 we have Western province overwhelmingly going for Raila with  Mudavadi as the Vice Presidential candidate. There is nothing to show that the voter turnout will be any different in Coast.

Finally on North Eastern Province. My thinking is that there will be a lower turnout and I am marking this for 55% this year against 58% for 2002. As in the case of Rift Valley (and even if we retained the Rift Valley turnout) President Kibaki will still come out top with a margin of some 40,000 votes.

So on an overall basis if my assumptions are correct, then Kibaki would win the vote 3,612,469 against Raila's 3,540,394. As the Steadman people said yesterday, what is important is to get people to go out and vote!

{swfremote}http://static.scribd.com/docs/h8d9deq5cpepf.swf{/swfremote} 

For a larger rendering of this chart, please click here , or here for a larger chart tracking the changes since the September 28th poll, click here .