you guys keep bringing up dead and buried stuff...why now....

what do the two idiots want after "their" results?

@guru - Some people have problems!!! Just leave them to wallow in their stupidity! Even an idiot knows that 100% is more accurate than a SAMPLE of any size!

Exit polls are done to ATTEMPT to predict the outcome of the election. NOT the reverse!

And it would be EXTREMELY difficult to conduct exit polls in Kenya! Here are the reasons:-

1) While sampling, you have to cater for the voting patterns expected in that particular election. I can take a sample from Mombasa, Kisumu and Machakos while someone else takes from Kiambu, Eldoret and Nakuru. Of course the results would be different like day and night! Even in the same city like Nairobi, a sample from Kibra would yield totally different results from a sample from Kasarani!

2) Weighted averages! You have to consider the number of registered voters in a certain area while deciding the relative number of respondents. You cannot select 10 respondents from area A with 2000 registered voters and that predominantly supports party X and 10 respondents from area B that has 100,000 registered voters and that predominantly supports party Y and then you tells us that your "results" show that the two parties are tied!

And even the other polling companies in Kenya don't take this into consideration when doing their sampling!!!

The truth of the matter is that you would need a HUGE budget and human resource to conduct even a remotely respectable exit poll. You need to capture all constituencies - different polling stations - and weight you sample sizes depending on the number of registered voters!!! The age groups and gender distribution should also reflect the distribution in the voter register!

I am a firm believer of the fact that "Numbers don't lie"! during the 2013 March Election we had a total number of 12 million people voting and if they conducted an exit poll of 6000 of whom 12% refused to answer thus the findings are from 5280 voters - which is a good sample size as it gives a 1.77 assuming uniform response distribution. I would assume that their sample was distributed across the country based on numbers of people who registered to vote in those areas and not by overrall population if they didn't they would suffer the tyranny of numbers on their sample - however I assume they did. I would also like to know if they suffered a similar fate to Tom Wolf (people lied to them - very difficult to tell) secondly if their sample size was representative of turnout or did they just pick a fixed number of people per station prior to election day and stuck to that regardless of how many people showed up.

63% of the Kambas voted for RAO and 12% voted for UMK.

Whom did the remaining 25% vote for?