I was making my suggestions for improvement then you decided it will infringe on the pre-determined results and rejected the suggestions. That proved my case that your MOAS started from the "outcome" down to the factors leading to that. So you adjust the factors to fit the outcome.
A better process would be to let the factors freely give you whatever outcome and you accept it.
For instance there is no way Raila's vote will shrink by 1 -400 % while Uhuru's expands by the same.
Then there are places in RV where you have chosen to ignore all evidence of a shift. To say Raila will get 14% in Bomet and clearly base that not on current events but largely on the 2013 outcome, would make sense, if you were also applying the same to Uhuru! It looks like you have (quite aptly I would say) Double Standards or to be exact two standards: One of Uhuru and a rather bad one for Raila.
A MOAS does not have to disregard all the rules of qualitative research. I would say you have allowed bias to influence and direct your analysis of the data.
Maybe they got everything wrong then? Trust my MOAS.
Lusaka is going nowhere and Jubilee will get some Mt. Elgon votes. Take it to the bank.
The polls got it for Turkana but not Bungoma.
The possible surprise will come from Busia where Amos Wako may be floored probably by Omtatah.