Liv wrote:Kabando highlighted the differences in regional results between the pollsters... sometimes the difference in a region was higher than 10%.
Another point is that .... the pollsters have been changing the sampling methodology along the way in the last 3 weeks after his criticism. e.g. the first poll did not have the data by counties which was added later. Sample sizes have been changed along the way too.
IPSOS agreed that the differences in regions between the pollsters could not be avoided as the sample size per region is too small and hence the margin error must be high.
IPSOS also agreed that the polls could be misleading with respect to gender and age of the sample. The IEBC has not released details about the age and gender of registered voters per region. So each pollster is making their assumptions with regard to this...which could be very different from the actual population of registered voters.
The pollsters owned up that the 95% of people they contacted claimed to be registered voters yet only 30% of the country population is registered. (14 / 40 million). That means there could be an issue of respondents cheating they are registered when they are not.... thus affecting the poll results. some Kenyans in the poll are definitely cheating.
The above are real issues with the polls. Others factors like the terrain of Africa and illiteracy levels make it difficult to have credible polls.
In USA where the communities are more homogeneous and literate, you would expect pools to be precise and better, yet in 2012 the polls were showing a dead heat between Obama and Romney (too close to call) up to the last minute. Yet the final election results was like a landslide for Obama.... quite different from what the polls were showing.
That is why I have been saying none of the poll results could be relied on. So I support what Kabando was saying. These poll results could not be relied upon.
Why are you following Kabando into that hole, the disparities between pollsters at regional level may exist but at the end of the day when they tally their national polls, they arrive at the same conclusions at the national level so let it go. Its like trying to argue that the polls prediction of a UHURUTO sweep in Central is inaccurate.
As to those registered, my take is anyone who didn't register is unlikely to again fall into the pollsters net, they are somewhere stuck in some hole. Plus you have to consider that our population is 60% under 18 years of age.
"The purpose of bureaucracy is to compensate for incompetence and lack of discipline." James Collins