Methodology and Approach:
Our sample sizes are usually determined using the estimated voting population (18.5 million), our
preferred margin of error, usually +- 1.7% and 99% confidence level.
❑Samples are then allocated to the various regions using EC 2020 votes register as a frame.
❑30% of the constituencies in each of 16 regions are then selected randomly using simulating model
❑The regional quota is then allocated to the randomly selected constituencies based on the voting
population of each of the constituencies selected relative the others.
❑Samples allocated to each constituency is further allocated to up a maximum of 3 electoral areas in
the constituencies based on voting population of the electoral areas selected.
❑Samples of less than 50 is assigned to one electoral area, sample between 50-100 is allocated to
two electoral areas, samples between 100-150 is assigned to three electoral areas.
❑ Voters are selected randomly in the communities to respond to the polls and polls are not
conducted among any groups of people.
❑The presidential polls is based on “likely voters” model, meaning the views of undecided and I will
not vote have been excluded in the analysis