We selected three cooperatives from the regional sample in
two steps. First we interviewed experts knowledgeable on the
Kenyan coffee sector and reviewed secondary sources to establish a shortlist of potential cooperatives to be taken up in the
study. Second, we interviewed the management committees of
the shortlisted cooperatives and made a selection based on
similarities in location, agro ecological conditions and
resource endowments. Hereafter, a random household sample
was taken from the membership listings of each cooperative.
The sampling frame prepared for each of the cooperatives
was based on all members delivering coffee to the wet mills
belonging to the respective cooperative. To facilitate suitable
matching, the number of coffee trees owned by members was
used as an indicator to guarantee uniformity in wealth. We
randomly selected households proportionately to the distance
to the cooperative wet-mills and their tree density.