Value Chain Dynamics et Impact of Collective Action on Smallholder Livelihood Improvement in Kenya

Factors Influencing Changing Boundaries and/or Routes of Fulani Herders in the North Eastern Corridors of Ghana: A Count Model Approach

“…A herder's decision to change herding route is represented by the number of times a herder has relocated since arriving in Ghana. The number of times a herder has relocated or changed route are discrete events and can best be modeled through a count modeling framework [25]. In that case, since the number of times a herder has changed route or relocated is a count, the Poisson regression model, which assumes equi-dispersion of the dependent variable, is usually used [26].…”

“…This means that the Poisson model was not appropriate for modelling this count-dependent variable. The distribution of the discrete variables showed evidence of overdispersion and therefore the Negative Binomial model, which relaxes the assumption of equidispersion, replaces the Poisson model to estimate the determinants of changing routes among the Fulani herders of the study area [25]. In overdispersed data, the Negative Binomial model, which also estimates the parameter (α) of overdispersion, is appropriate [28].…”

“…In overdispersed data, the Negative Binomial model, which also estimates the parameter (α) of overdispersion, is appropriate [28]. The restrictive assumption of the Poisson model is relaxed in the case of the Negative Binomial and, for that matter, since the variance is not equal to the mean, following Mutonyi [25], the variance can be given as;…”