d) H0 βCropLivestockEthiopia β CropLivestockKenya Versus HA :вCropLivestockEthiopia ≠ βCropLivestockKenya
e) H0 βCropLivestockKenya βPastoralKenya Versus HA βCrop LivestockKenya ≠ βPastoralKenya
For instance, results from hypothesis test d, indicate that the crop livestock systems in Kenya
and Ethiopia are statistically different and consequently should not be pooled together:
LEthiopia =-366 and LKenya =-692. LEthiopia+ LKenya =-1059 while the restricted Lpooled
crop livestock
=-1223 with a χ72 = 329 which is much larger than the critical value of 14.1 for the
conventional one tailed test with probability of type I error of 5%. In the same way, the other
hypotheses for preference stability were rejected. Consequently, the MNL estimations were
done separately for the pastoral system in Kenya and the crop livestock systems of Kenya and
Ethiopia.
4. Results and discussions
Maximum likelihood estimates for the multinomial logit models estimated for bulls
and cows from the choice experiment data is presented in Table 2. Since the traits had 2-3
levels each, one level was left out as base during estimation. A total of 2,783 choices made by
253 households was collected for bulls and 3,036 choices made by another 253 households
collected for cows. Most of the trait coefficients are statistically significant and have the
expected signs, though their magnitude varies by the type of production system. For instance,
trypano-tolerance trait coefficient has the expected positive sign across all production systems,
indicating that respondents prefer trypano-tolerant cattle relative to trypano-susceptible ones.
Traction potential for bulls is strongly positive and significant for crop livestock systems,
indicating a high contribution of good traction potential trait in bulls to the crop-livestock
farmers’ utility function. In the pastoral systems, trait coefficients associated with fecundity
(high fertility and reproductive potential) is strongly positive. The trait for liveweight is
positive and strongly significant for bulls in crop-livestock and pastoral systems in Kenya. It
is however not significant in the crop-livestock system in Ethiopia. Supplementary purchased