As is evident, the relative risk ratio is calculated without reference to
the remaining two adopter groups. This shows the underlying assumption
of themodel of the independence of irrelevant alternatives. Although
statistical tests are available to test for this proposition, their use is not
recommended due to unreliable test results (Cheng and Long, 2005; Long
and Freese, 2006). Thus, the use of amultinomial logitmodel in this study
is based on the recommendation by Amemiya (1981) that the model
works wellwhen the alternatives are dissimilar. Overall, themodel helps
to reveal significant differences between early, medium and late adoption
of organic drystock farming in Ireland, relative to non-adoption.