The candidate model structures vary in the specification of preference heterogeneity
(that is, whether the b parameters are constant across individuals or
choice observations, or whether subsets of identical preferences are specified
with probabilistic allocation of individuals or respondents across these subsets)
and the nature of the error term (that is, whether the error term is solely
distributed independently and identically Extreme Value I, or whether there are
additional, individual or choice-observation-specific forces represented within
the error term). Still, each model structure shares the same goal of identifying
the trade-offs made by respondents in the sample as effectively as possible