5.1. Estimation model
Conditional logit (CL) models,which assume independent and identical
distribution (IID) of random terms, have been widely used in past
studies. However, the property of independency from the irrelevant alternatives
(IIA) derived from the IID assumption of the CL model is too
strict to allow flexible substitution patterns.5 The most prominent
scheme is a mixed logit (ML) model that accommodates differences in
the variance of random components (or unobserved heterogeneity).
ML models have sufficient flexibility to overcome the limitations of CL
models by allowing random taste variation, unrestricted substitution
patterns, and correlation of random terms over time (McFadden and
Train, 2000) Accordingly,we can demonstrate a variety of parameters at the individual
level through themaximumsimulated likelihoodmethod by setting
100 Halton draws.6 Furthermore, since a respondent answers eight
questions in the questionnaire for conjoint analysis, the data thus obtained
constitute a panel that can be used for a standard random