Findings and Discussion
We now turn to our initial results from the empirical survey. Our sample con- sists of 70 respondents representing shippers conducting activity along at least one of the three corridors within the study. After removing three observations in which respondents accidentally submitted choices prematurely, the resulting estimation sample includes 557 choice observations. Despite both the presence of three unique corridors in the study and a range of organisational roles (for example, retailer versus freight forwarder), preferences estimated within exploratory models indicated a lack of distinct classes of sensitivities within the study. This made the latent class model, which centres on the identification of groups with distinct preferences, an ineffective framework for our final model. Rather, we turned to the mixed logit model to reveal the degree of preference heterogeneity within the sample. We found that preference heterogeneity was most effectively identified with respect to the freight rate alone. The primary implication from this is that, after accounting for differences in sensitivities to