Methodologically, this research confirms the merit of using optimal
experimental designs both in freight settings, in general, and in studies involving
proportional choice variables, in particular. The model outputs generally
had strong statistical significance and consistent behavioural implications
throughout the exploratory modelling phase as new data were received. Under
a standard orthogonal design, we would have expected weak statistical significance
and potentially variable behavioural implications over much of the
data collection process until a sufficiently large sample size was obtained.
The role of the proportional choice variable was likewise essential in this study.
Not only did the specification of a non-binary choice variable enable respondents
to make behaviourally meaningful choices within the study, but the
resulting choice models were also able to capture the presence of strong latent
relative preferences for modal alternatives that may have been misrepresented
through the use of a binary choice variable.