Exploration of a Box Cox logit model of intercity freight mode choice

During the past years, the linear logit model has been used extensively in modal choice analysis. More recently, the introduction of Box–Cox transformations on the explanatory variables in passenger studies have generally shown the superiority of the Box–Cox logit over the linear logit. Nevertheless, we have found only one such application in freight transportation. This study is devoted to filling this gap by testing different configurations of the Box–Cox logit over the linear logit. Our results confirm the usefulness of the nonlinear Box–Cox specification found in passenger studies. We have used an original data bank developed for Canada. The bank contains, for 64 commodity groups, the 1979 domestic flows among 67 geographical zones, by three transportation modes: truck (private and for hire), rail and ship. To the authors' knowledge, this is the first Canadian study on mode choice that takes into account private trucking; since Canadian private trucking is at least as important as for hire trucking, this could have contributed to the results obtained.