Analysing Freight Shippers’ Mode Choice Preference Heterogeneity Using Latent Class Modelling

This paper describes a study to improve understanding of the decision-making process of New Zealand firms, freight shippers and agents when making freight transport mode choice decisions. Such studies, despite their importance, are relatively scarce due to issues related to data confidentiality, restraining firms from taking part in such studies. To achieve the objective, we use latent class (LC) modelling, which postulates that firms’ behaviour depends on two components: 1) some observable attributes, such as travel distance and size of operations; and 2) unobserved latent heterogeneity. The latter is taken into account by sorting firms into a number of classes based on similarities in their characteristics. Subsequently, the behaviour of firms in each class is explained by a set of parameter estimates, which differs from the sets assigned to other classes. In this study, data were gathered using stated preference surveys from 190 NZ firms, freight shippers and agents. Based on their freight operations, participants were grouped into: 1) long-haul and large shipments and 2) long-haul and small shipments. Furthermore, as each participant evaluated 18 choice scenarios, the data set contains 3,420 choice records. The results of the LC modelling allow policy makers to design more appropriate strategies and policies for different segments of the population to improve intermodal transport and to attract the largest latent class for both cases. In addition, the LC model indicates that the potential improvement in modal shift, which can be achieved by applying different policy options, varies with both transport distance and the size of shipments. Furthermore, in order to promote sustainable freight transport, one policy would be to increase the reliability of both the rail and sea freight transport services.


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01642168
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jul 27 2017 10:05AM