International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
A hybrid DCM application in freight mode choice set in the Port of Ghent (Belgium): do environmental attitudes matter?
Louise M. F. De Tremerie, Stephane Hess, Gerard C de Jong

Last modified: 28 March 2017


This paper proposes to use stated preference (SP) techniques to study the preferences of freight shipper companies in the port of Ghent (Belgium). The use of SP techniques and choice modelling in such a context is more difficult than in most applications given the small sample of real decision makers, the confidential nature of their business decisions, and the potential lack of alignment between the survey respondent’s preferences and those of his/her company.

The SP survey was conducted with shipper companies in spring 2016, a pilot was previously conducted in autumn 2015. The design of the survey was that of an experimental design (D-efficient) following Rose and Bliemer (2009), where the derived priors from the pilot served as the initial priors for the final survey. The logistical managers of the companies were targeted, as conventionally done in the majority of freight mode choice surveys in the literature (Tavasszy & de Jong, 2014). Therefore, the firm is simplified to a single-unit decision maker. Each respondent was confronted with 32 choice tasks (CTs). This is clearly a much higher number of CTs than is commonly used, but is necessary in the face of very small samples of potential respondents. A second reason for the high number of CTs is that not only a within-mode stated choice part (WM) is included, but also a between-mode stated choice part (BM).  In addition, each part is divided into eight CTs for in-going and eight CTs for out-going typical transports. The WM is characterised by three unlabelled alternatives, each with three attributes: cost (tariff), transit time (door-to-door) and a green attribute expressed as varying levels of CO2 emissions (g/tonne-km) around an European average based on data coming from the European Environment Agency (2015). The BM adds reliability (%arriving on time) and a loss & damage (% of total cargo) attribute to the previous three attributes. In contrast, the alternatives are labelled here according to the main mode: road, rail, IWW (inland waterways), SSS (short sea shipping). For example for IWW, the mode is conventionally an intermodal transport demanding an access and egress transport. The time and cost attribute are pivoted around reference scenario values given by the respondent. In a simultaneous estimation on the within-mode and between-mode SP parts, we controlled for scale effects between these by using scaling techniques (as in for example the work of Bradley and Daley, 1994).

We analyse the data with the help of a hybrid choice model accounting for latent attitudes towards the environment. While hybrid applications are increasingly applied in passengers transport literature (although not without critique as for example in Chorus and Kroesen (2014)), applications in the freight mode choice literature are scarce. An example can be found in the work of Ben-Akiva, Bolduc and Park (2008), who focus on service quality. Initial origins in the freight literature can be traced back to the work of Park (1995), in which the author included a behavioural theoretical framework in freight transport modelling.

The work in this paper sets itself apart from previous applications by introducing two latent constructs. One latent constructs controls for the respondent’s personal environmental attitudes while the other latent construct is derived from indicators that aim to measure the corporate environmental attitudes or outlook (as perceived by the respondent). The attitudinal questions on the level of the respondent where based on topics identified from passengers mode choice studies using both scales of “agreement” as in Bahamonde-Birke and Ortuzar (2015), Bahamonde-Birke and Hanappi (2015), Bahamonde-Birke et al. (2015), Hess et al. (2013) as well as scales of problem evaluation as used in Daziano and Bolduc (2013). In contrast, no examples were easily available for the corporate level. The corporate attitudinal questions were therefore developed with the help of stake-holder group analysis. An additional motivation to use a hybrid DCM framework in this study is the small sample size, remembering that in their key study of hybrid choice models, Vij and Walker (2015) show that HDCM can lead to gains in efficiency (i.e. lower standard error terms) as compared to standard DCM models, by making use of more data at the level of each individual. While this is of benefit in any application, it matters especially in the case of contexts with small target samples, such as with our freight sample.

Apart from the standard coefficients and derived willingness-to-pay measures, results gained from the hybrid extension are expected to give insight to what extent environmental attitudes play a role in the mode choice. However, more interestingly will be the interaction and differences between the latent construct on the individual’s level and the latent construct on the corporate level (e.g. the relative importance of both). It is envisioned that these initial results will serve as guidance for further research to look at this interaction within the firm and how this affects the final choice made.

see pdf file (submission)


Conference registration is required in order to view papers.