Modelling intermodal traveller behaviour in mega-city regions: simultaneous versus sequential estimation frameworks

The sustained expansion of mega-city regions and the development of multimodal transport networks have catalysed intercity mobility, thereby restructuring regional travel demand patterns. This study aims to interpret the behaviour of intermodal travellers in a short-haul intercity context within mega-city regions. A comparative modelling framework, utilising both simultaneous and sequential estimation methods, is proposed based on stated preference survey data collected in the Beijing-Tianjin-Hebei region, China. The simultaneous estimation framework examines the integrated measurement of the perceived utility of multiple stages of travel using cross-nested logit models. In contrast, the sequential estimation framework systematically investigates the bidirectional interactions associated with the intercity mode decision and decisions related to access and egress modes in a stepwise manner. The latter quantifies the accessibility of transport hubs and destinations to assess the implicit cost of feeder trips in the intercity mode decision. It validates the sequential impact on feeder mode choice preferences. In addition to identifying behavioural determinants, the models’ relative performance is assessed regarding behaviour prediction accuracy for diverse groups of travellers categorised by travel purpose, fellow traveller, baggage size, and travel frequency. Statistically, the weighted prediction errors for access, intercity, and egress mode choices are 1.12%, 1.33%, and 0.89% under the simultaneous estimation framework. In contrast, under the sequential estimation framework, these errors are reduced to 0.81%, 0.63%, and 0.50%, respectively. The results suggest the superior applicability of the latter in interpreting intermodal mobility patterns.