Modelling multi-stage behavioural responses to air travel disruptions with information completeness and herd effects

Intermodal transport offers a proactive strategy for managing air travel disruptions, yet travellers’ responses to such alternatives remain insufficiently understood. Existing research presents two main gaps. First, most studies focus on rail as a single-mode substitute, with limited attention to intermodal solutions. Second, even studies of intermodal alternatives often overlook behavioural factors associated with bounded rationality, particularly travellers’ emotional states, herd effects, and information completeness, despite their relevance to uncertain and time-pressured emergency decisions. This study conducted a multi‑stage stated‑preference survey to capture travellers’ emotional and behavioural responses under vague and detailed social information conditions. Drawing on Regret Theory and Social Learning Theory, we developed a dynamic Bayesian hybrid utility‑regret model to characterise the dynamics of decision‑making under air travel disruptions. The results show that, compared with travellers reporting peace, those reporting anger, sadness, or fear are generally more sensitive to social information and exhibit stronger herd-effect intensity. Anger is associated with broader shifts towards alternative modes or trip cancellation, sadness with rebooking and cancellation, and fear mainly through its interactions with business travel purpose and disruption duration. Compared with vague information, detailed information reduces the concentration of preferences on rail and initially favoured intermodal alternatives and leads to a broader redistribution of choices. These emotion-related patterns should, however, be interpreted in relation to the contexts in which they occur, as emotional states are also significantly associated with contextual variables, including trip purpose, traveller location, disruption scope, and disruption duration. The findings provide implications for emotion-adaptive services, context-specific information provision, and integrated emergency management across the pre-, during-, and post-disruption stages.