Estimating a seat choice model for Dutch suburban train users

Abstract

Passenger behaviour and preferences on-board buses and trains have become more relevant after the COVID outbreak. Surprisingly, the impact of seat and individual characteristics on passengers’ seat choice in public transport vehicles remained unknown. We estimate an en-route seat choice model for Dutch suburban trains, using a mixed logit with error components. Our model accounts for seat attributes (windows/aisle, group-of-4, etc.), crowdedness level, and individual characteristics. Our findings indicate that: (i) passengers clearly favour sitting alone, are influenced by distance to the entry, and show preferences for window/aisle seats, and (ii) level of crowdedness changes some seat preferences (e.g. passengers choose window seats when the train is empty but aisle seats if the train is highly crowded). Personal characteristics have a moderate influence on seat choice and we could not find significant evidence of COVID perception on seat choice preferences.