Not All Travelers Think Alike: Segmenting Travel Behavior Under Air Pollution Exposure Using a Hybrid Latent Class and Discrete Choice Approach

Abstract

Commuters face their highest exposure to air pollution while traveling; however, behavioral responses to this risk remain poorly understood in urban transport research. This study investigates how air quality awareness, sociodemographic factors, and travel context influence mode choice. Household survey data were collected from 723 commuters in Kolkata, India, during winter, when ambient pollution levels are elevated. A latent class cluster analysis was conducted to uncover unobserved behavioral segments based on attitudes toward air quality information (AQI) and travel behavior. This yielded three distinct commuter profiles: (1) low awareness and passive (24%); (2) aware but inactive (57%); and (3) proactive AQI responders (19%). These segments were further characterized using sociodemographic and travel attributes to understand profile-specific tendencies. Then, a multinomial logit model was estimated to examine how latent class membership, along with factors such as income, trip purpose, distance, and reported health symptoms, influenced travel mode decisions. The results show that access to real time AQI information, particularly through mobile apps and public displays, was associated with reduced use of high-exposure modes, such as two-wheelers and auto-rickshaws. However, many commuters who reported health symptoms, such as eye irritation, continued using these modes, indicating limited flexibility or a lack of alternatives. The metro was preferred by lower-income individuals, while higher-income, AQI-aware commuters tended to choose private cars, probably viewing them as more protective. These findings highlight the need for integrated urban transport and environmental strategies that pair risk communication with equitable access to clean, safe, and protective travel options.