Transportation Demand Management (TDM) measures are essential for promoting sustainable travel choices and enhancing the efficiency of transportation networks. However, TDM strategies often assume homogeneity of traveller behaviour, leading to "one-size-fits-all" approaches. These approaches frequently overlook the diverse preferences and sensitivities of different traveller groups, resulting in limited success. Therefore, it is crucial to develop strategies tailored to the specific needs and preferences of various groups. This thesis explores how to design differentiated behavioural guidance from three key aspects: data application, preference analysis, and policy design.
Regarding data, the thesis investigates how to leverage multiple types of data to overcome the limitations of a particular type of data source. It emphasises the combination of data types and the use of emerging data sources across various analyses. For instance, revealed preference (RP) survey data and self-reported psychological metrics are used to examine how perceived risk affects mode choice during the pandemic. Additionally, RP data from smart cards is employed to investigate the long-term changes in residential locations and commuting patterns during the pandemic. Further, the prospect of integrating smart card (passively generated RP) data with stated preference (SP) survey data to get more robust residential relocation choice models. This joint use of SP and passive-RP data addresses the limitations of using a single data type, such as the lack of real consequences in SP and the lack of socioeconomic variables in smart card data. Finally, virtual reality (VR) data is used to bridge traditional SP and RP. This emerging data allows for the simulation of realistic travel scenarios and the testing of complex interventions that would be difficult to implement in real-world settings or traditional SP experiment environments.
For preference analysis, the thesis presents mathematical models of travel behaviour that explicitly consider the impact of internal and external factors on traveller preferences. From a temporal perspective, the research models the short-term, long-term, and dynamic decision-making of travellers. In the context of the pandemic, the thesis investigates how risk perception influences short-term travel mode choices. The dissertation also identifies key factors behind long-term relocation choices using the lens of the ‘Push-Pull-Mooring’ theory and quantifies the relative impact of location, built environment, and level-of-service factors. Further, the dynamic decision-making and learning processes of travellers when subjected to soft-policy interventions have been explored in the context of the shift to green modes of transport. Three advanced choice modelling frameworks are utilised to better model traveller preferences: an MDCEV model combined with latent variable model, an SP-RP joint model, and a dynamic hybrid model. These enable effective integration of diverse data for a comprehensive understanding of the internal and external factors affecting travel behaviour.
Regarding policy design, the thesis discusses how to create interventions tailored to the heterogeneous preferences of travellers with different socioeconomic attributes, geographical features, and psychological profiles. In the context of the pandemic, the research designs risk communication strategies to guide travellers with different social demographics and attitudes towards risk information. These strategies aim to help travellers form rational risk perceptions and make informed travel mode choices. In the context of long-term relocation behaviour of travellers, differentiated policies are proposed based on the varying commuting characteristics of the travellers. Additionally, using a VR experiment, the research investigates the effectiveness of innovative nudge interventions in encouraging travellers to opt for green travel modes.
Overall, this thesis expands approaches to capturing behaviours and analysing traveller preferences using emerging data and experimental methods. Practically, the use of multi-source datasets to analyse heterogeneous traveller preferences and design differentiated policies can inspire policymakers to move beyond "one-size-fits-all" approaches, thereby improving policy efficiency and effectiveness.
