Abstract
The transportation system is experiencing disruptive changes with the development of emerging transportation technologies like ride-hailing, electric vehicles (EVs), and automated vehicles (AVs). These technologies are double-edged swords, whose impacts on the transportation system, urban forms, energy consumption, and emissions are highly uncertain. To start to quantify the uncertainty, it is critical to study consumers’ adoption preferences for these emerging technologies, which informs policy opportunities to support sustainable outcomes in deployment. This dissertation examines consumer preferences for the three innovative technologies in transportation (ride-hailing, EVs, and AVs), with a particular focus on preference heterogeneity.
Preference heterogeneity has been well studied in the research on the traditional transportation system. With the disruption of emerging technologies, recent studies highlight the importance of preference heterogeneity in the adoption of new technologies. This dissertation contributes to the existing literature in the following three aspects: 1) examine the disparity in ride-hailing usage under various spatial contexts, 2) examine the heterogeneous preferences for EVs using advanced discrete choice models which allow for random preference heterogeneity, and 3) examine the heterogeneity in AV mode choice preferences that can be linked to latent attitudinal constructs.
First, based on the 2017 National Household Travel Survey data, the ride-hailing analysis shows the disparities in ride-hailing usage. For example, seniors (compared to younger) and low-income (compared to high-income) travelers are less likely to use ride-hailing services. Such disparities between age groups and between income groups widen in urban and rural areas, respectively. Moreover, ride-hailing services are found to fill mobility gaps for non-vehicle owners from public transport desert communities. Findings provide insights for future ride-hailing research to consider the interplay between socio-economic characteristics and spatial contexts, rather than examining these two elements independently.
Second, the EV preference heterogeneity analysis develops mixed logit (MXL), latent class (LC), and latent class-mixed logit (LC-MXL) models based on stated choice experiments data collected in Virginia in 2018 (n = 837). Model results suggest that monetary incentives are the most effective in increasing EV market share, followed by deploying more charging infrastructure, while improvement in battery range is found to be least effective. Moreover, the comparison across the three statistical models shows that no one model is unanimously superior to the other models in uncovering consumer preference heterogeneity in EV adoption. Rather, altogether they provide a more comprehensive picture of the complex EV preference structure.
Third, AV mode choice preferences are examined using the integrated choice and latent variable (ICLV) model based on stated preference surveys distributed in the Seattle (n = 511) and Kansas City regions (n = 558) in 2020 and 2021. Model results suggest the importance of latent attitudes (e.g., attitudes towards AV technology, willingness to share travel with strangers) and mode-specific attributes (e.g., trip cost, trip time) in explaining AV mode choice outcomes. Additionally, the sensitivity to in-vehicle travel time in private AVs can be significantly lower than in human-driven vehicles, suggesting the potential of induced vehicle miles traveled in the AV era.
In sum, this dissertation investigates three distinct aspects of preference heterogeneity (observed, unobserved, and latent attitudes) in the adoption of three emerging transportation technologies (ride-hailing, EVs, and AVs). Findings provide technology-specific policy insights for sustainable deployment. Moreover, as the three technologies are in different stages of adoption, insights from preferences for more mature technologies have implications for the newer technologies. For example, the findings of the ride-hailing analysis have implications for future studies on the deployment of AVs once real-world usage data is available. The potential spatial heterogeneity in using shared AVs and shared AVs with pooling modes should be explicitly considered in AV policy-making, ensuring that the benefits of new technologies can be shared by all groups of people. Lastly, considering that the transportation system keeps evolving with the introduction of new technologies, the study framework in this dissertation can also apply to future research on other emerging transportation technologies beyond ride-hailing, vehicle electrification, and automation.
