Description
TitleTravel behavior of ridesourcing
Date Created2021
Other Date2021-05 (degree)
Extent1 online resource (xv, 149 pages)
DescriptionThe emerging ridesourcing services provided by Transportation Network Companies (TNCs) such as Uber, Lyft, and DiDi are widely and increasingly used across the world. Yet, little is known about the travel behavior and decisions made by those who use ridesourcing. While many travelers are attracted by the convenience and low cost, ridesourcing has raised widespread concerns about its adverse effects on cities and society, including congestion and social inequality. Governments and planners are urged to impose regulations on the operation of ridesourcing. However, little data-driven empirical research has been conducted for understanding, planning, and policymaking about this emerging mobility option. This dissertation fills this literature gap by making use of large-scale ridesourcing trip data.
In my dissertation, I ask three interrelated questions about travel behavior of ridesourcing: (1) What affects the travel demand of ridesourcing trips? (2) What are the factors associated with the decision to share a ridesourcing trip? And (3) does ridesourcing provide equitable accessibility to people across space and population groups? I conduct three analytical studies to disentangle each of these three questions: (1) variation in ridesourcing trip generation, (2) factors associated with the decision to share a ridesourcing trip, and (3) equity of ridesourcing accessibility. I use trip-level DiDi data in Chengdu, China, in the first study, and TNC data in Chicago, Illinois in the second and third studies. Data about these two study cities, including the built environment factors, demographic, socioeconomic, transportation variables, points of interest, crime, and business licenses, are obtained or web-scraped from various sources.
In the first analysis, I investigate various spatial, economic, and land-use factors associated with the generation of ridesourcing trips in Chengdu. I characterize the unique pattern of TNC ridesourcing trips over space and for different time periods. I examine the association between the generation of ridesourcing trips and spatial characteristics, including population density, floor-area ratio, housing prices, road networks, the proximity of public transit, land use mix, and points of interest. I estimate "global" linear regression models and "local" geographically weighted regression models that account for the spatial variation of each factor on trip generation. Results suggest that population density, local road density, floor-area ratio, housing price, and the proximity to subways have positive associations with ridesourcing trip generation. I also explore the spatial variation associated with the built environment effects on trip generation throughout the city.
In the second study, I investigate the temporal and spatial distribution of authorized ride-splitting (i.e., shared) trips in Chicago. I find that the willingness to share TNC trips differs across neighborhoods with different demographics, socioeconomic status, and built environment characteristics. The willingness to share is related to price and trip duration. I estimate logistic regression and random forest models to determine the marginal price and time effects on the decision to share. The results indicate the probability of authorizing a ride-splitting trip is highly sensitive to the changes in price per mile holding other trip characteristics constant, and the random forest model had better predictive accuracy than the logistic model. Additionally, I examine the importance and marginal effects of total price and trip duration. I use two data preprocessing methods to address rounding in the price data and demonstrate the robustness of the results, despite this limitation. The coefficient of each variable on the odds of authorizing a shared trip ranges from -2.58 to -2.62, for price per mile, from -2.03 to -2.29 and for trip duration from 0.59 to 0.94.
The third study provides a comprehensive understanding of ridesourcing accessibility. I measure ridesourcing accessibility by applying gravity-based metrics based on realized ridesourcing trip-making. Opportunity attractiveness and the travel impedance factor are included in the ridesourcing accessibility estimation. Travel impedance factors are computed based on the real travel times. Both the employment volume by sector and points of interest locations are used to measure the opportunity attractiveness. I estimate accessibility to three categories of nonwork destinations, i.e., healthcare facilities, restaurants, and grocery stores, and compare the accessibility between ridesourcing and transit with a consistent measure. Given the presence of spatial autocorrelation, I estimate ordinary least squares models and spatial autoregressive models to examine the associations between ridesourcing accessibility and census tract-level demographic and socioeconomic indicators. I find that ridesourcing is a more inequitable mode compared to transit. Ridesourcing accessibility has a significant positive association with median household income and a significant negative association with the percentage of non-white populations in most models. Ridesourcing accessibility is also negatively associated with the percentages of three age groups (0 – 5, 6 – 18, and 65 plus) but positively associated with the percentage of zero-vehicle households.
The analytic studies in this dissertation provide evidence and insights of ridesourcing travel behavior from three perspectives. Overall, the behavioral characteristics of using ridesourcing services (e.g., travel demand, sharing decision, and access to different opportunities) appear to vary across space and are significantly affected by local contexts, including the built environment, demographics, and socioeconomic factors. Racial and social disparities in these behaviors are also illustrated in these analyses. Based on the findings, my dissertation discusses policy implications related to ridesourcing in big cities in terms of leveraging the travel demand, encouraging shared rides, and improving the accessibility of disadvantaged populations and neighborhoods.
NotePh.D.
NoteIncludes bibliographical references
Genretheses, ETD doctoral
LanguageEnglish
CollectionSchool of Graduate Studies Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.