2021

Junxiang XU, Southwest Jiaotong University
Jin Zhang, Southwest Jiaotong University
Jingni Guo, Southwest Jiaotong University

Abstract: In this paper, considering the heterogeneity of travelers’ decision-making behavior caused by travel environment factors, thus affecting the choice of travel path, the theories and methods pf travel path choice based on improved cumulative prospect theory (ICPT) in complex environment were proposed. On the basis of cumulative prospect theory (CPT), the value function was improved, and the parameter value range was enlarged. The nonlinear curve of value function and weight function of cumulative prospect theory was fitted through thousands of data tests and experiments. Then according to the decision preference, the decision makers were divided into different categories and the reference point value relationship of heterogeneous decision makers was found. In this paper, fuzzy travel time reference point and periodic dynamic risk degree reference point were set up, and a dynamic path selection model based on heterogeneous double reference point was established to improve the cumulative prospect value. Taking the highway network in Sichuan-Tibet region for example, the optimal path selection scheme of heterogeneous travel groups under the complex environmental factors such as debris flow and landslide in each time stage was studied, and the influence of preference parameter of travel time and risk degree on path choice was analyzed, and then the parameter sensitivity in the cumulative prospect theory (ICPT) was analyzed. T

Kyusik Kim, Florida State University
Kyusang Kwon, Korea Research Institute for Human Settlements
Mark Horner, Florida State University

It is important to analyze factors that influence travel mode choice and predict individual mode choice because travel mode choice shapes people's movement and it determines their level of mobility. While there have been studies investigating how built environment elements are associated with travel mode choice, most efforts have neglected evaluating the heterogeneity of effects that the built environment has on travel mode choice across different age groups. This study aims to examine the effects of the built environment influencing travel mode choice across age groups in Seoul, Korea, using random forest approaches. Our random forest model demonstrates what factor is important and how it is associated with the effect on travel mode choice. As a result, the built environment has a greater impact on the subway selection for older adults than other age groups and the random forest approach captures the non-linear relationships between certain predictors and travel mode choices. Applying this approach to the travel mode choice analysis, we can examine the heterogeneous effects of the built environment on travel mode choice across different age groups.

Jonas De Vos (jonas.devos@ucl.ac.uk), University College London
Patrick Singleton, Utah State University
Tommy Gärling, University of Gothenburg

Many travel behavior studies have analyzed how travel attitudes and travel mode choice are related with each other, and recently also found that the chosen travel mode can influence travel satisfaction. However, the desire and intention of using certain travel modes – which can be considered as important predictors of mode choice – have received limited attention. In this paper we introduce the travel mode choice cycle, which is composed of attitudes, desire, intention, behavior and satisfaction, and is inspired by existing psychological theories such as the theory of planned behavior, the theory of cognitive dissonance and the model of goal-directed behavior. We argue that the elements in this cycle are strongly connected with each other and that this cycle can provide valuable insights for policy makers to stimulate the use of desired travel modes, such as public transport and active travel.

Sung Hoo Kim (skim3020@gatech.edu), Georgia Institute of Technology (Georgia Tech)
Patricia Mokhtarian, Georgia Institute of Technology (Georgia Tech)

This study examines long-distance travel behavior in the state of Georgia. Based on a survey conducted in 2017-2018, we modeled number of domestic leisure long-distance (LD) trips over the past 12 months by air and car modes. We posited that there are two types of zero trips – structural zeros (by people who essentially never travel LD) and incidental zeros (by people who simply happened not to have traveled LD within the past year) – and used zero-inflated negative binomial models to endogenously segment people into a structural zero-trip regime versus a trip-making regime. Selected demographics, attitudes, and geographical characteristics played important roles in explaining the segmentation into regimes and the amount of long-distance travel. We present separate models by mode, and they show different sensitivities to the pertinent factors. In particular, the presence of children and distance to nearest major airport had different roles in the two models. For example, the presence of children acted as a barrier to belonging to the trip-making regime for air travel, but it was a facilitator of doing so for car travel. However, it was negatively associated with the number of trips in both modes. As distance to major airport increased, both entry into the trip-making regime and number of trips were inhibited for air travel, but car travel exhibited the opposite effects. Finally, we discuss some implications for the contexts of autonomous vehicles and COVID-19, and suggest some avenues of future research.

Patrick Loa, University of Toronto
Sk Md Mashrur, University of Toronto
Khandker Nurul Habib (khandker.nurulhabib@utoronto.ca), University of Toronto

The continued growth and utilization of ride-sourcing services have reshaped traditional perceptions of urban mobility. As the popularity of this relatively novel mode of travel has continued to grow, there has been a greater focus on the effect that the adoption and use of ride-sourcing have on the utilization of more traditional modes of transportation. With a few notable exceptions, prior studies on the topic often rely on descriptive statistics to understand the modes that are replaced by ride-sourcing for specific trips. This study utilizes structural equation models to investigate the factors that influence the decision to choose public transit and taxi as their so-called substitute alternative mode for a recent ride-sourcing trip, using data from a web-based survey of Toronto residents. The results indicate that the selection of the substitute alternative mode is influenced by socio-economic and trip-related attributes, and to a lesser extent, attitudes towards ride-sourcing services. In particular, students, persons from lower-income households, and transit pass owners were found to be more likely to substitute ride-sourcing for public transit, while older respondents and persons from wealthier households were more like to select taxi instead. These results shed light on the varied impact that the use of ride-sourcing services can have on the utilization of other modes.

Yuqiao Huang, Shanghai Jiao Tong University
Linjie Gao, Shanghai Jiao Tong University
Anning Ni, Shanghai Jiao Tong University
Xiaoning Liu, Shanghai Jiao Tong University

Effective Transportation Demand Management (TDM) strategies depend on the accuracy of forecasting in travel demand. Research on joint choice behavior of travel mode choice and trip chain pattern as well as their relationship is helpful for travel demand forecast. However, previous studies believed commuting trips remained unchanged, and did not consider the dynamic changes of multi-day commuting trips. This paper proposes two Nested Logit (NL) models with different structures (trip chain above travel mode structure and travel mode above trip chain structure) to investigate joint choice behavior using multi-day travel data from Shanghai, China. Based on the best structure, Monte Carlo (MC) method is used to simulate the switch of trip chain and travel mode under possible policies changes. We arrive at conclusions: (1) the nesting structure of trip chain above travel mode is the best, which means trip chain pattern decision precedes travel mode choice; (2) decrease in car ownership makes commuters switch from cars to public transit and increase in bicycle ownership makes commuters switch from public transit to bicycles; (3) complex trip chains are more dependent on cars and simple trip chains are more dependent on public transit; (4) decrease in car ownership and increase in bicycle may have a mutually restrictive relationship. These findings are beneficial to implement TDM strategies to develop sustainable transportation systems and optimize urban trip structure.

Kaili Wang, University of Toronto
Sanjana Hossain, University of Toronto
Khandker Nurul Habib, University of Toronto

The outbreak of coronavirus disease 2019 (COVID-19) spreads globally, disrupting every aspect of normal activities. In the post - COVID era, as pre-emptive health and safety measures, all major post-secondary institutions in the Greater Toronto Area (GTA) decided to go virtual in their 2020 fall semester. Post-secondary students will become telecommuters. This imposed telecommuting will shift their daily routines. This study first empirically investigates students' activity-travel behavior during normal conditions using a dynamic econometric activity-based model (CUSTOM). Then it uses the estimated model to simulate student’s activity-travel patterns in typical and telecommuting scenarios. Students’ behavioral differences between the two scenarios are analyzed. It seems that students’ average school activities increase from 0.86 to 1.06 per day, with increased time allocated due to telecommuting. Students’ average number of total activities per day also increases from 1.89 to 2.13. However, trip rates will decrease from 2.98 to 1.80, due to eliminating commute needs to school. Morning peak hours will no longer exist for students. But this induces a spread afternoon & evening peak period by being active for discretionary travel during the afternoon. Besides, due to telecommuting, students will travel shorter distances for discretionary activities. As a result, densely populated zones will become popular destinations for their discretionary activities. This finding raises the flag that policy attention is needed for small businesses in densely population zones, as they will be challenged with practice health and safety guidelines for COVID-19 with the added demand from telecommuting students.

Hesham El Masry, Ain Shams University
Mohamed El Esawey, British Columbia Ministry of Transportation and Infrastructure
Khaled El-Araby, Ain Shams University
Ahmed Osama, Ain Shams University

Private cars are typically designed to accommodate more than one person. However, they are often used to transport one or two individuals only. A single-occupancy vehicle (SOV) is a private vehicle whose only occupant is the driver. Private vehicles are not fully utilized when only a single occupant is using it. That leads to unjustified traffic congestion and shortage in available parking spaces especially in mega cities. Recently, some car manufacturers developed small-sized vehicles (1-2 seaters) that can serve different classes of users. Small vehicles (SV) can occupy less space, which is beneficial for cities suffering from limited space where constructing new roads or upgrading the existing ones is infeasible. Furthermore, many SVs are electrically powered which makes them environment friendly. Small vehicles offer other advantages compared to motorcycles and bikes; they provide shelter from adverse weather conditions, allow for storage space for carrying personal belongings, and can be more culturally accepted in conservative societies such as Egypt, where this study took place. This study examines the public willingness to purchase and use SVs using Multinomial Logit (MNL) models. The data was collected using an internet-based survey where the sample size included 317 valid responses. The results showed that females, younger individuals, and people who do not own any private cars in the current time are more willing to purchase and use SVs. These results provide some useful insights into the willingness of individuals to purchase and use small vehicles.

Abdul Rahman Masoud (amasoud@outlook.com), University of British Columbia, Okanagan
Ahmed Idris, Arab Academy for Science, Technology, Maritime Transport
Gordon Lovegrove, University of British Columbia

Auto-dependency has triggered many public health challenges for North American citizens including increases in physical inactivity rates, road collisions, and Greenhouse Gas (GHG) emissions. As a result, there has been a growing interest in addressing these challenges by promoting more sustainable and livable neighbourhood planning. This research developed an agent-based model that evaluates the impact of neighbourhood design on travel pattern by simulating agents’ daily trip activities. In addition, the results from the Capri-Landmark neighbourhood case study in Kelowna, BC are presented in this paper to demonstrate the developed model via an application to assess the impact of retrofitting the neighbourhood on travel pattern. The developed agent-based model employs a framework that integrates the traditional RUM based modelling technique with reinforcement learning concepts to account for the bounded rationality of human beings and knowledge learning process. Moreover, the model utilizes the diffusions of innovations theory to account for the impact of social interactions by simulating how agents share their knowledge and propagate information about their preferred travel mode across family members and co-workers. In addition, the model accounts for the iterative feedback process between agents’ actions and the environment. The results from the case study show that providing more accessibility to non-motorized users have more impact on modal share than restricting car use. In addition, increasing activity density would cause the most significant increase in transit modal share while combing increasing activity density with retrofitting the transportation network would achieve the most significant increase in non-motorized modal share.

Sangwan Lee (sangwan@pdx.edu), Portland State University
Michael Smart, Rutgers University
Aaron Golub, Portland State University

Recent changes to the transportation system, including the emergence of new mobility services, and the nature of immigration may have resulted in changes in how immigrants travel. A few studies, however, have examined immigrant’s travel behavior on emerging transportation modes, including car share programs, ride share applications, and bike share programs. We, thus, explore whether immigrants rely more on the alternative transportation modes than U.S born residents by examining the 2017 National Household Travel Survey. We develop the Zero-Inflated Negative Binomial Regression Models, which separate but linked equations; (1) a logit and (2) a negative binomial model. The logit models find the “immigrant effect” only for car sharing services and bike sharing programs; that is, relative to U.S. born residents, immigrants in the U.S. use car sharing and bike sharing services more frequently, while insignificant association found in ride sharing apps usage. The negative binomial models, however, suggest that immigrants use car sharing and ride sharing less frequently than would be expected. Immigrants who are in their first few years of living in the U.S. use smartphone rideshare app more frequently, confirming the “assimilation theory.” The results of the predicted frequency of the usage show that, with all other independent variables are held constant at its sample mean value, U.S. born residents would use car sharing and ride sharing services more frequently than immigrants, though the difference is marginal. Immigrants, however, would still tend to use bike share programs more frequently rather than U.S. born residents. We believe that from a policy perspective, the results of this paper can contribute to providing improved transportation policies that can respond to the needs of immigrants, as public policies require more reliable information on travel behavior.

Patrick Singleton, Utah State University
Antje Graul, Utah State University
Niranjan Poudel, Utah State University

Evolving mobility options including bike sharing and car sharing demand the introduction of novel frameworks in order to understand consumers’ personal transportation mode choices. On the basis of qualitative research (focus groups) and quantitative research (large scale survey), this paper proposes that in addition to considerations of instrumental attributes related to convenience, time-efficiency, and financial cost; non-instrumental attributes such as social and psychological motivations, safety concerns, health benefits, and environmental considerations play an important role in motivating personal travel mode choices related to evolving and more sustainable mobility options. In addition, we explore the effect that socio-demographic characteristics may have in influencing the importance attributed to such characteristics, and how they (together) influence transportation mode choice. Notably, we find that consumers who opt for non-motorized and shared transportation modes give comparably more importance to health, environment, well-being, and social attributes when considering transportation modes, but that positive ratings on these attributes may not overcome disadvantages in time and convenience enough to result in a different choice. In addition, demographic variables play an important role when considering the relative importance of transportation mode attributes. Specifically, gender, ethnicity, education, children in the household, and income levels significantly predict numerous modal attribute importance rankings (especially non-instrumental ones). Together, these findings contribute to prior literature on transportation mode choices from a consumer perspective and provide important insights for transportation managers on how to best target specific consumer segments to encourage sustainable transportation mode choices.

Katherine Asmussen, University of Texas, Austin
Aupal Mondal, University of Texas, Austin
Chandra Bhat (bhat@mail.utexas.edu), University of Texas, Austin

Understanding the “ if ” and “ when ” of autonomous vehicle (AV) adoption is of clear interest to car manufacturers in their positioning of business processes, but also to transportation planners and traffic engineers. In this paper, we examine the individual-level AV adoption and timing process, considering the psycho-social factors of driving control, mobility control, safety concerns, and tech-savviness. A ranked choice stated preference design is used to elicit responses from Austin area residents regarding AV adoption. Our results underscore the need to examine the adoption of technology through a psycho-social lens. In particular, technology developments and design should not be divorced from careful investigations of habits and consumption motivations of different groups of individuals in the population.  The findings from our analysis are translated to specific policy actions to promote AV adoption and accelerate the adoption time frame.

Muhammad Habib, Dalhousie University
Md Asif Hasan Anik, Dalhousie University
Caroline Robertson, Dalhousie University

Child-care centers are major trip generators for households with pre-school aged children (<5 years). This paper presents a framework to analyze the factors affecting daily travel choices and challenges of child-care going households. Through rigorous literature review, this study first identifies important variables affecting mode choice, child-care center choice, transit use, and intention to drive. Next, it develops and distributes an online survey questionnaire to the registered child-care centers in Halifax Regional Municipality, Canada, to get data on individual household’s travel choices, challenges, factors affecting those choices, importance level of those factors, as well as socio-demographic information. The study collects data from over 200 households, which is then analyzed using structural equation modelling (SEM) to examine the effects of the identified factors, such as, safety, cost, distance, location, affordability, and  convenience on day-to-day child-care trip decisions. The findings suggest that most of the families rely on private vehicles (73%) to make child-care trips as a result of family time, travel freedom, and parking availability. Other variables that motivate them to drive are travel safety, convenience, and opportunity to do trip chaining. Results show that 57% of the households group their child-care trip with work trip. Distance to nearest bus stop, carrying child- belongings, length of trip to child-care are considered major barriers for transit use. The results of this study are expected to assist policy makers to understand travel patterns and barriers of households with pre-school aged children and help in designing communities to promote sustainable travel behavior.

Elisa Bin (ebin@kth.se), KTH Royal Institute of Technology
Claudia Andruetto, KTH Royal Institute of Technology
Yusak Susilo, University of Natural Resources and Life Sciences
Anna Pernestål, KTH Royal Institute of Technology

By using 781 responses from an online questionnaire, this study investigates how individuals have changed their activity-travel patterns, during the COVID-19 pandemic period. The roles of the digitalisation solutions in replacing physical activities, and which behavioural changes that may be kept after the pandemic period are investigated. Case studies from Sweden, Italy and India are in particular analysed and compared in order to investigate the plausible impacts of the restriction measurements to the behavioural changes observed. The results show that the opportunity and possibility to change the behaviour matter. The ones who made conscious decisions not to travel for certain activities (whether it was imposed on them by external actors or by self-conscience) are the ones who consistently had a significant reduction in their trips. This is where different levels of restrictions of movement matter during the restriction period. However, the estimation results do not show any strong indication of countries’ influence (and their restriction policy) on one’s likelihood to adopt the (new/online based) behaviours for all the activities after the restriction period. The acceptance and long-term adoption of using technology alternatives tie more to the personality and socio-demographic group of the given person, which highlights the importance of promoting alternatives as a part of longer-term behavioural and lifestyle changes.

Juan Gomez, Universidad Politécnica de Madrid
Alvaro Aguilera-Garcia, Universidad Politecnica de Madrid
Felipe Dias, University of Texas, Austin
Chandra Bhat, University of Texas, Austin
Jose Manuel Vassallo, Universidad Politecnica de Madrid

New app-based mobility services are revolutionizing urban transport. Particularly, ride-hailing has experienced a worldwide boom in the last decade since it provides a convenient, on-demand door-to-door service for urban trips. In parallel, an increasing number of studies have analyzed individuals’ behavior towards this transport option, mobility patterns, as well as ride-hailing effects on urban sustainability. Nevertheless, the majority of these contributions focus on US cities, while almost no efforts have been devoted to other geographic areas, such as Europe. Cities in this continent present some particular characteristics that make them a case worth investigating, such as a higher presence of public transport modes or a greater public concern on environmental issues. The aim of this paper is to explore travel behavior towards ride-hailing services in a European city. Based on the information collected from a survey campaign in the city of Madrid (Spain), we estimate a Generalized Heterogeneous Data Model approach to identify the key factors motivating ride-hailing adoption and frequency of use. The paper identifies a higher adoption of ride-hailing services among young, well-educated, wealthy individuals, who are familiar with new technologies. More interestingly, the research suggests a noticeable role played by environmental consciousness in ride-hailing frequency of use, compared to US cities. Particularly, individuals with lower environmental consciousness are more car-oriented, which is also linked to a more intense use of ride-hailing. By contrast, individuals with a higher environmental consciousness tend to reduce their use of ride-hailing, which reflects their propensity towards public transport in a transit-intensive background.

Yang Zhou (txzhpy@gmail.com), Tongji University
Quan Yuan, Tongji University
Chao Yang, Tongji University
Yinhai Wang, University of Washington

Activity patterns clustering is crucial in understanding the regularity of human mobility scheduling and daily lives. There still remains the challenge of measuring similarity between pairs of activity patterns that are in the form of the categorial time series sequences. The existing studies measured similarity with binary vector or edit distance, but these methods cannot fully characterize routine arrangement and time scheduling of daily activities. To overcome these problems, we cluster activity patterns with a Markov-chain-based mixture model, which captures features of activity scheduling by Markov transition matrix and measures similarity with probability distribution. The logistic regression models are further built to analyze the relationships between activity patterns and socio-demographic characteristics. Results show that there exist three main human activity patterns – work-related pattern, recreational pattern, and child-care pattern, from the perspectives of daily routine and activity chains. The child-care pattern not belongs but highly related to the other two patterns. Each pattern exhibits significant difference in daily activity behaviors and corresponds to a specific group of people. The results reveal the three main kinds of life styles and could contribute to urban transportation planning.

Maike Puhe, Karlsruhe Institute of Technology (KIT)
Jens Schippl, Karlsruhe Institute of Technology
Torsten Fleischer, Karlsruhe Institute of Technology
Peter Vortisch, Karlsruhe Institute of Technology (KIT)

Large-scale changes are expected for urban mobility systems, triggered by digitalization and various other factors such as climate concerns, urbanization and others. For researchers and planners, it is therefore becoming increasingly important to understand the determinants of changeability and stability of mobility decisions. The motivation for the study is that in transportation research and modeling frameworks, travel choices usually derive from individual traits and accessibility variables. What is underrepresented by such an approach is that decisions are also socially embedded. We postulate that mobility patterns are strongly interwoven with the way people configure their daily life. In this paper, we introduce and discuss an empirical approach to investigate the social embeddedness of mobility decisions. Basic premise of the approach is that social network configurations provide an important setting for individual travel decisions. Analysis is based on a three-phase interview study, conducted in Karlsruhe, Germany. The analytical approach reveals that a substantial part of travel decisions are only loosely coupled to generalized costs of transport. Instead, the motivational degree linked to certain relationships largely influences willingness to travel and the relative stability of everyday life decisions. Relationships, which are internally satisfying, appear extremely persistent, while more extrinsically motivated relationships are maintained as long as external rewards are effective. This finding is valuable for a longer-term perspective. In a short- to medium-term perspective, it appears important to note that relationships are established and maintained through numerous interactions that differ in terms of their spatio-temporal flexibility and their degree of familiarity.

Erik Almlöf (ealmlof@kth.se), KTH Royal Institute of Technology
Isak Rubensson, Region Stockholm
Matej Cebecauer, KTH Royal Institute of Technology
Erik Jenelius, KTH Royal Institute of Technology

The COVID-19 pandemic has changed travel behaviour and reduced the use of public transport throughout the world, but the reduction has not been uniform. In this study we analyse the propensity to stop travelling by public transport during COVID-19 for the holders of 1.8 million smart cards in Stockholm, Sweden. We suggest two models for explaining the change in travel pattern, linking socioeconomic data with the probability to stop travelling. We find that education level, income and age are strong predictors, but that workplace type also substantially affect the propensity of public transport travel. Furthermore, we use clustering to divide the population into five separate social groups, serving as a more intuitive understanding of how the pandemic has affected different citizens’ propensity to use public transport. The results can guide policy makers on how to better tail e.g. bus supply to local demand, either through an increased understanding of differences based on the results or by further incorporating the results into a transport simulation models.

Daniel Reck (daniel.reck@ivt.baug.ethz.ch), ETH Zurich
Sergio Guidon, Eidgenossische Technische Hochschule Zurich
He Haitao, Loughborough University
Kay Axhausen, Eidgenossische Technische Hochschule Zurich

Shared micromobility services (e-scooters, bikes, e-bikes) have rapidly gained popularity in the past few years, yet little is known about their usage. While most previous studies have analysed datasets from single providers, only few comparative studies of two modes exist and none so-far have analysed competition or mode choice at a high spatiotemporal resolution for more than two modes. To this end, we develop a generally applicable methodology to model and analyse shared micromobility competition and mode choice using widely accessible vehicle location data. We apply this methodology to estimate the first comprehensive mode choice models between four different micromobility modes using the largest and densest empirical shared micromobility dataset to-date (~169M vehicle locations collected in Zurich over two months). Our results suggest that mode choice is nested and dominated by distance and time of day. Docked modes are preferred for commuting. Hence, docking infrastructure for currently dockless modes could be vital for bolstering micromobility as an attractive alternative to private cars to tackle urban congestion during rush hours. Furthermore, our results reveal a fundamental relationship between fleet density and usage. A "plateau effect"is observed with decreasing marginal utility gains for increasing fleet densities. City authorities and service providers can leverage this quantitative relationship to develop evidence-based micromobility regulation and optimise their fleet deployment, respectively.

Yuchen Song, Southeast University
Dawei Li (lidawei@seu.edu.cn), Southeast University

In amidst the popularity of Mobility as a Service (MaaS), the purpose of this paper is to extend the conventional intermodal trip-based path planning to the tour-based planning combining the activity-based analysis structure. The extension raises two challenges including reliable mode chain generation and the consideration of the necessary constraints in the path planning algorithm. A dynamic discrete choice model (DDCM) is proposed using purpose and mode choice as actions to capture the mode chain choice behavior of travelers. With the DDCM model results and individual-specific travel demand, the refined shortest hyperpath algorithm is applied to design full-day travel paths for users, which can consider mode and parking constraints. In the case study, DDCM is estimated according to a household travel survey, and two reliable hyperpaths are generated for the traveler to choose on the basis of his preference. Further still, the computation time of the proposed algorithm shown in seconds has a good performance.

Zhiqiu Jiang (zj3av@virginia.edu), University of Virginia
Max Zheng, University of Virginia
Andrew Mondschein, University of Virginia

While driverless shuttles systems and technologies are being unveiled in cities around the world, researchers still understand little about the factors behind driverless shuttle acceptance. In particular, it is unknown how attitudes towards driverless shuttles may change as the technology becomes more widespread. Using data collected in April 2020 from respondents of eight US cities, we investigate the factors associated with residents’ attitudes towards driverless shuttles. These eight US cities form a quasi-experimental design of four “pilot” cities and four “non-pilot” cities. Using exploratory factor analysis, we establish four attitudinal dimensions: confidence in the safety of driverless shuttles, privacy concerns, technology familiarity and interest, and preference for human control when traveling. We develop an ordered probit model to estimate the determinants of residents’ levels of adoption of driverless shuttles in pilot cities and non-pilot cities. We find that individuals in pilot cities demonstrate a greater awareness and a more acceptance of driverless shuttle programs, compared with those in non-pilot cities. Beyond greater awareness, the motivations for accepting driverless shuttles are different in pilot cities, including demographic broadening of support from frequent transit users and low-income individuals. Pilot programs of driverless shuttles positively impact residents’ confidence in data security and reduce privacy concerns. These findings provide transit operators, planners, policymakers with early evidence as to how driverless shuttles may be accepted by the broader population and what factors must be addressed if this new technology is to be an effective part of urban public transportation.

Fei Xue, Beijing Jiaotong University
Enjian Yao (enjyao@bjtu.edu.cn), Beijing Jiaotong University

Travel-related carbon emissions have been researched by many scholars, but few studies have focused on the impact of residential relocation on it. This paper adopts the Fifth Travel Survey of Beijing Inhabitants and the corresponding land-use data, building a structural equation model with latent variables to explore the effects of changes of the residential built environment, commute distance, and household social-economic attributes on changes of vehicle ownership and use and the related carbon dioxide emissions before and after the relocation. It is concluded that changes in commuting distance, private vehicle purchase, the number of electric bicycles, and the number of smart traffic cards in households have direct effects on changes in emissions. And relocating direction, changes in the convenience of living, the basic education children, and family size are found to have indirect effects. Besides, this paper discusses recommendations for reducing carbon emissions from motorized travel in response to the trend of suburbanization in big cities. This paper can help urban planners and managers understand the impact of residential relocation on vehicle’s ownership and use and the carbon emissions they generate, and can also provide a reference to formulate policies on reducing emissions and low-carbon travel.

Hamidreza Asgari, Florida International University
Rajesh Gupta, University of Lucknow
Xia Jin (xjin1@fiu.edu), Florida International University

This study examined the impacts of generation cohort and attitudinal factors on autonomous vehicle (AV) adoption and willingness to pay (WTP). A confirmatory factor analysis was performed to identify attitudinal factors and explore the differences across the generation groups. A structural equations model (SEM) was developed to capture the causal effects of socioeconomic and demographic attributes and attitudinal factors on adoption and WTP for AVs. Interaction effects between the attitudes and generational cohorts were further investigated to assess the potential sources of heterogeneity among Millennials. Model results revealed that regardless of the age cohort, technology savviness, the desire for driving assistance and safety offered by autonomous features, and the benefits of mobility of non-drivers encouraged both adoption and WTP for AVs. In addition, WTP was also highly affected by socioeconomic and demographic determinants including income, employment status, and previous experience with online shopping and ridesharing. Older Millennials (35-39 years old), and those who favored on-demand features were highly prone to AV adoption. On the other hand, lifestyle attributes such as short trips or limited number of daily trips might have negative impacts on AV adoption. Trust issues with the technology and data privacy also presented barriers toward AV adoption.

Jun Liu, University of Alabama
Kara M. Kockelman (kkockelm@mail.utexas.edu), University of Texas, Austin

Car-sharing and ride-sharing offer travelers another mode of transport in and between cities. To provide decision-makers reasonable information about mode splits, congestion, fleet operations, parking shifts, and other impacts of these new modes, transportation planners and researchers need microsimulation for advanced travel modeling approaches. Activity-based models (ABMs) typically anticipate travel choices at the level of individuals, and normally offer greater temporal and spatial details than traditional aggregate models such as four-step travel demand models whose units of analysis are traffic analysis zones (TAZs). ABM takes the individual’s daily activities chained by a series of travel trips, also called “tour” if the last trip ends where the first trip starts, e.g., home, as the travel demand input. The input can be simply summarized as “4Ws”: Who this person is, where he/she lives and works, what daily activities he/she undertakes, and when he/she plans to perform activities. This study delivers a methodological framework to prepare the “4W” inputs, taking advantage of existing travel model data (including travel survey data) and open-source data (like OpenStreetMaps). It provides a series of coded algorithms that output a complete synthetic population, with locations for all activities, trip chains or tours, and travel schedules. The tool is particularly useful for state, regional and local transportation-planning agencies and their consultants, who already have the network and travel data and seek to convert their existing TAZ-based models to microscopic (person-based) travel chains and ABMs that better support simulations of emerging travel behaviors modes.

Chandra Bhat (bhat@mail.utexas.edu), University of Texas, Austin
Aupal Mondal, University of Texas, Austin
Katherine Asmussen, University of Texas, Austin
Aarti Bhat, Pennsylvania State University

In this paper, we propose, for the first time, a closed-form multiple discrete-grouped extreme value model that accommodates grouped observations on consumptions rather than continuous consumptions. For example, in a time-use context, respondents tend to report their activity durations in bins of time (for example, 15-minute intervals or 30-minute intervals, depending on the duration of an activity). Or when reporting annual mileages driven for each vehicle owned by a household, it is unlikely that households will be able to provide an accurate continuous mileage value, and so it is not uncommon to solicit mileages in grouped categories such as 0-4,999 miles, 5000-9,999 miles, 10000-14,999 miles, and so on. Our proposed model is applied to study the weekend time-use choices of individuals.

Joshua Ward, Utah State University
Patrick Singleton (patrick.singleton@usu.edu), Utah State University
Antje Graul, Utah State University
Mackenzie Garrison, Utah State University

In this study, we qualitatively investigate instrumental and non-instrumental factors influencing travel mode choices, utilizing content analysis of discussions from a set of six mode-specific focus groups sessions. Users of more active/sustainable transportation modes (walking, bicycling) tended to discuss more non-instrumental (including affective or symbolic) attributes—such as health, emotions, and social influences—and placed greater importance on those attributes in their mode choices than those in the motorized mode (automobile, transit) sessions. Safety and security concerns also motivated travelers, but in different ways for different mode users. Situational differences also affected the degree to which people considered various mode choice attributes. Our work contributes to an understanding of the needs and motivations that underlie travel (and activity) choices, highlights the important role of non-instrumental factors on active travel mode decisions, and suggests a need for research to better measure and model these factors in order to inform policies to encourage sustainable transportation behaviors.

Alireza Rahimi, Florida International University
Ghazaleh Azimi, Florida International University
Hamidreza Asgari, Florida International University
Xia Jin, Florida International University

This paper presents a comprehensive analysis of the propensity toward autonomous vehicles (AVs). The main hypothesis of this study is that individuals’ decisions toward AVs vary by their modality style, and it is possible to identify a distinct set of attitudes toward AVs among people with distinct mobility profiles. A latent class clustering analysis model was applied to the survey data, and three distinct user classes (sub-datasets) were identified, including auto-dependent users, all-mode users, and non-drivers (passengers and transit users). Separate structural equation models were developed to identify a distinct set of attitudes for each user class and estimate their propensity toward AV technology. The results showed that attitudes play a critical role in users’ behavior toward AV adoption and WTP. Moreover, the identified attitudes for each class and their contribution to the decisions were different among the three classes, confirming the necessity to develop separate models to account for the heterogeneity in their choice behavior. In view of attitudes, pro-technology showed significant positive impacts on both AV adoption and WTP for auto dependent users and non-drivers. While self-driving features might motivate auto dependent users to adopt AV technologies, driving assistance features seemed to be more important for all-mode users in their adoption and WTP decisions. Trip privacy and data privacy concerns presented potential barriers for auto users and all-mode users but did not show significant impacts for non-drivers, who were more likely to be discouraged to adopt AVs by joy of driving.

Alireza Rahimi, Florida International University
Ghazaleh Azimi, Florida International University
Xia Jin, Florida International University

This paper presents a study in investigating the generational gaps between Millennials and Generation X in terms of their mobility attitudes. A comprehensive analytical framework was proposed in this study and applied to data obtained from a SP survey in the U.S. Four modeling steps were involved, to measure the attitudes, identify generational gaps, recognize potential contributors to the attitudes, and decompose the contributions into Endowment , Coefficient , and Interaction effects. The Endowment effects measure how much of the generational differences in can be attributed to socioeconomic and demographic variables, while the Coefficient effects reflect the gap that are due to actual behavioral changes or attitudinal shifts between the generations. The findings of this study confirmed the existence of generational gaps in mobility attitudes and revealed that a dominant portion of the gaps stemmed from the different perspectives between the generational cohorts. This indicates that these attitudinal disparities are likely to persist and remain at significant magnitudes, reflecting the unique views and values of the Millennials. Particularly, the preferences for transit and alternative modes and less reliance on private vehicles among Millennials were more of a reflection of their preferences in lifestyle choices and not so much constrained by their socioeconomic status as the previous generation. This study provides empirical evidence of the generational gaps between Millennials and the previous generation in terms of their mobility preferences. The findings provide valuable inputs for policy development in promoting sustainable transportation and community design.

Stavroula Maria Mavrouli (mavrouli.1@osu.edu), Ohio State University
Andre Carrel, Ohio State University
PRIYAMVADA NATARAJAN, Ohio State University
Andrea Broaddus, Ford Motor Company
Richard Twumasi-Boakye, Ford Motor Company
Cathy Xia, Ohio State University

This study assesses factors that may influence demand for private on-demand microtransit commuter shuttle service. It reports results from a survey featuring a stated preference (SP) choice experiment with a hypothetical microtransit service for employees of the Ohio State University Wexner Medical Center (OSUWMC) in Columbus, Ohio. The survey was designed to assess employees’ current commute patterns and investigate explanatory factors for choosing microtransit. Respondents were presented hypothetical scenarios with varying travel time and cost attributes for the service. The data (n=755) were used to estimate a mixed logit model with error components. Results indicated that this population is receptive to using microtransit for commuting, due in part to how they value time. The disutility caused by the in vehicle travel time (IVTT) of microtransit (-0.0953) was lower than the IVTT for driving (-0.144), implying that individuals might generally perceive commuting by microtransit as more enjoyable or productive. Respondents were indifferent between walking 5 minutes or driving 3 minutes to reach a microtransit pick-up point. The larger the time window for the shuttle’s arrival or departure, the less likely it was to be selected, implying people avoid wait times at work before and after their shift. Medical center employees have special characteristics, such as working long shifts through 24 hours with strict on-time requirements. Yet results may be transferable to other medical centers, especially in urban locations where parking is difficult or expensive, and employees have high commute dissatisfaction.

Sungtaek Choi, Georgia Institute of Technology (Georgia Tech)
Joonho Ko, Hanyang University
Daejin Kim (daejin.kim@gwnu.ac.kr), Gangneung-Wonju National University
Yongseok Ko, Korea Research Institute for Human Settlements
Jonghan Park, Hanyang University

This study identifies how the adoption of autonomous vehicles (AVs) affects the value of travel time savings (VOTTS) for commuting trips under different trip distance scenarios (intra- and inter-city travel). Using a stated preference survey administered to Korean commuters in 2019, two multinomial logit mode choice models are developed. The models are stratified by trip length and include four alternatives: AV with a manual driving option (AVMD), AV with a self-driving option (AVSD), shared AV (SAV), and public transit (PT). The result shows that the value of in-vehicle travel time savings for AVSD (on average, $7.05/hr) is lower than for the other two AV modes (AVMD = $9.51/hr, SAV = $12.68/hr), suggesting that the private hands-free mode allows travelers to use their travel time more productively. In addition, changes in VOTTS for each mode vary by trip length. As travel distance increases, VOTTS for shared modes (SAV and PT) tends to decrease. Value of in-vehicle travel time for SAV is the highest among the four modes in the intra-city travel scenario but is considerably lower in the inter-city scenario. This difference potentially tells us that people need a certain amount of time to fully engage in their activities while traveling, highlighting the importance of trip length in determining VOTTS. Interestingly, VOTTS for PT is the lowest among the four modes, implying that there might not be a substantial decrease in VOTTS in a near AV future due to the potential risk of accidents, unfamiliarity with AVs, or unchanged mode preference.

Miloš Balać, Eidgenossische Technische Hochschule Zurich
Sebastian Hörl, IRT SystemX

This paper describes the steps to create a synthetic population of any region in California. By using only open data, and an open-source population synthesis pipeline, we ensure that the whole process can be easily repeated by others. This not only ensures reproducibility and transparency of the synthesis process, but also allows that studies using this population can be easily replicated. As agent-based models are gaining in popularity in recent times due to the rapid developments in the transportation sector, the need for convenient ways to generate synthetic individuals and their daily patterns has grown as well.         We present our approach for two regions: nine-county San Francisco Bay area and San Diego County. The validation results show that the methodology used is suitable to replicate socio-demograpahics and activity patterns of the population. However, it also points to some limitations due to the lack of data and the methods used. Nevertheless, the approach could be a good complement to the local and regional transportation models, as it allows easy access and can be readily used in agent-based models.

Jaehyung Lee, Yonsei University
Euntak Lee, Yonsei University
Jaewoong Yun, Yonsei University
Jin-Hyuk Chung, Yonsei University
Jinhee Kim (kim.jinhee@yonsei.ac.kr), Yonsei University

Autonomous driving technologies (ADTs) are transformative because they are expected to let drivers from driving. Since travel distance is closely related to the burden of driving, it is expected to affect the preference of ADTs. However, the effects of travel distance on people’s preference for ADTs have been largely ignored in the transportation literature. Therefore, the present study is designed to investigate the preference of ADT and its heterogeneity by travel distance. Hypothetical choice situations were designed to investigate people’s preference for ADT compared with human driving with varying travel distance. Four sets of choice profiles per distance were given to respondents, and each choice profile consists of four hypothetical choice situations with different travel time and cost. The experiments were designed using Bayesian D-efficient method for each choice profile per distance. Using the stated preference data collected via the experiments, four latent class models were estimated to account for taste heterogeneity with classes and across travel distance. The estimation results show that three classes are revealed by the preference of ADTs regardless of travel distance. In addition, in-vehicle activities by classes during autonomous driving are observed across class and travel distance. Sleep/Rest and Using ICTs are the two most preferred in-vehicle activities and preference for Sleep/Rest of those who prefer ADTs varies with travel distance.

David Palma (d.palma@leeds.ac.uk), University of Leeds
Chiara Calastri, University of Leeds
Jacek Pawlak, Imperial College London

With the increasing importance of mobile technologies and growing urban sprawl, ensuring productivity and enjoyment of time spent travelling has become a key issue in promoting public transport worldwide. Yet, the investigation of how people spend travel time is an emerging field in travel behaviour research. This paper contributes to this area of study by jointly analysing activity engagement and time use while travelling, with a particular focus on multitasking. The application of the newly developed eMDCEV model allows relaxation of the assumption of a defined time budget, and permits investigating substitutions and complementarity effects between different activities. The model results unveil the interplay between activity engagement, travel mode and travel purpose, and the forecasting exercise provides a compelling illustration of the synergies between different activities conducted while travelling.

Milad Ghasri (m.ghasri@unsw.edu.au), University of New South Wales
Akshay Vij, University of South Australia

The emerging technology of autonomous vehicles is envisaged to substantially alter transport systems and travel behaviour. However, this impact is conditional on the level and rate of AV market penetration. This study examines consumers consideration for AV while controlling for the influence of media commentary and social network sentiment. The data about individuals’ most used mass media and social media platforms is obtained from an online survey conducted in December 2019 on sample of 862 Sydneysiders in Australia. Respondents in this survey were asked to introduce a subset of their social contacts, which then was used to design a customised discrete-choice experiment. A latent class choice model is developed to model AV consideration and elicit taste towards various sources of information. Estimation results reveal eleven segments in the sample that differ from one another in terms of sociodemographic attributes, evaluation of social contact trust, and the relative importance they attach to different sources from which they receive information about AVs. The findings show word of mouth received from females are valued less, and positive word of mouth is valued more than negative word of mouth.

Ruiying Lu (1310212337@qq.com), Southeast University
Xiucheng Guo, Southeast University
Pei Liu, Southeast University
Cong Qi, Southeast University
Juchen Li, Southeast University

Researches on tourism mode choice will contribute to understanding travel behavior characteristics and developing specific tourist travel policies. This study aims to predict the mode choice of tourists using the travel behavior analysis approach based on activity chains. A travel survey from 521 respondents in Nanjing, China in 2020 was conducted to obtain tourist’s individual and household attributes, basic travel characteristics and detailed information of a one-day activity chain. The activity-chain-based travel mode choice model of tourists was proposed using a two-level Nested Logit (NL) model, where the upper level consists of activity chain patterns and the lower level represents travel mode choice. Model estimation results confirmed significantly positive effects of activity chain patterns on tourists’ mode choice. It can be indicated that the increasing complexity of activity chains will reduce the choice probability of  public transport. Results also clarified the impact of tourist’s individual, household attributes and activity chain characteristics on the two choice levels. According to the obtained influencing mechanism, policy implications for cities similar to Nanjing were put forward, so as to deal with the holiday traffic congestion in urban areas and attract more tourists to suburban scenic spots.

Alireza Dianat, University of Toronto
Khandker Nurul Habib, University of Toronto
Jason Hawkins, University of Toronto

The COVID-19 lockdown provided many individuals an opportunity to explore changes in their daily routines, particularly when considered in combination with an ever-changing Information and Communication Technology (ICT) landscape. These new routines and alternative activities have the potential to be continued in the post-COVID era. Understanding the changes in routine is essential for transport planners to accurately model activity-travel scheduling. This study seeks to understand the impacts of the COVID-19 pandemic lockdown on activity-travel behavior and the utilization of alternative solutions from ICT. A particular focus is anticipating the long-term implications of this disruption on activity-travel scheduling. Focusing on five of the most repetitive activities in the daily schedule (working, grocery and non-grocery shopping, preparing/eating meals, visiting family/friends), this study investigates the changes in frequency and mode of completing these activities during the lockdown and shortly after reopening. Through a combination of graphical and statistical analysis, we find an increased preference for home food preparation over online ordering and that individuals are wary of in-person shopping until a large proportion of the population has received a vaccine against the virus. Respondents prefer to work at home, provided they have access to suitable home office supplies (e.g., desk, chair, computer monitor). Those with children also require adequate childcare before considering a return to the workplace.

Konstadinos Goulias (goulias@geog.ucsb.edu), University of California, Santa Barbara
Rongxiang Su, University of California, Santa Barbara
Elizabeth McBride, University of California, Santa Barbara

In this paper, we develop a new joint pattern recognition method that combines network motif-based analysis with activity sequence-based analysis. We use the advantages of both methods in creating groups of patterns that have within them distinct pattern homogeneity and across-pattern heterogeneity. The first portion of the analysis here applies a more traditional approach to identify unique network motifs, with 16 of them capturing 83.05% of the 2017 NHTS-California workday data. Multivariate analysis of grouped motifs data shows different preference of motifs for students, part-time workers, retirees, telecommuters, drivers, women, and younger adults. In the second portion of the analysis, motifs are grouped into categories based on the number of locations a person visits in a day and their correlation with time use and travel is explored. Time use and travel are analyzed based on minute-by-minute time allocation pattern identification using sequence analysis and hierarchical clustering. The correlation between motifs group and sequence analysis finds substantial heterogeneity within the motif groups. The within motif group clusters of activity-based sequences show typical commuting, going to school, and resting patterns. We also find seven patterns that are not typical but have similarities across motifs in their temporal footprint and the variety of activities in each sequence.

Andrea Hamre (ahamre@gmail.com), Montana State University

Traveler segmentation may improve efforts to reduce car reliance by enabling policies and programs to be targeted towards those most likely to change behavior. Segmentation leverages sociopsychological constructs and lifestyle measures relating to values and attitudes, and has focused on the key indicators of modes and vehicles. This study employs factor identification and cluster analysis to segment travelers by modal orientations in order to evaluate vehicle ownership and mode use in the small urban area of Chittenden County, Vermont. The study utilizes a unique series of four travel surveys specifically designed around the collection of travel attitudes and transportation planning priorities, collected in 2000, 2006, 2012, and 2018. Ordered logistic regression for household vehicles and multinomial logistic regression for mode use indicate the modal orientation clusters are significantly associated with these outcomes, even after controlling for income, household size, and geographic setting. Transportation planners in small urban areas may focus reliance reduction on the Car Tolerant cluster, representing about half the regional population and characterized by a willingness to change travel behavior, concern about congestion and safety, and willingness to prioritize incentives for alternatives.

Shin Hyung Cho, Georgia Institute of Technology (Georgia Tech)
Eun Hak Lee, Seoul National University
Hosuk Shin, Seoul National University
Dong-Kyu Kim, Seoul National University
Seung-Young Kho (sykho@snu.ac.kr), Korea National University of Transportation

The development of an integrated transportation system has required various strategies to improve mobility in public transportation networks. The aim of this study is to construct a route choice model that incorporates heterogeneous travel behaviors and to recommend policy improvements that consider travel behavioral characteristics in transportation planning. The behavioral differences of elderly travelers are compared with young adult travelers by the parameters of travel time attributes, and the value of travel time savings and multiplier are applied to directly compare with previous research. The mixed multinomial logit model is used to estimate the heterogeneous route choice behaviors. The stated preference data are obtained to model the travelers’ route choices from a field survey in the metro network. The results show that travelers have the most sensitive characteristics to the transfer time, and the young adult travelers are more sensitive to access and waiting time than in-vehicle time. The value of travel time savings is used to understand the degrees of travel time attribute as a monetary value, and the results indicate that the transfer time of the elderly has the greatest value. The multipliers of the travel time attribute show the range from 0.7 to 2.0 or higher, but the multipliers are different from each other when the methodology, region, and the type of data are considered. The results also provide the implications of transport policy on different travelers’ segmentation in consideration of diminishing the specified travel time attributes on route travel in the metro network.

Fatemeh Mohajeri, Imam Khomeini International University
Babak Mirbaha, Imam Khomeini International University

Iran is located on a seismic belt and it is always in danger of earthquakes.  Therefore, focusing on the trip choice behavior in urban road network is essential to manage earthquake disaster and confront its destructive effects. Since choice behavior is a function of interaction between physical and behavioral factors, it is attempted to study the pattern of choice behavior in Qazvin transport network by considering physical and behavioral factors simultaneously. The required data were collected through the stated preferences(SP) survey for six designated earthquake scenarios and revealed preference(RP) survey for the earthquake occurred on December 20, 2017. The binary logit models were applied for analyzing the choice behavior in the stages of chosen action, mode choice and destination choice in response to earthquake disaster. The results indicate that socio-economic characteristics, familiarity with alternative routes, making decisions on the basis of experience, trust in others, cultural and religious beliefs, severity of earthquake and the time of earthquake occurrence are significant variables in trip choice behavior in response to earthquake disaster. After analyzing the results of questionnaires in Transcad software, it was found that in the time of evacuation, 52% of the network links are in oversaturated condition compared to only 17% of network links under normal condition. In our case study, it was found that Vehicle-Time traveled by all vehicles in the network increase 446.4%, due to trip choice behaviors of individuals during earthquake. Also, the fuel consumption will also increase 271.5% in the entire network.

Christopher Cummings, Northwestern University
Divyakant Tahlyan, Northwestern University
Hoseb Abkarian, Northwestern University
Hani Mahmassani, Northwestern University

This paper is focused on understanding the factors driving the satisfaction of the attendees at a large annual planned special event named AirVenture during the arrival and departure processes from the event site. The study uses data from a survey conducted amongst 627 attendees at AirVenture 2019, a large air show held in Oshkosh, Wisconsin. Survey respondents reported their experienced quality of transportation related services and satisfaction with the arrival and departure processes, along with other socio-demographic information. Using the available data, we use a bivariate ordered probit model to delineate the factors driving attendees’ satisfaction during the arrival and the departure processes. Results indicate that attendees’ satisfaction is largely driven by the quality of the transportation services available at the event site, including wait times to enter/exit from various parking lots, travel time to reach from parking lots to the main gate of the event site (and visa-versa) etc. Further, the results also suggest that the past experience with the event during previous event years or during the arrival process sets the expectation for the quality of transportation services at future time periods and thus affects satisfaction. Lastly, the estimated model suggests the presence of significant systematic heterogeneity in the effect of various factors on arrival and departure satisfactions. The insights provided from these results can help improve transportation planning for special events by identifying ways to increase attendees’ satisfaction with the system.

Abdulfatah Abu (abdulfatah.abu16@imperial.ac.uk), Imperial College London
Chenyang Wu, Southeast University
Aruna Sivakumar, Imperial College London

The joint nature of decisions at the same level has become a point of interest in the transportation modelling literature and models that account for the simultaneity in the decisions have been under development. In this paper, we explore what we refer to as mobility portfolio models that aim to jointly model the decision to use multiple mobility resources. We develop a multivariate ordered probit for jointly modelling vehicle availability, regular public transport usage, regular taxi and ridehailing usage, and car sharing subscriptions for London using the 2018/19 London Travel Demand Survey (LTDS). Model results show significant negative correlations in the unobserved effects between car availability and regular public transport use, and car availability and car sharing subscriptions. We also find a strong positive correlation in the unobserved effects between regular public transport use and regular ridehailing use. Finally, we explore how changes in household, individual and neighbourhood characteristics affect the probability of each outcome.

Marcelo Altieri (m.altieri@fe.up.pt), Universidade do Porto Faculdade de Engenharia
Cecília Silva, Universidade do Porto Faculdade de Engenharia
Shintaro Terabe, Tokyo University of Science

Extensive research has found that people are more likely to choose the transport alternative which offers shorter travel time. But few studies approached the role of travel time across different transport alternatives and cities. This research assesses the influence of competitive travel time between car and public transit in public transit modal share for commute trips. São Paulo, New York, and Tokyo were selected to perform the analysis. A Fractional Logistic Regression in the binary form was drawn, and a competitive travel time index was calculated based on the Car/Public transit travel time ratio weighted by the number of the employed population at the origin and jobs at destinations in the absence of an Origin-Destination matrix. Findings suggest that, though car ownership was identified as the major factor, Car/Public transit travel time ratio is positively associated with the increase of public transit modal share. Furthermore, the Car/Public transit travel time ratio effect in public transit modal share consistently increases as people get increased access to cars.

Marcelo Altieri (m.altieri@fe.up.pt), Universidade do Porto Faculdade de Engenharia
Cecília Silva, Universidade do Porto Faculdade de Engenharia
Shintaro Terabe, Tokyo University of Science

This research focuses on the empirical assessment of the influence of the competitive travel time (CTT) and competitive travel cost (CTC) on public transit modal share in an environment of high levels of public transit service, and high costs for car use. It addresses an important question to urban planners, transit operators, and policymakers in still developing cities: What kind of competitive environment public transit in developing megacities are expected to experience once they achieve high levels of service? Tokyo Metropolitan Area (TMA) was selected as study case since it has one of the most competitive and successful public transit services around the world, which accounts for about one-third of the trips and a half-part between motorized modes, according to the latest Household Travel Survey from Tokyo.A Generalized Estimation Equation (GEE) was used to examine the effects of both the selected competitive factors on public transit modal share for commuting trips. Data for this research was obtained from the 2008 Tokyo Person Trip Survey (TPTS) and the 2010 Japanese Population Census regarding the four prefectures that compose the TMA, namely Tokyo, Kanagawa, Chiba, and Saitama. Overall, the results find that transit use levels advance from faster and cheaper services, however, varying from different regions in TMA and whatever it is the commuting origin or destination. Furthermore, though each megacity is unique in its terms, findings from this research provide insights for the design of future transport policies in developing megacities.

Chang Liu (cliu28@ncsu.edu), North Carolina State University
Eleni Bardaka, North Carolina State University

The travel behavior of university students has become an important research topic in recent years. Previous studies have shown that although the majority of off-campus university students commute to school by personal vehicle, they also show great interest in sustainable travel modes with comparable accessibility and considerable benefits. However, there are very limited studies focusing on the travel behavior of university students who use sustainable travel modes. This study presents a preliminary analysis and comparison of the travel behavior of university student and the other young adult transit users aged between 16 and 35 in the Denver-Aurora Combined Statistical Area. We first present a descriptive analysis of socioeconomic characteristics, trip purposes, and access/egress modes of the young transit users. The results indicate that the university students have different travel characteristics compared to the younger adults; as an example, the university students prefer to travel by light rail more than the bus. Then, we specifically focus on access mode choice for light rail users. We use a multinomial logit model to identify the factors influencing the access mode choice and compare between the university students and their counterparts. Our results indicate that travel time and the built environment impact students' choices differently compared to younger adults when accessing light rail stations.

Divyakant Tahlyan (dtahlyan@u.northwestern.edu), Northwestern University
Amanda Stathopoulos, Northwestern University
Michael Maness, University of South Florida

This study is aimed at examining the multi-dimensional nature of social capital and the relationship between these dimensions and travel behavior. Prior research has shown connections between single social capital concepts such as social support, with aspects of travel behavior. In this work we expand the definition of social capital to cover multiple dimensions, modeled via multiple indicators. Specifically, we make use of data from the Pew Internet Networks and Community Survey to build a structural equation model dividing social capital into two latent dimensions: social support capital and social network resources capital and then study the relationship of these dimensions with social activity participation diversity and frequency. Moreover, broader measures of neighborhood and community engagement are included in the model to explain how such engagement can help accumulation of social capital. Our results indicate a positive relationship between both social capital dimensions and social activity participation. Further, the results also suggest absence of correlation between social support and network resources capital, strengthening the hypothesis that social capital is multi-dimensional. In terms of explaining the social capital accrual, we also find that both community and neighborhood engagement are associated with network resources capital but not with social support capital. This suggests that individuals predominantly rely on close-knit and stronger relationships for social/emotional support, while instead, neighborhood and community engagement significantly effects accumulation of social network resources.

Patrick Singleton (patrick.singleton@usu.edu), Utah State University
Niranjan Poudel, Utah State University

Tradeoffs between work, travel, and leisure have been made explicit through the value of travel time savings (VTTS) and its components, including the values of time assigned to work (VTAW) and to travel (VTAT). A few studies have recently been able to empirically decompose VTTS into VTAW and VTAT, but they usually rely on costly multiday activity-travel-expenditure diaries and surveys. In this paper, we use panel mixed logit modeling to analyze a very simple and easy-to-administer set of choices (work time and commute time tradeoffs) that allows us to quantify some (but certainly not all) aspects of these VTTS components: specifically, we find the signs (positive or negative) and relative magnitude of the marginal utilities of work time and commute travel time. In contrast to previous research, our results imply that VTAT is more negative than VTAW (commuting seems to be more onerous than working), except that the opposite is true for walk and bicycle commuters. Around 4% of our sample had positive marginal utilities of travel time. We discuss and link our findings to the broader literature around (positive) valuations of travel time.

Xiao Fu, Southeast University
Chen Yang, Southeast University
Zhiyuan Liu, Southeast University

With the rapid development of information and telecommunication technology, joint activity or joint travel constitutes an important share of individual’s daily activity-travel pattern. In recent years, joint activity-travel pattern (JATP) scheduling models have been developed to investigate individuals’ independent and joint activity/travel choice behavior. The additional benefit resulting from joint activity/travel, which is related to the length of joint episode, is identified of significant concern in individuals’ JATP scheduling. In previous JATP scheduling models, joint activity/travel benefit is generally modeled with simulated parameter. As a pioneering endeavor, this study is devoted to quantifying the relationship between joint activity/travel benefit and JATP utility with specific consideration of joint episode’s length based on activity-travel survey data. A two-stage framework is proposed to estimate joint activity/travel benefit in JATP scheduling model. At the first stage, a bi-level programming model is proposed to estimate the joint activity/travel benefit. At the second stage, Kalman Filter is used to reduce the influence of deviation of survey data on the accuracy of estimating joint activity/travel benefit. The proposed method is examined with the data of travel characteristics survey collected in Hong Kong. The results show that the proposed method is effective in estimating joint activity/travel benefit for JATP scheduling model.

Julene Paul (julenepaul@ucla.edu), University of California, Los Angeles
Evelyn Blumenberg, University of California, Los Angeles

Many public agencies and advocacy organizations are working to decrease the reliance of Americans on the private automobile, especially as high-profile climate change policies highlight the negative environmental consequences of driving. In this context, a better understanding of cohort influences on travel behavior can help predict future trends in vehicle miles traveled. In this paper, we examine the relationship between lifecycle, cohort and period effects and vehicle ownership in the U.S.  The analysis relies on data from the 2005, 2010, and 2015 one-year American Community Survey Public Use Microdata Samples (PUMS).  We use median polish analysis to test the presence of cohort effects and then model vehicle ownership status by generation and by year. The analyses show that the oldest cohort (those born prior to 1945) are less likely to live in households with cars than more recent cohorts. However, Millennials differ from previous generations as they too are less likely to live in households with automobiles.  Although additional data are needed to draw a definitive conclusion, the findings suggest that Baby Boomers may represent a generation of “peak” automobile users.   The potential cohort effect among Millennials may be encouraging to those interested in reducing automobile use. However, given the access benefits of automobiles, declining vehicle ownership may disadvantage some Millennials, a group that continues to fare worse economically than previous generations at similar ages. Finally, recent developments—new auto-based shared services and the current COVID-19 pandemic—complicate efforts to forecast automobile ownership and use moving forward.

Ke Wang (wxk@tongji.edu.cn), Tongji University
Xin Ye, Tongji University

This paper developed a mixed multinomial probit (MNP) model with alternative error specification and random coefficients to accommodate flexible covariance structure and taste variation. The model can be efficiently estimated with analytic approximations of multivariate normal cumulative distribution (MVNCD) functions, which avoids defects of simulation-based integration. Simulation experiments and empirical analysis of Shanghai commuters’ mode choice behavior were undertaken to examine the performance of mixed MNP models. Both simulation results and empirical results show that mixed MNP models can well accommodate flexible covariance structures (even with negative correlations) and taste variation reflected through random coefficients being associated with both generic and personal variables. A negative correlation between taxi and metro, when other insignificant correlations are fixed at zeros, implies that the correlation between taxi and metro is less than those between other modes. Unequal error variances reflect heterogeneities in unspecified factors in commute modes’ utilities. Random coefficients of “in-vehicle time of car” and “number of companions” indicate that taste heterogeneity does exist across individuals. All theoretically identifiable parameters in mixed MNP models can be correctly estimated with a large enough sample size in simulation experiments. When empirical data cannot allow for estimating all these identifiable parameters, some parameters may be fixed. Empirical results indicate that the mixed MNP model performs better than traditional discrete choice models, such as the multinomial logit (MNL), mixed MNL, and MNP models. The mixed MNP model will offer an alternative efficient way to accommodate taste heterogeneity and flexible error covariance structure in discrete choice models.

Ke Wang (wxk@tongji.edu.cn), Tongji University
Xin Ye, Tongji University
Cheng Shi, Tongji University

This paper develops a probit-based discrete-continuous model to analyze the relationship of household car ownership and the duration of commuters’ three non-work outdoor activities (Act1: shopping and dining, Act2: leisure and entertainment, and Act3: visiting relatives and friends) in Xiaoshan District, Hangzhou City, China. And the model is estimated using a full-information maximum likelihood (FIML) procedure. The simulation experiment verified that the probit-based model can accommodate a flexible error covariance structure (including negative covariance). When one element in the error covariance matrix is fixed, the remaining covariances are identifiable. In this paper, the standard deviation of the binary discrete choice is fixed to 1 in both simulation and empirical research. The empirical dataset was collected from a face-to-face household travel survey conducted in 2015. Empirical results indicate strong effects of individual and household socio-demographics, built environment attributes, work-related and activity-related characteristics on the car ownership decision, and the duration of three non-work activities. The analysis shows a positive correlation in unobserved components among the car ownership decision and the duration of Act1 and Act2, indicating a mutually promotive relationship. Similarly, a negative correlation between the duration of Act1 and Act2 shows that non-work activities’ duration is mutually substitutive. These findings will help to better understand commuters’ household car ownership decision and non-work outdoor activity behavior restricted by fixed work schedules in developing countries, which can in turn guide the formulation of transportation policies, the growth of car ownership, and the development of commuters’ activity-based travel demand models in developing countries.

Adam Weiss (adam.weiss3@carleton.ca), Carleton University
Sam Pollock, University of Calgary
Lina Kattan, University of Calgary, Schulich

Shared e-scooters are a new disruptive micro-mobility technology. Several Canadian cities, including Calgary, have launched pilot programs to test these micro-mobility options. As part of the Calgarian pilot, a public engagement survey was launched to collect information about the public’s perception of shared e-scooters. The survey includes a set of questions related to users’ perception of comfort using shared e-scooters on various surfaces: roads, bike lanes, sidewalks and dedicated pathways, as well as a question related to frequency of use of the e-scooters. Understanding how the safety of these e-scooters is perceived is an essential component of encouraging e-scooter ridership as a first/last mile alternative to automobile travel and is expected to have a high degree of correlation with the frequency of use. This research will present a joint ordered choice model examining how different sociodemographic characteristics and typical travel patterns of the survey respondents influence perceptions of comfort on these different surfaces as well as  their frequency of e-scooter use. The outcomes from this research shed light on policy recommendations around where e-scooters can be used and provide rationale for prioritizing the investment into more dedicated infrastructure for the e-scooters and/or other active modes of travel. This awareness will allow decision makers to tailor new policies to improve the overall usage of these modal options and maximize the benefits that these new technologies will provide.

Ke Wang (wxk@tongji.edu.cn), Tongji University
Xin Ye, Tongji University

Commute trips account for a large portion of travel demand in peak hours and significantly influence the operation of urban transportation systems. To make efficient travel demand forecasting for commute trips, a fully flexible multinomial probit (MNP) model was applied and compared with the more traditional travel demand models, including the multinomial logit (MNL), the cross-nested logit (CNL), the homoscedastic non-independent MNP (HONI-MNP), and the heteroscedastic independent MNP (HI-MNP). The two-variate bivariate screening (TVBS) method, a recently developed analytical evaluation of the multivariate normal cumulative distribution (MVNCD) function, was used to estimate a 6-alternative MNP model in this paper. The sample data were derived from a web-based travel survey conducted in Shanghai. The results show that the fully flexible MNP performs best, and the heterogeneity across individuals and the correlation among different modes should not be neglected. Compared to the MNP, other models tend to underestimate the value of travel time. The HI-MNP tends to greatly underestimate the bus’ variance and slightly overestimate the metro’s variance. The HONI-MNP and the CNL both underestimate the positive error correlation between car and metro, which indicates a substitutive relationship between the Shanghai metro and private cars. Findings also provide useful references for policymakers to encourage commuters to shift from private cars to public transit. And the TVBS method makes the evaluation of the high-dimensional integral of MVNCD more efficient and provides a feasible way to integrate the MNP model into a city-wide travel demand modeling system.

Junji Urata, University of Tokyo
Zhengtian Xu (zhengtian@gwu.edu), George Washington University
Jintao Ke, Hong Kong University of Science and Technology
Guojun Wu, Worcester Polytechnic Institute
Yafeng Yin, University of Michigan, Ann Arbor
Hai Yang, Hong Kong University of Science and Technology
Jieping Ye, DiDi AI Labs

Ride-sourcing drivers spend a significant portion of service time being idle, during which they can move freely over the spatial market to search for the next customer. Such customer-searching movements, while not directly controlled by ride-sourcing platforms, impose great impacts on the service efficiency and thus need to be better understood. To this purpose, we design a dynamic discrete choice framework by modeling drivers' customer search as absorbing Markov decision processes. The model enables us to differentiate three latent search movements of idle drivers, as they either remain motionless, cruise around without target areas, or reposition towards specific destinations. Our calibration takes advantage of large-scale empirical datasets from Didi Chuxing, including the transaction information of 5 million passenger requests and the trajectories of 32,000 affiliated drivers. The calibration results uncover the variations of drivers' attitudes in customer search across the time and space, and compare the differences of search behavior between full-time and part-time drivers.

Annesha Enam (aenam@anl.gov), Argonne National Laboratory
Ali Ardeshiri, University of New South Wales
Taha Rashidi, University of New South Wales
Joshua Auld, Argonne National Laboratory

In the last couple of years, the emergence of innovative technologies has offered interesting challenges to the travel behavior research community. The anticipation of novel modes such as connected and automated vehicles (CAV) has resulted in multifarious research topics including the understanding of the adoption and usage of this emerging mode. People’s attitudes towards this imminent mode have also received an attention, along with the perception of value of travel time (VOTT). The primary objective of the study is threefold: first, using attitudinal indicators, the paper identifies prevalent types of attitudes towards AV; second, the paper explores variation of those attitudes across different demographic groups; third, the paper quantifies the influence of these attitudes on the preference for AV. To this end, the paper uses a stated preference survey data collected among respondents of Chicago in 2019. The exploratory factor analysis (EFA) of the attitudinal indicators identifies four factors; namely “Cautious Travelers”, “AV Enthusiasts: Hedonists”, “AV Enthusiasts: Utilitarians”, and “AV Enthusiasts: Safety First”. Next, the paper develops an integrated choice and latent variable model to test the influence of these four types of attitudes on the preference for AV. The result indicates that, the cautious travelers (having a high value of the first factor) associate higher VOTT to AV while enthusiastic travelers (having a high value of the last three factors) associate lower VOTT to AV compared to a base model that does not account for such attitudes.

Tanjeeb Ahmed (ahmed.t@uci.edu), University of California, Irvine
Rezwana Rafiq, University of California, Irvine
Sarwar Jahan, Bangladesh University of Engineering and Technology

The urban life has become so involved that people try to integrate several tasks during their work trip to save time and money. These resulting chains of trips have been claimed to have significant impact on the mode choice behavior. Dhaka is one of the highly dense megacities with an intricate land use composition and a variety of transport mode options that has the potential to portray unique trip chaining pattern. This study is designed to investigate this travel behavior by creating worker classes based on trip chaining pattern and mode choice. Results indicate that women and married commuters are more inclined to make complex trip chains. It also indicates the presence of a distinct groups of complex trip chain makers who mostly use non-motorized transport modes, like walk or rickshaw. The findings from this paper can provide useful insights on the complex travel behavior of commuters, which can be used to develop necessary travel demand management policies.

Trang Luong (TrangLuong@usf.edu), University of South Florida
Michael Maness, University of South Florida

Transportation research has paid deeper attention to exploring factors affecting leisure activities and their induced travel. Motivated by the influence of social capital on leisure activity behavior, this paper proposes a theory that leisure activity variety is an instrumental outcome and thus mostly affected by instrumental social resources. The theory underlines two hypotheses that 1) social capital is an integral determinant of leisure activity participation, and 2) having access to instrumental social support promotes instrumental outcomes demonstrated by increased in leisure activity variety. This theory was comprehensively tested on the number of different unique leisure activities collected from 1,297 survey respondents. To the authors’ knowledge, this refined and specially designed survey is the first in the transportation literature to use both position generator and resource generator to measure social capital. Results from negative binomial regression models demonstrated that instrumental support indeed had the largest influence on predicting activity variety outcome. This study’s findings helped to reduce the biases and unobserved heterogeneity across various socioeconomic attributes. As social capital has distinct impacts even among homogeneous groups, transportation modelers can derive insights from social capital measures to build more socially and behaviorally realistic models.

Divyakant Tahlyan, Northwestern University
Hani Mahmassani, Northwestern University

This study presents a double-hurdle problem perspective to the household vehicle ownership modeling problem, where we argue that the traditionally used econometric models of vehicle ownership assume a single latent equation to express the household vehicle data generation process. However, these models ignore the fact that no vehicle owning state can correspond to two situations: 1) inability to own a vehicle due to financial constraints and 2) either voluntarily giving up owning a vehicle due to attitudinal reasons, even when there is an ability to own one or external constraints like medical condition that makes it impossible to own a vehicle. We propose to use a zero-inflated version of traditionally used ordered probit model to address this issue, which allows the zero states to be generated using two separate latent equations. Using 2017 National Household Travel Survey’s California add-on dataset, we show that the proposed zero-inflated ordered probit model fits the data better than a traditionally estimated ordered probit model and can help us to gain insights about an understudied segment of the population.

Ehsan Rahimi (erahim4@uic.edu), University of Illinois, Chicago
Ramin Shabanpour, University of North Florida
Ali Shamshiripour, University of Illinois, Chicago
Abolfazl Mohammadian, University of Illinois, Chicago

The COVID-19 pandemic has caused our daily routines to change quickly. The pandemic provokes public fear, resulting in changes in what modes of transport people use to perform their daily activities. It is imperative for transportation authorities to properly identify the different degrees of behavioral change among various focused groups. A major factor which can substantially explain individuals’ behavior changes is the individual-level risk perceptions toward using shared mobility solutions. Thus, this study aims for exploring the risk that individuals perceive while using public transit and ridesharing services (as the most widespread form of shared mobility) during the COVID-19 pandemic. To do so, we designed and implemented a multidimensional travel-behavior survey in Chicago which comprises socio-demographic information as well as retrospective questions related to attitudes and travel behavior before and during the pandemic. Utilizing a bivariate ordered probit modeling approach to better account for the correlation between unobserved factors, we simultaneously modeled the perceived risk of exposure to the novel coronavirus in case of riding transit and using ridesharing services. A wide range of factors are found to be influential on being risk-averse of risk-taker users of public transit or ridesharing services, including the socio-demographic, built environment, and the virus spread. Per the results, the mitigation strategies to increase the ridership of shared mobility services not only should be focused on the equality among minority groups but also should be adaptive considering the spatial variations.

Tsai-Yun Liao (tsaiyunliao@gmail.com), National Chiayi University
Yi-Zhou Chen, National Cheng Kung University
Ta-Yin Hu, National Cheng Kung University

With the increase of the percentage of older people, we must focus on the issues generated by older people. As the advanced of vehicle technology, autonomous vehicles are widely accepted to be a future transportation mode and different surveys have been conducted in the past few years to explore user preferences. In addition to vehicle technologies, smartphones have also become necessities for daily life in most of the countries. However, how older people adopt new mixed technologies in the future is still a critical issue. This study aims at exploring user preferences for adopting mixed technologies, by using mobile payment to request autonomous services. This study focuses on the middle-aged and elderly using LINE Pay to pay for the fee of shared autonomous vehicles (SAVs), finding what factors might attract elderly to use these new technologies. A hybrid choice model (HCM) is proposed to explore the factors that affect technology adoption for the elderly and to enhance the elderly adaption of new technologies through policy and strategies. In the HCM framework, a structural equation model (SEM) with the structure of technology acceptance model (TAM) is designed to specify suitable latent variables for the elderly’s new technology acceptance. Based on the results, HCM is constructed to investigate the elderly’s choice decision with adding those latent variables.

Sascha von Behren, Karlsruher Institut fur Technologie
Lisa Bönisch, Karlsruhe Institute of Technology (KIT)
Jan Vallée, Karlsruher Institut fur Technologie
Peter Vortisch, Karlsruhe Institute of Technology (KIT)

In dense urban areas, the pressure for policy makers to act increases due to high car use and the associated congestion and space consumption. In order to make socially fair decisions and take appropriate measures, a detailed understanding of car owners with regular use is of interest. The objective of this research is to provide a more profound differentiation of car owners using their psychographic profiles in three basic dimension of privacy, autonomy and car excitement. These profiles will also be investigated with regard to their mobility behavior in everyday and long distance travel. For the investigation we have collected data in Munich and Berlin (Germany). To segment respondents to latent profiles, we applied a latent class analysis. As a result, we found six different profiles. Besides Car Independents without any strong orientations, we received also classes with high privacy issues towards the public transit. Car Lovers form the largest class (21.2%). They do not currently use their cars very often. However, they have a strong car orientation, which they live out more under changed conditions. We suggest to address the different target groups (profiles) with specific measures to reduce ownership and use. Even if some groups, such as the Convinced Car users, can only be reached via strong restrictions. These restrictions such as road pricing or parking management must be socially fair.

Md. Shahadat Hossain, University of British Columbia
Khademul Haque, CDM Smith Inc
Mahmudur Fatmi, University of British Columbia

Understanding the interaction between in-home/out-of-home activity participation decision is important, particularly at a time when opportunities for out-of-home activities such as shopping, entertainment etc. are limited because of the COVID-19 pandemic. The travel restriction imposed due to the pandemic has made a massive impact on the out-of-home activities, consequently changing the in-home activities as well. This study investigates the in-home and out-of-home activity participation during the COVID-19 pandemic. Data comes from the COVID – 19 Survey for assessing Travel impact (COST). This study utilizes data for the Kelowna region of British Columbia, Canada to develop the following two models: a random parameter multinomial logit (RPMNL) model for out-of-home activity participation and, a hazard-based random parameter duration (HRPD) model for in-home activity participation. The model results suggest that significant interactions exist between out-of-home and in-home activities. For example, higher frequency of out-of-home work travel is more likely to result in longer duration of in-home work activities. Similarly, longer duration of in-home leisure activities might yield a lower likelihood of recreational travel. Health care workers are more likely to engage in work-related travel and less likely to participate in personal and household maintenance activities at home. The model confirms heterogeneity among the individuals. For instance, longer duration of in-home online shopping yields a higher probability for participation in out-of-home shopping activity. This variable shows significant heterogeneity with a large standard deviation, which reveals that sizable variation exist for this variable.