There is considerable interest in understanding the potential induced demand implications of the advent of automated vehicles. In an automated vehicle future, drivers and passengers are relieved of the driving task, thus rendering car travel more convenient and less onerous. As such, there is the possibility that people will undertake more trips in an automated vehicle future, raising the specter of induced demand. Induced demand may also arise from mode shifts, changes in trip lengths, and residential relocations. This study posits that induced demand resulting from the adoption of automated vehicles is inter-related to the adoption modality. Automated vehicles may be purchased and owned personally or used as a mobility-on-demand service (or both). This study aims to shed light on the relationship between automated vehicle adoption modality and likelihood of making additional trips in an automated vehicle future. A joint model of these two outcome variables, wherein automated vehicle adoption modality affects likelihood of making additional trips, is estimated and presented in this paper. The results show that, regardless of the adoption modality, the likelihood of making additional trips increases, with private ownership contributing more to induced demand than a service-based adoption modality. This finding suggests that efforts should be aimed at curbing private ownership of automated vehicles to limit unintended consequences.

This paper aims to quantify the impacts of telecommuting on transit use, with an intent to determine the extent to which transit use drops as telecommuting increases. Data for this analysis is derived from the 2019 and 2023 editions of the Puget Sound Regional Council (PSRC) household travel survey, and a joint model of telecommuting and transit use frequency is estimated that captures the differential influence of variables between the two years. The findings revealed a sideways U-shaped relationship between telecommuting and transit use. Lower transit frequency was observed at both high and low levels of telecommuting, while higher transit frequency was associated with medium levels of telecommuting. This pattern became more pronounced in 2023, with a notable shift effect among non-telecommuters. Based on the treatment effects calculations, it was found that transitioning from medium-level (hybrid) telecommuting to non-telecommuting resulted in a 21 percent decrease in transit use in 2019, which deepened to a 35 percent decrease in 2023. Similarly, moving from hybrid to frequent telecommuting led to a 6 percent reduction in transit use in 2019, increasing to a 9 percent reduction in 2023. These findings suggest that the loss in transit ridership in the post-pandemic era is likely to persist and that compelling workers to return to the workplace full-time is unlikely to yield significant gains unless transit agencies find innovative ways to attract non-telecommuters (full commuters) back to transit. Instead, embracing a hybrid work modality, along with employer-provided incentives to promote transit use, may yield greater benefits.

There are growing concerns about the representativeness of survey data in an era of rapidly emerging and evolving technology, low response rates, and increasingly diverse and heterogeneous populations. Because of the complexities and costs associated with conducting surveys using traditional mail and phone methods, researchers and practitioners are adopting new methods to sample respondents. This paper aims to provide a comprehensive assessment of the representativeness of the samples obtained from three survey sampling strategies utilized in the nationwide COVID Future Panel Survey: convenience sampling, email sampling, and online panel sampling. The three subsamples were statistically different from each other for all socio-economic and demographic variables except race, ethnicity, household size, and gender. However, these differences were ameliorated with the application of weights and the three subsamples converged to census distributions on many variables except educational attainment. Weighting was also able to reduce the differences between the subsamples for a variety of mobility variables except transit use frequency. Modeling the influence of survey sample recruitment strategy on measures of mobility shows that it is significant even after controlling for socio-economic and demographic variables in the model specification. It is likely that the survey sample recruitment strategy variable is accounting for unobserved traits such as attitudes and lifestyle preferences. It is, therefore, recommended to include attitudinal and lifestyle preference questions in transportation surveys so that these traits can be explicitly included in travel model specifications to enhance explanatory power and reduce bias.

The limitations of the current transportation funding system based on federal and state gas taxes have resulted in steep shortfalls in the Highway Trust Fund in the U.S., especially given the growing adoption of high fuel-efficient gasoline vehicles and electric vehicles. While the notion of mileage-based user fees (MBUF) has received attention as an alternative to the current gas tax, public support for MBUF is still low. In this study, we unravel the potential reasons for such low public support by examining the perceived fairness of the MBUF system through the conceptualization of fairness from two perspectives: the perspective that MBUF is fair because everyone pays equally for using the infrastructure, and the perspective that MBUF is less fair because those who adopt cleaner vehicles are not rewarded. Using a bivariate ordered model with attitudinal variables, and employing data from the first wave of the Transportation Heartbeat of America (THA) Survey conducted from October 2024 through January 2025, the effects of socio-economic, demographic, and attitudinal factors on each of the two perspectives of MBUF fairness are examined. The findings suggest substantial heterogeneity in fairness perceptions based on these factors, as well as based on the specific perspective of fairness considered, underscoring the importance of positioning MBUF-based policies with care and sensitivity to different groups of individuals to garner broad support.

Empirical research studies regularly encounter sampling-related challenges that can impact the validity and reliability of model estimation results. This paper presents a comprehensive examination of the implications of nonrandom sampling for estimator consistency and asymptotic efficiency. Through theoretical and simulation-backed support, we underscore the importance of adopting appropriate sampling and estimation methods in two broad scenarios. First, we demonstrate that achieving range variation in exogenous variables, rather than strict population representativeness, is crucial for estimating individual-level causal relationships when sampling is based only on observed exogenous variables. Second, we investigate the efficacy of weighting approaches when sampling is endogenous and use a joint modeling approach to accommodate unobserved self-selection effects where traditional weighting approaches prove inadequate. Our proposed approach accommodates unobserved correlations and successfully recovers true population parameters when the joint distribution of exogenous variables in the population is known. The methodology also shows improved performance compared to existing methods even when only the population marginal distribution of exogenous variables is available. Notably, our simulation experiments extend beyond the conventional linear regression framework to include binary outcomes, providing crucial insights for nonlinear choice modeling applications. The findings underscore the importance of carefully considering sampling mechanisms and their implications for model estimation, while offering practical guidance for researchers facing various sampling-related challenges in empirical studies.

While previous research has focused heavily on understanding the factors deriving alternative fuel vehicle adoption rates, there remains a significant gap in understanding how households distribute mileage across different powertrains. This study utilizes data from the 2022 Next Generation National Household Travel Survey to investigate vehicle miles traveled within a sample of 150 plug-in electric vehicle (PEV)-owning households (in which at least one battery electric vehicle is present), characterizing how different powertrains are integrated into daily mobility. Leveraging a Seemingly Unrelated Regression (SUR) framework the study jointly models the utilization of PEVs, hybrid electric vehicles (HEV), and internal combustion engine vehicles (ICEVs) while accounting for household-level substitution effects. The results provide evidence of an asymmetric substitution effect. In households with mixed-powertrain configurations, the ICEV captures a substantially higher share of household miles (compared with the PEV), acting as a utility sponge. Conversely, the model identifies specific socioeconomic and geographic cohorts that prioritize PEV as the primary household workhorse, indicating a systematic sorting effect. Although the sample size limits broader generalizability, these findings suggest that PEVs are used for frequent, specific routine-intensive roles, whereas the ICEV remains a specialized utility vehicle. These insights highlight distinct intrahousehold vehicle use behaviors that are often obscured by aggregate fleetwide statistics.

Neural network-based discrete choice models (neural DCMs) have demonstrated consistent gains in predictive accuracy over classical econometric specifications due to their flexible functional form. Yet, the reliability of their behavioral outputs for statistical inference has received little scrutiny. This paper argues that inference variance in neural DCMs arises from two conceptually distinct sources: sampling uncertainty from finite sample variability, and optimization uncertainty arising from non-convex loss surfaces and over-identification, where different initializations of neural weights converge to different utility function shapes. This decomposition is critical because the two sources have different remedies: sampling uncertainty shrinks with more data, but optimization uncertainty does not and instead threatens the reproducibility of behavioral inference. To separately quantify these two uncertainties, we apply a bootstrap decomposition framework across different model specifications under two Monte Carlo simulation studies of increasing complexity. The results confirm four hypotheses motivated by the structural properties of these architectures. First, Multinomial Logit (MNL) exhibits confident misspecification, with narrow uncertainty bands that conceal large systematic bias. Second, DCMs based on unconstrained neural networks suffer from substantial optimization uncertainty that renders single-run inferences unreliable. Third, regularization in the Bayesian Neural Network (BNN) reduces optimization uncertainty in simpler settings, but posterior uncertainty becomes increasingly miscalibrated as preference heterogeneity compounds; yet, wide credible bands fail to cover the bias induced by prior shrinkage. Fourth, the Monotonic Neural Network (MNN) achieves the best bias-variance trade-off across both studies, as behavioral constraints simultaneously enforce interpretability and stabilize the optimization landscape, yielding a more favorable decomposition than either classical econometric models or unconstrained neural networks. These findings directly address the need for rigorous uncertainty quantification in neural DCMs, where behavioral outputs such as willingness-to-pay and elasticities inform consequential policy decisions yet are routinely reported without accounting for optimization-induced variability.

Existing methodology on food accessibility predominately focuses on on-premise services, that is, dine-in and shopping at stores, which assumes a linear distance decay property (the closer, the higher accessibility). Access to delivery services is fundamentally different from that to on-premise stores. Stores with close proximity (within an inner boundary) are less desirable for delivery due to delivery fees, and there is an outer boundary beyond which deliveries are unavailable, both challenging the assumption of increasing impediment with distance. These two boundaries form a donut shape for delivery services. We propose a modified 2-step floating catchment area method that incorporates the donut shape, accounts for both demand and supply, and examines the diversity of food options. Using Seattle as a case study, our results show that delivery services increase restaurant and fast-food accessibility in areas where there is already good accessibility (e.g., downtown Seattle for restaurants and South Seattle for fast-food). Given South Seattle is where low-income and low-access households concentrate, the increase in accessibility to fast-food may not be desired. Interestingly, with delivery services, more low-income or low-access households (those who live far from grocery stores) have better accessibility to fresh produce from grocery stores compared to the rest of the population. And the newly created Supplemental Nutrition Assistance Program (SNAP) online program appears to miss low-access households. These findings have important implications for policymakers and stakeholders seeking to improve food accessibility in urban areas through delivery services.

Transportation planning and forecasting have long been essential in guiding investment and policy decisions and shaping transportation systems. However, rapid technological advancements, shifting societal values, and the lingering effects of the COVID-19 pandemic have challenged the relevance and accuracy of traditional forecasting models. This paper examines the evolving context of travel behavior, highlighting the limitations of existing models in capturing new trends such as telework, e-commerce, and induced demand. It advocates for the development of more dynamic, data-driven, and context-sensitive modeling approaches that can better accommodate these changes. The paper also discusses the need for planners and forecasters to engage with stakeholders to ensure that models remain effective tools for decision-making in an increasingly complex and uncertain world.

Introduction Walking is an important physical activity with significant health benefits. Despite the presence of an extensive body of research dedicated to understanding various aspects of walking, there is a need for a more holistic and comprehensive understanding of walking behaviors. With many countries facing an increasingly aging population, this issue is of particular importance for older adults for whom walking can provide significant physical and mental health benefits. This paper studies three key walking behaviors: walking frequency, purpose, and place/location. Methods The study utilizes data from the 2022 American Association of Retired Persons (AARP) walking survey and employs a multivariate ordered probit (MORP) approach to jointly model the three dimensions of walking. This method allows capturing exogenous variable effects and endogenous variable effects, while controlling for error correlations arising from the presence of unobserved traits that simultaneously affect multiple outcome variables. The survey also provides valuable data to examine other dimensions of walking behavior in older adults in a post-COVID environment, including duration, companionship, and perceived benefits and barriers. Results The findings indicate plenty of scope for enhancing purpose-driven walking through the provision of walk-friendly environments in and around residential neighborhoods. Significant socioeconomic disparities also hinder certain population segments from engaging in walking activities, particularly in their residential neighborhoods. Conclusions The multidimensional dataset and findings obtained from this study offer a rich resource for future research, and for informing the design of urban planning and public health interventions, to promote walking and enhance quality of life among older adults.

Modern transportation network modeling increasingly involves the integration of diverse methodologies including sensor-based forecasting, reinforcement learning, classical flow optimization, and demand modeling that have traditionally been developed in isolation. This paper introduces Flow Through Tensors (FTT), a unified computational graph architecture that connects origin destination flows, path probabilities, and link travel times as interconnected tensors. Our framework makes three key contributions: first, it establishes a consistent mathematical structure that enables gradient-based optimization across previously separate modeling elements; second, it supports multidimensional analysis of traffic patterns over time, space, and user groups with precise quantification of system efficiency; third, it implements tensor decomposition techniques that maintain computational tractability for large scale applications. These innovations collectively enable real time control strategies, efficient coordination between multiple transportation modes and operators, and rigorous enforcement of physical network constraints. The FTT framework bridges the gap between theoretical transportation models and practical deployment needs, providing a foundation for next generation integrated mobility systems.

There is growing interest in understanding the interaction between weather and transportation and the ability of communities and the nation’s infrastructure to withstand extreme conditions and events. This study aims to provide detailed insights on how people adjust and change their activity-travel and time use behaviors in the face of extreme heat conditions.By leveraging time use records integrated with weather data, the study compares activity-mobility patterns between extreme heat days and non-extreme days. A series of models are estimated to understand the impact of extreme heat even after controlling for other variables. The findings reveal that heat significantly impacts time use and activity-mobility patterns, with some groups exhibiting potentially greater vulnerability arising from the inability to adapt sufficiently to extreme heat. Designing dense, shaded urban environments, declaring heat days to facilitate indoor stays, and providing transportation vouchers for vulnerable populations can help mitigate the ill-effects of extreme heat.

The growing behaviors of work-from-home (WFH) and online shopping hold significant potential for reducing traffic congestion and emissions. Understanding the frequency and the interplay between these two behaviors is important for successful implementation. This study investigates the recent trends of WFH and online shopping and the underlying factors influencing individuals’ decisions on these two behaviors. Focusing on non-grocery online shopping, this study uses comprehensive survey data collected across the United States during October and November 2021. We develop a Generalized Structural Equation Model to jointly examine WFH and online shopping frequency and their interaction. Moreover, the study investigates the psychological aspects of WFH and online shopping, introducing four stochastic latent constructs—WFH comfort, WFH unproductiveness, online shopping enjoyment, and online shopping inconvenience using the attitudinal variables. Results indicate a positive causal relationship, suggesting that increased WFH promotes online shopping engagement. Perceived comfort and productivity at home affect WFH frequency shaped by factors like home workspace, commuting time, childcare responsibilities, and telecommunications with co-workers. Likewise, perceived convenience and enjoyment significantly affect online shopping, influenced by aspects such as timesaving, and the delivery and return process. Technological tools at home also play a role in WFH frequency. Demographic factors like age, race, income, physical disability, and mode choice habits correlate with WFH and online shopping incidence, while job category and employer flexibility influence WFH frequency. These insights can help policymakers to regulate remote work and online shopping activities as they continue to grow.

The COVID-19 pandemic is an unprecedented global crisis that has impacted virtually everyone. We conducted a nationwide online longitudinal survey in the United States to collect information about the shifts in travel-related behavior and attitudes before, during, and after the pandemic. The survey asked questions about commuting, long distance travel, working from home, online learning, online shopping, pandemic experiences, attitudes, and demographic information. The survey has been deployed to the same respondents thrice to observe how the responses to the pandemic have evolved over time. The first wave of the survey was conducted from April 2020 to June 2021, the second wave from November 2020 to August 2021, and the third wave from October 2021 to November 2021. In total, 9,265 responses were collected in the first wave; of these, 2,877 respondents returned for the second wave and 2,728 for the third wave. Survey data are publicly available. This unique dataset can aid policy makers in making decisions in areas including transport, workforce development, and more. This article demonstrates the framework for conducting this online longitudinal survey. It details the step-by-step procedure involved in conducting the survey and in curating the data to make it representative of the national trends.

How we spend our time and in what activities we participate vary from individual to individual. These individual decisions are characterized by the nature of our activity desires, and the where, with whom, when, and duration dimensions of desired participations. Taken together, these choices determine our activity involvements and movements over space and time. At the same time, our activity-travel environment itself is changing rapidly, thanks to the advent of new technologies that fundamentally alter the nature of our space-time interactions. Within this context, the current chapter will capture ongoing interdisciplinary directions of scholarly enquiry in the following directions: (1) time-use and activity-based approaches to travel demand analysis, (2) emerging modeling methods for the analysis of human activities, time-use, and travel behavior, (3) relevant developments in data collection methods, and (4) trends in time-use, activity, and travel behavior due to emerging travel technologies, access to information and communication, and unprecedented circumstances posed by the recent pandemic.

A sustainable transportation future is one in which people eschew personal car ownership in favor of using autonomous vehicle (AV)-based ridehailing services in a shared mode. However, the traveling public has historically shown a disinclination toward sharing rides and carpooling with strangers. In a future of AV-based ridehailing services, it will be necessary for people to embrace both AVs as well as true ridesharing to fully realize the benefits of automated and shared mobility technologies. This study investigated the factors influencing willingness to use AV-based ridehailing services in the future in a shared mode (i.e., with strangers). This was done through the estimation of a behavioral model system on a comprehensive survey data set that included rich information about attitudes, perceptions, and preferences pertaining to the adoption of AVs and shared mobility modes. The model results showed that current ridehailing experiences strongly influenced the likelihood of being willing to ride AV-based services in a shared mode. Campaigns that provide opportunities for individuals to experience such services firsthand would potentially go a long way to enabling a shared mobility future at scale. In addition, several attitudinal variables were found to strongly influence the adoption of future mobility services; these findings provide insights on the likely early adopters of shared autonomous mobility services and the types of educational awareness campaigns that may effect change in the prospects of such services.

This paper presents an examination of the interrelationship between household vehicle ownership and ridehailing use frequency. Both variables constitute important mobility choices with significant implications for the future of transport. Although it is generally known that these two behavioral phenomena are inversely related to one another, the direction of causality is rather ambiguous. Do vehicle ownership levels affect ridehailing use frequency, or does the adoption and use of ridehailing services affect vehicle ownership? If ridehailing services affect vehicle ownership, then it is plausible that a future of mobility-as-a-service would be characterized by lower levels of vehicle ownership. To explore the degree to which these causal relationships are prevalent in the population, a joint latent segmentation model system was formulated and estimated on a survey data set collected in four automobile-oriented metropolitan areas of the United States. The latent segmentation model system recognized that the causal structures driving the mobility choices of individuals were not directly observable. Model estimation results showed that 58% of the survey sample followed the causal structure in which ridehailing use frequency affected vehicle ownership. This finding suggests that there is considerable structural heterogeneity in the population with respect to causal structures and that ridehailing use does indeed hold considerable promise to effect changes in private vehicle ownership in the future.

With work arrangements experiencing dramatic changes over the past three years due to the COVID-19 pandemic, and the possibility that altered work arrangements may persist well into the future, the implications of teleworking on activity-travel behavior are potentially profound. This paper aims to substantially add to the body of knowledge about the present and future of telework in the wake of the pandemic through a rigorous analysis of telework arrangements between two distinct time periods. The paper focuses on three key aspects of telework, including whether to telework or not, frequency of telework, and location of telework. Behavioral data for this study is derived from a workplace location choice survey conducted across Texas in February-March 2022, which included a recall component to obtain workplace location choice information in the pre-pandemic period. The evolution of telework arrangements between the pre-and after-pandemic periods is explored through a joint model system estimated using a joint multivariate methodology. Results show that, After COVID, the population of workers is generally inclined toward a hybrid work arrangement, with an overall tendency to engage in a higher frequency of teleworking than Before COVID. Finally, teleworkers have a higher propensity to work only from home as opposed to working only from a third workplace or from a combination of home and a third workplace. Overall, our results indicate that telework arrangements may remain at an elevated level into the future, with home serving as the dominant telework location. These findings suggest that transportation demand forecasting models need to be updated to reflect higher levels of teleworking, as well as the heterogeneity across individuals in teleworking adoption, frequency, and location.

As transportation systems grow in complexity, analysts need sophisticated tools to understand travelers’ decision-making and effectively quantify the benefits of the proposed strategies. The transportation community has developed integrated demand–supply models to capture the emerging interactive nature of transportation systems, serve diverse planning needs, and encompass broader solution possibilities. Recently, utilizing advances in Machine Learning (ML) techniques, researchers have also recognized the need for different computational models capable of fusing/analyzing different data sources. Inspired by this momentum, this study proposes a new modeling framework to analytically bridge travel demand components and network assignment models with machine learning algorithms. Specifically, to establish a consistent representation of such aspects between separate system models, we introduce several important mathematical programming reformulation techniques—variable splitting and augmented Lagrangian relaxation—to construct a computationally tractable nonlinear unconstrained optimization program. Furthermore, to find equilibrium states, we apply automatic differentiation (AD) to compute the gradients of decision variables in a layered structure with the proposed model represented based on computational graphs (CGs) and solve the proposed formulation through the alternating direction method of multipliers (ADMM) as a dual decomposition method. Thus, this reformulated model offers a theoretically consistent framework to express the gap between the demand and supply components and lays the computational foundation for utilizing a new generation of numerically reliable optimization solvers. Using a small example network and the Chicago sketch transportation network, we examined the convergency/consistency measures of this new differentiable programming-based optimization structure and demonstrated the computational efficiency of the proposed integrated transportation demand and supply models.