Pub Date : 2026-02-02DOI: 10.1016/j.tbs.2026.101245
Yingnan Yan, Tianming Liu, Yafeng Yin
As a key advancement in artificial intelligence, large language models (LLMs) are set to transform transportation systems. While LLMs offer the potential to simulate human travelers in future mixed-autonomy transportation systems, their behavioral fidelity in complex scenarios remains largely unconfirmed by existing research. This study addresses this gap by conducting a comprehensive analysis of the value of travel time (VOT) of three popular LLMs. We employ a full factorial experimental design to systematically examine LLMs’ sensitivities to various transportation contexts, including the choice setting, travel purpose, and socio-demographic factors. Our results reveal a high degree of behavioral similarity between LLMs and humans. Some LLMs exhibit an aggregate VOT similar to that of humans, and all tested models demonstrate human-like sensitivity to travel purpose, income, and the time-cost trade-off ratios of the alternatives. Furthermore, the behavioral patterns of LLMs are highly consistent across varied contexts. However, while the behavior of every single model is highly robust, we also find some heterogeneity across models regarding the magnitude and direction of sensitivity to travel contexts and income elasticity. Overall, this study provides a foundational benchmark for the future development of LLMs as proxies for human travelers, demonstrating their robust decision-making capabilities while cautioning that misaligned magnitudes of economic trade-offs between humans and LLMs necessitate rigorous validation and additional conditioning of LLMs before their application.
{"title":"Valuing time in silicon: Can large language models replicate human value of travel time","authors":"Yingnan Yan, Tianming Liu, Yafeng Yin","doi":"10.1016/j.tbs.2026.101245","DOIUrl":"10.1016/j.tbs.2026.101245","url":null,"abstract":"<div><div>As a key advancement in artificial intelligence, large language models (LLMs) are set to transform transportation systems. While LLMs offer the potential to simulate human travelers in future mixed-autonomy transportation systems, their behavioral fidelity in complex scenarios remains largely unconfirmed by existing research. This study addresses this gap by conducting a comprehensive analysis of the value of travel time (VOT) of three popular LLMs. We employ a full factorial experimental design to systematically examine LLMs’ sensitivities to various transportation contexts, including the choice setting, travel purpose, and socio-demographic factors. Our results reveal a high degree of behavioral similarity between LLMs and humans. Some LLMs exhibit an aggregate VOT similar to that of humans, and all tested models demonstrate human-like sensitivity to travel purpose, income, and the time-cost trade-off ratios of the alternatives. Furthermore, the behavioral patterns of LLMs are highly consistent across varied contexts. However, while the behavior of every single model is highly robust, we also find some heterogeneity across models regarding the magnitude and direction of sensitivity to travel contexts and income elasticity. Overall, this study provides a foundational benchmark for the future development of LLMs as proxies for human travelers, demonstrating their robust decision-making capabilities while cautioning that misaligned magnitudes of economic trade-offs between humans and LLMs necessitate rigorous validation and additional conditioning of LLMs before their application.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101245"},"PeriodicalIF":5.7,"publicationDate":"2026-02-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146109861","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Developing gender sensitive transit infrastructure is essential as there exists difference in how commuters of different gender experience the transit system. This research highlights the differences in commuter experience based on socio-demographic and travel characteristics of commuters across gender, while using regional transit system of the National Capital Region (NCR), India. The perception of 1419 male and 792 female commuters of regional transit system was analysed by integrating Relative to an Identified Distribution Integral Transformation (RIDIT) analysis with Importance-Satisfaction Analysis (ISA) to reveal significant differences in their perception. Male commuters prioritized time, whereas cost was most critical for females. Last-mile connectivity along with the need to improve toilet facilities and cleanliness inside vehicle was crucial for both the commuters.
The findings of Spearman’s rank correlation analyses showcased heterogeneity among commuters of specific gender with different socio-demographic and travel characteristics, suggesting the need for targeted interventions. The approach can be used to formulate gender-based strategies for improving the service quality of regional transit systems. Developing a gender responsive transit system can assist planners and decision-making authorities in addressing the existing gaps and ensure an inclusive and efficient transit experience for all commuters.
{"title":"Unveiling gender dynamics in perception of commuters towards regional transit","authors":"Aditya Manish Pitale , Shubhajit Sadhukhan , Manoranjan Parida","doi":"10.1016/j.tbs.2026.101255","DOIUrl":"10.1016/j.tbs.2026.101255","url":null,"abstract":"<div><div>Developing gender sensitive transit infrastructure is essential as there exists difference in how commuters of different gender experience the transit system. This research highlights the differences in commuter experience based on socio-demographic and travel characteristics of commuters across gender, while using regional transit system of the National Capital Region (NCR), India. The perception of 1419 male and 792 female commuters of regional transit system was analysed by integrating Relative to an Identified Distribution Integral Transformation (RIDIT) analysis with Importance-Satisfaction Analysis (ISA) to reveal significant differences in their perception. Male commuters prioritized time, whereas cost was most critical for females. Last-mile connectivity along with the need to improve toilet facilities and cleanliness inside vehicle was crucial for both the commuters.</div><div>The findings of Spearman’s rank correlation analyses showcased heterogeneity among commuters of specific gender with different socio-demographic and travel characteristics, suggesting the need for targeted interventions. The approach can be used to formulate gender-based strategies for improving the service quality of regional transit systems. Developing a gender responsive transit system can assist planners and decision-making authorities in addressing the existing gaps and ensure an inclusive and efficient transit experience for all commuters.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101255"},"PeriodicalIF":5.7,"publicationDate":"2026-02-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146111043","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-02DOI: 10.1016/j.tbs.2026.101240
Sajjad Karimi , Robert Kluger , Abolfazl Karimpour
E-scooters have emerged as a popular means of commuting, specifically for short trips, within urban environments. Ensuring the availability of e-scooters within a shared fleet system is crucial for providing convenient services. However, increasing demand and service areas make it challenging to keep e-scooters available based on when and where users want to use them. This study aims to investigate the factors affecting e-scooter availability by specifically studying their idle patterns. Idle time refers to the duration between a vehicle becoming available for pick-up and being rented. Survival analysis was employed to analyze and visualize the probability of e-scooters remaining idle in a shared network in Louisville, Kentucky. Idle time was found to be influenced by several factors, including time of day, season, pick-up location, and operator. Fall and winter drop-offs have longer idle times despite fewer e-scooters indicating a large drop-off in demand. Morning drop-offs have longer idle times than other times of day. The survival analysis identifies optimal pick-up windows that operators can use to guide rebalancing, stage replacements, plan maintenance and charging, and offer incentives, customizing operations and policies to improve service reliability and efficiency.
{"title":"Investigating e-scooters idle time patterns: A survival analysis approach to understand availability","authors":"Sajjad Karimi , Robert Kluger , Abolfazl Karimpour","doi":"10.1016/j.tbs.2026.101240","DOIUrl":"10.1016/j.tbs.2026.101240","url":null,"abstract":"<div><div>E-scooters have emerged as a popular means of commuting, specifically for short trips, within urban environments. Ensuring the availability of e-scooters within a shared fleet system is crucial for providing convenient services. However, increasing demand and service areas make it challenging to keep e-scooters available based on when and where users want to use them. This study aims to investigate the factors affecting e-scooter availability by specifically studying their idle patterns. Idle time refers to the duration between a vehicle becoming available for pick-up and being rented. Survival analysis was employed to analyze and visualize the probability of e-scooters remaining idle in a shared network in Louisville, Kentucky. Idle time was found to be influenced by several factors, including time of day, season, pick-up location, and operator. Fall and winter drop-offs have longer idle times despite fewer e-scooters indicating a large drop-off in demand. Morning drop-offs have longer idle times than other times of day. The survival analysis identifies optimal pick-up windows that operators can use to guide rebalancing, stage replacements, plan maintenance and charging, and offer incentives, customizing operations and policies to improve service reliability and efficiency.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101240"},"PeriodicalIF":5.7,"publicationDate":"2026-02-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146109863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-31DOI: 10.1016/j.tbs.2026.101233
Seung Eun Choi , Ilsu Kim , Xinyi Wang , Jason Soria , Patricia L. Mokhtarian
Teleworking, significantly accelerated by the COVID-19 pandemic, has altered traditional work arrangements and commuting habits. To understand factors influencing telework behavior and their changes over time, we examine telework adoption trends using three waves of cross-sectional survey data collected in the springs of 2021–2023 (and including retrospective pre-COVID data) in the Dallas–Fort Worth–Arlington (DFA) and Washington (DC)–Arlington–Alexandria (WAA) regions. We categorize employees into non-teleworkers (NTWers), non-usual teleworkers (NUTWers), and usual teleworkers (UTWers) based on their frequency of full days working from home. Region- and year-specific multinomial logit (MNL) models are constructed to identify factors influencing telework patterns across different pandemic phases. Four groups of influential factors are revealed: those consistent in significance throughout (e.g., education and income), those fading across years (e.g., car-related attitudes), those strengthening over time (e.g., telework attitudes), and those with differential impacts between NUTW and UTW (e.g., household composition). For instance, in WAA, workers with young children were more likely than others to be NUTWers pre- and early-pandemic, but this factor faded in significance in later years. Conversely, the impact of having school-aged children on NUTWing strengthened in the late-pandemic period. Additionally, pro-car ownership attitudes decreased the propensity to UTW throughout the years, though this influence faded late in the pandemic. These findings highlight the dynamic interplay between work-family balance, job characteristics, and attitudes related to telework adoption. The results provide valuable insights for policymakers and employers aiming to optimize flexible work arrangements, refine telework policies, and address future workforce needs.
{"title":"How has the importance of factors influencing telework adoption changed over time? Observing pre- to late-pandemic trends using multi-year data from two US regions","authors":"Seung Eun Choi , Ilsu Kim , Xinyi Wang , Jason Soria , Patricia L. Mokhtarian","doi":"10.1016/j.tbs.2026.101233","DOIUrl":"10.1016/j.tbs.2026.101233","url":null,"abstract":"<div><div>Teleworking, significantly accelerated by the COVID-19 pandemic, has altered traditional work arrangements and commuting habits. To understand factors influencing telework behavior and their changes over time, we examine telework adoption trends using three waves of cross-sectional survey data collected in the springs of 2021–2023 (and including retrospective pre-COVID data) in the Dallas–Fort Worth–Arlington (DFA) and Washington (DC)–Arlington–Alexandria (WAA) regions. We categorize employees into non-teleworkers (NTWers), non-usual teleworkers (NUTWers), and usual teleworkers (UTWers) based on their frequency of full days working from home. Region- and year-specific multinomial logit (MNL) models are constructed to identify factors influencing telework patterns across different pandemic phases. Four groups of influential factors are revealed: those <em>consistent</em> in significance throughout (e.g., education and income), those <em>fading</em> across years (e.g., car-related attitudes), those <em>strengthening</em> over time (e.g., telework attitudes), and those with <em>differential impacts</em> between NUTW and UTW (e.g., household composition). For instance, in WAA, workers with <em>young</em> children were more likely than others to be NUTWers pre- and early-pandemic, but this factor <em>faded</em> in significance in later years. Conversely, the impact of having <em>school-aged</em> children on NUTWing <em>strengthened</em> in the late-pandemic period. Additionally, pro-car ownership attitudes decreased the propensity to UTW throughout the years, though this influence <em>faded</em> late in the pandemic. These findings highlight the dynamic interplay between work-family balance, job characteristics, and attitudes related to telework adoption. The results provide valuable insights for policymakers and employers aiming to optimize flexible work arrangements, refine telework policies, and address future workforce needs.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101233"},"PeriodicalIF":5.7,"publicationDate":"2026-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080726","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-30DOI: 10.1016/j.tbs.2026.101242
Jiahe Bian , Wei Li , Xiao Li , Sun Quan , Andong Chen , Sinan Zhong , Muhammad Usman , Samuel Dominic Castiglione Towne Jr. , Muyang Li , Bahar Dadashova , Xinyue Ye , Marcia G. Ory
Adults with physical impairments or disabilities that prevent physical activity (PID-PA) face significant transportation barriers to essential healthcare, often forgoing care despite higher healthcare needs. While on-demand ride-sourcing services (e.g., Uber and Lyft) may improve mobility, concerns remain about the current level of inclusivity and equity, especially for individuals with more complex needs. Whether on-demand ride-sourcing will facilitate mobility or further isolate certain people with PID-PA is largely unknown. This study examined the transportation barriers to healthcare among people with temporary and chronic PID-PA and assessed the role of alternative access strategies, with particular attention to small and rural communities where residences are dispersed and transit options are limited. A cross-sectional online survey was conducted in nine such communities in Texas, yielding 416 valid responses for analysis. Fisher’s exact tests, logistic regression models, and mediation analysis were used to assess associations between adults with PID-PA and variables such as forgone necessary healthcare due to lack of transportation, use of on-demand ride-sourcing, and alternative transportation options.
Among participants, people with PID-PA were more likely to use rides provided by others and telemedicine. However, logistic regression models showed that having chronic PID-PA and using on-demand ride-sourcing for healthcare were positively associated with forgone necessary medical care due to transportation barriers. Moreover, on-demand ride-sourcing use did not mediate the relationship between chronic PID-PA and forgone necessary healthcare. This result indicates that ride-sourcing services do not effectively reduce transportation barriers for individuals with chronic PID-PA. Instead, dependence on such services may be associated with forgoing necessary medical care. The study highlights substantial challenges to using on-demand ride-sourcing in small and rural communities, including limited physical/digital accessibility, affordability concerns, and unreliable service. To improve transportation equity for people with PID-PA, interventions must address broader systemic issues affecting the accessibility of ride-sourcing.
{"title":"Left behind: forgone medical care due to transportation barriers among adults with physical impairments and disabilities that prevent physical activity in small and rural communities","authors":"Jiahe Bian , Wei Li , Xiao Li , Sun Quan , Andong Chen , Sinan Zhong , Muhammad Usman , Samuel Dominic Castiglione Towne Jr. , Muyang Li , Bahar Dadashova , Xinyue Ye , Marcia G. Ory","doi":"10.1016/j.tbs.2026.101242","DOIUrl":"10.1016/j.tbs.2026.101242","url":null,"abstract":"<div><div>Adults with physical impairments or disabilities that prevent physical activity (PID-PA) face significant transportation barriers to essential healthcare, often forgoing care despite higher healthcare needs. While on-demand ride-sourcing services (e.g., Uber and Lyft) may improve mobility, concerns remain about the current level of inclusivity and equity, especially for individuals with more complex needs. Whether on-demand ride-sourcing will facilitate mobility or further isolate certain people with PID-PA is largely unknown. This study examined the transportation barriers to healthcare among people with temporary and chronic PID-PA and assessed the role of alternative access strategies, with particular attention to small and rural communities where residences are dispersed and transit options are limited. A cross-sectional online survey was conducted in nine such communities in Texas, yielding 416 valid responses for analysis. Fisher’s exact tests, logistic regression models, and mediation analysis were used to assess associations between adults with PID-PA and variables such as forgone necessary healthcare due to lack of transportation, use of on-demand ride-sourcing, and alternative transportation options.</div><div>Among participants, people with PID-PA were more likely to use rides provided by others and telemedicine. However, logistic regression models showed that having chronic PID-PA and using on-demand ride-sourcing for healthcare were positively associated with forgone necessary medical care due to transportation barriers. Moreover, on-demand ride-sourcing use did not mediate the relationship between chronic PID-PA and forgone necessary healthcare. This result indicates that ride-sourcing services do not effectively reduce transportation barriers for individuals with chronic PID-PA. Instead, dependence on such services may be associated with forgoing necessary medical care. The study highlights substantial challenges to using on-demand ride-sourcing in small and rural communities, including limited physical/digital accessibility, affordability concerns, and unreliable service. To improve transportation equity for people with PID-PA, interventions must address broader systemic issues affecting the accessibility of ride-sourcing.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101242"},"PeriodicalIF":5.7,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146080725","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-29DOI: 10.1016/j.tbs.2026.101251
Woojae Kim , Youngsang Cho , Taeho Park , Kyuho Maeng
As autonomous vehicles (AVs) are increasingly integrated into everyday mobility systems, ethically complex crash scenarios have become a critical issue. Therefore, trolley dilemmas have attracted significant attention. However, little is known about how moral programming influences consumers’ acceptance of an AV. This study investigated the relative impact of ethical decision logic, accident liability, and safety performance on AV adoption preferences. A stated-choice experiment was conducted with 1,032 Korean respondents, and a mixed logit model with the Bayesian estimation method was used to estimate heterogeneous utility parameters. The experiment included five attributes: whether the AV protects the driver or pedestrian, the party responsible for the accident, annual accident probability, algorithm personalization, and purchase price. Demographic characteristics were also examined. The results indicated that the attribute “whether an AV protects drivers or pedestrians” had no significant effect on consumer utility. By contrast, a lower accident probability and assigning responsibility to manufacturers or software developers rather than to drivers substantially increased AV acceptance. Male, urban, and lower-income respondents were more likely to prefer AVs that protect drivers and shift the liability toward institutional actors. These findings suggest that consumers prioritize measurable safety and institutional accountability over abstract ethical logic. For AV developers and policymakers, these results highlight the value of adaptive algorithmic frameworks and clearly defined liability structures. This study contributes to the design of socially acceptable AV systems that align with public expectations in the age of algorithmic decision-making.
{"title":"Hands off the wheel, hands off the choice? A discrete choice experiment on trolley dilemma in autonomous vehicles","authors":"Woojae Kim , Youngsang Cho , Taeho Park , Kyuho Maeng","doi":"10.1016/j.tbs.2026.101251","DOIUrl":"10.1016/j.tbs.2026.101251","url":null,"abstract":"<div><div>As autonomous vehicles (AVs) are increasingly integrated into everyday mobility systems, ethically complex crash scenarios have become a critical issue. Therefore, trolley dilemmas have attracted significant attention. However, little is known about how moral programming influences consumers’ acceptance of an AV. This study investigated the relative impact of ethical decision logic, accident liability, and safety performance on AV adoption preferences. A stated-choice experiment was conducted with 1,032 Korean respondents, and a mixed logit model with the Bayesian estimation method was used to estimate heterogeneous utility parameters. The experiment included five attributes: whether the AV protects the driver or pedestrian, the party responsible for the accident, annual accident probability, algorithm personalization, and purchase price. Demographic characteristics were also examined. The results indicated that the attribute “whether an AV protects drivers or pedestrians” had no significant effect on consumer utility. By contrast, a lower accident probability and assigning responsibility to manufacturers or software developers rather than to drivers substantially increased AV acceptance. Male, urban, and lower-income respondents were more likely to prefer AVs that protect drivers and shift the liability toward institutional actors. These findings suggest that consumers prioritize measurable safety and institutional accountability over abstract ethical logic. For AV developers and policymakers, these results highlight the value of adaptive algorithmic frameworks and clearly defined liability structures. This study contributes to the design of socially acceptable AV systems that align with public expectations in the age of algorithmic decision-making.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101251"},"PeriodicalIF":5.7,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072159","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-28DOI: 10.1016/j.tbs.2026.101252
Yue Liu, Guohua Liang, Ziyu Chen, Zhixiang Gao
Travel behavior modeling is essential for transportation demand analysis and policy-making, yet traditional discrete choice models often struggle with real-world data complexities, such as heavy-tailed distributions and strong feature correlations. This study proposes a novel neural network framework integrated with advanced statistical techniques to effectively address these issues. Specifically, a ParetoTail transformation is employed to normalize heavy-tailed travel attributes, such as travel time and cost, reducing the undue influence of extreme values. To explicitly capture complex dependencies among features, a Gaussian copula approach is integrated, improving the robustness of the model against traditional independence assumptions. Furthermore, a gating mechanism is introduced to dynamically balance the contributions of continuous and discrete features, incorporating random noise to account for preference heterogeneity across individual travelers. Extensive empirical analyses, initially on the Swissmetro dataset and validated in three additional diverse public datasets, demonstrate that the proposed model consistently and significantly outperforms the baseline models (MNL, MXL, L-MNL, E-MNL, EL-MNL) in terms of prediction accuracy, F1 score, and AUC values. Crucially, the interpretability of the model reveals nuanced behavioral insights, such as the heterogeneity of decision-making styles across the population and non-linear responses to cost in long-distance travel. Additional ablation studies underscore the essential roles of the ParetoTail, Gaussian copula, and gating components. In general, this integrated framework provides a flexible, robust, and generalizable approach to modeling travel behavior, offering transport planners a more accurate tool for policy evaluation in complex real-world scenarios.
{"title":"Neural integrated choice model with ParetoTail and Gaussian copula for travel behavior analysis","authors":"Yue Liu, Guohua Liang, Ziyu Chen, Zhixiang Gao","doi":"10.1016/j.tbs.2026.101252","DOIUrl":"10.1016/j.tbs.2026.101252","url":null,"abstract":"<div><div>Travel behavior modeling is essential for transportation demand analysis and policy-making, yet traditional discrete choice models often struggle with real-world data complexities, such as heavy-tailed distributions and strong feature correlations. This study proposes a novel neural network framework integrated with advanced statistical techniques to effectively address these issues. Specifically, a ParetoTail transformation is employed to normalize heavy-tailed travel attributes, such as travel time and cost, reducing the undue influence of extreme values. To explicitly capture complex dependencies among features, a Gaussian copula approach is integrated, improving the robustness of the model against traditional independence assumptions. Furthermore, a gating mechanism is introduced to dynamically balance the contributions of continuous and discrete features, incorporating random noise to account for preference heterogeneity across individual travelers. Extensive empirical analyses, initially on the Swissmetro dataset and validated in three additional diverse public datasets, demonstrate that the proposed model consistently and significantly outperforms the baseline models (MNL, MXL, L-MNL, E-MNL, EL-MNL) in terms of prediction accuracy, F1 score, and AUC values. Crucially, the interpretability of the model reveals nuanced behavioral insights, such as the heterogeneity of decision-making styles across the population and non-linear responses to cost in long-distance travel. Additional ablation studies underscore the essential roles of the ParetoTail, Gaussian copula, and gating components. In general, this integrated framework provides a flexible, robust, and generalizable approach to modeling travel behavior, offering transport planners a more accurate tool for policy evaluation in complex real-world scenarios.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101252"},"PeriodicalIF":5.7,"publicationDate":"2026-01-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072166","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-26DOI: 10.1016/j.tbs.2026.101243
César A. Merchán-Núñez , Mauricio Orozco-Fontalvo , Luisa F. Morales-Moreno , Aquiles Darghan , Sonia C. Mangones M. , Lenin A. Bulla-Cruz
Carpooling is a sustainable transportation alternative that allows users with similar destinations to share private vehicles, contributing to reductions in energy consumption, pollutant emissions, and traffic congestion. While not a comprehensive solution, carpooling can lower private car use and increase vehicle occupancy rates. However, most carpooling initiatives have been limited in scope, often operating on a small scale or within corporate frameworks, restricting their potential for widespread adoption. In Latin America, where ride-hailing services are popular despite regulatory issues, carpooling remains uncommon. In Bogotá, an exception is the informal service called “Wheels,” which operates through WhatsApp groups to coordinate rides, focusing on university communities. This service quickly became a preferred mode of transport for students, faculty, and staff. This study aims to identify the factors driving the success of this informal initiative. A survey of 470 university community members was conducted, incorporating a discrete choice experiment to evaluate the attributes influencing their stated likelihood of use. Analytical methods included multiple correspondence analysis, logit models, and machine learning techniques. Our findings reveal that the adoption of carpooling is significantly influenced by price sensitivity, safety perceptions (particularly among women), reluctance to share rides with strangers, and demographic factors such as age and socioeconomic status. These insights offer valuable guidance for enhancing the appeal and scalability of carpooling as a door-to-door, reliable transportation alternative, particularly in similar sociocultural contexts as our case study.
{"title":"Exploring peer-to-peer paid carpooling in Bogotá: A path to sustainable shared mobility","authors":"César A. Merchán-Núñez , Mauricio Orozco-Fontalvo , Luisa F. Morales-Moreno , Aquiles Darghan , Sonia C. Mangones M. , Lenin A. Bulla-Cruz","doi":"10.1016/j.tbs.2026.101243","DOIUrl":"10.1016/j.tbs.2026.101243","url":null,"abstract":"<div><div>Carpooling is a sustainable transportation alternative that allows users with similar destinations to share private vehicles, contributing to reductions in energy consumption, pollutant emissions, and traffic congestion. While not a comprehensive solution, carpooling can lower private car use and increase vehicle occupancy rates. However, most carpooling initiatives have been limited in scope, often operating on a small scale or within corporate frameworks, restricting their potential for widespread adoption. In Latin America, where ride-hailing services are popular despite regulatory issues, carpooling remains uncommon. In Bogotá, an exception is the informal service called “Wheels,” which operates through WhatsApp groups to coordinate rides, focusing on university communities. This service quickly became a preferred mode of transport for students, faculty, and staff. This study aims to identify the factors driving the success of this informal initiative. A survey of 470 university community members was conducted, incorporating a discrete choice experiment to evaluate the attributes influencing their stated likelihood of use. Analytical methods included multiple correspondence analysis, logit models, and machine learning techniques. Our findings reveal that the adoption of carpooling is significantly influenced by price sensitivity, safety perceptions (particularly among women), reluctance to share rides with strangers, and demographic factors such as age and socioeconomic status. These insights offer valuable guidance for enhancing the appeal and scalability of carpooling as a door-to-door, reliable transportation alternative, particularly in similar sociocultural contexts as our case study.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101243"},"PeriodicalIF":5.7,"publicationDate":"2026-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048469","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-23DOI: 10.1016/j.tbs.2026.101241
Zijian Guo , Mei-Po Kwan , Jian Liu , Xintao Liu
Shared bikes provide flexible mobility but expose riders to harmful outdoor environments, such as humid and oppressive heat, negatively impacting the travel experience. Reducing extra travel time, particularly for open-air transport, is an effective way to minimize unnecessary environmental exposure (UEE). However, the role of extra travel time in exposure studies has received limited attention, and the relationship between the built environment and UEE remains underexplored. This study addresses these gaps by constructing complex networks of UEE based on shared bike travel flows. We first calculate each Shenzhen bike-sharing trip’s extra travel time by comparing the optimal and actual travel time. UEE is then defined as a combination of this extra travel time and the corresponding “feels-like” temperature. For each origin–destination pair, numerous UEE values of trips form a distribution, from which the maximum probability point (EP) and fluctuation (EF) are extracted as two key indicators. These two indicators, along with traffic volume, serve as the weights of network edges. After the network aggregation, spatial hotspot comparisons are conducted, followed by the application of a GCN-LIME model to explain the contribution of the built environment to UEE. The results indicate that areas associated with work, education, and high diversity inhibit the UEE, while areas with food, shops, services, and hospitals promote it. Notably, laborer communities experience higher UEE and are sensitive to changes in the built environment, underscoring issues of spatial justice. These findings provide valuable insights for policymakers to identify high-exposure areas and optimize facilities to mitigate exposure.
{"title":"Investigating extra environmental exposure in bike-sharing trips: spatial patterns and built environment factors","authors":"Zijian Guo , Mei-Po Kwan , Jian Liu , Xintao Liu","doi":"10.1016/j.tbs.2026.101241","DOIUrl":"10.1016/j.tbs.2026.101241","url":null,"abstract":"<div><div>Shared bikes provide flexible mobility but expose riders to harmful outdoor environments, such as humid and oppressive heat, negatively impacting the travel experience. Reducing extra travel time, particularly for open-air transport, is an effective way to minimize unnecessary environmental exposure (UEE). However, the role of extra travel time in exposure studies has received limited attention, and the relationship between the built environment and UEE remains underexplored. This study addresses these gaps by constructing complex networks of UEE based on shared bike travel flows. We first calculate each Shenzhen bike-sharing trip’s extra travel time by comparing the optimal and actual travel time. UEE is then defined as a combination of this extra travel time and the corresponding “feels-like” temperature. For each origin–destination pair, numerous UEE values of trips form a distribution, from which the maximum probability point (EP) and fluctuation (EF) are extracted as two key indicators. These two indicators, along with traffic volume, serve as the weights of network edges. After the network aggregation, spatial hotspot comparisons are conducted, followed by the application of a GCN-LIME model to explain the contribution of the built environment to UEE. The results indicate that areas associated with work, education, and high diversity inhibit the UEE, while areas with food, shops, services, and hospitals promote it. Notably, laborer communities experience higher UEE and are sensitive to changes in the built environment, underscoring issues of spatial justice. These findings provide valuable insights for policymakers to identify high-exposure areas and optimize facilities to mitigate exposure.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101241"},"PeriodicalIF":5.7,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146032755","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-23DOI: 10.1016/j.tbs.2026.101235
Ningzhe Xu , Jun Liu , Steven Jones
Shared Autonomous Vehicles (SAVs) are viewed as a promising next generation mobility solution; however, most existing research has focused on their use in routine daily travel. Individuals’ preferences for using SAVs during emergencies remain largely underexplored. Gaining insight into these preferences is essential for designing effective emergency mobility strategies that can leverage advanced mobility technologies when available. This study investigates SAV usage preferences in emergency contexts across the United States using data from a nationally distributed online survey (N = 1,015). Specifically, the study examines (1) differences in SAV usage preferences between daily and emergency contexts, (2) variations in these preferences across segments defined by emergency preparedness levels, and (3) the factors that influence SAV adoption behavior within each preparedness segment. Methodologically, chi-squared tests were used to assess shifts in SAV usage preferences between contexts, while latent class analysis (LCA) was employed to classify respondents into two preparedness segments: under-prepared and well-prepared. Within each segment, Light Gradient Boosting Machine (LightGBM) models were developed and interpreted using feature importance rankings and partial dependence plots. Findings show increased willingness to use SAVs during emergencies, especially among individuals who were hesitant or unwilling in daily settings. SAV usage preferences also varied by preparedness segment. Several factors, including residential duration, vehicle access, race, and ethnicity, showed consistent effects across groups, while others, such as land use, household with children, home ownership, and household size, displayed divergent patterns. These results highlight the moderating role of preparedness in SAV adoption and caution against directly applying daily-use assumptions to emergency contexts. Policy efforts should consider preparedness-based segmentation to support effective SAV deployment during emergencies.
{"title":"Beyond daily travel: understanding shared autonomous vehicle usage behavior across emergency preparedness segments in the United States","authors":"Ningzhe Xu , Jun Liu , Steven Jones","doi":"10.1016/j.tbs.2026.101235","DOIUrl":"10.1016/j.tbs.2026.101235","url":null,"abstract":"<div><div>Shared Autonomous Vehicles (SAVs) are viewed as a promising next generation mobility solution; however, most existing research has focused on their use in routine daily travel. Individuals’ preferences for using SAVs during emergencies remain largely underexplored. Gaining insight into these preferences is essential for designing effective emergency mobility strategies that can leverage advanced mobility technologies when available. This study investigates SAV usage preferences in emergency contexts across the United States using data from a nationally distributed online survey (N = 1,015). Specifically, the study examines (1) differences in SAV usage preferences between daily and emergency contexts, (2) variations in these preferences across segments defined by emergency preparedness levels, and (3) the factors that influence SAV adoption behavior within each preparedness segment. Methodologically, chi-squared tests were used to assess shifts in SAV usage preferences between contexts, while latent class analysis (LCA) was employed to classify respondents into two preparedness segments: under-prepared and well-prepared. Within each segment, Light Gradient Boosting Machine (LightGBM) models were developed and interpreted using feature importance rankings and partial dependence plots. Findings show increased willingness to use SAVs during emergencies, especially among individuals who were hesitant or unwilling in daily settings. SAV usage preferences also varied by preparedness segment. Several factors, including residential duration, vehicle access, race, and ethnicity, showed consistent effects across groups, while others, such as land use, household with children, home ownership, and household size, displayed divergent patterns. These results highlight the moderating role of preparedness in SAV adoption and caution against directly applying daily-use assumptions to emergency contexts. Policy efforts should consider preparedness-based segmentation to support effective SAV deployment during emergencies.</div></div>","PeriodicalId":51534,"journal":{"name":"Travel Behaviour and Society","volume":"44 ","pages":"Article 101235"},"PeriodicalIF":5.7,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146015816","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号