Using longitudinal data, it is possible to separate the effects of growing older (ageing) from the differences between generations (age cohorts) when modeling travel behavior. This approach has been used by the author and others to investigate whether Millenials traveled fewer miles by auto than other age cohorts, using data from the 1995, 2001 and 2008 National Household Travel Survey datasets. In this presentation, the earlier analyses are updated to also utilize the 2012-13 and 2022-23 NHTS datasets to see whether any differences have persisted over a longer period. The analysis is also expanded to also include Generation Z, and, more importantly, to also focus on the time spent walking and biking. Although ageing has a clear effect, with the use of active transportation decreasing with age, once the ageing effect is taken into account, each successive age cohort can be shown to walk and bike more than previous cohorts, all else equal. The analyses include descrete choice modeling of mode choice, as well as censored regression analysis of the distance and time spent using specific modes. The models are multi-variate, controlling for the effects of income, gender, employment status, household size and composition, and other socio-demographic variables.  

In the rapidly evolving transportation landscape, a lack of timely and reliable data presents a major challenge for planning and modeling. Traditional household travel surveys, while valuable, are costly and infrequent, leaving gaps in understanding emerging mobility trends. Big data sources offer an opportunity to bridge this gap, but their reliability must be thoroughly assessed before they can be used to inform decision-making. Ensuring that such data accurately represents real-world conditions is critical for advancing the state of the practice in travel demand modeling.

TDM23, adopted in June 2023 as part of the region’s Long-Range Transportation Plan, was developed using a 2019 base year to project travel patterns for 2050. However, the pandemic has brought notable shifts in commuting habits, transit usage, and trip-making behavior, creating an urgent need to reassess how well the model reflects current conditions. As an accelerated solution until the results of the 2024–25 Massachusetts Travel Survey become available, this study uses Replica, a big data fusion platform, to evaluate post-pandemic changes in travel behavior and tests TDM23’s Work-From-Home component as a potential approach to simulate post-pandemic travel patterns. 

Overall, this study seeks to answer the following questions:

• How have mobility patterns shifted in the post-pandemic era?

• To what extent is Replica data valid and effective for calibrating TDM23?

• Does activating TDM23’s WFH component adequately reflect post-pandemic travel behavior?

To meet these objectives, we compare Replica estimates from Fall 2019 (pre-pandemic) and Fall 2022 (early post-pandemic stabilization) across key model components—Vehicle Availability, Trip Generation, Trip Distribution, Mode Choice, and Time of Day—as well as system-level trends such as Highway Vehicle-Miles Traveled (VMT). The findings were documented in a memo, which served as a diagnostic step to determine whether Replica-estimated changes aligned with expectations and to highlight model elements most affected by post-pandemic behavior.

Building on these findings, the study then tests whether activating the Work-from-Home (WFH) component in TDM23 can account for the observed changes. Differentiated WFH rates are applied by geography for workers and by job sector for employment. Using summary and validation reports, we evaluate whether these adjustments are sufficient to reproduce post-pandemic conditions or if additional refinements to the model are required.

Updating travel models to reflect evolving post-pandemic travel behaviors and forecasting challenges.

One enduring effect of the COVID-19 pandemic has been the popularity of telecommuting: To this day, 23% of the salaried workers continue to work from home, according to the U.S. Bureau of Labor Statistics. Using three waves of the household travel survey data from 2017, 2019 and 2021 in the Puget Sound Region, WA, this study examines how telecommuting, which also means the removal of the workplace as an anchor point from one’s daily activity and travel pattern, affects the generation and rescheduling of maintenance and discretionary trips that are previously conducted around home and workplaces. The associated consequences including changes in modes of transportation used and vehicle miles traveled (VMT) are also investigated. We found that though telecommuting resulted in reduced number of trips and VMT in general, there is a significant increase in the number of maintenance and discretionary trips. Additionally, telecommuters also exhibited less complex trip chaining behavior, characterized by simpler tours with shorter trips, fewer stops, and lower mode diversity compared to non-telecommuters. Spatially, telecommuters conducted maintenance and discretionary trips closer to home; temporally, and the departure times of these trips are more spread out with emerging peaks such as late morning, and mid-day. These results have significant policy and modeling implications relating to transportation service provision, local economy, and travel demand forecasting models.

Authors: Grace Jia, Kaitlyn Ng, Ekin Ugurel, Brian Lee, Ram Pendyala, Cynthia Chen.

Transport Policy paper: https://authors.elsevier.com/a/1lkgX,L-HRtsAI (free access before 28 Oct 2025)