Preferences for on-demand transportation as a last-mile mode considering weather and time urgency: A stated choice approach

• Contextual effects like weather and time urgency influence last-mile mode choices. • Stated choice data from rail commuters ensure realistic last-mile mode assessment. • Demand for on-demand transport is more elastic than buses for key attributes. • Cost elasticity of on-demand transport varies widely depending on contexts. • Context-aware demand modeling supports better design of last-mile mobility services. On-demand transportation (ODT) has been increasingly adopted to address first- and last-mile challenges, with the aim of integration into public transportation systems. However, previous studies have not sufficiently examined how contextual factors influence preferences for this service as a last-mile mode. This study analyzes preferences for ODT as an activity-end last-mile mode, focusing on weather and time urgency. Data were collected through a stated choice experiment conducted among railway commuters traveling to Yokohama National University from five nearby stations. Respondents evaluated hypothetical scenarios comparing ODT with buses or walking under four contextual scenarios (sunny/rainy × low/high time urgency). A panel mixed logit model was estimated to analyze last-mile mode choices and quantify the effects of cost, wait time, and travel time, including interactions with weather and time urgency. The results show that travel time has a greater negative impact under time urgency than under ordinary rainy conditions. Demand for ODT is more elastic than for buses concerning cost, wait time, and travel time in most contextual scenarios, and cost elasticity varies considerably depending on the scenario. The value of travel time for ODT is the highest in “rainy and high time urgency” scenarios, reaching up to 5.6 times the baseline value. Simulations reveal that weather strongly influences demand levels for ODT when costs are low, while time urgency becomes more influential when costs are high. The findings provide quantitative insights into context-sensitive preferences, supporting the design and evaluation of last-mile mobility services.