Methodological issues in modelling time-of-travel preferences

We address three methodological issues that arise when modelling time-of-travel preferences: unequal period lengths, schedule delay in the absence of desired time-of-travel data and the 24-hour cycle. Varying period length is addressed by using size variables. Schedule delay is treated by assuming either arrival or departure time sensitivity and using market segment specific utility functions of time-of-travel, or using distributions of the desired times-of-travel. The 24-hour cycle is modelled by using a trigonometric utility functional form. These methodologies are demonstrated in the context of a tour-based travel demand model using the 2000 Bay Area travel survey.