Accounting for Systematic Heterogeneity across Commuters in Response to Multiple TDM Policies: Application to a Megalopolis

摘要

Modeling commuters choice behavior in response to Transportation DemandManagement (TDM) helps predicting the policiesconsequences. Although research in thechoice behavior has been evolved to investigate preference heterogeneity in response tofactors influencing their choice, none of them have considered taste variation acrosscommuters in response to multiple TDM policies. This paper investigates the presence ofsystematic preference heterogeneity (which can be explained by heterogeneity in bothcommuters and trip characteristics) across commuters, in response to the TDM policies thatcan be explained by their socio-economic or commuting-related characteristics. The analysisis based on the results of a stated preferences survey developed through the design ofexperiments approach. Five policies including increasing parking cost, increasing fuel cost,cordon pricing, transit time reduction and transit access improvement are assessed in order tostudy their impact on commuters considerations of six modes of transportation, to travel towork. For the sake of assessing both systematic and random preference heterogeneity acrosscar commuters, an extended form of Mixed Multinomial Logit (MMNL) model wasdeveloped for the 366 individuals who regularly commute to their workplace in the (center ofthe city of Tehran) city center of Tehran, Iran, as a developing country. The likelihoodfunction of this model shows more than three percent improvement, compared to the baseMNL model, including same variables. In addition to a number of commuting and contextualvariables, the model shows that the taste variation across commuters results in differences ineffectiveness of increasing parking cost, transit time reduction and transit accessimprovement policies. The analysis includes examining several distributions for randomparameters to test the impacts of restricting distributions to allow only normality, confirmingthe potential to improve model fitness with alternative distributions.