Modelling Mode Choice in Short Trips: Shifting from Car to Bicycle
K Halldorsdottir, L Christensen, T Jensen, C Giacomo Prato, Technical University of Denmark, DK
This paper presents mixed logit estimation and scenario simulation of mode choice models in short trips in order to evaluate policies encouraging the shift from car to bicycle and in general sustainable transport modes.
Over the years, production, income, number of trips and travel distances have increased in Denmark as all over the world. Demand for faster and more flexible transport has consequently grown and the use of motorized private mode has steadily increased. As a result, road congestion has become a major problem, especially in the Greater Copenhagen Area. Measures are required toward the decreasing of the congestion and the solution of the related health problems through the promotion of sustainable transport modes, with particular emphasis on the shifting from private transport to bicycling.
The Department of Transport at the Technical University of Denmark, in collaboration with the Danish Road Directorate, investigated short trips by car and questioned whether it is possible to make car drivers shift to biking or walking. The study presented an MNL model of trips shorter than 22 km on the basis of data from the Danish National Travel Survey and demonstrated how three types of conditions influence the choice of transport mode:
? Conditions concerning the road user and the purpose of the trip, where car ownership and number of children in the family are the most important factors.
? Conditions concerning the circumstances of the trip, where differences of hilliness and temperature have proved to be greatly relevant.
? Conditions concerning the travelling conditions where the project describes the effect of speed for car drivers and cyclists, the parking conditions as well as a general effort to promote biking.
The study also showed that 90 % of the short car trips would be transferred to bicycle, in the case travellers would transfer to sustainable transport modes. Cycling policies should therefore guide the effort to reduce the short trips by car in favour of bicycling.
The present study extends the choice model of the previous one in two directions. First, public transport was not considered in the MNL model of the previous study because of the lack of data about the level of service of the public transport service. The present study considers indeed public transport within the choice model, since clearly the evaluation of the transfer from car to bicycle could be biased by the exclusion of public transport as possible option. Second, the MNL model of the previous study did not allow considering heteroscedasticity across alternatives and heterogeneity across travellers. The present study estimates a mixed logit model able to account for heteroscedasticity and to capture taste variations through a specification that expresses randomly distributed parameters and differentiates the travel time parameters across modes to express different values of time for different modes. Hence, this study provides a fresh and up-to-date perspective with respect to the previous study.
Data about short trips in the Greater Copenhagen Area are available from the Danish National Transport Survey that is being conducted by DTU Transport on behalf of the Ministry of Transport and other governmental departments. The survey identifies the travel behaviour of the Danish population through interviews collecting travel diaries and socio-economic variables of a representative sample of the population. 14,747 observations of short trips in the Greater Copenhagen Area constitute the sample for model estimation. Five alternatives are considered according to the classification in the Danish National Transport Survey (i.e., car driver, car passenger, public transport, walk and obviously bike), and level of service variables are calculated through assignment procedures available for each period of the day in which the trip was conducted. The calculation of level of service variable allows considering congestion conditions similar in average to the ones encountered by the travellers.
A mixed logit model allows investigating the effect of level of service variables, individual characteristics of the travellers, purpose of the trips, environmental conditions (e.g., weather, temperature, and slope) and accidents. Results suggest factors able to make cycling more attractive, for example improving accessibility, enhancing infrastructures, addressing specific population groups for specific trip purposes. As the objective is the promotion of bicycle use with emphasis on everyday cycling, modelling the choice of transport modes allows understanding which policies might be effective in influencing the choice of sustainable transport modes in order to reduce car traffic. Scenario simulations allow further evaluation of the effects of possible policy instruments intending to convert short car trips to cycling or walking.
Association for European Transport