Modelling student travel mode choice: A case study in Semarang, Indonesia

Abstract

The paper presents an investigation of university students' travel mode choice behaviour for commuting to campus. It uses a dataset collected through a survey among students of four universities in Semarang, i.e., Unnes, UINW, Udinus, and Unissula. Eight hundred and thirty-seven respondents completed the questionnaires. Multinomial logit (MNL) models are used for investigating four transport mode choices, i.e., walk, motorcycle, car, and bus. The independent variables are schools, genders, origin, residential types, vehicle ownership, driving license ownership, monthly allowance, transportation budget, and commuting distance. The model shows that all independent variables, except monthly allowance, are significant predictors. Overall, the model prediction accuracy is about 90 percent. A motorcycle was the foremost prevalent transport for school trips. Empirical models reveal that the mode of choice behaviour of non-native students differs significantly from those that originate in Semarang. The log odds of driving a car, riding a motorcycle, and walking versus riding a bus are higher when students' origin is not from Semarang compared to Semarang.