COMPARISON OF SENSITIVITIES FOR BUS TRAVELERS AND RAIL TRAVELERS USING STATED PREFERENCE METHODS Case Study: West Yorkshire, United Kingdom
ARIF KURNIAWAN, Prof. Dr-Ing. Ir. Ahmad Munawar, M.Sc
2013 | Tesis | S2 Mag. S. & T.Transportasi-
This research involves the estimation of discrete choice models on this dataset. The objective is to evaluate commuter sensitivities between two groups of commuters according to several attributes, such as travel time, crowding, fares, reliability, and information availability. More detailed study of the differences in willingness to pay measures between the two groups was carried out. An existing stated preference dataset containing 391 observations for bus and rail passengers was used in this research as part of a study looking at current travel behaviour. The dataset was obtained from online survey in the United Kingdom in 2011. Passengers identified as choice users were presented with a set of choice options illustrating the comparison between current modes condition regarding to service attributes and two alternatives options. The purpose of stated preference survey is to assess how passengers’ choices vary in different hypothetical situations. Random utility concept was used to reflect unobservable elements of choice behaviour. Multinomial Logit Model (MNL) was applied to start with, as this method is the mostly applied technique. Further estimation was used using advance models such as nested and mixed multinomial logit model. Logit analysis was used to analyse the discrete choice data. The estimation was typically based on the statistical principle of likelihood maximisation and statistical measurement. Biogeme v1.8 was used for the model estimation. Mixed MNL was considered as best fit models instead of the other. Further model estimation along with socio-demographic segmentation was carried out. Statistically significant results were obtained from these segmentations. Most of the estimated coefficients are statistically significant using 95% of confidence interval and satisfies the priori belief with correct signs. Sensitivity analysis confirms that bus and rail passengers almost have no significant differences in sensitivity toward service attributes, except for reliability attributes. Socio-demographic segmentation results for sensitivity are shown similar trends. Male passengers are the most sensitive groups towards all of service attributes. For income segmentation, the higher income groups tend to have higher sensitivity. Value of time result clarifies that male passenger have higher value of time, higher income level have higher value of time except for rail passengers. The values vary between £0.007/min up to £0.024/min across socio-demographics segmentations.
Kata Kunci : Mixed multinomial logit model, sensitivity, service attributes, value of time, socio-demographics, segmentation
