6. Discussions and conclusions
This paper applied Logistic regression models to understand the nearest station choice behaviours of transit users. The study revealed that the nearest station choice depended on the location of station, and characteristics of stations and transit users. When the chosen station was located at the end or near the end of a train line (captive stations), this left transit users with much less station choice than a station located along the line (non-captive stations). This also means that less variables influence the nearest station choice. For example, only two variables—distance and station location in terms of destination direction (further-away or non-further away)—were relevant for the captive model. Based on our survey, the reasons why users choose a station further away from their origin and destination are seat- and parking-availability, particularly the former. This suggested that crowding on trains was becoming an issue in Western Australia, which can be managed by increasing train capacity and frequency (Li and Hensher, 2012), providing better service and design, such as improving air quality and circulation (Thompson et al., 2012)