It is important to measure public transport accessibility to help improve the sustainability of transport systems in metropolitan areas. Although many studies have defined different approaches for measuring public transport accessibility, there have been limited methods developed for measuring accessibility levels that incorporate spatial aspects. Population density is an important distributional indicator that has also been ignored in previous methods developed for quantifying accessibility. This paper outlines the research context for measurement of public transport accessibility and then describes a methodology developed as well as an application the Public Transport Accessibility Index in Melbourne area, Australia. Using the Victorian Integrated Survey of Travel and Activity dataset, we applied separate-ordered logit regression models to examine how the new index performs with a series of predictor variables compared with two existing approaches. Key findings indicate that there is a higher probability of public transport patronage in areas with higher levels of accessibility. Furthermore, it was found using statistical modelling that the new index produces better results compared with previous approaches. This paper presents the results of a study aimed at objectively measuring PT accessibility by considering population density in the metropolitan area of Melbourne, Australia. The study contains four main parts. The first part after a review of previous research describes the calculation process for estimating the accessibility index. The following section presents the methodology, analysis and results of the models. The final section discusses the key findings and the implications of the approach.
BACKGROUNDAccessibility measures have been generally categorized into three groups, access to PT stops, duration of journeys by PT modes and access to destinations by PT modes [9]. A large number of studies measuring accessibility have focused on the proximity to a PT stop/station [10][11][12][13]. Some of these studies have measured accessibility levels by considering an administrative division to a PT stop. Currie [11] claimed that using an administrative division as an alternative for homes of all residents within a selected boundary can bias the results. To address this problem, some studies have measured accessibility from dwelling units to PT stops [10,14,15]. A key component in modelling access to PT stops is the walking distance. Typically, the maximum acceptable walking distance is considered as 400 and 800 m for PT stops or stations [7,11,16].Although physical access to PT stops is important, the time taken to travel between an origin and destination by PT modes can be considered as another significant factor [1]. Along with studies that focus on access to PT stops, some studies focus on the duration of a journey undertaken by PT modes [17,18]. O'Sullivan et al.[17] measured PT accessibility generating maps of accessible areas with the same travel time. In another study, Cheng and Agrawal [19] introduced...