Chapter 4 Data and methodology
4.3 Research strategy for Objective 1
The census data in Australia is readily purchasable from the Australian Bureau of Statistics (ABS). Given the research goal of Objective 1, the compilation of the census data involves the consideration of three major issues, namely, the year of census, data to be extracted and comparability of the data.
First, considering that Brisbane’s BRT has been operated since the year 2000, census data before and after 2000 is apparently needed. A search of census years shows that census data of 1996, 2001, 2006 and 2011 are accessible and suitable to meet this cross-lagged consideration. For the second issue, in order to fulfil the examination of modal share patterns, for each census, the method for travel to work (MTW) data that stores the modal shares of various transport modes (e.g., private car, public transport and walk) for work trips are extracted. In addition, socio-demographic variables are also needed to model the relationships between modal share patterns and socio-demographic characteristics, which are available on the ABS website.
Compared to the first two issues, the third issue is the most challenging one due to the changes in the geographic unit systems between each of the censuses. Specifically, there have been some minor changes on the smallest census unit (i.e., Collection District or CD) boundaries between 1996, 2001 and 2006 censuses. For the 2011 census, there was a major shift from CD to Statistical Area 1 (SA1) as the smallest census unit. Such changes largely hinder the comparability between raw census data of different census years. To overcome this issue, census data need to be concorded to the same spatial units to enable the comparability between different censuses. Given the major shift in the 2011 census, its concordance with 1996, 2001 and 2006 census data (at the smallest spatial unit) is not possible. Therefore, concorded MTW data by place of enumeration of 1996, 2001 and 2006 on 2006 CD boundaries was purchased from the ABS. This constitutes the basis for the time-series analysis for the modal share patterns of BRT catchments of the study context.
The 2011 census data was drawn separately from the Table Builder of the ABS as the basis for investigating the relationship between modal share patterns and socio-demographic characteristics of BRT catchments. Through reviewing previous literature examining public transport use and socio-demographic characteristics (Boarnet and Crane, 2001; Kitamura et al., 1997; Stead, 2001; Bagley and Mokhtarian, 2002), 20 variables were further extracted to reflect the key socio-demographic dimensions of commuters at the SA1 level
80 (i.e., gender, age, income, education, household composition, education, motor vehicle ownership and population density), which is detailed in Chapter 5.
4.3.2 Defining BRT catchments
To assist the examination of modal share patterns of BRT catchments, three levels of catchment areas based on the access distance to the BRT stations are applied, i.e., 800-metre, 1,600-metre and 3-kilometre distance radiating from the BRT stations opened before year 2011 (Figure 4.3).
Common to many transport studies, an 800-metre radial distance from an identified UPT station is considered as the primary catchment area for a UPT service. The logic behind this is that for most people an 800-metre distance is the walkable distance to a transit station (Stringham, 1982; Guerra et al., 2012). Empirical evidence has been found to support this notion, that a large proportion of UPT passengers walk to their transit stations within 800-metre (Stringham, 1982; O'Sullivan and Morrall, 1996; Cervero, 2007), however it is noted that there have been some exceptions to this distance (Zhao et al., 2003). Given this, in addition to the 800-metre catchment, two additional distances (i.e., 1,600-metre and 3-kilometre) were also adopted here to investigate the change of travel patterns for work trips under the impacts of BRT (Figure 4.3). The 1,600-metre distance represents the maximum walking distance to transit service following previous studies, e.g., Zhao et al (2003), while the 3-kilometre distance is found to be a limit for drive-in/bus-in distance to a transit service (Norley, 2010). By comparing the travel patterns on these three bands, it is expected that a clearer picture of the impacts of BRT on travel behaviour of its catchments can be obtained.
81 Figure 4.3 Catchments of BRT stations prior to 2011
82
4.3.3 Analytical methods for census data
For Objective 1, the reported modal share patterns are mainly concentrated on the mode shares of train, bus, car (car driver, car passengers and total), walk and bicycle for work trips. Given that the mode shares of other methods including ferry, taxi, motorcycle and mixed methods were low individually at both CD and city-wide levels, they were added together as the ‘other’ group. Total number of people who made work trips excludes people who did not go to work, worked at home, or did not state their travel methods.
Two sets of analysis were carried out: (1) time-series analysis of modal share patterns for work trips before and after BRT implementation; and (2) regression analysis of socio-demographic characteristics and modal share patterns. For the first task, the MTW data of 1996, 2001 and 2006 are aggregated to calculate and compare the modal shares for the BRT catchments. Three catchment areas, i.e., 800-metre, 1,600-metre and 3-kilometre areas, are adopted to distinguish the changes in modal share patterns at different BRT catchments. Next, stepwise regression is applied to the 2011 census data at the SA1 level to model the relationship between modal shares (mainly focusing on the modal shares by bus and private cars) and socio-demographic characteristics. The detailed results are provided in Chapter 5.