Introduction

Active school transport

Active School Transport (AST) includes any active mode of travelling to or from school, the most common being walking or riding a bike. AST represents one of the key contributors to overall physical activity in children (Larouche et al., 2014). Other health benefits of AST in children include improved emotional health (Ramanathan et al., 2014), increased social health (Hunter et al., 2015), improved cognition (Phansikar et al., 2019) and increased independence (Herrador-Colmenero et al., 2017). Furthermore, reductions in car use associated with AST lead to reduced traffic congestion, improved pedestrian safety, improved air quality, reduced noise pollution, and increased community liveability (Garrard, 2011). Economic benefits of AST include reductions in costs associated with traffic (i.e. road crashes and travel time) and health (WA Department of Transport, 2020). Thus, the more students that use AST, the more health, environmental, and economic benefits are realised.

Concerningly, fewer students walk and cycle to school than ever before (Aubert et al., 2018), with several Australian studies documenting a steady decline in AST over the past 40 years (Booth et al., 2019; van der Ploeg et al., 2008). In Perth, Western Australia (WA), rates of AST appear to be among the lowest of all capital cities within Australia and internationally, with only 28% of Perth primary school students reporting to regularly use AST (defined as 6 trips or more per week) in 2005 (Trapp et al., 2010). More recent preliminary travel survey data in Perth from 2018 shows that a mere 20% of students use AST, despite the majority of students living less than one kilometre away from school (WA Department of Transport, 2020). Clearly, there is considerable room for improvement in student’s AST participation.

Perth’s low levels of AST participation can be explained in part by built environment characteristics that have perpetuated dependency on private vehicles and discouraged more active forms of transport such as walking and cycling (Ewing et al., 2004). Perth is characterised by conventionally designed neighbourhoods with large block sizes and low network connectivity (Kelobonye et al., 2019). Perth’s population density (346 people per km2) is lower than Sydney’s (428 people per km2) and Melbourne’s (503 people per km2) population densities (Australian Bureau of Statistics, 2022a). Perth does not have any areas classified as high population density (> 5000 people per km2), whereas the total areas in Sydney and Melbourne classified as high population density are 193km2 and 77 km2 respectively (Australian Bureau of Statistics, 2022a).

The high level of reliance on private motor vehicles in Perth has substantial repercussions, with the estimated annual economic cost of car travel to school currently over $186 million (WA Department of Transport, 2020). Therefore, adapting the built environment to encourage more active forms of transport in cities like Perth has the potential to reduce costs as well as improve health.

The built environment and AST

Having a built environment that is supportive of AST is fundamental to encouraging AST participation (D’Haese et al., 2015; Pont et al., 2009). School walkability is an index measure of environmental factors that have been shown to be associated with AST participation, typically including two components, pedestrian connectivity and traffic exposure (Giles-Corti et al., 2011). The pedestrian connectivity (a measure of how well the routes that pedestrians can travel along are connected) is an important component of walkability as it impacts route directness and the overall distance required to travel to school (Fig. 1) (Braza et al., 2004; Panter et al., 2010). Traffic exposure is a second important component of walkability, whereby exposure to high volumes or speeds of traffic can negatively impact AST by increasing safety concerns of both parents and children (Giles-Corti et al., 2011; Kim & Lee, 2020; Panter et al., 2010). Modifying the walkability of built environments surrounding schools to ensure streets are highly connected with low traffic exposure would increase the potential for AST participation.

Fig. 1
figure 1

Examples of disconnected street network (left) versus grid-like street network (right), showing how pedestrian connectivity can impact walkability (figure adapted from (Neighbourhood Streets Project Stakeholders, 2000))

Full size image

Prior research has found a positive association between school walkability and AST participation (D’Haese et al., 2015; Lee et al., 2020; Macdonald et al., 2019). For example, an international systematic review of associations between the built environment and AST found 9 of the 13 studies that have evaluated walkability found a positive association with walking and/or cycling to school (D’Haese et al., 2015). Only three studies have been published since this review that have measured an association between school walkability and AST. All three studies found a positive association between school walkability and AST (Jacobs et al., 2021a; Lee et al., 2020; Macdonald et al., 2019).

Evidence of a link between the walkability of neighbourhoods surrounding Perth Government primary schools and primary school students’ AST habits was found in the 2005 TRavel Environment and Kids (TREK) study (Giles-Corti et al., 2011). Primary school students were 7.4% more likely to regularly use AST if they attended a school located in a highly walkable neighbourhood compared to students attending a school located in a low walkable neighbourhood (Giles-Corti et al., 2011). Increasing the walkability of low walkable school neighbourhoods (through improving pedestrian connectivity and reducing traffic exposure) and siting new schools in highly walkable neighbourhoods therefore presents an opportunity to increase AST participation in Perth.

School walkability and school type

To-date, no research has compared school walkability by school type (i.e. primary vs secondary vs K-12; government vs non-government). It is plausible that differences exist in the walkability of different school types given the differences in policies that guide where schools should be located in a neighbourhood (i.e. adjacent to different land use classifications and road types) and also when a neighbourhood should set aside a location for a school (i.e. at the conceptual phase or after a community has been established) (Department of Planning, Lands, and Heritage & Western Australian Planning Commission, 2020; McDonald, 2010). For example, WA State policy recommends primary schools be located central to a neighbourhood (which may encourage more walking) whereas secondary schools can be located on a neighbourhood’s periphery to facilitate access by vehicle and public transport (which may discourage walking) (Department of Planning, Lands, and Heritage & Western Australian Planning Commission, 2020). Additionally, although more than one third of WA students (33%) attend non-government schools (Australian Bureau of Statistics, 2022b), WA State policy encourages, but does not require, the siting of non-government schools at the structure planning stages of new residential developments (Department of Planning, Lands, and Heritage & Western Australian Planning Commission, 2020). This is because the demand for non-government schools does not normally materialise until neighbourhoods are well established.

School walkability and SES

Current international and Australian evidence suggests that patterns of walkability may vary by socioeconomic status (Giles-Corti et al., 2011; Jacobs et al., 2021b; Macdonald et al., 2016; Zhu & Lee, 2008). For example, an evaluation of walkability surrounding primary schools in Scotland (n = 937) found that schools located in high SES areas of Glasgow were more walkable than in lower SES areas (Macdonald et al., 2016). In Australia, a recent national study measuring the walkability surrounding all primary and K-12 schools (n = 7,133) found schools located in high SES areas were significantly more walkable than schools located in low SES areas (Jacobs et al., 2021a, b). In Perth, the 2005 TREK study found primary schools located in high SES areas were significantly more walkable than primary schools located in low SES areas (Giles-Corti et al., 2011). This disparity is important because students attending schools located in low SES areas generally have less access to private vehicles (McDonald, 2008), and lower levels of physical activity than children attending schools located in high SES areas (Yang et al., 2019). Thus they may be at a double disadvantage. However, no study to-date has investigated SES differences in walkability for all school types (i.e., primary, secondary and K-12 schools), hence more research is needed to assess whether this disparity extends to all schools or just certain types. If disparities in school walkability are shown to exist, they can be targeted by future interventions.

School walkability and subregion

Different planning policies have been introduced to guide the development and urban growth patterns across Perth (Curtis, 2006; Kelobonye et al., 2019; Western Australian Planning Commission, 2015). This has resulted in different urban forms (e.g., transport networks, housing lot size, zoning) across the metropolitan area and its four sub-regions (Central, South Metropolitan, North-East and North-West) (Department of Planning, Lands, and Heritage, 2018). The Central subregion, containing the Central Business District, has some of the oldest neighbourhoods in Western Australia and is characterised by well-connected rectangular street patterns (Western Australian Department of Planning, 2013). Development in Perth’s South Metropolitan, North-East and North-West subregions emphasises low density sprawling neighbourhoods (Kelobonye et al., 2019; Newman, 2016), which may contribute to a lack of well-connected school neighbourhoods. Urban growth has followed the coastline in the North-West and South Metropolitan subregions, while the North-East subregion is characterised by more rural land (Curtis, 2006; Department of Planning, Lands, and Heritage, 2018). Therefore, variations in regional planning policies might lead to disparities in the walkability of schools, although this potential effect has not been explored to date.

Aims

The aims of this study were to: (1) assess the walkability of all schools located in the Perth metropolitan area (n = 651) in 2021; and (2) investigate whether school walkability differs by school socioeconomic status (low vs medium vs high), school type (primary vs secondary vs K-12; government vs non-government), and subregion.

Method

Identification of schools

The names and addresses of all Perth metropolitan schools were identified from the WA Department of Education’s, “Current list of Western Australian Schools” for 2021 (WA Department of Education, 2021) and geocoded in ArcGIS Pro (version 2.7.0) (Esri Inc, 2022). The following schools were excluded from analysis: Community Kindergartens, Education Support Centres, Specialist service schools (i.e. residential colleges, detention centres, re-engagement schools), and schools that have subsequently closed. Schools were classified as being government or non-government based on the classifications outlined in the 2021 “Current list of Western Australian Schools”. “Primary” schools had students in Year 6 or below, “Secondary” schools had students in Years 7 and above (to year 12), and “K-12” schools had students across primary and secondary year groups. The subregion for each school (to account for different urban forms) was identified using the Perth and Peel at 3.5 million Planning Framework (Department of Planning, Lands, and Heritage, 2018).

School SES

The Index of Community Socio-Educational Advantage (ICSEA) score for each school was obtained from the MySchool website profile (Australian Curriculum, Assessment, and Reporting Authority, 2021) and used to measure school SES (Australian Curriculum Assessment & Reporting Authority, 2015). The ICSEA score provides a measure of relative advantage or disadvantage of the overall student community within each school and is based on student family background data, including parent occupation, parent education, school geographic location and proportion of Indigenous students. Schools were separated into tertiles based on their distribution of ICSEA scores (among the sample of schools used for the present study) and classified as “Low” (scores 653–1009), “Medium” (scores 1010–1072) or “High” (scores 1073–1237) SES.

School walkability indices

The walkability within a 2 km radius of all Perth schools (n = 651) was computed in a Geographical Information System (GIS). The School walkability index consisted of two components: (1) pedestrian connectivity; and (2) road traffic exposure, and replicated the methodology used in the 2005 TREK Project (Giles-Corti et al., 2011). This methodology is also consistent with walkability indices in other active transport studies involving children (Christiansen et al., 2014; Lee et al., 2020; Trapp et al., 2012).

Pedestrian connectivity

Pedestrian connectivity is a ratio of the pedestrian network area to the maximum possible area within a defined Euclidean (straight line/as the crow flies) distance (Chin et al., 2008).

A pedestrian network centreline dataset was created using the most recent version of the Open Street Map “roads” data layer (Geofabrik Download Server, 2021). This data layer contains roads, pathways and cut throughs (e.g., laneways and cut throughs through parks). Freeways and freeway entrance ramps were excluded from the layer due to pedestrian access limitations.

Pedestrian connectivity was calculated using a three-step procedure within GIS. First, a ‘walkable service area’ polygon was generated extending out from the school point 2 km along the pedestrian network in all directions. The area of each walkable service area was calculated.

Second, a Euclidean distance circular buffer was created using a 2 km “as-the-crow-flies” buffer around the perimeter of each school cadastre, and the area within this buffer computed (referred to as the ‘circular buffer area’). Figure 2 shows an example of a ‘walkable service area’ and a ‘circular buffer area’ surrounding a school. A buffer distance of 2 km is considered a valid threshold for AST to capture the school catchment area and is aligned with previous studies (Chillón et al., 2015; Giles-Corti et al., 2011). All 2 km ‘circular buffer areas’ were clipped to major water bodies (> 0.5 km2) due to these areas being inaccessible by pedestrians (i.e. the Swan River, the Indian Ocean, etc.) (Department of Biodiversity, Conservation, and Attractions, 2021; Landgate, 2021).

Fig. 2
figure 2

Example of the 2 km ‘circular buffer area’ and 2 km ‘walkable service area’ along the pedestrian network surrounding a school

Full size image

Finally, the pedestrian connectivity score was calculated as the ratio of the ‘walkable service area’ divided by the ‘circular buffer area’. A pedestrian connectivity score closer to 1 indicates a more connected street network. The WA Planning Commission’s (WAPC) Liveable Neighbourhood Guidelines, which specify how to design developments to encourage active transport, have set a pedestrian connectivity score of > 0.60 as the target for a walkable catchment area, meaning a minimum of 60% of the maximum possible 2 km area around a school is accessible within a 2 km walk or cycle (Western Australian Planning Commission, 2015).

Traffic Exposure

Traffic exposure was defined using the most current version of the Main Roads WA’s Functional Road Hierarchy data layer (Main Roads WA, 2022). Within this dataset, roads are classified as either Primary Distributors (> 15,000 vehicles/day), District Distributors (> 6,000 vehicles/day), Local Distributors (< 6,000 vehicles/day), and Access Roads (< 3,000 vehicles/day) (Roads (LGATE-012), 2021). Traffic exposure was calculated as the ratio of kilometres of Primary, District, plus Local Distributor roads to kilometres of Access Roads (roads suitable for bicycles and pedestrians) within the 2 km ‘walkable service area’ of each school. This replicates the measure of traffic exposure from the 2005 TREK Project and provides an indicator of how suitable the roads around the school are for walking and cycling (Fig. 3).

Fig. 3
figure 3

Example of how school traffic exposure is calculated based on Main Roads WA’s Functional Road Hierarchy (Main Roads WA, 2022) within the ‘walkable service area’ of a school

Full size image

Overall Walkability Score

The pedestrian connectivity and traffic exposure ratio scores were combined to create a walkability score for every Perth school (Giles-Corti et al., 2011). The pedestrian connectivity ratios were collapsed into deciles (1 = least pedestrian connectivity and 10 = most pedestrian connectivity). Traffic exposure ratios were also collapsed into deciles and reverse-coded so that (1 = most traffic exposure and 10 = least traffic exposure) (Giles-Corti et al., 2011). The overall walkability score was the sum of the pedestrian connectivity and the traffic exposure deciles, resulting in a total walkability score ranging from 2 (least walkable) to 20 (most walkable).

Statistical Analysis

All spatial analysis was conducted using ArcGIS Pro (version 2.7.0) (Esri Inc, 2022) and all data analyses were conducted in R (version 4.1.2) (R Core Team, 2022) using the tidyverse package (Wickham et al., 2019). Descriptive statistics (mean, standard error, and proportion) were used to describe the proportion of schools that meet the WAPC’s target pedestrian connectivity ratio of 0.60 as well as pedestrian connectivity and traffic exposure ratios and walkability scores. Statistical tests (one-way ANOVAs with post-hoc comparisons with Tukey’s adjustment) were used to investigate any significant differences in: (a) pedestrian connectivity ratios; (b) traffic exposure ratios; and (c) walkability scores; by school SES, school type (i.e., primary/secondary/K-12, government/non-government) and subregion. Odds ratios were used to compare the proportion of schools that met the WAPC pedestrian connectivity score target walkability scores by school SES and school type (i.e., primary/secondary/K-12, government/non-government) and subregion. Schools that did not have an ICSEA score (n = 10) (i.e. new schools or schools with small student populations where not enough data were available) were excluded from analyses involving SES as a measure.

Results

Table 1 presents the pedestrian connectivity, traffic exposure and overall walkability results for all Perth schools (n = 651).

Table 1 Pedestrian connectivity, traffic exposure and walkability of Perth schools in 2021 (n = 651). Note: Bolding denotes a significant difference was found. The significant differences are described in the footnotes
Full size table

Pedestrian connectivity

Overall (all schools combined)

The mean pedestrian connectivity ratio across all schools was 0.51, ranging from 0.08 to 0.80. Only 26.2% of schools met or exceeded the 0.60 WAPC target for a walkable catchment area.

School Type

Pedestrian connectivity did not significantly differ between primary schools (mean 0.52, range 0.09–0.80), secondary schools (mean 0.51, range 0.20–0.72), and K-12 schools (mean 0.49, range 0.08–0.73). There were no significant differences in the proportion of primary vs secondary vs K-12 schools that met the WAPC target for a walkable catchment area. Government school pedestrian connectivity (mean 0.51, range 0.09- 0.76) was not significantly different from non-government school pedestrian connectivity (mean 0.52, range 0.08–0.80). Non-government schools had significantly higher odds of meeting the WAPC target for a walkable catchment area than government schools (32.7% versus 23.0%; OR 1.63, 95% CI 1.13, 2.35). Non-government primary schools had significantly higher odds of meeting the WAPC target for a walkable catchment area than government primary schools (39.4% versus 23.4%; OR 2.1, 95% CI 1.3, 3.4). Non-government secondary schools had significantly higher odds of meeting the WAPC target for a walkable catchment area than government secondary schools (41.4% versus 20.8%; OR 2.7, 95% CI 1.1, 6.8).

SES

Overall, high SES schools had significantly higher pedestrian connectivity ratio scores (mean 0.58, range 0.17 – 0.80) compared with both medium (mean 0.49, range 0.08—0.73, p < 0.001) and low SES schools (mean 0.48, range 0.09—0.68, p < 0.001). High SES primary schools had significantly higher pedestrian connectivity scores (mean 0.59, range 0.19- 0.80) compared with medium SES primary schools (mean 0.50, range 0.17 – 0.69, p < 0.001) and low SES primary schools (mean 0.48, range 0.09 – 0.68, p < 0.001). High SES secondary schools had significantly higher pedestrian connectivity scores (mean 0.59, range 0.42 – 0.71) compared with medium SES secondary schools (mean 0.50, range 0.20 – 0.72, p = 0.007) and low SES secondary schools (mean 0.49, range 0.25—0.67, p < 0.001).

High SES schools also had significantly higher odds of meeting the WAPC target for a walkable catchment area than medium SES schools (52.1% versus 17.1%, OR 5.3, 95% CI 3.4,8.2) and low SES schools (52.1% versus 10.3%, OR 9.5, 95% CI 5.7,15.9). Medium SES schools had significantly higher odds of meeting the WAPC target for a walkable catchment than low SES schools (17.1% versus 10.3%, OR 1.8, 95% CI 1.0,3.2). High SES primary schools had significantly higher odds of meeting the WAPC target than medium SES primary schools (54.0% versus 17.9%, OR 5.4, 95% CI 3.2, 9.1) and low SES primary schools (54.0% versus 9.6%, OR 11.0, 95% CI 5.9, 20.5). Medium SES primary schools had significantly higher odds of meeting the WAPC target than low SES primary schools (17.9% versus 9.6%, OR 2.1, 95% CI 1.0, 4.1). Among secondary schools, high SES secondary schools had significantly higher odds of meeting the WAPC target than medium SES secondary schools (68.2% versus 18.8%, OR 9.3, 95% CI 2.6, 32.8) and low SES secondary schools (68.2% versus 14.0%, OR 13.2, 95% CI 4.0, 43.8).

Subregion

Central schools (mean 0.59, range 0.28–0.80) had significantly higher mean pedestrian connectivity than North-East schools (mean 0.43, range 0.17–0.61, p < 0.001), North-West schools (mean 0.53, range 0.08–0.73, p < 0.001), and South schools (mean 0.45, range 0.09–0.68, p < 0.001). North-West schools (mean 0.53, range 0.08–0.73) had significantly higher mean pedestrian connectivity than North-East Schools (mean 0.43, range 0.17–0.61, p < 0.001). South schools (mean 0.45, range 0.09–0.68) had significantly higher mean pedestrian connectivity than North-East schools (mean 0.43, range 0.17–0.61, p < 0.001).

Central schools had a significantly higher odds of meeting the WAPC target than North-East schools (48.0% versus 1.1%, OR 82.1, 95% CI 11.3, 598.6), North-West schools (48.0% versus 23.3%, OR 3.0, 95% CI 1.8–5.0), and South schools (48.0% versus 11.7%, OR 7.0, 95% CI 4.2, 11.5). North-West schools had a significantly higher odds of meeting the WAPC target than North-East schools (23.3% versus 1.1%, OR 27.0, 95% CI 3.6, 203.0) and South schools (23.3% versus 11.7%, OR 2.3, 95% CI 1.2, 4.2). South schools had a significantly higher odds of meeting the WAPC target than North-East schools (11.7% versus 1.1%, OR 11.8, 95% CI 1.6, 88.5).

Traffic exposure

Overall (all schools combined)

The mean traffic exposure ratio for all schools combined (n = 651) was 0.52 and ranged between 0.09 and 3.54. Half of all Perth schools (51.5%) had a traffic exposure ratio of less than 0.50, meaning that the length of low traffic roads (< 3,000 vehicles per day) was more than twice the length of the high traffic roads (> 3,000 vehicles per day). Among all schools, 2.5% had a traffic exposure ratio greater than 1, meaning that the length of high traffic roads exceeded the length of low traffic roads.

School Type

Non-government secondary schools had significantly higher traffic exposure ratio scores (mean 0.80, range 0.32–2.34) compared with non-government primary schools (mean 0.52, range 0.20–1.65, p < 0.001) and government secondary schools (mean 0.49, range 0.16 – 1.07, p < 0.001). Secondary schools (mean 0.58, range 0.16–2.34) had significantly higher traffic exposure ratio scores than primary schools (mean 0.50, range 0.09–1.65, p = 0.026). Non-government schools (mean 0.58, range 0.13–3.54) had significantly higher traffic exposure ratio scores than government schools (mean 0.50, range 0.09–1.42, p < 0.001).

SES

For all schools combined (n = 651), there were no statistically significant differences in traffic exposure by level of school SES. Low SES secondary schools were found to have significantly lower traffic exposure ratio scores (mean 0.51, range 0.16 – 0.99) compared with high SES secondary schools (mean 0.75, range 0.32 – 2.34, p = 0.015).

Subregion

Central schools (mean 0.56, range 0.25–2.34) had significantly higher traffic exposure than South schools (mean 0.48, range 0.09–3.54, p = 0.013).

Overall school walkability scores

Overall (all schools combined)

For all schools combined (n = 651), the mean school walkability score was 11, ranging from 2–20.

School Type

No significant differences in walkability scores were found when comparing primary (mean 11.2, range 2–20), secondary (10.6, range 2–20) and K-12 schools (10.6, range 2–19). No significant differences were found when comparing government school walkability (mean 10.9, range 2–20) and non-government school walkability (mean 11.1, range 2–20).

SES

On average, High SES schools had a significantly higher walkability score (mean 12.4, range 3–20) than medium SES (mean 10.5, range 2–20, p < 0.001) and low SES (mean 10.1, range 2–20, p < 0.001) schools. High SES primary schools had a significantly higher walkability score (mean 13.0, range 3–20) than medium SES primary schools (mean 10.5, range 2–19, p < 0.001) and low SES primary schools (mean 10.1, range 2–20, p < 0.001). SES and overall walkability for all schools is displayed in Fig. 4.

Fig. 4
figure 4

Walkability and socioeconomic status among all schools (n = 651) in Perth displayed by subregion (a). North-West, (b). North-East, (c). Central, (d), South Metropolitan). Each dot represents a school. A larger dot represents a school with a higher walkability score while a smaller dot represents a school with a smaller walkability score

Full size image

Subregion

School walkability is displayed by subregion in Fig. 4. Central schools (mean 12.3, range 3–20) had significantly higher overall walkability than North-East schools (mean 8.4, range 2–18, p < 0.001), North-West schools (mean 11.0, range 2–18, p = 0.024), and South schools (mean 10.5, range 2–20, p < 0.001). North-West schools (mean 11.0, range 2–18) had significantly higher overall walkability than North-East Schools (mean 8.4, range 2–18, p < 0.001). South schools (mean 10.5, range 2–20) had significantly higher overall walkability than North-East schools (mean 8.4, range 2–18, p < 0.001).

Discussion

This study assessed the walkability of all schools located in the Perth metropolitan area (n = 651) in 2021 and whether it differed by school socioeconomic status, school type and subregion. Overall, this study found the majority of Perth schools were surrounded by street networks that are disconnected. For example, nearly three quarters of Perth schools (73.8%) had pedestrian connectivity ratios below the WAPC’s target ratio for a walkable catchment area. This finding is consistent with the 2005 TREK study where the majority of Perth government primary schools had pedestrian connectivity ratios below the WAPC target ratio (Christiansen et al., 2014). The lack of well-connected school neighbourhoods was significantly more prevalent in the North-West, North-East and South subregions of Perth. Compared to the Central subregion, the other subregions were developed predominantly under planning policies with poorly connected neighbourhoods with winding and dead end streets (i.e. cul-de-sacs) (Kelobonye et al., 2019), resulting in poor connectivity typical of that seen in some other Australian and U.S. metropolitan areas (Currie et al., 2009; Nechyba & Walsh, 2004). Schools located in the older suburbs of the Central subregion were significantly more walkable due to their traditional connected street block layouts.

This study also found evidence of a socio-economic gradient in school neighbourhood walkability, whereby neighbourhoods surrounding high SES schools were significantly more walkable than low or medium SES schools. This finding was consistent with the 2005 TREK Study, which found high SES government primary schools were located in more walkable school neighbourhoods than low SES schools in Perth (Giles-Corti et al., 2011). This finding was also consistent with a national Australian study which found a trend of increasing walkability with increasing SES among high SES primary and K-12 schools in Perth, Melbourne, Adelaide, Sydney and Brisbane (Jacobs et al., 2021a, b). Similarly, an evaluation of walkability in Glasgow primary schools found that schools located in high income areas were more walkable than schools located in low-income areas (Macdonald et al., 2016).

This study found non-government schools had significantly more traffic exposure than government schools. Given this was the first study to evaluate walkability of all non-government schools, there are no other studies to compare this finding to. Reasons for this difference require further investigation, but might relate to the different WA policies regarding how and when non-government schools are sited.

Overall, the results highlight the importance of modifying the built environment to increase the walkability of school neighbourhoods in order to increase the opportunities for children to use an active form of transport to school. Existing schools located in neighbourhoods with poor pedestrian connectivity and/or high traffic volume should be retrofitted with interventions to improve their walkability. Built environment interventions to improve pedestrian connectivity in existing neighbourhoods are associated with increased AST (Boarnet et al., 2005; McDonald et al., 2013). Examples include cut-throughs (i.e. paths though parks or between houses), footpaths on both sides of the road, pedestrian underpasses/overpasses, and pedestrian crossings. Interventions to reduce traffic speed and volume are also associated with increased AST (Boarnet et al., 2005; Rothman et al., 2021; Smith et al., 2020). Examples include speed humps, speed limit signage, chicanes, changing the appearance of the road surface (i.e. painting speed limits, brick surfaces instead of pavement), narrowing traffic lanes, temporary road closures to vehicles surrounding the school, and reducing parking availability (to disincentivise driving). Results from this study indicate that connectivity improvements should be prioritised around low-SES schools. Students in low SES areas often rely on active transport to school, so making these areas more walkable is particularly important (Vincent et al., 2017; Yang et al., 2019).

To avoid expensive retrofits of existing neighbourhoods, it is more efficient to plan future neighbourhoods to have higher street connectivity and to site schools away from high traffic roads. Supportive policies are crucial to facilitate the placement of schools in walkable areas. An example of a supportive policy to reduce traffic exposure is evident in New South Wales and Victoria policies that explicitly prohibit the location of schools on arterial (high-traffic) roads (New South Wales Department of Education, 2020; Victoria Department of Education, 2022). In contrast, the current policy in Western Australia mandates that at least one street frontage of secondary schools should be an arterial road (Department of Planning, Lands, and Heritage & Western Australian Planning Commission, 2020). Aligning WA policy with policies in other states may benefit school walkability. Future research should focus on evaluating policy and practice interventions to improve connectivity and reduce traffic exposure and then measuring impacts on AST participation. These interventions should focus on retrofitting existing school neighbourhoods as well as planning for future school neighbourhoods.

This is the first study within Australia and internationally to assess SES differences in walkability for all school types (i.e., primary, secondary and K-12 schools). Another strength of this study is that it is the first to assess the walkability of all schools across an entire metropolitan area. A potential limitation of this study is that some of the stratifications of schools (e.g. school SES and school type) had small numbers of schools, so these results should be interpreted with caution. However, statistical analyses were only conducted when the sample of schools was larger than 20. Another potential limitation of this study is that the walkable service area may not align with actual school catchment areas and as such, may misrepresent actual student travel. However, the walkable service areas used in this study was more accurate/suitable than using Euclidian buffers around schools and standardises the areas around schools. The walkable service areas were also based on Open Street Map data which is open-source data with the potential to be limited in completeness (Barrington-Leigh & Millard-Ball, 2017). However, Open Street Map data has the advantage of including pedestrian pathways to create a more complete pedestrian network compared to using a road network. Open Street Map data is also reproducible, more cost-efficient than manual data collection and has enormous potential for assessing road networks without technical geospatial expertise (Klinkhardt et al., 2021; Moradi et al., 2022).

In conclusion, this study found a large proportion of Perth schools are surrounded by street networks that are disconnected, reducing the potential for more children to walk or cycle to school. The results highlight the importance of modifying the built environment surrounding existing schools to increase opportunities for active school transport, particularly in low and medium socio-economic areas. Moreover, future neighbourhoods need to ensure that new schools are sited away from high traffic roads and in areas with high pedestrian connectivity.