Sage Journals: Discover world-class research

Abstract

Active transportation modes such as walking and biking are gaining popularity for their extensive health and environmental benefits, yet scholars know little about how place-based accessibility varies by area sociodemographic composition. This study is among the first to examine sociodemographic disparities (by both race and socioeconomic status) in bikeability while allowing for heterogeneity in disparities. Consideration of bikeability disparities is particularly critical within the framework of urban planning concepts that promote equitable accessibility and reduced dependency on automobiles, such as the 15-minute city. Geographically Weighted Regressions examined associations between census tract-level bikeability (using an index that combines five components), socioeconomic status, and percentage non-White residents (controlling for age of structures in tracts). Findings showed that the strength and directionality of associations between bikeability and race/socioeconomic status varied throughout the county, providing targeted information on where greater concentrations of low socioeconomic status and non-White residents were associated with lower bikeability.

Keywords

bikeability transportation demography sociology of population sociology social sciences disparities education sustainability

Introduction

Advocates in fields ranging from public health to conservation are increasingly encouraging individuals to use “active transportation” modes such as walking or biking due to their environmental, social, economic, and health benefits (see Giles-Corti et al., 2010 for review). Active transportation is associated with reduced air pollution, road congestion, and noise, as well as increased neighborhood social interactions, which have economic benefits (see Giles-Corti et al., 2010 for review). Personal and public health benefits (Mueller et al., 2015) of active transportation include increasing individuals’ engagement in regular physical activity, thus reducing obesity (Bassett et al., 2008) and mortality risk (Andersen et al., 2000). Cycling may be among the most beneficial modes of active transportation, as it can be reasonably used to replace automobile trips (Winters et al., 2013), which could improve air quality thereby reducing risk of chronic respiratory disease (Nieuwenhuijsen, 2016).

Use of active transportation is influenced by characteristics of the built environment that affect accessibility (Handy et al., 2002; Vale et al., 2015, for review). Unfortunately, at the structural level, accessibility-based approaches to engineering are competing with auto-centric approaches (Abdullah et al., 2022). While accessibility-based approaches focus on non-motorized transport and are not generally speed dependent, traditional car-based approaches prioritize a focus on speed and improvement of vehicular traffic flow. And yet, accessibility-based approaches have gained ground during the pandemic and post-pandemic eras. Increasingly, city planners are considering alternatives to speed-dependent mobility, for example, the “15-minute city” concept (Ali et al., 2021; Gaglione et al., 2022). Key features of this model are that city dwellers are able to get to points of interest, including shopping, entertainment, healthcare, recreation, education, home and work within 15 min, either through walking, public transportation, or cycling (Abdullah et al., 2022; Chen & Crooks, 2021).

Review of the Literature

Studies find greater active travel use among urban dwellers in areas with greater perceived accessibility, which includes measures such as perceived access to bike lanes and number of destinations (Hoehner et al., 2005). Studies also report more engagement in active travel (Freeman et al., 2013) and more minutes of physical activity among those living in areas with greater measures of accessibility such as mixed land use and street connectivity (Frank et al., 2005). Much less research has examined bikeable infrastructure specifically (although there are some common characteristics between bikeable and walkable areas), but studies that do so also find that accessibility affects cycling rates. Indeed, among the most important factors that encourage or deter cycling, particularly cycling for transportation rather than recreation (Porter et al., 2020; Yu, 2014), is how bike-friendly the built environment is, including land use and connectivity (Saelens et al., 2003). For example, a study of 26 Canadian and United States cities found that bikeability was positively associated with cycling rates (Winters et al., 2016).

Despite evidence that place-based access influences individuals’ active transportation use, researchers have paid far less attention to disparities in place-based accessibility, which likely contribute to the disparate rates of cycling in differing communities and population groups. Considering calls for individuals to use active transportation, it is important for scholars, planners, and public health professionals to examine whether there are disparities in access to active transportation, as those could ultimately contribute to disparities in health and other outcomes. A large literature finds substantial sociodemographic disparities in neighborhood built environment, (Gordon-Larsen et al., 2006; Moore et al., 2008) which suggests the same may be true for access to active transportation infrastructure. In terms of walkable infrastructure, studies find mixed evidence. Some report lower walkability in low-income and minority neighborhoods, such as one study in the St. Louis, MO, metro area that found unevenness and obstruction of sidewalks impeded walkability (Kelly et al., 2007). Yet other studies, including a study conducted in Phoenix, AZ, have found greater walkability in high minority neighborhoods areas, but that other neighborhood disamenities impede use even though infrastructure is walkable (Cutts et al., 2009). Of the few studies focused on bikeability disparities, most have focused on income disparities, with some reporting low access to adequate bikeable infrastructure in neighborhoods with more low-income residents (Fuller & Winters, 2017).

There is a dearth of research examining racial/ethnic disparities in bikeability, despite the fact that minority groups are more likely to depend on cycling for transportation than White, college-educated men (Porter et al., 2018). Scholars have noted that residential segregation by race contributes to population health disparities by limiting access to health-enhancing resources (Williams & Collins, 2001), including the built environment. Diverse groups of residents may also have different needs when it comes to their points of interest in the built environment, which requires that urban inequity is understood in order to plan a successful 15-minute city (Chen & Crooks, 2021). Thus, it is important to understand how concentration of residents by both socioeconomic status (SES) and racial/ethnic background are associated with access to bikeable infrastructure to access points of interest. Despite race and socioeconomic status being “fundamental causes” of population health disparities (Phelan et al., 2010), these variables are understudied or missing from research on the 15-minute city (Chen & Crooks, 2021; Gaglione et al., 2022).

Finally, the few previous studies examining disparities in place-based accessibility to active travel have used traditional, global regression approaches, which overlook how geographic processes (such as residential sorting and bikeability) are affected by local differences in historical, administrative, political, and planning context—factors that contribute to disparities in neighborhood environments (Gilbert & Chakraborty, 2011). It is plausible that there are neighborhoods in which higher bikeability is associated with higher SES and greater concentration of White residents, while in other areas there is no association, or an inverse association. For instance, neighborhood disadvantage theories might focus on how areas with a higher concentration of low SES and minority residents have lower collective efficacy (Sampson et al., 1997), making it difficult to advocate successfully for neighborhood improvements in bikeable infrastructure and safety. On the other hand, while this may characterize older, urban neighborhoods because of the historical factors such as macroeconomic changes and suburbanization that have shaped residential segregation in those areas (Wilson, 2012), it may not hold true in more recently developed areas, such as planned communities where planners may have designed neighborhoods with active transportation in mind. Thus, studies are needed to assess not only whether there is an overall association between area sociodemographic composition and bikeability, but more specifically, in which areas sociodemographic disadvantage is associated with lower area bikeability. This information could help urban planners better target their efforts to reduce inequities in place-based access to bikeable infrastructure.

Current Study

This study builds on existing active transportation literature by examining sociodemographic disparities in access to bikeable infrastructure. We address several large gaps in the literature. First, we are among the few to examine bikeability access disparities by both race/ethnicity and socioeconomic status. Second, we examine a larger geographic area than most previous studies, which often focused on one city (Cowie et al., 2016; Winters et al., 2013)—this allows us to assess how these associations vary across areas with different socio-political and planning histories. Ours is the first study of this kind, to our knowledge, to use Geographically Weighted Regression to assess potential geographic variation in the nature of these disparities across a large study area. The study objectives were twofold: (a) to determine whether there is an association between residents’ sociodemographic composition and area-level bikeability (net of neighborhood age), and (b) to examine geographic variation (i.e., non-stationarity) in these associations across a large geographic area. We examine census tracts in Orange County, CA, the sixth most populous county in the United States at over 3 million residents, and one of the most diverse in terms of population sociodemographic characteristics (U.S. Census Bureau, 2018). Given this sociodemographic diversity, as well as varied topography and urbanicity, Orange County is a useful context in which to study potential disparities in active transportation access in a mixture of urban and suburban settings.

Methods

Study Area and Data Sources

We acquired data on sociodemographic characteristics of Orange County from the American Community Survey (ACS) 5-year estimates (2012–2016) for race/ethnicity, education, and income, and from CalEnviroScreen 3.0 (which uses a special analysis of ACS data) for the housing burden (2009–2013), linguistic isolation (2011–2015), and unemployment (2011–2015) measures. Data on roads, census tract boundaries, and urban areas definitions came from the TIGER/Line shapefiles (United States Census Bureau, 2017). Data on infrastructure age also came from the ACS 5-year estimates (2013–2017). Of the 582 total census tracts in Orange County, CA, our final analysis excluded census tracts that were coastal waters, those that were not census-defined urbanized areas, and areas missing data on any of the socioeconomic indicators, leaving a final analytic sample size of 526 census tracts.

Dependent Variable: Census Tract Bikeability

We combined five components to create an index of bikeability, based on measures used in a previous study (Winters et al., 2013): density of bike routes, separation of bike routes from motor vehicle traffic, connectivity between bike-friendly routes, density of destinations cyclists are most likely to visit, and topography. Data came from numerous sources: bikeways and arterial highways came from publicly available Orange County Transportation Authority data (2015); elevation data came from the United States Geological Survey (2016); and land use data came from the Southern California Association of Governments (2012).

We created raster surfaces for each component, then overlaid them to create a final index. A detailed account of our methodology is provided below. All geoprocessing was done in ArcMap 10.3.

Generating Raster Surfaces

Dependent Variable: Bikeability

Bicycle Route Density

Using bikeway data from Orange County Transportation Authority, we isolated all existing (as opposed to prospective) designated on-street and off-street bikeways to create a bike route shapefile, then converted this shapefile to a raster surface using the Line Density tool.

Bicycle Route Separation

Bike routes are separated into three main classes depending on their separation from the road: Class III bikeways share a lane with motor vehicle traffic, Class II bikeways occupy a painted bike lane, and Class I bikeways are paved pathways completely separated from motor traffic. We selected all existing Class I bikeways from our bike route shapefile used above, and converted this new shapefile to a raster file using the Line Density tool.

Connectivity of Bike-Friendly Routes

Using arterial highway data from Orange County Transportation Authority, we isolated all non-arterial streets, with the assumption that since arterial highways tend to have higher speed limits and higher traffic volumes, they are less safe and therefore less bikeable than non-arterial routes (Winters et al., 2013). For the purpose of this study, these roads were deemed bike-friendly roads. We used the Intersect tool with a point output to create junctions between these bike-friendly roads and bikeways, and did the same to create junctions between bikeways and other bikeways. These two point shapefiles were combined using the Merge geoprocessing function, then converted to a raster surface using the Point Density tool.

Destination Density

Based on land uses positively associated with cycling in previous studies (Winters et al., 2013), we used land use data and tables from Southern California Association of Governments to create a shapefile of the following potential destinations for cycling: general office, commercial services, select facilities, education, transportation, mixed commercial and industrial, mixed residential and commercial, and select open space and recreation uses. We used the Feature To Point tool to convert the resulting polygons to points, then created a raster surface using the Point Density tool.

Topography

Using the Slope tool with percentage rise, we created a slope raster from an existing elevation raster to indicate the percent increase or decrease in elevation across the surface.

Creating Bikeability Index

To create the final bikeability index, we first reclassified each of the five raster surfaces described above to be on a scale of 1 to 10, where 10 indicated a high level of bikeability and 1 indicated low bikeability. The five components were overlaid and summed with equal weighting to create a raster surface index of bikeability (Figure 1). We then used the Zonal Statistics tool to compute average census tract bikeability from the raster data, which resulted in the final dependent measure, mean bikeability score by census tract (where 1 is low bikeability and 10 is high bikeability).

Figure 1.

Simplified visual representation of methods.

Independent Variables: Sociodemographic Characteristics

For our socioeconomic status (SES) index, we included measures shown in previous studies to be associated with neighborhood context and infrastructure: unemployment, income, housing burden, linguistic isolation, and educational attainment. Unemployment is defined as percent of the census tract population above 16 years of age that is eligible for the labor force but not currently employed (active military, students, and those no longer seeking work are excluded). Income is measured using median household income over the past 12 months. Housing burden refers to the percentage of the census tract that spends over 50% of income on housing. Linguistic isolation describes the percent of households where no one over the age of 14 speaks English well. Educational attainment is measured as the percent of the population with less than a Bachelor’s degree. To create the index of low socioeconomic status, we calculated the z-scores for each of the five indicator variables (median household income was reversed to be consistent with the directionality of the other indicators).

The final low socioeconomic status index was created by summing the z-scores for each of the five indicator variables above, with higher values indicating lower relative SES. We also examined the percentage of non-White residents by census tract. Finally, we included median year built of structures within the census tract as a covariate to control for the possible confounding factor of infrastructure age.

Analyses

All statistical analyses were conducted in ArcMap 10.3, and all layers were projected to the same projected coordinate system. Only census tracts with non-missing data were included in regressions (N = 526). We first conducted diagnostic analyses to determine the most appropriate model for our data, beginning with an Ordinary Least Squares (OLS) linear regression. The high level of clustering of the regression residuals indicated significant spatial autocorrelation (Moran’s I = 0.42, p < .001), meaning that the residual values were more similar in tracts that are near each other, which violates the independence assumption of linear regression. This suggested that spatial models would be more appropriate to model these associations.

We next used a spatial regression, Geographically Weighted Regression (GWR), to assess the independent associations between bikeability, low SES, and percent non-White population while controlling for age of the neighborhood. GWR is an ideal modeling approach because it accounts for spatial dependence and allows for the possibility that the association between census tract sociodemographic factors and bikeability varies across the study area. This method creates a separate regression equation for each observation in the study area, rather than generating a single regression equation averaged across the entire county; thus, we present both a table comparing the OLS and GWR regression results and a table comparing two selected census tracts to illustrate how GWR reveals geographic variation in associations.

We used an adaptive bandwidth (rather than fixed), which enabled the regression to adapt to the variable density of data due to varying sizes of census tracts in the study area. As a sensitivity analysis, we examined a reduced model including income rather than the socioeconomic index as a predictor—goodness-of-fit indicators showed better fit of the model using the SES index (AIC = 1306.41 versus 1370.81). To reduce multicollinearity, all variables were standardized.

Results

Table 1 presents the census tract characteristics in Orange County, CA. The mean bikeability across census tracts in the study area is 5.1 out of 10, with 10 representing high bikeability. The county is racially/ethnically diverse, with the mean percentage of non-White residents within census tracts being 55.7%. Median household income is approximately $82,500 for the county, with census tracts ranging from $22,700 to nearly $250,000. On average, 62.6% of residents report having less than a Bachelor's degree. Across census tracts in the county, the mean unemployment rate is 7.7%, housing burden prevalence is 18.8%, and linguistic isolation prevalence is 10%.

Table 1.

Census Tract Characteristics (Orange County, CA; N = 549).

Characteristic	M	SD	Min.	Max.
Dependent variable
Bikeability index^a	5.1	1.1	2	8.5
Independent variables
Non-White (%)	55.7	24.2	5.9	100
Socioeconomic index components
Unemployment (%)	7.7	2.9	0.9	19.1
Housing burden (%)	18.8	7.4	3.4	46.7
Linguistic isolation (%)	9.9	8.9	0	41.7
Less than Bachelor’s education (%)	62.6	18.6	10	97.9
Median household income ($)	82,545	31,673	22,736	249,597

Bikeability Index ranges from 1 to 10, with higher numbers indicating greater bikeability.

Figure 2 reveals that bikeability varies substantially across the county, and tends to be clustered, with areas of high bikeability being near each other, and areas of low bikeability being near each other. The Moran’s I of 0.44 (p < .01) confirms that there is spatial autocorrelation in bikeability.

Figure 2.

Bikeability in Orange County, CA, urban census tracts (N = 526). Bikeability index components: bicycle route density, route separation, connectivity, destination density, and topography.

To examine whether bikeability is associated with residents’ sociodemographic characteristics, we conducted several regressions, beginning with traditional linear regressions. Table 2 compares results from the OLS regression to results from a GWR. The OLS coefficients suggest that, across census tracts in the study area, there are independent effects of percentage non-White population and of low SES on bikeability, even controlling for neighborhood age. Specifically, there is a negative association between percent non-White and bikeability (b = −0.10) and a small, positive association between low SES and bikeability (b = 0.02); however, the model adjusted R² is very small. Model diagnostics for OLS regression revealed that assumptions of OLS regression are violated, suggesting the need for a spatial model that accounts for spatial dependence in the dependent variable. Specifically, the Moran’s I of 0.42 (p < .001) on the OLS residuals shows evidence of spatial autocorrelation, which violates the assumption that the errors are independent. We then conducted GWR, which allows non-stationarity of the association across the study area; in other words, a GWR model computes regressions for each census tract.

Table 2.

Summary of Ordinary Least Squares and Geographically Weighted Regression of Sociodemographics and Bikeability (N = 526).

	OLS		GWR coefficients
	Coefficient	Mean	Median	Min.	Max.
% Non-White	−0.10	0.030	−0.052	−1.902	1.977
Low SES index	0.02	0.041	0.037	−0.286	0.421
Intercept	5.08	5.472	5.382	4.122	8.120
R ²	.003	.245	.217	.002	.646
Adjusted R²	−.001

Note. All covariates are standardized. GWR model includes “median age built of structures” as a covariate to control for the age of surrounding infrastructure as a confounding variable. Non-urban tracts were excluded from analysis. Low SES index is a sum of the z-scores for five index items; higher numbers indicate lower socioeconomic status.

In the GWR model, 51% of census tracts have a negative regression coefficient for percent non-White, and 36.3% of tracts have a negative coefficient for low SES index. This negative association indicates that greater concentrations of lower socioeconomic status residents and a higher percentage of non-White residents are associated with lower bikeability. Conversely, the remaining positive coefficients indicate a positive association, where lower socioeconomic status and higher non-White population are associated with higher bikeability. Since neither proportion of positive to negative coefficients are overwhelming, this highlights the spatial nature of the association between census tract bikeability and sociodemographic indicators. Overall, the GWR model shows that the model (including race-ethnicity, socioeconomic status index, and neighborhood age) explains roughly 65% of the variability in bikeability across the county (model global R² = .652, global adjusted R² = .542).

Moreover, there is substantial geographic variation in the predictive strength of the model. Figure 3 visualizes the local R² values throughout the county, revealing that the model explains a high proportion of variation in bikeability (>40%) in only 16.7% of census tracts. Specifically, the model appears to perform better in the southern parts of the county, a small portion of the northern part of the border of the county, and the southern urban region near the central portion of the county, while it performs rather poorly in the central northern parts of the county.

Figure 3.

Geographically Weighted Regression (GWR) local R² values in Orange County, CA, urban census tracts (N = 526).

Figure 4 illustrates how the association between census tract bikeability and the sociodemographic predictors varies in both direction and significance throughout the county, using the GWR coefficients for percent non-White and the low SES index. Negative coefficient values and corresponding red shading indicate a negative association between sociodemographic indicators and bikeability, meaning low socioeconomic status and high non-White population are associated with low bikeability. Blue shaded areas indicate the opposite association; specifically, that low bikeability is associated with high socioeconomic status and low minority population.

Figure 4.

Independent effects of concentration of low SES and non-White residents on census tract bikeability.

Table 3 compares the GWR results for two census tracts that were selected to illustrate how the associations between bikeability and sociodemographic factors vary locally. We selected these tracts because they are geographically proximal (approximately 3.5 miles apart) and comparable in terms of urbanicity, bikeability levels (approximately 4.8), and predictive strength of the model (R² = >.40) yet these areas show distinct associations between population sociodemographic composition and bikeability. In the Garden Grove census tract (for reference, FIPS code: 06059088903), bikeability is negatively associated with concentration of non-White and low SES residents, while in the Westminster tract (FIPS code: 06059099601), percent non-White population is negatively associated with bikeability (and the coefficient is substantially larger than for the Garden Grove tract) while low SES is positively associated with bikeability. These distinct independent effects of sociodemographic composition on bikeability evidence the utility of a GWR versus global regression approach in revealing nuanced patterns that may be useful in targeting areas in greatest need.

Table 3.

Comparison of Two Orange County (OC), CA census tracts: Geographically Weighted Regression results.

	Tract in city of Garden Grove (FIPS code: 06059088903)		Tract in city of Westminster (FIPS code: 06059099601)
	b	SE	b	SE
% Non-White	−0.74	0.54	−1.59	0.48
Low SES index	−0.06	0.10	0.32	0.12
Intercept	5.99		4.85
R ²	.40		.40
Mean observed bikeability	4.83		4.86

Note. All covariates are standardized. GWR model includes “median age built of structures” as a covariate to control for the age of surrounding infrastructure as a confounding variable. Low SES index is a sum of the z-scores for five index items; higher numbers indicate lower socioeconomic status. These census tracts are approximately 3.5 miles apart.

Discussion

Accessibility-based approaches to the urban built environment, such as the 15-minute city, make sense across a variety of socioeconomic thresholds. For example, transitioning to active transportation can improve life satisfaction and health among residents and reduce dependence on fossil fuels, both individually and collectively, thus saving individuals money and reducing harmful impacts on climate at the city-level. While the 15-minute city has been gaining popularity, there is still a lack of prioritization for accessibility-based approaches for a wide variety of reasons (Abdullah et al., 2022). One piece of the puzzle toward reduced dependency on motorized transport at the city—and county—level is understanding how different groups of people are able to access non-motorized travel such as cycling. To attend to this gap, we build on existing active transportation literature by examining sociodemographic disparities in bikeability.

Our study focused on census tracts in Orange County, CA, the sixth most populous Unites States county, and a diverse one in terms of racial/ethnic and socioeconomic composition, topography, and planning history/land use composition. We contribute novel information from spatial analyses which revealed that sociodemographic disparities in bikeability are non-stationary; in some areas, there is lower bikeability in areas with greater concentration of lower SES or non-White residents, while in other areas there is no association or the opposite is true. These findings have both methodological and substantive implications for understanding disparities in access to bikeable infrastructure in many settings, and could provide insight to urban planners concerned with equity issues in access to active travel.

One of our most novel contributions is to show that the association between census tract bikeability and population sociodemographic composition is non-stationary. We extend the literature by providing evidence that race-ethnicity and socioeconomic status (and neighborhood age), together, account for a substantial proportion of variability in bikeability, but that the strength and direction of these associations varies considerably across the county. In some areas, our results confirmed previous studies’ findings—in a non-trivial number of census tracts, we found that greater concentrations of low SES and non-White residents were independently associated with lower bikeability. Specifically, we found lower bikeability in areas with larger percentages of non-White population in the more urban regions in the northern part of the county, with the strongest associations around the cities of Buena Park, Westminster, and some parts of Irvine. Areas where lower socioeconomic status was associated with lower bikeability included the central urban region of the county, parts of Irvine, and the coastal region of Newport Beach. The finding of lower bikeability in lower SES areas is consistent with Fuller and Winters’ (2017) study of several Canadian cities, in which they found income disparities in bike lane access and overall bikeability score, with higher income areas having greater bikeability. This also aligns with some of the evidence about the association between neighborhood sociodemographic attributes and suitability of infrastructure for active transportation (Kelly et al., 2007) and between race-ethnicity, income, and access to recreational facilities (Moore et al., 2008).

However, in some parts of the county there was no association between bikeability and sociodemographic composition. Other studies, including one from Australia (Cowie et al., 2016), have similarly found no significant association between neighborhood socioeconomic status and active travel access, which could be due to contextual and policy variations between the United States and Australian context. Given the mixed evidence from studies in different geographic regions and countries, more research is needed to understand the contexts and conditions in which population sociodemographic composition is associated with placed-based access to active transportation.

Interestingly, the independent effects of concentration of low SES and non-White residents in some locations were in opposite directions. For instance, in some areas, low SES was associated with low bikeability (net of racial/ethnic composition) and in those same areas there was no independent effect or an opposite effect of racial/ethnic composition on bikeability. In the southern portion of the county, for example, although the predictive strength of our model was relatively high, there was a positive association between bikeability and percentage non-White population (indicating greater bikeability in areas with larger minority population), and no strong association between bikeability and SES. Future studies can assess more nuanced associations between these sociodemographic characteristics by, for example, examining a possible interaction between non-White population and SES. The independent effects examined here may be useful in targeting resources to the areas in greatest need, for example those in which both greater concentrations of low socioeconomic status residents and of non-White residents are associated with lower bikeability.

The non-stationarity of the association between bikeability and sociodemographic composition may be explained by several area-specific characteristics, including topography, population or land use characteristics, and political or incorporation attributes. For context, Orange County consists of both older, denser urban areas and newer, planned communities that have not only influenced residential segregation processes across the county, but may have also influenced bikeability due to planned development considerations. Although our analyses control for neighborhood age by including median year built for structures, we may not capture all the effects of new, planned communities in contrast to older, more high-density urban neighborhoods. Topography also has an effect on bikeability, meaning that areas in the southern region of the county that have attracted a high income population to the hills are very low in bikeability despite being high in socioeconomic status. Of course, topography is not modifiable, and therefore planners must take into account the more modifiable components of bikeability in order for active transportation improvements to be effective in light of topography constraints.

In light of these findings, we propose that more geographically nuanced analyses should be used to better assess where there are inequities in access to active transportation. Our findings highlight the geographic variability in the nature of such spatial data, and is in line with the limited literature available that leverages GWR to examine environmental health and urban planning-related issues. For instance, Maroko et al. (2009) found that race/ethnicity and socioeconomic factors played an important role in predicting access to parks in various regions across New York City, but that disparities varied across the study area with regard to strength and directionality. Studies of exposure of minority populations to air toxics in Florida (Gilbert & Chakraborty, 2011) and New Jersey (Mennis & Jordan, 2005) reported similar trends. Thus, our findings have methodological implications for future research on active transportation, underscoring the need for and utility of analytic methods (such as GWR) that allow for geographic variation in associations when studying relationships between the built environment, physical health, and sociodemographic attributes of a larger region. Studies using global regression approaches may obscure disparities within the study region.

Conclusion

This study provides evidence of inequities in place-based access to active transportation, which may ultimately translate into individual and public health disparities. These findings have important implications for both research and practice. In terms of research, results highlight the importance of using spatial methods, especially those such as GWR, that allow for nuanced examination of disparities by modeling non-stationarity in associations when considering urban bikeability and/or the 15-minute city. For applied purposes, understanding the association between bikeability and population sociodemographic characteristics is important for planners working on sustainable design and healthy communities, and for interventionists seeking to increase physical activity through active travel. Residents of neighborhoods with greater concentrations of low socioeconomic status individuals may be more likely to rely on cycling as a form of commute (Yu, 2014), especially if such populations do not have financial means of owning an automobile (Rachele et al., 2018). Thus, to ensure equity in active transportation, there may be a more pressing need to increase safety and access to bikeable infrastructure in low-income and minority regions than in more affluent, predominantly White regions. Moreover, inequity in the built environment has a greater effect on cycling rates than sociodemographic attributes (Yu, 2014), and greater access to bikeable infrastructure is associated with higher rates of cycling (Fraser & Lock, 2010; Goodman et al., 2013; Porter et al., 2020; Saelens et al., 2003). Thus, disparities in active transportation may be more effectively addressed through changes in land use planning than personal health interventions (de Nazelle et al., 2011).

Considering that many land use planning decisions are carried out on a local government level, geographic variations in active travel disparities provide valuable insight into formulating appropriate solutions for each area. Likewise, the complexity of how bikeability of different regions is affected by sociodemographic composition emphasizes the importance of addressing disparities with local regulations versus larger regional intervention strategies. Indeed, critics have pointed to the fact that planning does not affect all racial/ethnic groups in the same way, and specifically that historical and political processes shape inequities in access to bicycle culture, including bikeable infrastructure (Hoffmann, 2016). Considering the increasing interest in equity among planners and others (American Planning Association, 2019), this study is timely and provides insight into potential inequities in bikeability access. Our findings highlight the utility of using GWR in analysis of sociodemographic disparities in active transportation, and that changes in land use planning should occur on the local level to account for nuances in the strength and direction of association throughout the county. These findings suggest that, in order to address disparities in active transportation and physical activity, local governments and urban planners should target resources and funding opportunities to improve bikeable infrastructure especially in low-resource areas where bikeability is low and the population is particularly disadvantaged. This might encourage and enable more equitable access to active transportation, particularly cycling, as a mode of commute, and ultimately reduce population health disparities. Investing in bikeable infrastructure can also increase non-vehicular access to other more sustainable forms of transportation, such as public transit, for longer trips. This is particularly useful for populations dependent on non-vehicular modes of transportation and further reduces reliance on automobiles.

Future studies can build on this work in several ways. First, although we include measures of bike route separation (as a proxy for safety), we were unable to directly measure bicycle safety through measures of bicycle crashes. Second, our study used a bikeability index with equal weighting of scale items, based on survey research of factors that encourage or deter cycling behavior and following previous research using a similar index (Winters et al., 2013); future studies can build on this study by determining whether different weighting of items on the bikeability scale is warranted. Finally, our study focuses on census tract characteristics because residents hoping to use active transportation for commuting must, by default, pass through their immediate “neighborhoods” (through their census tract) on their commute; however, a growing literature advocates for broader definitions of “exposure,” (Perchoux et al., 2016) thus future studies can examine these issues using different geographic units.

While robust and aggregated quantitative analyses cannot explain individuals' choices to engage in active mobility options, recognized patterns of use can help city planners and engineers encourage non-motorized transportation across diverse sociodemographic categories. In particular, our work helps shed light on how the racialized and classed use of active transportation varies by geographical location. Investing in accessibility-based approaches to urban planning is timely, progressive, and savvy design—even when demand may be hard to predict or the demographic composition of the city population is more heavily reliant on motorized options. With work and attention to sociodemographic factors, forward-thinking city planners and engineers can ensure that in the future, bike lanes will not be white lanes (Hoffmann, 2016).

Footnotes

Acknowledgements

The authors would like to thank anonymous reviewers for comments on early versions of this manuscript.

Authors’ Note

This paper uses secondary data and is exempt from IRB review.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Jeanette Gritton

Megan Thiele Strong

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

References

Abdullah

Ali

Javid

M. A.

Hussain

(2022). Awareness and knowledge levels of engineering and planning students and practitioners about the 15-minute city concept in a developing country. Journal of Urban Mobility, 2, 100037. https://doi.org/10.1016/j.urbmob.2022.100037

Ali

Abdullah

Javid

M. A.

(2021). Accessibility-based approach: Shaping travel needs in pandemic situation for planners’ perspectives. Engineering Journal, 25(1), 16–22. https://doi.org/10.4186/ej.2021.25.1.15

American Planning Association. (2019). Planning for equity policy guide.

Andersen

L. B.

Schnohr

Schroll

Hein

H. O.

(2000). All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work. JAMA Internal Medicine, 160(11), 1621–1628. https://doi.org/10.1001/archinte.160.11.1621

Bassett

D. R.

Pucher

Jr. Buehler

Thompson

D. L.

Crouter

S. E.

(2008). Walking, cycling, and obesity rates in Europe, North America, and Australia. Journal of Physical Activity and Health, 5(6), 795–814. https://doi.org/10.1123/jpah.5.6.795

Chen

Crooks

A. T.

(2021). Delineating a 15-minute city’: An agent-based modeling approach to estimate the size of local communities. In Proceedings of the 4th ACM SIGSPATIAL International Workshop on GeoSpatial Simulation, GeoSim’ 21, New York, NY, 2021 (pp. 29–37). Association for Computing Machinery.

Cowie

C. T.

Ding

Rolfe

M. I.

Mayne

D. J.

Jalaludin

Bauman

Morgan

G. G.

(2016). Neighbourhood walkability, road density and socio-economic status in Sydney, Australia. Environmental Health, 15(1), 58. https://doi.org/10.1186/s12940-016-0135-y

Cutts

B. B.

Darby

K. J.

Boone

C. G.

Brewis

(2009). City structure, obesity, and environmental justice: An integrated analysis of physical and social barriers to walkable streets and park access. Social Science and Medicine, 69(9), 1314–1322. https://doi.org/10.1016/j.socscimed.2009.08.020

de Nazelle

Nieuwenhuijsen

M. J.

Antó

J. M.

Brauer

Briggs

Braun-Fahrlander

Cavill

Cooper

A. R.

Desqueyroux

Fruin

Hoek

(2011). Improving health through policies that promote active travel: A review of evidence to support integrated health impact assessment. Environment International, 37(4), 766–777. https://doi.org/10.1016/j.envint.2011.02.003

10.

Frank

L. D.

Schmid

T. L.

Sallis

J. F.

Chapman

Saelens

B. E.

(2005). Linking objectively measured physical activity with objectively measured urban form: Findings from smartraq. American Journal of Preventive Medicine, 28(Suppl. 2), 117–125. https://doi.org/10.1016/j.amepre.2004.11.001

11.

Fraser

S. D. S.

Lock

(2010). Cycling for transport and public health: A systematic review of the effect of the environment on cycling. European Journal of Public Health, 21(6), 738–743. https://doi.org/10.1093/eurpub/ckq145

12.

Freeman

Neckerman

Schwartz-Soicher

Quinn

Richards

Bader

M. D. M.

Lovasi

Jack

Weiss

Konty

Arno

(2013). Neighborhood walkability and active travel (walking and cycling) in New York City. Journal of Urban Health, 90(4), 575–85. https://doi.org/10.1007/s11524-012-9758-7

13.

Fuller

Winters

(2017). Income inequalities in bike score and bicycling to work in Canada. Journal of Transport & Health, 7, 264–268. https://doi.org/10.1016/j.jth.2017.09.005

14.

Gaglione

Gargiulo

Zucaro

Cottrill

(2022). Urban accessibility in a 15-minute city: A measure in the city of Naples, Italy. Transportation Research Procedia, 60, 378–385. https://doi.org/10.1016/j.trpro.2021.12.049

15.

Gilbert

Chakraborty

(2011). Using geographically weighted regression for environmental justice analysis: Cumulative cancer risks from air toxics in Florida. Social Science Research, 40(1), 273–286. https://doi.org/10.1016/j.ssresearch.2010.08.006

16.

Giles-Corti

Foster

Shilton

Falconer

(2010). The co-benefits for health of investing in active transportation. NSW Public Health Bulletin, 21(6), 122–127.

17.

Goodman

Panter

Sharp

S. J.

Ogilvie

(2013). Effectiveness and equity impacts of town-wide cycling initiatives in England: A longitudinal, controlled natural experimental study. Social Science and Medicine, 97, 228–337. https://doi.org/10.1016/j.socscimed.2013.08.030

18.

Gordon-Larsen

Nelson

M. C.

Page

Popkin

B. M.

(2006). Inequality in the built environment underlies key health disparities in physical activity and obesity. Pediatrics, 117(2), 417. https://doi.org/10.1542/peds.2005-0058

19.

Handy

S. L.

Boarnet

M. G.

Ewing

Killingsworth

R. E.

(2002). How the built environment affects physical activity: Views from urban planning. American Journal of Preventive Medicine, 23(2), 64–73. https://doi.org/10.1016/S0749-3797(02)00475-0

20.

Hoehner

C. M.

Brennan Ramirez

L. K.

Elliott

M. B.

Handy

S. L.

Brownson

R. C.

(2005). Perceived and objective environmental measures and physical activity among urban adults. American Journal of Preventive Medicine, 28(Suppl. 2), 105–116. https://doi.org/10.1016/j.amepre.2004.10.023

21.

Hoffmann

M. L.

(2016). Bike lanes are White lanes: Bicycle advocacy and urban planning. University of Nebraska Press.

22.

Kelly

C. M.

Schootman

Baker

E. A.

Barnidge

E. K.

Lemes

(2007). The association of sidewalk walkability and physical disorder with area-level race and poverty. Journal of Epidemiology and Community Health, 61(11), 978. https://doi.org/10.1136/jech.2006.054775

23.

Maroko

A. R.

Maantay

J. A.

Sohler

N. L.

Grady

K. L.

Arno

P. S.

(2009). The complexities of measuring access to parks and physical activity sites in New York City: A quantitative and qualitative approach. International Journal of Health Geographics, 8(1), 34. https://doi.org/10.1186/1476-072x-8-34

24.

Mennis

J. L.

Jordan

(2005). The distribution of environmental equity: Exploring spatial nonstationarity in multivariate models of air toxic releases. Annals of the Association of American Geographers, 95(2), 249–268. https://doi.org/10.1111/j.1467-8306.2005.00459.x

25.

Moore

L. V.

Diez Roux

A. V.

Evenson

K. R.

McGinn

A. P.

Brines

S. J.

(2008). Availability of recreational resources in minority and low socioeconomic status areas. American Journal of Preventive Medicine, 34(1), 16–22. https://doi.org/10.1016/j.amepre.2007.09.021

26.

Mueller

Rojas-Rueda

Cole-Hunter

de Nazelle

Dons

Gerike

Götschi

Panis

L. I.

Kahlmeier

Nieuwenhuijsen

(2015). Health impact assessment of active transportation: A systematic review. Preventive Medicine, 76, 103–114. https://doi.org/10.1016/j.ypmed.2015.04.010

27.

Nieuwenhuijsen

M. J.

(2016). Urban and transport planning, environmental exposures and health-new concepts, methods and tools to improve health in cities. Environmental Health, 15(1), S38. https://doi.org/10.1186/s12940-016-0108-1

28.

Orange County Transportation Authority Data. (2015). OCTA GIS open data portal: Bikeways. https://www.arcgis.com/home/item.html?id=ecdb6438f07945479d92bca2c82980ba#overview

29.

Perchoux

Chaix

Brondeel

Kestens

(2016). Residential buffer, perceived neighborhood, and individual activity space: New refinements in the definition of exposure areas – the record cohort study. Health & Place, 40, 116122. https://doi.org/10.1016/j.healthplace.2016.05.004

30.

Phelan

J. C.

Link

B. G.

Tehranifar

(2010). Social conditions as fundamental causes of health inequalities: Theory, evidence, and policy implications. Journal of Health and Social Behavior, 51(Suppl. 1), S28–S40. https://doi.org/10.1177/0022146510383498

31.

Porter

A. K.

Kohl

H. W.

Pérez

Reininger

Gabriel

K. P.

Salvo

(2020). Bikeability: Assessing the objectively measured environment in relation to recreation and transportation bicycling. Environment and Behavior, 52(8), 861–894. https://doi.org/10.1177/0013916518825289

32.

Porter

A. K.

Salvo

Pérez

Reininger

Kohl

H. W.

III. (2018). Intrapersonal and environmental correlates of bicycling in U.S. adults. American Journal of Preventive Medicine, 54(3), 413–418. https://doi.org/10.1016/j.amepre.2017.10.008

33.

Rachele

J. N.

Sugiyama

Turrell

Healy

A. M.

Sallis

J. F.

(2018). Automobile dependence: A contributing factor to poorer health among lower-income households. Journal of Transport & Health, 8, 123–128. https://doi.org/10.1016/j.jth.2017.11.149

34.

Saelens

B. E.

Sallis

J. F.

Frank

L. D.

(2003). Environmental correlates of walking and cycling: Findings from the transportation, urban design, and planning literatures. Annals of Behavioral Medicine, 25(2), 80–91. https://doi.org/10.1207/s15324796abm2502_03

35.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277(5328), 918–924. https://doi.org/10.1126/science.277.5328.918

36.

Southern California Association of Governments. (2012). SCAG GIS: Open data portal. https://gisdata-scag.opendata.arcgis.com/

37.

United States Census Bureau. (2017). TIGER/Line shapefiles. https://www.census.gov/cgi-bin/geo/shapefiles/index.php

38.

United States Census Bureau. (2018). New census bureau population estimates show Dallas-fort worth-Arlington has largest growth in the United States.

39.

United States Geological Survey. (2016). ScienceBase catalog. https://www.sciencebase.gov/catalog/

40.

Vale

D. S.

Saraiva

Pereira

(2015). Active accessibility: A review of operational measures of walking and cycling accessibility. Journal of Transport and Land Use, 9(1), 209–235.

41.

Williams

D. R.

Collins

(2001). Racial residential segregation: A fundamental cause of racial disparities in health. Public Health Reports, 116(5), 404–416. https://doi.org/10.1093/phr/116.5.404

42.

Wilson

W. J.

(2012). The truly disadvantaged: The inner city, the underclass, and public policy. University of Chicago Press.

43.

Winters

Brauer

Setton

E. M.

Teschke

(2013). Mapping bikeability: A spatial tool to support sustainable travel. Environment and Planning B: Planning and Design, 40(5), 865–883. https://doi.org/10.1068/b38185

44.

Winters

Teschke

Brauer

Fuller

(2016). Bike score^®: Associations between urban bikeability and cycling behavior in 24 cities. International Journal of Behavioral Nutrition and Physical Activity, 13(1), 18. https://doi.org/10.1186/s12966-016-0339-0

45.

C.-Y.

(2014). Environmental supports for walking/biking and traffic safety: Income and ethnicity disparities. Preventive Medicine, 67, 12–16. https://doi.org/10.1016/j.ypmed.2014.06.028

Bikeability Disparities in Orange County,California: Intersection of Place and Demographics

Abstract

Keywords

Introduction

Review of the Literature

Current Study

Methods

Study Area and Data Sources

Dependent Variable: Census Tract Bikeability

Generating Raster Surfaces

Dependent Variable: Bikeability

Bicycle Route Density

Bicycle Route Separation

Connectivity of Bike-Friendly Routes

Destination Density

Topography

Creating Bikeability Index

Independent Variables: Sociodemographic Characteristics

Analyses

Results

Discussion

Conclusion

Footnotes

Acknowledgements

Authors’ Note

Declaration of Conflicting Interests

Funding

ORCID iDs

Data Availability Statement

References