Advertisement

Social and Economic Differences in Neighborhood Walkability Across 500 U.S. Cities

Open AccessPublished:June 06, 2021DOI:https://doi.org/10.1016/j.amepre.2021.03.014

      Introduction

      Neighborhood walkability has been established as a potentially important determinant of various health outcomes that are distributed inequitably by race/ethnicity and sociodemographic status. The objective of this study is to assess the differences in walkability across major urban centers in the U.S.

      Methods

      City- and census tract–level differences in walkability were assessed in 2020 using the 2019 Walk Score across 500 large cities in the U.S.

      Results

      At both geographic levels, high-income and majority White geographic units had the lowest walkability overall. Walkability was lower with increasing tertile of median income among majority White, Latinx, and Asian American and Native Hawaiian and Pacific Islander neighborhoods. However, this association was reversed within majority Black neighborhoods, where tracts in lower-income tertiles had the lowest walkability. Associations varied substantially by region, with the strongest differences observed for cities located in the South.

      Conclusions

      Differences in neighborhood walkability across 500 U.S. cities provide evidence that both geographic unit and region meaningfully influence associations between sociodemographic factors and walkability. Structural interventions to the built environment may improve equity in urban environments, particularly in lower-income majority Black neighborhoods.

      INTRODUCTION

      The prevalence of numerous chronic conditions, including obesity and diabetes, has been increasing in the U.S., and researchers have observed persistent disparities in these conditions by sociodemographic factors, such as race/ethnicity and poverty.
      National Center for Health Statistics (U.S.)
      Health, United States, 2015: with special feature on racial and ethnic health disparities.
      Regular physical activity reduces the risk for these chronic conditions,
      • Pucher J
      • Buehler R
      • Bassett DR
      • Dannenberg AL.
      Walking and cycling to health: a comparative analysis of city, state, and international data.
      and a growing body of research suggests that neighborhood walkability or the extent to which the built environment supports walking is associated both with higher physical activity
      • Sallis JF
      • Cerin E
      • Conway TL
      • et al.
      Physical activity in relation to urban environments in 14 cities worldwide: a cross-sectional study.
      • Kerr J
      • Emond JA
      • Badland H
      • et al.
      Perceived neighborhood environmental attributes associated with walking and cycling for transport among adult residents of 17 cities in 12 countries: the IPEN Study.
      • Rundle A
      • Neckerman KM
      • Freeman L
      • et al.
      Neighborhood food environment and walkability predict obesity in New York City.
      and with lower prevalence of chronic conditions.
      • Rundle A
      • Neckerman KM
      • Freeman L
      • et al.
      Neighborhood food environment and walkability predict obesity in New York City.
      ,
      • Coffee NT
      • Howard N
      • Paquet C
      • Hugo G
      • Daniel M.
      Is walkability associated with a lower cardiometabolic risk?.
      Because walkability is a potentially important health determinant and because there are significant disparities in health outcomes associated with it, researchers have explored whether there are corresponding differences in neighborhood-level characteristics that would influence the walkability of these environments.
      • Neckerman KM
      • Lovasi GS
      • Davies S
      • et al.
      Disparities in urban neighborhood conditions: evidence from GIS measures and field observation in New York City.
      • Kelly CM
      • Schootman M
      • Baker EA
      • Barnidge EK
      • Lemes A.
      The association of sidewalk walkability and physical disorder with area-level race and poverty.

      Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

      • Duncan DT
      • Aldstadt J
      • Whalen J
      • White K
      • Castro MC
      • Williams DR.
      Space, race, and poverty: spatial inequalities in walkable neighborhood amenities?.
      • Bereitschaft B.
      Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities.
      • Thornton CM
      • Conway TL
      • Cain KL
      • et al.
      Disparities in pedestrian streetscape environments by income and race/ethnicity.
      • King KE
      • Clarke PJ.
      A disadvantaged advantage in walkability: findings from socioeconomic and geographical analysis of national built environment data in the United States.
      However, studies that have examined the sociodemographic differences in neighborhood walkability have provided inconsistent results, with some having observed lower walkability levels among high-poverty or predominantly non-White neighborhoods,
      • Neckerman KM
      • Lovasi GS
      • Davies S
      • et al.
      Disparities in urban neighborhood conditions: evidence from GIS measures and field observation in New York City.
      • Kelly CM
      • Schootman M
      • Baker EA
      • Barnidge EK
      • Lemes A.
      The association of sidewalk walkability and physical disorder with area-level race and poverty.

      Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

      • Duncan DT
      • Aldstadt J
      • Whalen J
      • White K
      • Castro MC
      • Williams DR.
      Space, race, and poverty: spatial inequalities in walkable neighborhood amenities?.
      • Bereitschaft B.
      Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities.
      whereas others have observed higher walkability levels among more disadvantaged neighborhoods.

      Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

      ,
      • Thornton CM
      • Conway TL
      • Cain KL
      • et al.
      Disparities in pedestrian streetscape environments by income and race/ethnicity.
      ,
      • King KE
      • Clarke PJ.
      A disadvantaged advantage in walkability: findings from socioeconomic and geographical analysis of national built environment data in the United States.
      These disparate results across studies may in part reflect the lack of consensus on how to measure and define walkability. Studies have used a wide variety of variables to define walkability, with evidence suggesting that both built environmental factors, such as street connectivity or proximity to amenities,
      • Frank LD
      • Kerr J
      • Sallis JF
      • Miles R
      • Chapman J.
      A hierarchy of sociodemographic and environmental correlates of walking and obesity.
      • Duncan M
      • Mummery K.
      Psychosocial and environmental factors associated with physical activity among city dwellers in regional Queensland.
      • Lee C
      • Moudon AV.
      The 3Ds+ R: quantifying land use and urban form correlates of walking.
      • Gauvin L
      • Richard L
      • Craig CL
      • et al.
      From walkability to active living potential: an “ecometric” validation study.
      • Handy S
      • Cao X
      • Mokhtarian PL.
      Self-selection in the relationship between the built environment and walking: empirical evidence from Northern California.
      and social factors, such as perceived safety,
      • Tucker-Seeley RD
      • Subramanian SV
      • Li Y
      • Sorensen G.
      Neighborhood safety, socioeconomic status, and physical activity in older adults.
      ,
      • Hooker SP
      • Wilson DK
      • Griffin SF
      • Ainsworth BE.
      Perceptions of environmental supports for physical activity in African American and white adults in a rural county in South Carolina.
      are associated with higher physical activity. In general, studies using walkability measures that incorporate elements of safety or quality, such as rates of felony complaints or sidewalk unevenness, have observed lower walkability levels among higher-poverty or non-White neighborhoods.
      • Neckerman KM
      • Lovasi GS
      • Davies S
      • et al.
      Disparities in urban neighborhood conditions: evidence from GIS measures and field observation in New York City.
      • Kelly CM
      • Schootman M
      • Baker EA
      • Barnidge EK
      • Lemes A.
      The association of sidewalk walkability and physical disorder with area-level race and poverty.

      Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

      ,
      • Thornton CM
      • Conway TL
      • Cain KL
      • et al.
      Disparities in pedestrian streetscape environments by income and race/ethnicity.
      By contrast, studies that have used accessibility-focused measures that are derived from objective measures of the built environment, such as intersection density and retail floor area ratio, have observed higher walkability levels among higher-poverty or non-White neighborhoods.

      Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

      ,
      • Thornton CM
      • Conway TL
      • Cain KL
      • et al.
      Disparities in pedestrian streetscape environments by income and race/ethnicity.
      ,
      • King KE
      • Clarke PJ.
      A disadvantaged advantage in walkability: findings from socioeconomic and geographical analysis of national built environment data in the United States.
      Walk Score, a proprietary metric quantifying multiple dimensions of the built environment including walking routes, distances to different categories of amenities, population density, road segment lengths, and intersection density,

      Walk Score methodology. Walk Score. https://www.walkscore.com/methodology.shtml. Updated November 19, 2019. Accessed April 15, 2021.

      is one commonly used accessibility-focused measure of walkability that has been validated previously.
      • Carr LJ
      • Dunsiger SI
      • Marcus BH.
      Walk Score™ as a global estimate of neighborhood walkability.
      • Carr LJ
      • Dunsiger SI
      • Marcus BH.
      Validation of Walk Score for estimating access to walkable amenities.
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • Melly SJ
      • Gortmaker SL.
      Validation of walk score for estimating neighborhood walkability: an analysis of four U.S. metropolitan areas.
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • Melly SJ.
      Validation of Walk Scores and Transit Scores for estimating neighborhood walkability and transit availability: a small-area analysis.
      • Duncan DT
      • Méline J
      • Kestens Y
      • et al.
      Walk score, transportation mode choice, and walking among French adults: a GPS, accelerometer, and mobility survey study.
      In addition, most existing studies have been conducted within a single city or have been inconsistent in their selection of geographic levels, leading to a lack of comparability. Previous research has shown that the geographic unit of an analysis can influence the magnitude or direction of the observed spatial association.
      • Duncan DT
      • Kawachi I
      • Subramanian SV
      • Aldstadt J
      • Melly SJ
      • Williams DR.
      Examination of how neighborhood definition influences measurements of youths’ access to tobacco retailers: a methodological note on spatial misclassification.
      • Lee BA
      • Reardon SF
      • Firebaugh G
      • Farrell CR
      • Matthews SA
      • O'Sullivan D
      Beyond the census tract: patterns and determinants of racial segregation at multiple geographic scales.
      • Hameed SM
      • Bell N
      • Schuurman N.
      Analyzing the effects of place on injury: does the choice of geographic scale and zone matter?.
      This issue, known as the modifiable areal unit problem,
      • Fotheringham AS
      • Wong DW.
      The modifiable areal unit problemin multivariate statistical analysis.
      highlights the importance of selecting the appropriate geographic unit for a given analysis. Multiple studies that have explored the differences in walkability across small samples of cities also suggest that the associations may not be uniform across geographic areas.
      • Bereitschaft B.
      Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities.
      ,
      • Thornton CM
      • Conway TL
      • Cain KL
      • et al.
      Disparities in pedestrian streetscape environments by income and race/ethnicity.
      ,
      • Hirsch JA
      • Grengs J
      • Schulz A
      • et al.
      How much are built environments changing, and where?: patterns of change by neighborhood sociodemographic characteristics across seven U.S. metropolitan areas.
      ,
      • Bartzokas-Tsiompras A
      • Tampouraki EM
      • Photis YN.
      Is walkability equally distributed among downtowners? Evaluating the pedestrian streetscapes of eight European capitals using a micro-scale audit approach.
      In particular, a study that compared the associations between social vulnerability and walkability found that differences were highly variable across 3 U.S. cities, with stronger associations observed in the city in the South than in those in the Northeast and West.
      • Bereitschaft B.
      Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities.
      However, with only 3 cities included in this study, a clear pattern by geographic region cannot be discerned.
      Given the limitations of previous research and in light of the remaining unaddressed questions regarding health inequity and walkability, this study seeks to address several limitations of previous studies by:
      • 1.
        using a nationally standardized measure of neighborhood walkability to examine univariable and multivariable associations between sociodemographic factors and walkability across 500 major urban centers in the U.S., allowing for consistent inferences that are representative of all large cities across the nation;
      • 2.
        contrasting the associations observed at the city level versus those observed at the census tract level to understand the potential variations in the associations by geographic unit; and
      • 3.
        comparing the associations observed across different geographic regions of the U.S. to understand the potential variations in the associations by geographic area.
      The primary hypothesis is that larger differences in walkability will be observed when using census tract definitions of neighborhoods because these smaller geographic areas will better capture granular variation in walkability levels within cities. The secondary hypothesis is that the associations will not be uniform across U.S. regions because regions have distinct urban infrastructures, histories, and cultures that could influence these associations.
      • Frey WH
      • Myers D.
      Racial segregation in U.S. metropolitan areas and cities, 1990-2000: patterns, trends, and explanations.
      ,
      • Bereitschaft B
      • Debbage K.
      Regional variations in urban fragmentation among U.S. metropolitan and megapolitan areas.

      METHODS

      Study Sample

      In this cross-sectional study, 2 levels of analysis were incorporated: neighborhoods (defined as census tracts), and cities (defined as census incorporated places, whose boundaries correspond to municipal government jurisdictions). Data were analyzed on the 497 of the most populous cities in the U.S. and the 3 highest population cities in Vermont, West Virginia, and Wyoming as of 2010 to allow for representation from all 50 states.
      • Gourevitch MN
      • Athens JK
      • Levine SE
      • Kleiman N
      • Thorpe LE.
      City-level measures of health, health determinants, and equity to foster population health improvement: the City Health Dashboard.

      Measures

      Walkability was measured at the city and census tract level using the 2019 Walk Score.

      Walk Score methodology. Walk Score. https://www.walkscore.com/methodology.shtml. Updated November 19, 2019. Accessed April 15, 2021.

      ,
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • Melly SJ
      • Gortmaker SL.
      Validation of walk score for estimating neighborhood walkability: an analysis of four U.S. metropolitan areas.
      Walkability data were acquired through a data use agreement between Walk Score and the City Health Dashboard, a free online resource providing 37 public health–related measures for U.S. cities.
      • Gourevitch MN
      • Athens JK
      • Levine SE
      • Kleiman N
      • Thorpe LE.
      City-level measures of health, health determinants, and equity to foster population health improvement: the City Health Dashboard.
      Walk Score considers ratings from 0 to 49 to be car dependent and scores from 50 to 100 to be somewhat walkable to highly walkable.

      Walk Score methodology. Walk Score. https://www.walkscore.com/methodology.shtml. Updated November 19, 2019. Accessed April 15, 2021.

      To calculate city walkability, Walk Score divides cities into blocks of roughly 500 feet on each side and calculates a population density–weighted average for all blocks within the city.

      Walk Score methodology. Walk Score. https://www.walkscore.com/methodology.shtml. Updated November 19, 2019. Accessed April 15, 2021.

      ,
      • Fotheringham AS
      • Wong DW.
      The modifiable areal unit problemin multivariate statistical analysis.
      To calculate neighborhood walkability, Walk Score calculates walkability across the entirety of each census tract, removing places where people do not live.
      • Spoer B
      • Lampe T.
      Updated neighborhood walkability values available for all cities.
      Walkability was analyzed on a continuous scale ranging from 0 to 100.
      Sociodemographic factors explored in these analyses (income and majority race/ethnicity) were measured using the 2018 American Community Survey 5-year estimates. Both income and majority race/ethnicity were measured at the city and census tract level, and these were classified as categorical variables. Income was categorized as low, middle, and high using tertiles of median household income for the geographic area (from Table DP03: Selected Economic Characteristics). Majority race/ethnicity was defined by whether a racial/ethnic group comprised >50% of the total resident population for the geographic area (from Table DP05: American Community Survey Demographic and Housing Estimates). Majority race/ethnicity was categorized as non-Hispanic Asian American (AA) and Native Hawaiian and Pacific Islander (NH&PI), non-Hispanic Black (Black), Hispanic/Latino (Latinx), and non-Hispanic White (White) or None. Owing to small population sizes, areas identified as AA and NH&PI majority included both AA and NH&PI residents in the AA and NH&PI category, which limits the ability to distinguish patterns across heterogeneous Asian subgroups. Areas listed as None majority reflected areas with no racial/ethnic group representing >50% of the resident population. To analyze the potential geographic variation in the associations between sociodemographic factors and walkability, cities were classified into the 4 Census Bureau–designated regions of the U.S.: the Northeast, Midwest, South, and West.
      U.S. Census Bureau
      Census regions and divisions of the United States.

      Statistical Analysis

      Descriptive statistics of walkability by income tertile, majority race/ethnicity, and the region at the city and census tract level were analyzed. Next, univariable associations between the sociodemographic factors and walkability at the city level using linear regression models were assessed. Because walkability and sociodemographic factors are likely to vary across neighborhoods within a given city, univariable associations at the census tract level were assessed using random-intercept multilevel linear regression models to account for the clustering of census tracts within cities. Using census tract–level data, adjusted associations were assessed using a random-intercept multilevel linear regression model, including terms for income level, majority race/ethnicity, and the interaction of income level and majority race/ethnicity (F[828,492]=13.3, p<0.01). This multivariable model was replicated, including terms for the interaction of income level and U.S. region (F[628,096]=40.6, p<0.01) and the interaction of majority race/ethnicity and U.S. region (F[928,181]=6.7, p<0.01), to assess whether there was significant regional variation in these associations. Regional models excluded AA and NH&PI majority census tracts given small cell sizes. Census tracts that were missing data were excluded from these analyses. All analyses were conducted using SAS, version 9.4.2.

      RESULTS

      Of the 500 cities included in this study, 93 (18.6%) were located in the Midwest, 54 (10.8%) were located in the Northeast, 158 (31.6%) were located in the South, and 195 (39.0%) were located in the West. Of the 29,305 census tracts within these cities, 115 were missing Walk Score data, 203 were missing majority race/ethnicity data, and 380 were missing income data. As a result, 28,130 census tracts (98%) were included in the final sample, with an average number of census tracts per city of 62.1 in the Midwest, 78.7 in the Northeast, 61.5 in the South, and 50.3 in the West.
      Cities within the Northeast had, on average, higher city Walk Scores than those in the other regions of the U.S. Cities located within the South had lower variability in city Walk Scores, with >90% of Southern cities classified as car dependent (scores from 0 to 49). There was an inverse relationship between income and mean Walk Score, with low-income cities having the highest average Walk Score and high-income cities having the lowest average Walk Score (Table 1). This inverse relationship between income level and Walk Score was more pronounced at the neighborhood level than at the city level, with low-income neighborhoods having an average Walk Score of 55.5 (SD=22.6) compared with 40.6 (SD=27.9) for high-income neighborhoods. On average, majority AA and NH&PI cities had the highest average Walk Score, whereas majority White cities had the lowest average Walk Score at both the city and neighborhood level.
      Table 1Mean Walk Score by Income Level, Majority Race/Ethnicity, and Region at the City- and Neighborhood-Level for 500 U.S. Cities in 2019
      City levelNeighborhood level
      MetricsnMean (SD)nMean (SD)
      Total50042.9 (15.79)28,79147.6 (26.2)
      Income
       Low16644.2 (16.0)9,63255.5 (22.6)
       Mid16742.6 (15.4)9,62846.6 (25.8)
       High16741.8 (16.3)9,63140.6 (27.9)
      Majority race/ethnicity
       AA and NH&PI657.3 (9.7)56261.4 (25.7)
       Black2341.5 (14.7)3,92955.9 (23.3)
       Latino6350.1 (17.8)4,94655.9 (23.3)
       White24539.0 (13.8)13,60041.2 (26.6)
       None16345.6 (16.7)5,85449.7 (25.6)
      Region
       Midwest9340.5 (12.2)5,52747.2 (23.9)
       Northeast5465.6 (13.3)4,25078.7 (18.8)
       South15834.4 (11.8)9,46436.6 (22.5)
       West19544.7 (14.4)9,65044.9 (23.0)
      AA, Asian American; NH&PI, Native Hawaiian and Pacific Islander.
      When neighborhoods were stratified by majority race/ethnicity, the inverse relationship between income level and Walk Score was observed for majority AA and NH&PI, White, and Latinx neighborhoods and for neighborhoods with no racial/ethnic majority, but this relationship was reversed for majority Black neighborhoods. High-income, majority Black neighborhoods had an average Walk Score of 67.7 (SD=22.5) compared with 53.1 (SD=21.9) for low-income, majority Black neighborhoods (Figure 1).
      Figure thumbnail gr1
      Figure 1Average neighborhood-level walkability by income and majority race/ethnicity for 500 U.S. cities in 2019.
      In the univariable city-level model, there were minor differences in city walkability by income (Table 2). Majority White cities had the lowest estimated Walk Score. Majority AA and NH&PI cities had a mean Walk Score that was 18.3 units higher than that of majority White cities (95% CI=5.9, 30.8), and majority Latinx cities had a mean Walk Score that was 11.1 units higher than that of majority White cities (95% CI=6.8, 15.3). Differences between majority Black and majority White cities were minor (β=2.5, 95% CI= −4.1, 9.0).
      Table 2City- and Neighborhood-Level Model Coefficients and 95% CIs From Univariable Regression Models for 500 U.S. Cities in 2019
      MetricsCity levelNeighborhood level
      Income
       Low2.4 (−1.0, 5.8)18.2 (17.7, 18.8)
       Mid0.8 (−2.6, 4.2)9.0 (8.5, 9.5)
       Highrefref
      Majority race/ethnicity
       AA and NH&PI18.3 (5.9, 30.8)6.5 (4.7, 8.2)
       Black2.5 (−4.1, 9.0)2.4 (1.7, 3.1)
       Latino11.1 (6.8, 15.3)11.1 (10.4, 11.9)
       Whiterefref
       None6.6 (3.6, 9.7)4.7 (4.1, 5.3)
      Region
       Midwest−4.2 (−7.4, −0.9)−2.1 (−5.4, 1.2)
       Northeast20.9 (16.9, 24.9)25.7 (21.7, 29.7)
       South−10.3 (−13.1, −7.5)−7.8 (−10.6, −5.0)
       Westrefref
      AA, Asian American; NH&PI, Native Hawaiian and Pacific Islander.
      In the univariable tract-level model with random intercepts for cities, there was a more pronounced inverse association between Walk Score and median household income (Table 2). Low-income neighborhoods had an 18.2-unit higher mean Walk Score than high-income neighborhoods (95% CI=17.7, 18.8), and middle-income neighborhoods had a 9.0-unit higher mean Walk Score than high-income neighborhoods (95% CI=8.5, 9.5). Majority Latinx neighborhoods had the highest estimated neighborhood Walk Score (48.4, 95% CI=46.9, 49.8) than other racial/ethnic groups. Majority White and majority Black neighborhoods had the lowest mean Walk Score (White: 37.2, 95% CI=35.8, 38.6; Black: 39.6, 95% CI=38.1, 41.1).
      In the multilevel, tract-level model, the inverse relationship between income level and Walk Score was observed within all racial/ethnic groups. The largest differences by income were observed for majority AA and NH&PI and majority White neighborhoods, with low-income neighborhoods having a mean Walk Score 23 units higher than high-income neighborhoods. The smallest differences by income were observed for majority Black neighborhoods, with low-income neighborhoods having a mean Walk Score 13.4 units higher than high-income neighborhoods.
      Within each tertile of neighborhood median income, majority Black neighborhoods had lower mean Walk Scores than other majority racial/ethnic neighborhoods. For low-income neighborhoods, majority Black neighborhoods had an estimated 10.6-unit lower mean Walk Score than majority White neighborhoods. For high-income neighborhoods, majority Black neighborhoods had an estimated 1.5-unit lower mean Walk Score than majority White neighborhoods. Low-income majority Latinx neighborhoods also had lower mean Walk Scores than low-income majority White neighborhoods (White/low-income: 54.3, 95% CI=52.7, 55.9; Latinx/low-income: 51.8, 95% CI=50.2, 53.3).
      The magnitude of the associations between these sociodemographic factors and walkability varied by U.S. region. Cities within the South had the largest differences in estimated Walk Score by majority race/ethnicity, and cities within the South and West had the largest differences by income level. Cities within the Northeast had the smallest differences in estimated Walk Scores by both majority race/ethnicity and income (Figure 2).
      Figure 2
      Figure 2Neighborhood-level Walk Score estimates by income, majority race/ethnicity, and region of the U.S. for 500 U.S. cities in 2019, multivariable linear regression.

      DISCUSSION

      In this study, cross-sectional city- and census tract–level data were used to describe the differences in neighborhood walkability by income, racial/ethnic composition, and geographic region of the U.S. This is the first study that examines urban walkability at this scale across 500 cities from all regions of the U.S. Large differences were observed in neighborhood walkability by both income level and majority race/ethnicity across the regions. In particular, an inverse relationship was found between income level and walkability, with low-income neighborhoods associated with higher average Walk Scores than high-income neighborhoods. Overall, majority White neighborhoods had, on average, lower Walk Scores than other racial/ethnic majority neighborhoods or neighborhoods with no racial/ethnic majority.
      Evidence suggests that people whose incomes are below the federal poverty threshold tend to live in dense urban areas that are more walkable. One practical reason for this may be to have access to public transit owing to the costs of car ownership.
      • Rappaport J
      • Glaeser EL
      • Kahn ME.
      Why do the poor live in cities?.
      Indeed, there are large racial gaps in car ownership owing to reduced access to resources.
      • Raphael S
      • Stoll MA
      • Small KA
      • Winston C.
      Can boosting minority car-ownership rates narrow inter-racial employment gaps?[with comments].
      This may contribute to the overall higher walkability of low-income and non-White majority neighborhoods. However, within majority Black neighborhoods, a reversed association was observed between income and walkability compared with the relationship within other racial/ethnic majority neighborhoods where tracts in low-income neighborhoods had the lowest walkability. This association may be due to the geographic distribution of high-income majority Black neighborhoods because the association reversed after accounting for city and regional effects in the multilevel model (i.e., in adjusted models, low-income neighborhoods were also associated with higher walkability in majority Black neighborhoods). In addition, after accounting for these effects, both majority Black and Latinx low-income neighborhoods had estimated Walk Scores that were lower than those of majority White low-income neighborhoods.
      To explore how the associations are influenced by geographic unit and place, city-level to census tract–level models were compared, and regional interaction terms were included in the models. In these analyses, city-level models showed no significant association between income level and walkability. These results highlight the importance of the modifiable areal unit problem, specifically that larger geographic aggregations may cause spatial smoothing that reduces variation and unpredictably biases the results of multiple regression analyses.
      • Fotheringham AS
      • Wong DW.
      The modifiable areal unit problemin multivariate statistical analysis.
      Therefore, smaller geographic areas may provide a better representation of neighborhoods than city or county boundaries, particularly because many community-level variables vary significantly within these larger geographic areas.
      • Bustamante-Zamora D
      • Maizlish N.
      Cross-sectional analysis of two social determinants of health in California cities: racial/ethnic and geographic disparities.
      In addition, the magnitude of the associations varied significantly by U.S. region, with cities in the Northeast having the weakest association between walkability and these sociodemographic factors and cities in the South and West having the strongest associations. These differences in associations by geographic region may support previous studies that found stronger differences in walkability for a city in the South
      • Bereitschaft B.
      Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities.
      and found no significant differences for a city located in the Northeast.
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • White K
      • Castro MC
      • Williams DR.
      Space, race, and poverty: spatial inequalities in walkable neighborhood amenities?.
      Regions of the U.S. generally correspond with historical periods in urban development and immigration, which may contribute to the differences in social composition and street infrastructure by region.
      • Frey WH
      • Myers D.
      Racial segregation in U.S. metropolitan areas and cities, 1990-2000: patterns, trends, and explanations.
      ,
      • Bereitschaft B
      • Debbage K.
      Regional variations in urban fragmentation among U.S. metropolitan and megapolitan areas.
      However, further research is needed to understand the underlying reason for these regional differences in sociodemographic disparities in walkability.

      Limitations

      This study leverages Walk Score to measure walkability, which has the strengths of being an established measure of walkability that is regularly used in the literature and is standardized across the nation.
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • Melly SJ
      • Gortmaker SL.
      Validation of walk score for estimating neighborhood walkability: an analysis of four U.S. metropolitan areas.
      However, there are limitations to this measure that may influence the results. Walk Score is an accessibility-focused measure that incorporates distances to amenities, intersection density, block length, land use, and walking routes into their index.

      Walk Score methodology. Walk Score. https://www.walkscore.com/methodology.shtml. Updated November 19, 2019. Accessed April 15, 2021.

      However, this index does not include the physical condition or safety elements, such as sidewalks or lighting, which would influence a pedestrian's willingness or ability to walk in a neighborhood.
      • Tucker-Seeley RD
      • Subramanian SV
      • Li Y
      • Sorensen G.
      Neighborhood safety, socioeconomic status, and physical activity in older adults.
      ,
      • Hooker SP
      • Wilson DK
      • Griffin SF
      • Ainsworth BE.
      Perceptions of environmental supports for physical activity in African American and white adults in a rural county in South Carolina.
      ,
      • Spence JC
      • Plotnikoff RC
      • Rovniak LS
      • Martin Ginis KA
      • Rodgers W
      • Lear SA
      Perceived neighbourhood correlates of walking among participants visiting the Canada on the Move website.
      As seen in previous studies, the use of accessibility-focused walkability measures has generally provided researchers with the counterintuitive result of higher walkability scores in lower-income neighborhoods.

      Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

      ,
      • Thornton CM
      • Conway TL
      • Cain KL
      • et al.
      Disparities in pedestrian streetscape environments by income and race/ethnicity.
      ,
      • King KE
      • Clarke PJ.
      A disadvantaged advantage in walkability: findings from socioeconomic and geographical analysis of national built environment data in the United States.
      The results were consistent with these previous findings. In many cities, lower-income neighborhoods may be located in close proximity to downtown areas that have higher intersection densities, whereas higher-income neighborhoods may be located in outer portions of the city that have a more suburban infrastructure.
      • Madden JF.
      Changes in the distribution of poverty across and within the U.S. metropolitan areas, 1979-89.
      ,
      • Cooke T
      • Marchant S.
      The changing intrametropolitan location of high-poverty neighbourhoods in the U.S., 1990-2000.
      Accessibility-focused walkability measures would be greatly influenced by these infrastructure differences and may overlook the differences in quality or aesthetics that would be important to the neighborhood's residents. Although objectively built environment factors are important indicators of walkability,
      • Frank LD
      • Kerr J
      • Sallis JF
      • Miles R
      • Chapman J.
      A hierarchy of sociodemographic and environmental correlates of walking and obesity.
      • Duncan M
      • Mummery K.
      Psychosocial and environmental factors associated with physical activity among city dwellers in regional Queensland.
      • Lee C
      • Moudon AV.
      The 3Ds+ R: quantifying land use and urban form correlates of walking.
      • Gauvin L
      • Richard L
      • Craig CL
      • et al.
      From walkability to active living potential: an “ecometric” validation study.
      • Handy S
      • Cao X
      • Mokhtarian PL.
      Self-selection in the relationship between the built environment and walking: empirical evidence from Northern California.
      an improved walkability measure would also account for the physical condition of roads and sidewalks or crime because these factors have been shown to influence pedestrian walking habits.
      • Tucker-Seeley RD
      • Subramanian SV
      • Li Y
      • Sorensen G.
      Neighborhood safety, socioeconomic status, and physical activity in older adults.
      ,
      • Hooker SP
      • Wilson DK
      • Griffin SF
      • Ainsworth BE.
      Perceptions of environmental supports for physical activity in African American and white adults in a rural county in South Carolina.
      ,
      • Spence JC
      • Plotnikoff RC
      • Rovniak LS
      • Martin Ginis KA
      • Rodgers W
      • Lear SA
      Perceived neighbourhood correlates of walking among participants visiting the Canada on the Move website.
      Another key limitation of this study was that census tracts were the smallest geography used for defining neighborhoods in the available walkability data. Census tracts represent administrative boundaries of approximately 2,500–8,000 people and are commonly used to serve as neighborhood proxies.
      • Neckerman KM
      • Lovasi GS
      • Davies S
      • et al.
      Disparities in urban neighborhood conditions: evidence from GIS measures and field observation in New York City.
      ,
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • White K
      • Castro MC
      • Williams DR.
      Space, race, and poverty: spatial inequalities in walkable neighborhood amenities?.
      ,
      • King KE
      • Clarke PJ.
      A disadvantaged advantage in walkability: findings from socioeconomic and geographical analysis of national built environment data in the United States.
      However, research suggests that census tracts may not always align with natural neighborhoods
      • Sperling J.
      The tyranny of census geography: small-area data and neighborhood statistics.
      and may represent too large of geographic areas for walkability studies, particularly for less densely populated areas.
      • Duncan DT
      • Aldstadt J
      • Whalen J
      • Melly SJ
      • Gortmaker SL.
      Validation of walk score for estimating neighborhood walkability: an analysis of four U.S. metropolitan areas.
      Refining definitions of neighborhood boundaries or using smaller administrative units (i.e., census blocks or address-based Walk Score) may improve the validity of these results.

      CONCLUSIONS

      Significant differences in neighborhood walkability across 500 U.S. cities were observed, and evidence was suggestive that both geographic unit and geographic place meaningfully influence the associations between sociodemographic factors and walkability. City leaders increasingly aspire to enact data-driven policies to improve equity in their jurisdictions. Although structural interventions to the built environment, such as modifying street layouts, may be difficult to change, interventions such as Open Streets
      • Kuhlberg JA
      • Hipp JA
      • Eyler A
      • Chang G.
      Open streets initiatives in the United States: closed to traffic, open to physical activity.
      or changes in zoning codes may prove valuable and attainable public health solutions for addressing health disparities related to worse neighborhood environments.

      Take action. City health Dashboard, NYU Langone Health. https://www.cityhealthdashboard.com/take-action. Updated November 19, 2019. Accessed April 15, 2021.

      City leaders and policymakers can leverage neighborhood walkability data on the City Health Dashboard website to better plan and allocate resources for these sorts of interventions.
      • Gourevitch MN
      • Athens JK
      • Levine SE
      • Kleiman N
      • Thorpe LE.
      City-level measures of health, health determinants, and equity to foster population health improvement: the City Health Dashboard.
      On the basis of these observations, future studies should use small geographic areas when analyzing walkability measures. It is also warranted to explore the inclusion of elements such as safety and the condition of sidewalks and the examination of how car ownership influences the relationship between sociodemographic factors and walkability.

      ACKNOWLEDGMENTS

      The authors would like to thank Miriam Gofine, Taylor Lampe, and Pei Yang Hsieh on the City Health Dashboard team for their contributions to this work.
      This work was supported by the Robert Wood Johnson Foundation (Grant Number 794S43).
      No financial disclosures were reported by the authors of this paper.

      REFERENCES

        • National Center for Health Statistics (U.S.)
        Health, United States, 2015: with special feature on racial and ethnic health disparities.
        National Center for Health Statistics, Hyattsville, MD2016
        • Pucher J
        • Buehler R
        • Bassett DR
        • Dannenberg AL.
        Walking and cycling to health: a comparative analysis of city, state, and international data.
        Am J Public Health. 2010; 100: 1986-1992https://doi.org/10.2105/AJPH.2009.189324
        • Sallis JF
        • Cerin E
        • Conway TL
        • et al.
        Physical activity in relation to urban environments in 14 cities worldwide: a cross-sectional study.
        Lancet. 2016; 387 ([published correction appears in Lancet. 2016;387(10034):2198]): 2207-2217https://doi.org/10.1016/S0140-6736(15)01284-2
        • Kerr J
        • Emond JA
        • Badland H
        • et al.
        Perceived neighborhood environmental attributes associated with walking and cycling for transport among adult residents of 17 cities in 12 countries: the IPEN Study.
        Environ Health Perspect. 2016; 124: 290-298https://doi.org/10.1289/ehp.1409466
        • Rundle A
        • Neckerman KM
        • Freeman L
        • et al.
        Neighborhood food environment and walkability predict obesity in New York City.
        Environ Health Perspect. 2009; 117: 442-447https://doi.org/10.1289/ehp.11590
        • Coffee NT
        • Howard N
        • Paquet C
        • Hugo G
        • Daniel M.
        Is walkability associated with a lower cardiometabolic risk?.
        Health Place. 2013; 21: 163-169https://doi.org/10.1016/j.healthplace.2013.01.009
        • Neckerman KM
        • Lovasi GS
        • Davies S
        • et al.
        Disparities in urban neighborhood conditions: evidence from GIS measures and field observation in New York City.
        J Public Health Policy. 2009; 30 (suppl 1): S264-S285https://doi.org/10.1057/jphp.2008.47
        • Kelly CM
        • Schootman M
        • Baker EA
        • Barnidge EK
        • Lemes A.
        The association of sidewalk walkability and physical disorder with area-level race and poverty.
        J Epidemiol Community Health. 2007; 61: 978-983https://doi.org/10.1136/jech.2006.054775
      1. Franzini L, Taylor W, Elliott MN, et al. Neighborhood characteristics favorable to outdoor physical activity: disparities by socioeconomic and racial/ethnic composition. Health Place. 2010;16(2):267‒274. https://doi.org/10.1016/j.healthplace.2009.10.009.

        • Duncan DT
        • Aldstadt J
        • Whalen J
        • White K
        • Castro MC
        • Williams DR.
        Space, race, and poverty: spatial inequalities in walkable neighborhood amenities?.
        Demogr Res. 2012; 26: 409-448https://doi.org/10.4054/DemRes.2012.26.17
        • Bereitschaft B.
        Equity in neighbourhood walkability? A comparative analysis of three large U.S. cities.
        Local Environ. 2017; 22: 859-879https://doi.org/10.1080/13549839.2017.1297390
        • Thornton CM
        • Conway TL
        • Cain KL
        • et al.
        Disparities in pedestrian streetscape environments by income and race/ethnicity.
        SSM Popul Health. 2016; 2: 206-216https://doi.org/10.1016/j.ssmph.2016.03.004
        • King KE
        • Clarke PJ.
        A disadvantaged advantage in walkability: findings from socioeconomic and geographical analysis of national built environment data in the United States.
        Am J Epidemiol. 2015; 181: 17-25https://doi.org/10.1093/aje/kwu310
        • Frank LD
        • Kerr J
        • Sallis JF
        • Miles R
        • Chapman J.
        A hierarchy of sociodemographic and environmental correlates of walking and obesity.
        Prev Med. 2008; 47: 172-178https://doi.org/10.1016/j.ypmed.2008.04.004
        • Duncan M
        • Mummery K.
        Psychosocial and environmental factors associated with physical activity among city dwellers in regional Queensland.
        Prev Med. 2005; 40: 363-372https://doi.org/10.1016/j.ypmed.2004.06.017
        • Lee C
        • Moudon AV.
        The 3Ds+ R: quantifying land use and urban form correlates of walking.
        Transp Res D Transp Environ. 2006; 11: 204-215https://doi.org/10.1016/j.trd.2006.02.003
        • Gauvin L
        • Richard L
        • Craig CL
        • et al.
        From walkability to active living potential: an “ecometric” validation study.
        Am J Prev Med. 2005; 28 (suppl 2): 126-133https://doi.org/10.1016/j.amepre.2004.10.029
        • Handy S
        • Cao X
        • Mokhtarian PL.
        Self-selection in the relationship between the built environment and walking: empirical evidence from Northern California.
        J Am Plann Assoc. 2006; 72: 55-74https://doi.org/10.1080/01944360608976724
        • Tucker-Seeley RD
        • Subramanian SV
        • Li Y
        • Sorensen G.
        Neighborhood safety, socioeconomic status, and physical activity in older adults.
        Am J Prev Med. 2009; 37: 207-213https://doi.org/10.1016/j.amepre.2009.06.005
        • Hooker SP
        • Wilson DK
        • Griffin SF
        • Ainsworth BE.
        Perceptions of environmental supports for physical activity in African American and white adults in a rural county in South Carolina.
        Prev Chronic Dis. 2005; 2: A11
        https://pubmed.ncbi.nlm.nih.gov/16164815/
        Date accessed: April 3, 2019
      2. Walk Score methodology. Walk Score. https://www.walkscore.com/methodology.shtml. Updated November 19, 2019. Accessed April 15, 2021.

        • Carr LJ
        • Dunsiger SI
        • Marcus BH.
        Walk Score™ as a global estimate of neighborhood walkability.
        Am J Prev Med. 2010; 39: 460-463https://doi.org/10.1016/j.amepre.2010.07.007
        • Carr LJ
        • Dunsiger SI
        • Marcus BH.
        Validation of Walk Score for estimating access to walkable amenities.
        Br J Sports Med. 2011; 45: 1144-1148https://doi.org/10.1136/bjsm.2009.069609
        • Duncan DT
        • Aldstadt J
        • Whalen J
        • Melly SJ
        • Gortmaker SL.
        Validation of walk score for estimating neighborhood walkability: an analysis of four U.S. metropolitan areas.
        Int J Environ Res Public Health. 2011; 8: 4160-4179https://doi.org/10.3390/ijerph8114160
        • Duncan DT
        • Aldstadt J
        • Whalen J
        • Melly SJ.
        Validation of Walk Scores and Transit Scores for estimating neighborhood walkability and transit availability: a small-area analysis.
        GeoJournal. 2013; 78: 407-416https://doi.org/10.1007/s10708-011-9444-4
        • Duncan DT
        • Méline J
        • Kestens Y
        • et al.
        Walk score, transportation mode choice, and walking among French adults: a GPS, accelerometer, and mobility survey study.
        Int J Environ Res Public Health. 2016; 13: 611https://doi.org/10.3390/ijerph13060611
        • Duncan DT
        • Kawachi I
        • Subramanian SV
        • Aldstadt J
        • Melly SJ
        • Williams DR.
        Examination of how neighborhood definition influences measurements of youths’ access to tobacco retailers: a methodological note on spatial misclassification.
        Am J Epidemiol. 2014; 179: 373-381https://doi.org/10.1093/aje/kwt251
        • Lee BA
        • Reardon SF
        • Firebaugh G
        • Farrell CR
        • Matthews SA
        • O'Sullivan D
        Beyond the census tract: patterns and determinants of racial segregation at multiple geographic scales.
        Am Sociol Rev. 2008; 73: 766-791https://doi.org/10.1177/000312240807300504
        • Hameed SM
        • Bell N
        • Schuurman N.
        Analyzing the effects of place on injury: does the choice of geographic scale and zone matter?.
        Open Med. 2010; 4: e171-e180
        • Fotheringham AS
        • Wong DW.
        The modifiable areal unit problemin multivariate statistical analysis.
        Environ Plann A. 1991; 23: 1025-1044https://doi.org/10.1068/a231025
        • Hirsch JA
        • Grengs J
        • Schulz A
        • et al.
        How much are built environments changing, and where?: patterns of change by neighborhood sociodemographic characteristics across seven U.S. metropolitan areas.
        Soc Sci Med. 2016; 169: 97-105https://doi.org/10.1016/j.socscimed.2016.09.032
        • Bartzokas-Tsiompras A
        • Tampouraki EM
        • Photis YN.
        Is walkability equally distributed among downtowners? Evaluating the pedestrian streetscapes of eight European capitals using a micro-scale audit approach.
        Int J Transp Dev Intergr. 2020; 4: 75-92https://doi.org/10.2495/TDI-V4-N1-75-92
        • Frey WH
        • Myers D.
        Racial segregation in U.S. metropolitan areas and cities, 1990-2000: patterns, trends, and explanations.
        Population Studies Center, University of Michigan, Institute for Social Research, Ann ArborApril 2005 (MI) (PublishedAccessed April 15, 2021)
        • Bereitschaft B
        • Debbage K.
        Regional variations in urban fragmentation among U.S. metropolitan and megapolitan areas.
        Appl Spat Anal Policy. 2014; 7: 119-147https://doi.org/10.1007/s12061-013-9092-9
        • Gourevitch MN
        • Athens JK
        • Levine SE
        • Kleiman N
        • Thorpe LE.
        City-level measures of health, health determinants, and equity to foster population health improvement: the City Health Dashboard.
        Am J Public Health. 2019; 109: 585-592https://doi.org/10.2105/AJPH.2018.304903
        • Spoer B
        • Lampe T.
        Updated neighborhood walkability values available for all cities.
        City Health Dashboard, NYU Langone Health, New York, NYMay 27, 2020 (PublishedAccessed July 17, 2020)
        • U.S. Census Bureau
        Census regions and divisions of the United States.
        U.S. Census Bureau, Suitland-Silver Hill, MD2010
        • Rappaport J
        • Glaeser EL
        • Kahn ME.
        Why do the poor live in cities?.
        NBER Work Pap Ser. 2000; : 35https://doi.org/10.3386/w7636
        • Raphael S
        • Stoll MA
        • Small KA
        • Winston C.
        Can boosting minority car-ownership rates narrow inter-racial employment gaps?[with comments].
        Brookings Wharton Pap Urban Aff. 2001; : 99-145https://doi.org/10.1353/urb.2001.0013
        • Bustamante-Zamora D
        • Maizlish N.
        Cross-sectional analysis of two social determinants of health in California cities: racial/ethnic and geographic disparities.
        BMJ Open. 2017; 7e013975https://doi.org/10.1136/bmjopen-2016-013975
        • Spence JC
        • Plotnikoff RC
        • Rovniak LS
        • Martin Ginis KA
        • Rodgers W
        • Lear SA
        Perceived neighbourhood correlates of walking among participants visiting the Canada on the Move website.
        Can J Public Health. 2006; 97: S39-S44https://doi.org/10.1007/BF03405363
        • Madden JF.
        Changes in the distribution of poverty across and within the U.S. metropolitan areas, 1979-89.
        Urban Stud. 1996; 33: 1581-1600https://doi.org/10.1080/0042098966510
        • Cooke T
        • Marchant S.
        The changing intrametropolitan location of high-poverty neighbourhoods in the U.S., 1990-2000.
        Urban Stud. 2006; 43: 1971-1989https://doi.org/10.1080/00420980600897818
        • Sperling J.
        The tyranny of census geography: small-area data and neighborhood statistics.
        Cityscape. 2012; 14: 219-223
        • Kuhlberg JA
        • Hipp JA
        • Eyler A
        • Chang G.
        Open streets initiatives in the United States: closed to traffic, open to physical activity.
        J Phys Act Health. 2014; 11: 1468-1474https://doi.org/10.1123/jpah.2012-0376
      3. Take action. City health Dashboard, NYU Langone Health. https://www.cityhealthdashboard.com/take-action. Updated November 19, 2019. Accessed April 15, 2021.