Income disparities in perceived neighborhood built and social environment attributes

Health & Place 17 (2011) 1274–1283

Contents lists available at ScienceDirect

Health & Place journal homepage: www.elsevier.com/locate/healthplace

Income disparities in perceived neighborhood built and social environment attributes James F. Sallis a,n, Donald J. Slymen a, Terry L. Conway a, Lawrence D. Frank b,c, Brian E. Saelens d, Kelli Cain a, James E. Chapman c a

Department of Psychology, San Diego State University, 3900 Fifth Avenue, Suite 310, San Diego, CA 92103, USA School of Community and Regional Planning, University of British Columbia, 231-1933 West Mall, Vancouver, Canada BC V6T 1Z2 c Urban Design for Health, P.O. Box 85508, Seattle, WA 98145, USA d Seattle Children’s Hospital & the University of Washington, UW-CHI NE 74th St, Box 354920, Seattle, WA 98101, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 3 May 2010 Received in revised form 13 February 2011 Accepted 24 February 2011 Available online 9 March 2011

The present study explored whether perceived neighborhood environmental attributes associated with physical activity differ by neighborhood income. Adults aged 20–65 years (n ¼ 2199; 48% female; mean age ¼ 45 years; 26% ethnic minority) were recruited from 32 neighborhoods from the Seattle, WA and Baltimore, MD regions that varied in objectively measured walkability and neighborhood income. Perceived built and social environment variables were assessed with the Neighborhood Environment Walkability Scale. There were neighborhood income disparities on 10 of 15 variables. Residents from high-income neighborhoods reported more favorable esthetics, pedestrian/biking facilities, safety from traffic, safety from crime, and access to recreation facilities than residents of low-income areas (all p’s o 0.001). Low-income neighborhoods may lack amenities and safety attributes that can facilitate high levels of physical activity for both transportation and recreation purposes. & 2011 Published by Elsevier Ltd.

Keywords: Built environment Physical activity Obesity Health disparities Environmental justice

1. Introduction Multiple reviews have documented consistent associations of multiple attributes of the built environment, especially neighborhood walkability (defined by residential density, proximity of shops and services, and street connectivity) and proximity to parks and recreation facilities, with physical activity for transportation and recreation purposes (Bauman and Bull, 2007; Gebel et al., 2007; Frank et al., 2005; Owen et al., 2004; Saelens and Handy 2008; Transportation Research Board–Institute of Medicine, 2005). These neighborhood characteristics also have been related to obesity (Black and Mackinto, 2007; Papas et al., 2007). However, inconsistent associations of walkability with physical activity and obesity across gender, racial, and income groups raise questions about the generalizability of findings. For example, an Atlanta region study found associations between neighborhood walkability and physical activity (Frank et al., 2005) and overweight/obesity (Frank et al., 2004) for non-Hispanic whites but not for African

n

Corresponding author. Tel.: þ1 619 260 5534; fax: þ1 619 260 1510. E-mail addresses: [email protected] (J.F. Sallis), [email protected] (D.J. Slymen), [email protected] (T.L. Conway), [email protected] (L.D. Frank), [email protected] (B.E. Saelens), [email protected] (K. Cain), [email protected] (J.E. Chapman). 1353-8292/$ - see front matter & 2011 Published by Elsevier Ltd. doi:10.1016/j.healthplace.2011.02.006

Americans. Further analyses suggested that walkability ranged from being the most powerful variable predicting walking among white men, to among the least significant factors in explaining walking for lower income and non-white residents (Frank et al., 2008). A study in New York City found similar inconsistencies among low-income, low-education, and non-white subgroups (Lovasi et al., 2009b). In contrast, a study in two regions of the U.S. reported associations of walkability with physical activity and overweight/obesity did not differ by income group (Sallis et al., 2009). In yet another study, access to parks and recreation facilities was positively related to physical activity among African Americans and Hispanics but not among non-Hispanic whites (Diez Roux et al., 2007). Thus, further study is needed to determine how built environment attributes may support physical activity in a variety of subgroups, particularly those shown to suffer from health disparities (LaVeist, 2005). An Australian study found neighborhood walkability partially explained income disparities in walking for transportation (Cerin et al., 2009a,b). Identifying people in high-risk sociodemographic groups who also live in the least health-promoting environments provides a way to geographically focus scarce resources on those most in need. It could be that population subgroups are differentially responsive to built environment impacts. Alternatively, lack of amenities such as sidewalks and crosswalks may interact with social factors such as fear of crime and create barriers that reduce

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

potential benefits from activity-supportive neighborhoods. Evidence is mounting that physical activity-supportive environmental attributes are not equitably distributed. It is important to consider the degree to which specific environmental attributes are modifiable and the amount of time required for change to occur. It is essential to know which ‘‘policy levers’’ are most effective to change specific aspects of the built environment. Bronfenbrenner (1979) identified three levels of social environments, with ‘‘micro’’ referring to interactions in specific settings, such as with work groups; ‘‘meso’’ referring to interactions among settings, such as family, school, and work; and ‘‘macro’’ or ‘‘exo’’ referring to the larger social system such as economic forces and cultural values. These terms also have been used to categorize built environment characteristics (Bauman and Bull, 2007; McMillan et al., 2010). For current purposes, ‘‘macro’’ refers to elements of overall community design related to walkability. These attributes of street connectivity, residential density, and mixed land use reflect land use and transportation policies, and with a few exceptions they are difficult to change quickly. ‘‘Macro’’ variables include access to specific land uses including common destinations such as retail and food stores as well as leisure-related uses like parks and private recreation facilities. ‘‘Micro’’ refers here to built environment factors that represent details that are smaller in scale and generally changeable more rapidly and with less cost, such as pedestrian/cycling facilities, street-crossing characteristics, traffic volume and speed, crime, incivilities like graffiti. We also classify social environment characteristics like traffic volume and speed, crime, and incivilities like graffiti as ‘‘micro’’ variables because they refer to characteristics of specific neighborhoods rather than the larger society. Lower income communities have less disposable income to support local shops, services, and restaurants. Therefore the breadth and depth of destinations, including food stores and restaurants, is often related to sociodemographic factors (Frank et al., 2009; Lovasi et al., 2009a). A similar pattern of disparities has been found in which public and private recreation facilities are generally less common in low-income and racial/ethnic minority communities (Estabrooks et al., 2003; Giles-Corti and Donovan, 2002; Gordon-Larsen et al., 2006; Moore et al., 2008; Powell et al., 2006), though there are some exceptions (Abercrombie et al., 2008). There is growing evidence that disadvantaged groups have less-favorable ‘‘micro’’ environments even when ‘‘macro’’ walkability characteristics are favorable (Lovasi et al., 2009a). For example, a study in Austin, TX found low-income and Hispanic neighborhoods were more walkable than high-income, mostly non-Hispanic white neighborhoods, when considering ‘‘macro’’ or structural attributes such as residential density, street connectivity, and mixed land use (Zhu and Lee, 2008). However, moredetailed or ‘‘micro’’ environmental attributes were inequitably distributed, with low-income and Hispanic neighborhoods having worse maintenance of sidewalks, roads, and buildings; worse esthetics such as tree shade; and higher crash and crime rates. Even among walkable neighborhoods in New York City, poorer neighborhoods had significantly fewer street trees and clean streets, and higher rates of felony complaints and vehicular crashes than higher income areas (Neckerman et al., 2009). Based on the evidence to date, it appears low-income areas are disadvantaged in ‘‘micro’’ features such as esthetics, traffic safety infrastructure, and crime safety (Lovasi et al., 2009a), as well as selected ‘‘macro’’ attributes of breadth of desirable land uses, such as reduced access to food stores and places to exercise that are particularly relevant to health. These unfavorable attributes could blunt or negate the beneficial effects of neighborhoods deemed walkable based on ‘‘macro’’ attributes. Unfavorable safety and esthetic factors could also discourage businesses from entering or staying in the neighborhoods, reducing economic

1275

opportunities, access to walkable destinations, and access to local healthy foods. The purpose of the present study was to extend previous studies of disparities in access to activity-supportive environments by examining a broader range of built and social perceived environment ‘‘micro’’ variables. Proximity to a variety of specific land uses that are expected to be related to physical activity for transport and recreation purposes, as well as dietary behaviors, was investigated. The present study filled gaps in the literature because some previous studies had a narrow range of walkability (Lovasi et al., 2009b) or had confounding of walkability and demographic characteristics (Zhu and Lee, 2008). The current study’s design was well suited for present analyses because neighborhoods were systematically selected to balance neighborhood income across high and low levels of objectively measured macro-environmental walkability. This design permitted assessment of the distribution of ‘‘micro’’ environmental and specific land use variables across income in both high- and low-walkable neighborhoods.

2. Methods 2.1. Study design The Neighborhood Quality of Life Study (NQLS) was an observational epidemiologic study designed to compare multiple health outcomes among residents of neighborhoods stratified on ‘‘walkability’’ based on Geographic Information System-based (GIS) characteristics and median household income (Sallis et al., 2009; Frank et al., 2010). Participants were recruited from two metropolitan areas in the United States (King County-Seattle, WA and Baltimore, MD-Washington DC regions). Data were collected from adults living in 32 neighborhoods: 16 from Seattle-King County and 16 from Baltimore-Washington DC regions. Table 1 illustrates the study design in which selected neighborhoods were categorized into quadrants representing low versus high walkability and low versus high median income. The study was approved by Institutional Review Boards at participating academic institutions, and participants gave written informed consent. 2.2. Neighborhood selection A ‘‘walkability index’’ was computed (Frank et al., 2010) based on earlier conceptual work (Frank and Engelke, 2001) and empirical literature (Cervero and Kockelman, 1997; Saelens et al., 2003a,b). The walkability index was a weighted sum of four standardized measures in GIS computed at the census block group level: (a) net residential density (ratio of residential units to the land area devoted to residential use); (b) retail floor area ratio (retail building square footage divided by retail land square footage, with higher values indicating pedestrian-oriented design); (c) land use mix (diversity of 5 types of land uses— residential, retail, entertainment, office, institutional); and (d) intersection density (connectivity of street network measured as the ratio of number of intersections to land area). Detailed descriptions of the walkability index and its computation are provided in Frank et al. (2010) where the walkability index was validated by associations with journey to work travel data from the U.S. census. The index has been used to predict total physical activity and walking for transportation in the NQLS study (Sallis et al., 2009) and others (Frank et al., 2005; Owen et al., 2007). Block groups were used as the unit for assessing walkability and median household income, because they are the smallest geographic unit for which sociodemographic information is available. Data from block groups were used to approximate ‘‘real’’

1276

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

Table 1 Neighborhood Quality of Life Study design: neighborhood walkability and median household income by quadrant.a

Low income Seattle-King County (S-K) BaltimoreWashington DC (B-W)

Walkability index Neighborhood median household income Walkability Index Neighborhood median household income

High income Seattle-King County (S-K) BaltimoreWashington DC (B-W)

Walkability index Neighborhood median household income Walkability index Neighborhood median household income

Low walkability Mean (SD)

High walkability Mean (SD)

8 neighborhoods (4 per region) S-K n ¼311 B-W n¼ 222 0.03 (1.23) $47,531 ($3,679)

8 neighborhoods (4 per region) S-K n¼ 316 B-W n¼211 5.36 (2.68) $36,562 ($4,275)

 0.51 (0.19) $42,636 ($1,577)

1.42 (0.99) $37,258 ($3,047)

8 neighborhoods (4 per region) S-K n ¼325 B-W n¼ 241  1.92 (0.71) $74,576 ($8980)

8 neighborhoods (4 per region) S-K n¼ 335 B-W n¼238 2.93 (1.24) $70,546 ($10,493)

 0.74 (0.16) $80,098 ($8180)

1.55 (1.44) $72,013 ($9634)

a Walkability Index in z-score units; neighborhood median household income from 2000 Census data for the selected block groups (see Sallis et al., 2009 and Frank et al., 2009 for detailed descriptions).

neighborhoods from contiguous block groups with relatively homogeneous characteristics and minimal internal barriers. Block groups were ranked separately for each region and divided into deciles based on the walkability index scores. The 7th, 8th, 9th, and 10th deciles and the 1st, 2nd, 3rd, and 4th deciles represented ‘‘high’’ versus ‘‘low’’ walkability areas, respectively; block groups in the 5th and 6th deciles were omitted to create separation between categories. Block groups with 2000 Census-defined median household incomes less than $15,000 or greater than $150,000 were excluded to avoid outliers in neighborhood incomes; then block groups were ranked and divided into deciles, such that those in the 2nd, 3rd, and 4th deciles constituted the ‘‘low income’’ category; those in the 7th, 8th, and 9th deciles made up the ‘‘high income’’ category; those in the 5th and 6th deciles were omitted to create separation between the categories. The ‘‘walkability’’ and ‘‘income’’ characteristics of each block group were crossed (low/high walkability by low/high income) to produce a list of block groups that met definitions of study ‘‘quadrants’’. These were examined to identify clusters of contiguous block groups that might be used to define a ‘‘neighborhood’’. The goal was to identify ‘‘neighborhoods’’ with at least 1000 households, so there would be a sufficient recruitment pool. It was recognized that low-walkable neighborhoods would be geographically larger than high-walkable neighborhoods, given the difference in residential density. Good geographic distribution of neighborhoods within regions was desired to enhance diversity of racial/ethnic composition, access to transit, housing types, and access to employment. Selections of the final 32 neighborhoods were based on observations by the study team to ensure that grossly different walkability features did not exist near the boundaries of the candidate neighborhoods. There were 8 neighborhoods in each of the four quadrants (see Table 1). Additional built environment and sociodemographic characteristics of study quadrants and neighborhoods are available in Frank et al. (2010). Each neighborhood had its own profile of walkability components, so there was considerable diversity within quadrants. In general, high-walkable neighborhoods had much more multifamily housing, grid-like street networks, and nearby shopping with entrances opening onto sidewalks instead of parking lots. Low-walkability neighborhoods followed suburban, more automobile-oriented designs, with single-family detached housing, poorly connected streets, and little or no nearby retail. Fig. 1 illustrates the types of neighborhoods represented in each quadrant with GIS images that show land uses and street

layouts in example neighborhoods. Photographs of example neighborhoods in each quadrant illustrate some of the income differences in amenities and esthetics. 2.3. Summary of associations between neighborhood walkability and physical activity Earlier publications from the present study documented associations between physical activity and environmental attributes measured objectively and by self-report. Comparisons between high- and low-walkability neighborhoods were significant for total physical activity assessed by objective accelerometers, self-reported walking for transportation and recreation, and overweight/obesity status (Sallis et al., 2009). Those analyses were adjusted for numerous sociodemographic variables, automobile access, and a measure of reasons for self-selection into the neighborhood. Two additional papers reported significant correlations between most of the self-reported environmental variables used in the present paper and reported physical activity (Cerin et al., 2006; Shigematsu et al., 2009). These previous papers established that the objective and perceived environment measures used in the present study are related to physical activity, consistent with numerous other studies (Gebel et al., 2007; Heath et al., 2006; Saelens and Handy, 2008). The previous findings support the significance of examining socioeconomic disparities in environmental attributes known to be related to physical activity and other health outcomes. 2.4. Participant characteristics The final sample was 2199 participants (1287 in King County, WA and 912 in the Baltimore-Washington, DC region). Mean age was 45.1 (SD¼11.0) years, and 48.2% were women. Education was relatively high with 8.8% completing only high school or less, 26.2% having some college or vocational training, and 65.0% having college or graduate degrees. The race/ethnic breakdown was 73.5% Caucasian/non-Hispanic, 13.2% African American, 5.0% Asian American, 3.7% Hispanic, 0.9% Pacific Islander, 0.8% American Indian/Alaskan Native, and 2.9% other. Most participants were married (56.1%) or living with a partner (6.0%), with 17.2% divorced, widowed, or separated, and 20.7% single/never married. The study participation rate (i.e., returned a survey/eligible contacts) was 26% overall, and did not differ significantly by quadrant (range of 23–29%). Compared with census data from study block

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

1277

Fig. 1. GIS images of land uses and photographs of example neighborhoods in each study quadrant from the Neighborhood Quality of Life Study. Quadrants are formed by high/low walkability X high/low median neighborhood income. GIS images and photographs are not necessarily from the same neighborhoods. Thick black lines in the GIS images indicate census block group boundaries. Note: LL¼ low walkability/low income; HL¼ high walkability/high income; LH ¼low walkability/high income; HH¼ high walkability/high income. Photography: James Salis.

groups, the study sample was older, had fewer females, more whites, fewer Hispanics, and higher household incomes (all p’s o0.01). Participant demographic characteristics stratified by income and walkability quadrants are provided in Sallis et al. (2009). 2.5. Recruitment and assessment procedures Recruitment and data collection were conducted during two 18-month periods: May 2002–November 2003 in King County,

WA; and December 2003–June 2005 in Baltimore-Washington, DC region. Contact information for adults residing within selected neighborhoods was obtained from marketing companies. Records were randomly selected for order of recruitment within neighborhoods, and a letter introducing the project was mailed to heads of households, followed by telephone calls. Eligibility was defined as being between 20–65 years old, not residing in a group living establishment (e.g. nursing home, dormitory), ability to complete written surveys in English, and absence of a medical condition that interfered with walking. After returning a signed

1278

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

informed consent, participants were mailed the measurement instruments. 2.6. Measures Demographic covariates. Demographic variables assessed by survey were gender, age, education (5 levels from less than high school to graduate degree), ethnicity (re-categorized as nonHispanic white or non-white), number of motor vehicles per adult in household, marital status (re-categorized as married/ living together or other), number of people in household, and length of time at current address. Study region (Seattle/King County vs. Baltimore/Maryland sites) was included as a covariate. Neighborhood Environment Walkability Scales (NEWS). The NEWS survey (Saelens et al., 2003a,b) was composed of eight multi-item subscales to assess participants’ perceived attributes of their neighborhood environments. Table 2 presents the subscale names, number of items, and brief description or sample items. The subscales are categorized as either ‘‘macro’’ or ‘‘micro’’ environment variables. ‘‘Macro’’ environment variables are related to community design and correspond to land use indicators used to define walkability, plus seven specific land use types that were analyzed separately. ‘‘Micro’’ environmental variables refer to details of the built environment as well as social environment variables expected to be related to physical activity. Higher scores on all the NEWS scales were presumed to indicate more favorable environments for physical activity. Good testretest reliability and validity have been reported for the NEWS in multiple studies (Saelens et al., 2003a,b; Cerin et al., 2006, 2009a,b). Information on item wording, response formats, and scoring can be found at: http://www.drjamessallis.sdsu.edu/mea sures.html. All scales, with the exception of residential density and land use mix-diversity, were rated on a 4-point Likert scale from strongly disagree (1) to strongly agree (4); scale scores were

calculated as means. Residential density items asked about the frequency of various types of residential buildings in the neighborhood, from single-family detached homes to 13-story or higher apartments/condominiums, with a response range of 1 (none) to 5 (all). Residential density items were weighted relative to the average density of single-family detached residences (e.g., 7- to 12-story apartments and condominiums were considered to be 50 times more person-dense than single-family residences); weighted values were summed to create the residential density scale score. Land use mix-diversity was ordinally coded with respect to walking time from home to various types of destinations, ranging from a 1–5-min walking distance (coded as 5) to Z30-min walking distance (coded as 1); responses for 23 types of facilities were summed to form the scale score. Higher scores on land use mix-diversity indicated closer average proximity for walking to a larger number of facilities. Seven additional ‘‘macro’’ neighborhood environment measures were constructed from NEWS responses to provide a more detailed examination of access to various types of destinations (see Table 2 for descriptions). Six of the seven were taken from the NEWS land use mix-diversity scale. The final variable, convenient recreation facilities, was based on a separate survey of recreation facilities that has been shown to be reliable (adapted from Sallis et al., 1997). 2.7. Statistical analyses Each of 15 built and social environment perceptions of neighborhoods was examined as a dependent variable in generalized linear mixed models analysis (using SAS PROC GLIMMIX). Primary exposures tested in the models were the objectively derived neighborhood factors of high- versus low-walkability and high- versus low-income. The approach was first to examine the neighborhood income-by-walkability interaction. If significant, then separate tests were conducted for neighborhood income

Table 2 Neighborhood Environment Walkability Survey (NEWS) scales and variables classified into ‘‘macro’’ and ‘‘micro’’ categories. Scale or variable name

Number of items

Description and sample items

‘‘Macro’’ built environment variables that indicate walkability of neighborhood design and access to specific land uses Residential density 6 Frequency of different types of residences (e.g., single-family houses to high-rise apartments) Land use mix-diversity 23 Number of minutes to walk from home to nonresidential land uses, such as restaurants and retail stores Land use mix-access 7 Ease of access to nonresidential uses (e.g., can do most shopping in local shops) Street connectivity 5 Reflects directness of routes (e.g., number of 4-way intersections; long block length) Total types of destinations (derived from land use 23 Count of 23 types of stores or facilities within a 20-min walk from the mix-diversity scale) participant’s home Diversity of retail (derived from land use 14 Count of 14 types of retail stores or facilities within a 20-min walk mix-diversity scale) Healthy food access (derived from land use 3 Count of 3 healthy food retail outlets (e.g., supermarket, fruit/vegetable market, mix-diversity scale) non-fast food restaurant) within a 20-min walk Unhealthy food access (derived from land use 2 Count of 2 food retail types (convenience/small grocery and fast food mix-diversity scale) restaurant) within a 20-min walk Park access (derived from land use mix-diversity 1 Park within a 20-min walk scale) Gym/fitness facility access (derived from land use 1 Gym/fitness facility within 20-min walk mix-diversity scale) Convenient recreation facilities (adapted from 19 Computed as a count of ‘‘yes’’ responses to 19 activity-related facilities (e.g., Sallis et al., 1997) basketball court, aerobics/dance studio, tennis courts) that were either on a frequently traveled route (e.g., to and from work) or within a 5-min drive or 10-min walk from a participant’s home or work ‘‘Micro’’ built environment variables of a smaller scale, plus social environment Walking/cycling facilities 6 Esthetics 6 Pedestrian/traffic safety 11 Safety from crime 4

Infrastructure such as sidewalks and pedestrian/bike trails Trees that provide shade, attractive sights Amount and speed of nearby street traffic, crosswalks for safe street crossing Safe to walk at night, neighborhood crime rates

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

(high vs. low) stratified by walkability. If the interaction was not significant, then the main effect for neighborhood income was examined, since a non-significant interaction would suggest that whatever effects were present for neighborhood income did not vary significantly by neighborhood walkability. All models were adjusted for participants’ gender, age, education, ethnicity, number of motor vehicles per adult in household, marital status, number of people in household, length of time at current address, and region (Seattle vs. Baltimore sites) as in a previous paper (Sallis et al., 2009). Neighborhood was included as a random effect to adjust for clustering. In addition, the interactions of region by neighborhood income and walkability were examined but found to be non-significant in all models. Appropriate error distributions and link functions were selected for the 15 outcomes depending on their characteristics. Distributions included normal, Poisson, negative binomial, and binary. A significance level of p r0.05 was used for all tests; however, marginal significance was noted if the p-value was between 0.05 and 0.10. All analyses were carried out using SAS version 9.1.3.

3. Results To aid interpretation of income differences in perceived environment attributes, it is useful to first examine whether the objectively measured walkability index differed between low and high income areas in this study. The neighborhood-level walkability index was used as the dependent variable in a 2 (high/low income)  2 (high/low walkability) ANOVA. As designed, the walkability main effect was highly significant (F (1,31)¼36.313, p o0.0001). The income main effect was marginally significant (F (1,31)¼ 3.517, p o0.071), with the low income neighborhoods (mean¼1.575, SD¼2.747) tending to have higher walkability scores than high income neighborhoods (mean¼0.456, SD¼2.158). Most importantly, the income X walkability interaction was not significant (F (1,31)¼0.003, po0.960), indicating that higher income high walkable neighborhoods were not more walkable overall than the lower income high walkable areas in this study. 3.1. Neighborhood environment walkability scales and variables Three of the four ‘‘macro’’ environment NEWS scales that parallel components of the objectively measured walkability index had significant neighborhood income-by-walkability interactions: land use mix-diversity (p ¼0.0051), land use mix-access (p¼ 0.0087), and street connectivity (p ¼0.0078) (see Table 3). Among high walkable neighborhoods, land use mix-diversity was higher in high-income compared to low-income neighborhoods (3.7 vs 3.4, p¼0.034); among low-walkable neighborhoods, land use mix-diversity was marginally higher in the low- versus highincome neighborhoods (2.9 vs 2.7, p ¼0.068). Among high-walkable neighborhoods, perceived land use mix-access was higher in high-income compared to low-income neighborhoods (3.32 vs 3.11, p ¼0.031); among low-walkable neighborhoods, no significant differences were found. Regarding perceived street connectivity, no significant differences were found among high-walkable neighborhoods; among low-walkable neighborhoods, perceived street connectivity was higher in low-income compared to highincome neighborhoods (2.62 vs 2.38, p ¼0.012). For perceived residential density, neither the interaction between income and walkability nor the neighborhood income main effects were significant. There was no perceived measure of retail floor area ratio that was a component of the GIS-based walkability index. Also shown in Table 2, there were significant (or marginally significant) income-by-walkability interactions for 3 of the 7 selfreported counts of convenient recreation facilities and specific types

1279

of proximal land uses to participants’ homes: total types of land-uses (p¼0.020), diversity of retail access (p¼0.059), and healthy food access (p¼0.0076). Among high-walkable neighborhoods, self-reports of the total number of land uses within a 20-min walk were marginally higher (p¼0.076) in high-income compared to lowincome neighborhoods (18.4 vs 16.7); among low-walkable neighborhoods, there was no significant difference. For self-reported access to healthy food outlets, both stratum-specific tests were marginally significant. Among high-walkable neighborhoods, participants living in high-income neighborhoods had marginally higher reports of access to healthy food outlets (2.56 vs 2.37, p¼0.072) than those living in low-income neighborhoods; however, among low-walkable neighborhoods, participants living in low-income neighborhoods had marginally higher reports (1.98 vs 1.59, p¼0.052). For self-reported diversity of retail access within a 20-min walk, although the interaction was marginally significant, neither stratum-specific test was significant. Of the four other environmental attributes that did not show significant income-by-walkability interactions, two had neighborhood income main effects: count of convenient recreation facilities (p ¼0.0001), and access to a gym/fitness facility within a 20-min walk (p¼0.051). Regardless of neighborhood walkability, high-income neighborhoods had higher convenience and access than low-income neighborhoods. For the measures of unhealthy food and park access within a 20-min walk from home, neither the interaction nor income main effect was significant. Of the four perceived ‘‘micro’’ built and social environment NEWS scales, no significant neighborhood income-by-walkability interactions were found. The neighborhood income main effect, however, was significant for all four measures (p¼0.029 for walking/cycling facilities and po0.0001 for neighborhood esthetics, pedestrian/traffic safety, and safety from crime). Regardless of neighborhood walkability, participants in high-income neighborhoods had higher perceived ratings on all four ‘‘micro’’ factors than did participants in low-income neighborhoods.

4. Discussion Of the 15 perceived neighborhood built and social environment attributes examined, 10 were found to have significant income disparities. Residents of low-income neighborhoods perceived they had less favorable environments for physical activity on multiple dimensions and a relative advantage only in street connectivity (and then only for low-walkability neighborhoods). Most of the income disparities in environmental variables applied to both low-walkability and high-walkability neighborhoods. Present results provide further documentation of income-based disparities in neighborhood environments that may increase risk of chronic diseases and support an evidence-based call to action in this area of environmental justice. The goals of environmental justice are to ensure that vulnerable communities have equal access to environments that support health, education, and economic development (Schaeffer and Sclar, 1985; Bullard, 2007; Taylor et al., 2006). Among the ‘‘macro’’ environment variables that often define walkability and are most likely to be available in GIS databases (Frank et al., 2010), only perceived net residential density was unrelated to income. For all three other indicators of walkability, there were significant interactions with income. Two mixed land use variables showed the same pattern. In high-walkable neighborhoods, land use mix was perceived to be greater in the highincome areas, suggesting there are more destinations to stimulate walking for transportation. More shops and services in higherincome areas are consistent with basic economic principles, but more shops would result in a more favorable environment for

1280 Table 3 Perceived neighborhood environment attributes: adjusted means by design quadrants and tests of hypotheses for neighborhood income-by-walkability interaction, income main effect if non-significant (NS) interaction, and stratified test if interaction significant. Outcome

Adjusted means (SE)a

Tests of significance (P-value) High walkability

Income

Income

Low

High

Low

2.87 3.16 2.87 3.54

(0.11) (0.09) (0.05) (0.08)

2.94 2.85 2.87 2.97

Income main effect (for NS interaction)

High

‘‘Macro’’ built environment variables that indicate walkability of neighborhood design and access to specific land uses NEWS Residential density 222.3 (13.4) 204.3 (13.4) 249.8 (13.4) NEWS Land use mix-diversity 2.90 (0.09) 2.67 (0.09) 3.43 (0.09) NEWS Land use mix-access 2.75 (0.07) 2.59 (0.07) 3.11 (0.07) NEWS Street connectivity 2.62 (0.07) 2.38 (0.07) 3.07 (0.07) Convenient recreation facilities: count up to 18 9.28 (0.38) 10.48 (0.39) 9.59 (0.38) Total types of destinations within 20 min walk: count up to 23 13.7 (0.61) 11.9 (0.61) 16.7 (0.61) Diversity of retail access within 20 min walk: count up to 14 8.82 (0.48) 7.27 (0.48) 10.67 (0.48) Healthy food access within 20 min walk: count up to 3 1.98 (0.09) 1.59 (0.09) 2.37 (0.09) Unhealthy food access within 20 min walk: count up to 2 1.62 (0.06) 1.36 (0.06) 1.75 (0.06) Park access within 20 min walk: 0 (No) or 1 (Yes). 0.73 (0.034) 0.78 (0.034) 0.93 (0.034) Gym/fitness facility access within 20 min walk: 0 or 1. 0.36 (0.072) 0.43 (0.072) 0.55 (0.072) ‘‘Micro’’ built environment variables of a smaller scale, plus social environment NEWS walking/cycling facilities 2.64 (0.11) NEWS neighborhood esthetics 2.82 (0.09) NEWS pedestrian/traffic safety 2.70 (0.05) NEWS safety from crime 3.18 (0.08)

Income  Walk interaction

(0.11) (0.09) (0.05) (0.08)

255.6 (12.8) 3.70 (0.09) 3.32 (0.07) 3.18 (0.07) 11.29 (0.37) 18.4 (0.59) 11.80 (0.47) 2.56 (0.09) 1.80 (0.06) 0.96 (0.033) 0.75 (0.070) 3.19 3.44 3.12 3.54

(0.11) (0.08) (0.05) (0.09)

Income stratified by walkability (for significant interaction) Walkability Low

High

0.22 0.0051 0.0087 0.0078 0.51 0.020 0.059 0.0076 0.14 0.89 0.36

0.56 – – – 0.0001 – – – 0.26 0.34 0.051

– 0.068 0.122 0.012 – 0.13 0.14 0.052 – – –

– 0.034 0.031 0.23 – 0.076 0.24 0.072 – – –

0.89 0.14 0.48 0.24

0.029 o 0.0001 o 0.0001 o 0.0001

– – – –

– – – –

a All models were adjusted for gender, age, education, ethnicity, number of motor vehicles per adult in household, marital status, number of people in household, length of time at current address, and site. In addition, neighborhood clustering was adjusted for in all models.

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

Low walkability

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

walking among residents of high-income neighborhoods. In lowwalkable neighborhoods, the street network was significantly more connected for low-income areas, which should be more supportive for physical activity because the streets provide direct routes. Older more-central areas of U.S. cities were built to enable transit and walking as primary modes of travel before car dependence became so extreme (Frumkin et al., 2004). At least in the present study, older suburbs were more likely to be inhabited by low income populations than newer, more affluent suburbs with less connectivity. Because neighborhoods were selected based on parcel-level land use and objective street network data to be roughly balanced on these ‘‘macro’’ environment variables by neighborhood income, present findings may not be representative across all levels of walkability and income. Overall, perceived ‘‘macro’’ walkability indicators were similar across income categories with no difference in density, advantage to high-income on mixed land use (in high-walkability neighborhoods), and advantage to lowincome on street connectivity (in low-walkability neighborhoods). These perceived results are generally consistent with comparisons of the objective walkability index which was not significantly different by income categories but tended to favor the low-income neighborhoods. Examination of specific land uses within a 20-min walk of participants’ homes revealed a variety of patterns. The number of land use categories within a 20-min walk had a significant walkability by income interaction. Follow-up tests for the high-walkability neighborhoods showed a trend for more destinations in the high-income areas. A more specific measure of 14 types of retail stores within a 20-min walk had no significant income differences. Access to a summary score of 19 types of recreation facilities was significantly lower in low-income neighborhoods, consistent with previous studies (Estabrooks et al., 2003; Giles-Corti and Donovan, 2002; Gordon-Larsen et al., 2006; Moore et al., 2008; Powell et al., 2006). Seemingly in contradiction to the general measure of access to recreation facilities, perceived access to parks in particular was not related to income. Perhaps most of the disparities were in access to private recreation facilities, and this interpretation was partly supported by the finding of a trend for highincome neighborhoods to have more access to a health club or fitness facility. A limitation of the present study was that quality of parks was not assessed. The perceived ‘‘micro’’ built and social environment variables were more favorable for high-income neighborhoods regardless of walkability. This pattern applied to walking/cycling facilities, esthetics, pedestrian/traffic safety, and crime safety. Present findings are consistent with previous findings of disparities in quality of walking/cycling facilities and esthetic features (Kelly et al., 2007; Lovasi et al., 2009a; Neckerman et al., 2009; Zhu and Lee, 2008), but pedestrian/traffic safety and crime safety have been inconsistently related to physical activity for either recreation or transport purposes (Bauman and Bull, 2007; Saelens and Handy, 2008). Thus, in the present study, low-income neighborhood residents perceived they had systematically less-supportive environments for physical activity on a wide variety of attributes that have been less-studied than ‘‘macro’’ walkability indicators. Some of the disadvantages found in low-income neighborhoods could be remedied in the short term with specific policies. Improvements in amenities like sidewalks, bicycle facilities, streetlights, and street crossings could make walking and bicycling safer. Neighborhood esthetics could be enhanced by planting trees, picking up litter, and painting over graffiti. Public investments in parks and incentives for private recreation facilities to locate in low-income neighborhoods may be needed to remove disparities in access to recreation facilities. Comprehensive longerterm approaches are likely needed for the other solutions: traffic

1281

calming to improve pedestrian/traffic safety, policing strategies to improve crime safety, and economic development to renovate dilapidated buildings. The NEWS included measures of proximity to several types of food establishments. There were no income effects for sources of less-healthy foods (i.e., convenience stores, fast food restaurants) within a 20-min walk. This finding seems inconsistent with other studies showing fast food restaurants were concentrated in lowincome areas (Block et al., 2004; Cummins et al., 2005; Reidpath et al., 2002; Powell et al., 2007), but the present study did not measure the number of less-healthy food outlets. There was a significant walkability by income interaction for nearby healthierfood sources (i.e., supermarkets, restaurants). Follow-up tests for income effects within walkability strata were not significant, but the trends are worth noting. Within low-walkability neighborhoods, there were slightly more types of healthier-food sources in low-income neighborhoods. Within high-walkability neighborhoods, there were slightly more types of healthier-food sources in high-income neighborhoods. Present findings are generally inconsistent with previous studies indicating fewer supermarkets in low-income and high-minority neighborhoods (Lovasi et al., 2009a; Zenk et al., 2005; Larsen and Gilliland, 2008; Morland et al., 2006), but prior studies did not take walkability into account. The current results raise the possibility of different patterns of disparities in access to healthier foods, depending on neighborhood walkability. However, prior studies used GIS data on food sources instead of self-report of proximity, so present findings need to be replicated with objective food outlet proximity. The present study design maximized environmental variability and attempted to balance neighborhood income within walkability strata. This makes the design well suited to investigate income disparities in neighborhood environment variables. Because objectively defined walkability, based on ‘‘macro’’ variables, was not significantly different across income categories, the ‘‘micro’’ environment differences that emerged were not severely confounded by ‘‘macro’’ walkability differences. However, the use of the multi-component walkability index may not reveal income differences in specific components. A potential limitation is that participants did not base their reports of perceived neighborhood characteristics on the same geographic definition used in the neighborhood selection process (within a 20-min walk vs. contiguous census block groups, respectively). Though environmental variables were measured by self-report, the validity of the NEWS has been supported in several studies (Cerin et al., 2006; DeBourdeaudhuij et al., 2003; Leslie et al., 2005; Saelens et al., 2003a,b), including one study suggesting reports of walking time to destinations were relatively accurate (Adams et al., 2009). Few of the ‘‘micro’’ environmental variables assessed in the present study are readily available in GIS, or existing datasets have not been validated (Matthews et al., 2009), so it has not been feasible to study these variables objectively. Improved access to ‘‘micro’’ environment measures through publicly available GIS databases could benefit research and practice. Many of the ‘‘micro’’ environment attributes can be assessed by direct observation (Brownson et al., 2009), though at much higher cost than selfreports. Limitations of the present study included a modest recruitment rate and lack of representativeness of the study sample on several demographic characteristics. In summary, ‘‘macro’’ environmental variables used to define walkability did not vary systematically across income strata. However, low-income neighborhoods had consistently less access to recreation facilities and less favorable levels of perceived ‘‘micro’’ built and social environment attributes, including walking/cycling facilities, esthetics, pedestrian/traffic safety, and crime safety. Some of these attributes are expected to be supportive of active transportation and others supportive of active recreation

1282

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

(Sallis et al., 2006). Previous studies documented ‘‘micro’’ environment disparities only in walkable neighborhoods (Neckerman et al., 2009) or confounded ‘‘macro’’ and ‘‘micro’’ disparities (Zhu and Lee, 2008). In the present study, low-income disparities usually applied to both low-walkable and high-walkable neighborhoods. Thus, walkability defined only by macro variables would logically be an incomplete construct for describing the activity-supportiveness of neighborhoods. A broader concept and measures of ‘‘physical activity-supportiveness’’ need to be developed. Physical activity-supportiveness could include pedestrian infrastructure (sidewalks) and their quality, esthetics, protection from traffic dangers, crime safety, recreation facilities, and bicycling facilities, in addition to walkability. Studies of environmental correlates or interventions are encouraged to measure or intervene on the ‘‘micro’’ environment attributes and proximity to recreation facilities that often are related to physical activity (Bauman and Bull, 2007; Saelens and Handy, 2008). The ‘‘micro’’ built and social environment attributes appear to reduce the activity-friendliness of low-income neighborhoods, but are infrequently studied. Deficiencies in ‘‘micro’’ environment attributes may be amenable to relatively low-cost and short-term solutions. Planting trees, removing graffiti, building or improving sidewalks and bicycle facilities, adding streetlights, street crossing improvements, and street furniture such as benches may provide some near-term practical remedies to environmental disparities. These improvements may be an important part of an overall strategy that also addresses social factors, such as crime, traffic, and social capital, required to create activity-friendly environments where they are most needed.

Acknowledgments This study was supported by grant HL67350 from the National Heart, Lung, and Blood Institute. We acknowledge the important contributions of the office of the County Executive in King County, WA, and multiple state and county agencies in Maryland. Carrie Geremia contributed to the manuscript. References Abercrombie, L.C, Sallis, J.F., Conway, T.L., Frank, L.D., Saelens, B.E., Chapman, J.E., 2008. Income and racial disparities in access to public parks and private recreation facilities. American Journal of Preventive Medicine 34, 9–15. Adams, M.A., Ryan, S., Kerr, J., Sallis, J.F., Patrick, K., Frank, L.D., Norman, G.J., 2009. Validation of the Neighborhood Environment Walkability Survey (NEWS) items using geographic information systems. Journal of Physical Activity and Health 6 (Suppl. 1), S113–S123. Bauman, A.E., Bull, F.C., 2007. Environmental Correlates of Physical Activity and Walking in Adults and Children: A Review of Reviews. National Institute of Health and Clinical Excellence, London, UK /http://www.nice.org.uk/gui dance/index.jsp?action=download&o=34740S (accessed July 15, 2009). Black, J.L., Macinko, J., 2007. Neighborhoods and obesity. Nutrition Reviews 66, 2–20. Block, J.P., Scribner, R.A., DeSalvo, K.B., 2004. Fast food, race/ethnicity, and income: a geographic analysis. American Journal of Preventive Medicine 27, 211–217. Bronfenbrenner, U, 1979. The Ecology of Human Development, Cambridge, MA. Harvard University Press. Brownson, R.C., Hoehner, C.M., Day, K., Forsyth, A., Sallis, J.F., 2009. Measuring the built environment for physical activity: state of the science. American Journal of Preventive Medicine 36 (4, Suppl. 1), S99–S123. Bullard, R.D. (Ed.), MIT Press, Cambridge, MA. Cerin, E., Leslie, E., Owen, N., 2009a. Explaining socio-economic status differences in walking for transport: an ecological analysis of individual, social, and environmental factors. Social Science and Medicine 68, 1013–1020. Cerin, E., Saelens, B.E., Sallis, J.F., Frank, L.D., 2006. Neighborhood Environment Walkability Scale: validity and development of a short form. Medicine and Science in Sports and Exercise 38, 1682–1691. Cerin, E., Conway, T.L., Saelens, B.E., Frank, L.D., Sallis, J.F., 2009. Cross-validation of the factorial structure of the Neighborhood Environment Walkability Scale (NEWS) and its abbreviated form (NEWS-A). International Journal of

Behavioral Nutrition and Physical Activity 6, 32 /http://www.ijbnpa.org/ content/pdf/1479-5868-6-32.pdfS. Cervero, R., Kockelman, K.M., 1997. Travel demand and the 3Ds: density, diversity, and design. Transportation Research—Part D 2, 199–219. Cummins, S.C.J., McKay, L., MacIntyre, S., 2005. McDonald’s restaurants and neighborhood deprivation in Scotland and England. American Journal of Preventive Medicine 29, 308–310. De Bourdeaudhuij, I., Sallis, J.F., Saelens, B.E., 2003. Environmental correlates of physical activity in a sample of Belgian adults. American Journal of Health Promotion 18, 83–92. Diez Roux, A.V., Evenson, K.R., McGinn, A.P., Brown, D.G., Moore, L., Brines, S., Jacobs, D.R., 2007. Availability of recreational resources and physical activity in adults. American Journal of Public Health 97, 493–499. Estabrooks, P.A., Lee, R.E., Gyurcsik, N.C., 2003. Resources for physical activity participation: does availability and accessibility differ by neighborhood socioeconomic status? Annals of Behavioral Medicine 25, 100–104. Frank, L.D., Andresen, M.A., Schmid, T.L., 2004. Obesity relationships with community design, physical activity, and time spent in cars. American Journal of Preventive Medicine 27, 87–96. Frank, L.D., Engelke, P.O., 2001. The built environment and human activity patterns: exploring the impacts of urban form on public health. Journal of Planning Literature 16, 202–218. Frank, L.D., Kerr, J., Miles, R., Sallis, J.F., 2008. A hierarchy of sociodemographic and environmental correlates of walking & obesity. Preventive Medicine 47, 172–178. Frank, L., Kerr, J., Saelens, B., Sallis, J., Glanz, K., Chapman, J., 2009. Food outlet visits, physical activity, and body weight: variations by gender and raceethnicity. British Journal of Sports Medicine 43, 124–131. Frank, L.D., Sallis, J.F., Saelens, B.E., Leary, L., Cain, K., Conway, T.L., Hess, P.M., 2010. The development of a walkability index: application to the Neighborhood Quality of Life Study. British Journal of Sports Medicine 44, 924–933. Frank, L.D., Schmid, T.L., Sallis, J.F., Chapman, J., Saelens, B.E., 2005. Linking objectively measured physical activity with objectively measured urban form: findings from SMARTRAQ. American Journal of Preventive Medicine 28 (2, Suppl. 2), 117–125. Frumkin, H., Frank, L., Jackson, R., 2004. The Public Health Impacts of Sprawl. Island Press, Washington, DC. Gebel, K., Bauman, A.E., Petticrew, M., 2007. The physical environment and physical activity: a critical appraisal of review articles. American Journal of Preventive Medicine 32, 361–369. Giles-Corti, B., Donovan, R.J., 2002. Socioeconomic status differences in recreational physical activity levels and real and perceived access to a supportive physical environment. Preventive Medicine 35, 601–611. Gordon-Larsen, P., Nelson, M.C., Page, P., Popkin, B.M., 2006. Inequality in the built environment underlies key health disparities in physical activity and obesity. Pediatrics 117, 417–424. Heath, G., Brownson, R., Kruger, J., Miles, R., Powell, K.E., Ramsey, L.T., the Task Force on Community Preventive Services, 2006. The effectiveness of urban design and land use and transport policies and practices to increase physical activity: a systematic review. Journal of Physical Activity and Health 3 (Suppl. 1), S55–S76. Kelly, C.M., Schootman, M., Baker, E.A., Barnidge, E.K., Lemes, A., 2007. The association of sidewalk walkability and physical disorder with area-level race and poverty. Journal of Epidemiology and Community Health 61, 978–983. Larsen, K., Gilliland, J., 2008. Mapping the evolution of ‘food deserts’ in a Canadian city: supermarket accessibility in London, Ontario, 1961–2005. International Journal of Health Geography 18, 7–16. LaVeist, T.A., 2005. Minority Populations and Health: An Introduction to Health Disparities in the United States. Jossey-Bass, San Francisco. Leslie, E., Saelens, B., Frank, L., Owen, N., Bauman, A., Coffee, N., et al., 2005. Residents’ perceptions of walkability attributes in objectively different neighbourhoods: a pilot study. Health and Place 11, 227–236. Lovasi, G.S., Hutson, M.A, Guerra, M., Neckerman, K.M., 2009a. Built environments and obesity in disadvantaged populations. Epidemiologic Reviews 31, 7–20. Lovasi, G.S., Neckerman, K.M., Quinn, J.W., Weiss, C., Rundle, A., 2009b. Individual or neighborhood disadvantage modifies the association between neighborhood walkability and body mass index. American Journal of Public Health 99, 279–284. Matthews, S.A., Vernez Moudon, A., Daniel, M., 2009. Work group II: using geographic information systems (GIS) for enhancing research relevant to policy on diet, physical activity, and weight. American Journal of Preventive Medicine 36 (4 Suppl.), S171–S176. McMillan, T.E., Cubbin, C.C., Parmenter, B., Medina, A.V., Lee, R.E., 2010. Neighborhood sampling: how many streets must an auditor walk? International Journal of Behavioral Nutrition and Physical Activity 7, 20 /http://www.ijbnpa.org/ content/pdf/1479-5868-6-32.pdfS. 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, 16–22. Morland, K, Diez Roux, A.V., Wing, S., 2006. Supermarkets, other food stores and obesity: the Atherosclerosis Risk in Communities Study. American Journal of Preventive Medicine 30, 333–339. Neckerman, K.M., Lovasi, G.S., Davies, S., Perciel, M., Quinn, J., Feder, E., Raghunath, N., Wasserman, B., Rundle, A., 2009c. Disparities in urban neighborhoodconditions: evidence from GIS measures and field observation in New York City. Journal of Public Health Policy 30, S264–S285.

J.F. Sallis et al. / Health & Place 17 (2011) 1274–1283

Owen, N., Humpel, N., Leslie, E., Bauman, A., Sallis, J.F., 2004. Understanding environmental influences on walking: review and research agenda. American Journal of Preventive Medicine 27, 67–76. Owen, N., Cerin, E., Leslie, E., duToit, L., Coffee, N., Frank, L.D., Bauman, A.E., Hugo, G., Saelens, B.E., Sallis, J.F., 2007. Neighborhood walkability and the walking behavior of Australian adults. American Journal of Preventive Medicine 33, 387–395. Papas, M.A., Alberg, A.J., Ewing, R., Helzlsouer, K.J., Gary, T.L., Klassen, A.C., 2007. The built environment and obesity. Epidemiologic Reviews 29, 129–143. Powell, L.M., Chaloupka, F.J., Bao, Y., 2007. The availability of fast-food and fullservice restaurants in the United States: associations with neighborhood characteristics. American Journal of Preventive Medicine 33 (4 Suppl.), S240–S245. Powell, L.M., Slater, S., Chaloupka, F.J., Harper, D., 2006. Availability of physical activity-related facilities and neighborhood demographic and sociodemographic characteristics: a national study. American Journal of Public Health 96, 1676–1680. Reidpath, D.D., Burns, C., Garrard, J., Mahoney, M., Townsend, M., 2002. An ecological study of the relationship between social and environmental determinants of obesity. Health and Place 8, 141–145. Saelens, B.E., Handy, S.L., 2008. Built environment correlates of walking: a review. Medicine & Science in Sports & Exercise 40 (7 Suppl.), S550–S566. Saelens, B.E., Sallis, J.F., Black, J.B., Chen, D., 2003a. Neighborhood-based differences in physical activity: an environment scale evaluation. American Journal of Public Health 93, 1552–1558. Saelens, B.E., Sallis, J.F., Frank, L.D., 2003b. Environmental correlates of walking and cycling: findings from the transportation, urban design, and planning literatures. Annals of Behavioral Medicine 25, 80–91.

1283

Sallis, J.F., Saelens, B.E., Frank, L.D., Conway, T.L., Slymen, D.J., Cain, K.L., Chapman, J.E., Kerr, J., 2009. Neighborhood built environment and income: examining multiple health outcomes. Social Science and Medicine 68, 1285–1293. Sallis, J.F., Johnson, M.F., Calfas, K.J., Caparosa, S., Nichols, J., 1997. Assessing perceived physical environment variables that may influence physical activity. Research Quarterly for Exercise and Sport 68, 345–351. Sallis, J.F., Cervero, R.B., Ascher, W., Henderson, K.A., Kraft, M.K., Kerr, J., 2006. An ecological approach to creating more physically active communities. Annual Review of Public Health 27, 297–322. Schaeffer, K.H., Sclar, E., 1985. Access for All: Transportation and Urban Growth. Penguin, Baltimore, MD. Shigematsu, R., Sallis, J.F., Conway, T.L., Saelens, B.E., Frank, L.D., Cain, K., Chapman, J.E., King, A.C., 2009. Age differences in the relation of perceived neighborhood environment to walking. Medicine and Science in Sports and Exercise 41, 314–321. Taylor, W.C., Poston, W.S.C., Jones, L., Kraft, M.K., 2006. Environmental justice: obesity, physical activity, and healthy eating. Journal of Physical Activity and Health 3 (Suppl. 1), S30–S54. Transportation Research Board and Institute of Medicine, 2005. Does the built environment influence physical activity? Examining the evidence, Special Report 282. National Academies Press, Washington, DC. Zenk, S.N., Schulz, A.J., Israel, B.A., James, S.A., Bao, S., Wilson, M.L., 2005. Neighborhood racial composition, neighborhood poverty, and the spatial accessibility of supermarkets in metropolitan Detroit. American Journal of Public Health 95, 660–667. Zhu, X., Lee, C., 2008. Walkability and safety around elementary schools: economic and ethnic disparities. American Journal of Preventive Medicine 34, 282–290.