Environmental Measures of Physical Activity Supports Perception Versus Reality Karen A. Kirtland, PhD, Dwayne E. Porter, PhD, Cheryl L. Addy, PhD, Matthew J. Neet, MPH, Joel E. Williams, MPH, Patricia A. Sharpe, PhD, MPH, Linda J. Neff, PhD, C. Dexter Kimsey, Jr, PhD, MSEH, Barbara E. Ainsworth, PhD, MPH Background: Perceptions of the environment and physical activity have been associated using survey methods, yet little is known about the validity of environmental surveys. In this study, perceptions of the environment at neighborhood and community levels were assessed (1) to determine validity by comparing respondent perceptions to objective measures and (2) to determine test–retest reliability of the survey. Methods:
A telephone survey was administered to a stratified sample of Sumter County, South Carolina adults. Respondents’ home addresses were mapped using a geographic information system (GIS) (n ⫽1112). As an indicator of validity, kappa statistics were used to measure agreement between perceptions and objective measures identified at neighborhood and community levels using GIS. A second survey in an independent sample (n⫽408) assessed test–retest reliability.
Results:
When assessing perceptions of environmental and physical activity in a defined geographic area, validity and reliability for neighborhood survey items were ⫽⫺0.02 to 0.37 and rho⫽0.42 to 0.74, and for community survey items were ⫽⫺0.07 to 0.25 and rho⫽0.28 to 0.56.
Conclusions: Although causality between perception of access and safety and actual physical activity level cannot be assumed, those meeting national physical activity guidelines or reporting some physical activity demonstrated greatest agreement with access to recreation facilities, while those not meeting the guidelines demonstrated greater agreement with safety of recreation facilities. Factors such as distance and behavior may explain differences in perceptions at neighborhood and community levels. Using local environments with short distances in survey methods improves validity and reliability of results. (Am J Prev Med 2003;24(4): 323–331) © 2003 American Journal of Preventive Medicine
Introduction
R
ecent interventions to promote physical activity have targeted social and physical environments.1–3 Social environment correlates of physical activity include community supports such as active neighbors4 and safety,5,6 while the physical enviFrom the Prevention Research Center (Kirtland, Williams, Ainsworth), Department of Environmental Health Sciences (Porter), Department of Epidemiology and Biostatistics (Addy, Ainsworth), and Department of Exercise Science (Sharpe, Ainsworth), Norman J. Arnold School of Public Health, University of South Carolina, Columbia, South Carolina; Belle W. Baruch Institute for Marine Biology and Coastal Research (Neet), University of South Carolina, Columbia, South Carolina; and Cardiovascular Health Branch, Division of Adult and Community Health (Neff) and Physical Activity Branch, Division of Nutrition and Physical Activity (Kimsey), National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia Address correspondence and reprint requests to: Barbara E. Ainsworth, PhD, MPH, Prevention Research Center, Norman J. Arnold School of Public Health, University of South Carolina, 730 Devine Street, Columbia SC 29208. E-mail:
[email protected].
ronment correlates of physical activity include presence of sidewalks,4,5 trails,5,7 and recreation facilities.4,5,8 Although survey data have been used as evidence that the environment is conducive to physical activity,4,8 –10 little is known about the accuracy of environmental survey data.10,11 For example, in a study of convenient access to exercise facilities, objective environmental measures were related to physical activity, while selfreports of those same measures were not associated with physical activity.11 Thus, a systematic investigation in comparing perceived environments with objective measures of the environment is crucial for understanding the role of environmental supports for physical activity.4 Geographic information system (GIS) technology has recently been used for studying objective environmental data relative to physical activity.12 GIS has been commonly used as a mapping tool for public health planning13–16 and identification of objective environmental risk factors.17 Previous GIS-related research has investigated individual recall and found
Am J Prev Med 2003;24(4) © 2003 American Journal of Preventive Medicine • Published by Elsevier Inc.
0749-3797/03/$–see front matter doi:10.1016/S0749-3797(03)00021-7
323
Table 1. Validity (reported by physical activity levels) and reliabilitya values for neighborhood-related environmental survey itemsb Validity (kappa)
Neighborhood survey items Access Does your neighborhood have any sidewalks? Yes No Does your neighborhood have public recreation facilities? Yes No Characteristics How would you rate your neighborhood as a place to walk?d Very pleasant or somewhat pleasant Not very pleasant or not at all pleasant For walking in your neighborhood, would you say your sidewalks aree Very well maintained or somewhat maintained Not very well maintained or not at all maintained For walking in your neighborhood, would you say that unattended dogs are A big problem or somewhat of a problem Not very much of a problem Not a problem at all How would you rate the condition of your public recreation facilities?e Excellent Good Fair or poor For walking at night, would you describe the street lighting in your neighborhood as Very good or good Fair Poor or very poor Barriers How safe from crime do you consider your neighborhood to be? Extremely safe or quite safe Slightly safe or not at all safe Would you say that the motorized traffic in your neighborhood is Heavy Moderate Light Social issues Would you say that the people in your neighborhood are Very physically active or somewhat physically active Not very physically active or not at all physically active Would you say most people in your neighborhood can be trusted?d Yes No Thinking about how public money is spent on recreation facilities, which of the following statements is most accurate?d Always gets its fair share or often gets fair share Seldom gets its fair share or never gets its fair share
Active (nⴝ372)
Insufficiently active (nⴝ517)
Inactive (nⴝ214)
All (nⴝ1112)
Reliability (rho) total (nⴝ408)
0.39
0.35
0.38
0.37
0.74
0.35
0.33
0.16c
0.30
0.52
0.15
0.14
0.14
0.14
0.66
0.10
0.67
⫺0.02
0.69
0.13
0.42
⫺0.02
⫺0.03
⫺0.03
0.14
0.23
0.17
0.19
0.73
0.22
0.20
0.26
0.22
0.58
0.02
0.02
0.03
0.02
0.69
0.06
⫺0.03
0.17
0.03
0.47
0.17
0.17
0.28
0.20
0.56
0.19
0.25
0.13
0.21
0.55
(continued on next page)
324
American Journal of Preventive Medicine, Volume 24, Number 4
Table 1. (continued) Validity (kappa)
Neighborhood survey items
Active (nⴝ372)
Insufficiently active (nⴝ517)
Inactive (nⴝ214)
All (nⴝ1112)
Use For physical activity, do you use any private or membership‡ only recreation facilities?f Yes No
Reliability (rho) total (nⴝ408) 0.47
Note: Z statistic, p⬍0.05. Kappa levels of agreement22 follow: 0.21– 0.40 ⫽ fair; 0.00 – 0.20 ⫽ slight; and ⬍0.00 ⫽ poor. a For reliability values, all correlations p⬍0.05. b Nine respondents did not provide activity levels. c Active versus insufficiently active versus inactive. d Small sample size due to restriction of at least 5 individuals in respondent’s neighborhood. e Small sample size due to skip pattern. f Includes facilities outside of neighborhood.
that people misestimate distances relative to their place of residence.18,19 In this study, a community survey was used to assess perceptions of environmental supports for physical activity and to validate those perceptions using GIS methodology at the neighborhood and community level. Test–retest reliability of the environmental survey items was also determined in an independent sample of residents.
Methods Subjects and Sampling Design Residents of Sumter County, South Carolina (N ⫽1237; aged 18 to 96 years) were surveyed from January to February 2001 using a telephone survey method. Respondents were selected from a stratified random sample of households. Census tracts were used to stratify the county population to guarantee a balance in the race/ethnic profile and geographic distribution of the study sample and to facilitate GIS mapping (geocoding of the household addresses). Households were selected randomly within each census tract. Twenty-one census tracts were surveyed, with 2 to 80 respondents per tract (median⫽61; 25th to 75th inter quartile range⫽53 to 76) per tract. The number of households selected was proportional to the percentage of the county’s total population in that tract. Listed telephone numbers were purchased from a marketing company and used to sample the population. The response rate was 54%. At the end of the survey, respondents were asked to provide their home address to match their residential location with existing supports for physical activity. Ninety percent of the respondents (n ⫽1112) who completed the survey also provided valid addresses. Using a simple random sampling design, test–retest techniques were used to measure the reliability of survey items in an independent sample. In this sample, 552 Sumter County residents were surveyed in November 2001 using the randomdigit-dial method. Of those respondents, 408 were successfully re-interviewed 3 weeks later.
Perceptions of Environmental Supports Questionnaire Items for the questionnaire were developed from an extensive literature review,4,7,10,11 expert input, and community focus groups relating to supports and barriers to physical activity.20 Demographic information identified the respondent’s home address and length of residency, age, race/ethnicity, education level, and income level. Physical activity was measured using the 2001 Behavioral Risk Factor Surveillance System (BRFSS) physical activity module. Respondents completed two sets of items pertaining to social and physical environmental supports, characteristics, and barriers to physical activity. One set of 13 items focused on neighborhood-level variables, and the second set of 13 items focused on community-level variables. Neighborhood was defined as a 0.5-mile radius or a 10-minute walk from the respondent’s home, and community was defined as a 10-mile radius or a 20-minute drive from the respondent’s home. These definitions were provided to survey respondents and used in GIS validation methods. Neighborhood survey items with responses are presented in Table 1. A Likert-type scale was used to assess neighborhood characteristics, barriers to physical activity, social issues, and access (presence or absence) to environmental supports, with the lowest value indicating stronger endorsement. Community survey items with responses are presented in Table 2. Respondents indicated on a three-point scale whether they used, did not use, or did not have the environmental support for physical activity.
GIS Environmental Measures Objective measures of environmental supports for physical activity were collected using established databases, global positioning system (GPS) units, telephone interviews, and in-person audits. The data were then stored in a GIS database. Established databases used in this study were collected from state agencies, city and county offices, and private companies. Coordinates of public recreation facilities, shopping malls, and walking/bicycling trails were collected with GPS units.
Am J Prev Med 2003;24(4)
325
Table 2. Validity (reported by physical activity levels) and reliabilitya values for community-related environmental survey itemsb Validity (kappa)
Community survey items
Active (nⴝ372)
Access Please tell me if you use any of the following resources and facilities in your community: Walking or bike trails 0.01 Trails present (use or does not use) Trails absent Public swimming pools 0.10 Pools present (use or does not use) Pools absent Public recreation centers 0.01 Recreation centers present (use or does not use) Recreation centers absent Parks, playgrounds, or sports fields 0.00 Parks, etc. present (use or does not use) Parks, etc. absent Schools that are open for public recreation activitiesd Schools present (use or does not use) Schools absent Do you use a shopping mall for physical activity/walking programs? 0.25 Malls present (use or does not use) Malls absent Do you use physical activity programs and facilities at a place of 0.01 worship? Places of worship present (use or does not use) Places of worship absent Do you use nearby waterways for water-related physical ⫺0.06 activities? Waterways present (use or does not use) Waterways absent Barriers How safe are the public recreation facilities in your community?e ⫺0.03 Very safe or somewhat safe Somewhat unsafe or not at all safe Social issues How important are recreational/physical activity clubs, programs, or organized recreational events in your community?g Very important or somewhat important Not very important or not at all important Community does not have any physical activity clubs or programs Would you say that all people have equal access to public recreation facilities?g Yes No Community does not have any public recreation facilities Do concerns about safety at the public recreation facilities in your community influence your using them?e Yes No Community does not have public recreation facilities
Insufficiently active Inactive All (nⴝ517) (nⴝ214) (nⴝ1112)
0.11c
0.05
0.07
0.40
0.01
0.10
0.05
0.28
0.00
0.01
0.00
0.40
0.01
0.03
0.01
0.56
0.00
0.01
0.00
0.36
0.31
0.14
0.25
0.42
0.03
0.05
0.03
0.33
⫺0.07
⫺0.07
⫺0.07
0.36
0.20f
⫺0.05
0.08
Note: Z statistic, p⬍0.05. Kappa levels of agreement22 follow: 0.21– 0.40 ⫽ fair; 0.00 – 0.20 ⫽ slight; and ⬍0.00 ⫽ poor. a For reliability values, all correlations p⬍0.05. b Nine respondents did not provide activity levels. c Active versus insufficiently active. d Kappa not computed for “active” because all schools open for public recreation. e Questions not asked in reliability study. f Active versus insufficiently active versus inactive. g Environmental measures not available.
326
American Journal of Preventive Medicine, Volume 24, Number 4
Reliability (rho) total (nⴝ408)
0.41
0.31
Figure 1. Map of Sumter County, South Carolina, shows an example of a neighborhood, 0.5-mile distance or 10-minute walk from respondent’s home, and a community, 10-mile distance or 20-minute drive from respondent’s home. Half-mile radius represents distance “as the crow flies.” Ten-mile service area is based on the road network.
The sidewalks database was established using a combination of field surveys, community input, and GIS techniques. Telephone interviews with administrative personnel of schools and places of worship were used to obtain data about opportunities for physical activity at those facilities. A sidewalk checklist (inter-rater agreement, ⫽0.61) and recreation facility checklist (inter-rater agreement, ⫽0.80) were used to assess sidewalk maintenance and facility condition, respectively. Respondents’ home addresses and addresses of schools, places of worship, crime incidents, and locations of unattended dogs were geocoded or mapped in GIS software using a South Carolina 911 road file. Using GIS, neighborhood and community boundaries were placed around each respondent’s home (Figure 1). Presence or absence of each environmental support (e.g., recreation facility, sidewalk, mall, school, place of worship, and waterway) was identified in the neighborhood or community of each respondent. To provide the neighborhood or community environments with a rating (e.g., safe or unsafe neighborhood; heavy, moderate, or low traffic), the distribution of supports or barriers for physical activity was investigated with respect to relevant characteristics (e.g., violent or nonviolent crimes; traffic count). Because objective (GIS) data were not available to evaluate the respondent’s impression of their neighborhood in terms of pleasantness, trust, and spending of public money, an average of neighbors’ survey responses were used to compare individual perceptions for pleasant neighborhoods, trust of neighbors, equitable spending of public money, and active neighborhoods. This measure was calculated by averaging responses of neighboring respondents within a 0.5-mile radius. Only those respondents with at least five neighboring respondents were included in the analysis.
Analysis Sample weights were constructed following the protocol of the BRFSS to ensure that statistical analyses of the validity sample were generalizable to the population. Descriptive analyses (weighted proportions) of the validity sample were calculated using SAS-callable SUDAAN version 8.0 (Research Triangle Institute, Research Triangle Park NC). All other analyses were conducted using SAS version 8.2 (SAS Institute, Cary NC). Kappa statistics were computed for 21 environmental items (12 neighborhood, 9 community) to determine the agreement between survey responses and environmental measures in the validity study. Items with four or five response choices were collapsed into two (e.g., present or absent) or three (e.g., high, moderate, or low) response levels. Neighborhood items were collapsed based on the distribution of responses (e.g., combining adjacent categories when one had a low proportion) and to match the levels ascertained from the objective data. Some community items were collapsed based on the distribution. However, all access-related items were dichotomized to match the objective data (presence versus absence). Thus, survey responses that represent presence of environmental support (e.g., “present and uses” and “present and does not use”) were collapsed to represent presence of environmental support. Kappa statistics were also computed for 19 environmental items (10 neighborhood, 9 community) by the following three categories of physical activity: Active: Meets the national public health recommendations for moderate activity (five or more times per week, 30
Am J Prev Med 2003;24(4)
327
Table 3. Sociodemographic characteristics of Sumter County, South Carolina, residents surveyed January 2001–February 2001
Characteristic Gender Female Male Age group (yr) 18–29 30–44 45–64 65⫹ Race or ethnicity White Black Other Household income ($) ⬍10,000 10,000⬍20,000 20,000⬍35,000 ⱖ35,000 Education ⬍High school High school Some college ⱖCollege
Validity (nⴝ1112) Weighted sample %
Reliability (nⴝ408) Unweighted sample %
55.7 44.3
61.5 38.5
23.3 33.6 26.1 17.0
17.4 32.5 32.5 17.6
59.7 38.3 2.0
64.1 33.4 2.5
12.9 16.9 26.6 43.6
9.3 18.1 27.1 45.5
13.0 30.9 32.6 23.5
13.3 25.0 41.8 19.9
Neighborhood Items Kappas for the neighborhood variables comparing the survey responses to the GIS objective measures are shown in Table 1. As a validity indicator of survey items, kappa statistics ranged from ⫺0.02 to 0.37 for the total sample. Agreement was highest for access to sidewalks, access to public recreation facilities, safety/crime, equitable public spending on facilities, trust of neighbors, and streetlights (⫽0.19 to 0.37). In comparison by physical activity level, chi-square statistics revealed that kappas for access to recreation facilities (2⫽8.53, p ⫽0.014) were significantly different among the three levels of physical activity. Pairwise comparisons using z -test statistics revealed that kappas for access to public recreation facilities were significantly higher for active (z ⫽2.66, p ⫽0.008) and insufficiently active (z ⫽2.61, p ⫽0.009) respondents when compared to inactive respondents. Validity measures for other neighborhood variables were not statistically different. Spearman rhos for the test–retest reliability of the survey responses were reported in the range of 0.42 to 0.74 (Table 1). Highest reliability values were reported for access to sidewalks and streetlights.
Community Items minutes per activity)21 and/or vigorous activity (three or more times per week, 20 minutes per activity) Insufficiently active: Reports moderate or vigorous activity, but does not meet the recommended guidelines Inactive: Reports no activity. Validity for maintenance of sidewalks and condition of recreation facilities was assessed only in the total sample due to the smaller number who reported having the support. A chisquare statistic was used to determine if the kappa values were significantly different among all three categories of physical activity. A z-test statistic was then used to make pairwise comparisons of the kappa values (i.e., to detect statistical differences between two specific categories). Spearman rank correlation rhos were computed to determine the relationship between initial and follow-up responses and, thus, the reliability of 23 neighborhood and community environmental items. Original response categories were used in the correlations.
Results Sociodemographic characteristics were similar among respondents in the validity study and respondents in the reliability study (Table 3). Weighted proportions of physical activity levels among respondents in the validity study were as follows: 38.4% active, 43.8% insufficiently active, and 17.8% inactive. Similar unweighted proportions were identified for active respondents (37.8%) in the reliability study. In the reliability study, 52.2% were insufficiently active and 10.0% were inactive. 328
The validity and reliability of the community survey items are shown in Table 2. Kappa statistics ranged from ⫺0.07 to 0.25 for the total sample. Agreement was highest for access to malls for physical activity (⫽0.25). Spearman rhos were reported in the range of 0.28 to 0.56. The highest reliability value was reported for access to parks, playgrounds, and sports fields. In comparison by physical activity level, chi-square statistics revealed that kappas for access to trails (2⫽8.45, p ⫽0.015), and perceptions of recreation facility safety (2⫽12.70, p ⫽0.002) were significantly different among the three levels of physical activity. Pairwise comparisons using z -test statistics revealed that insufficiently active respondents had a higher kappa for safety of recreation facilities than either active (z ⫽3.17, p ⫽0.001) or inactive respondents (z ⫽2.83, p ⫽0.005). Insufficiently active respondents also had a higher kappa for access to trails when compared to active respondents (z ⫽2.91, p ⫽0.004).
Discussion Overall, kappa statistics reported in this study indicated fair to low agreement for neighborhood and community items based on a previous standard.22 Given the novelty of this study, it is unclear what level of kappa constitutes adequate agreement between environmental perceptions of supports for physical activity and objective observations of these supports. In the present
American Journal of Preventive Medicine, Volume 24, Number 4
study, kappas ranged from 0.19 to 0.37 for half of the neighborhood variables; the remaining kappas were lower. As kappa statistics assess agreement that occurs beyond chance,23 they are considered to be a robust measure of association between subjective and objective data.24 Similar levels of agreement between self-reported neighborhood characteristics and objective data were noted in a study of environmental conditions of neighborhoods.25 One explanation for the fair to low levels of agreement for neighborhood and community items would be the respondents’ inability to perceive distance accurately. This is a common problem in a variety of geographic settings.18,19 Misestimated distances may best explain the lower kappas reported for the community items where environmental measures were present; yet ⱖ80% of the respondents reported that the environmental support did not exist. Because the ideal distance from one’s home for the recall of environmental supports is unknown, establishing distances for use of GIS is problematic. In this study, predetermined boundaries of neighborhood (0.5 mile or 10-minute walk) and community (10 miles or 20-minute drive) were used to validate survey responses using GIS. It is possible that shorter distances may enhance perception of the presence of environmental supports in the neighborhood or community. Several psychological, cultural, and behavioral factors may explain the overall low level of agreement between perceptions and objective measures. People perceive their environments based on various types of lifestyle behaviors, including individual transportation routes, personal beliefs, and cultural values.18,25,26 Respondents in the current study lived in a southeastern county with rural towns and one small metropolitan area. Respondents from a large metropolitan area may not view their environment the same as respondents in smaller metropolitan areas.18,27 Other studies indicate that objective neighborhood data and perceptions of the neighborhood do not match because people judge the environment according to their own desires and expectations.28 –30 A study of perceptions of park safety found that users of parks perceived parks as safe and nonusers of parks perceived parks as unsafe.31 Thus, individual experiences relative to physical activity behavior may also explain some of the differences among respondents. An interesting finding in the present study was that active and insufficiently active respondents had the highest kappas for access to public recreation facilities. Accessibility to recreation facilities has been significantly associated with being physically active.4,5,8 Another notable finding was that insufficiently active respondents had the highest kappas for perceptions of safety or the likelihood of crime in recreation facilities. It is unknown if these respondents were insufficiently active due to their focus on safety- or crime-related
issues. This possibility is consistent with other findings that indicate less physical activity in unsafe environments.5,6 While we do not imply causality between these environmental features and physical activity behaviors, these results indicate that the accuracy of people’s perception of the environment is correlated with their physical activity behaviors. Test–retest reliability of the survey items ranged from r ⫽0.28 to 0.74. Recall of neighborhood supports was on the order of r ⫽0.42 to 0.74, and recall of community supports was on the order of r ⫽0.28 to 0.56. Reliability values at the neighborhood level were comparable to another study that measured test–retest reliability of neighborhood environmental items (r ⫽0.68).10 As indicated for validity measures, differences in reliability between neighborhood and community items may be affected by a variety of factors including individual behavior and perception of distance. A unique aspect of this study is the validation of physical-activity environmental survey items with objective measures using GIS. The use of GIS to study environmental supports for physical activity demonstrated versatility in this approach. Data from a wide variety of sources were integrated and analyzed to compare the actual environment at either local (neighborhood) or broad (community) scales with perceptions of the environment. Using GIS, comparisons were done at a greater level of accuracy than other methods such as visual observations or drawn maps.11 Further, a visual representation of the GIS data permitted an efficient identification of neighborhoods and communities with environmental supports for physical activity. Therefore, we consider GIS to be an important tool that can advance both survey and intervention research. A major strength of this study is that survey methods were combined with GIS methods to allow the researchers to examine and study the environment at different levels (e.g., perceptions versus objective measures) in contrast to an approach that validated neighborhood quality items using only perception.32 However, several limitations were uncovered in the present study regarding the use of GIS to validate survey data of environmental supports for physical activity. Specifically, objective measures that could be mapped by GIS were not available for social items and some of the barrier items, including unattended dogs and traffic volume. Data for unattended dogs collected by Sumter County Animal Control and traffic counts collected by the South Carolina Department of Transportation (DOT) may not have reflected the true environment. For example, Animal Control data may have under-represented unattended dogs because not everyone with an unattended dog problem would have contacted Animal Control. In addition, traffic volume was not measured by the DOT on all roads in Sumter County. Therefore, Am J Prev Med 2003;24(4)
329
we were unable to validate perceptions of traffic on all neighborhood roads with the desired level of precision. Nevertheless, as measurements accumulate in the United States increase and reveal differences in community satisfaction,33 this study will provide validity and reliability measures of environmental survey items representative of small metropolitan areas in the southeastern United States. Recommendations for future research include the use of short distances, such as a 3-mile or 5-mile radius for community, to define geographic areas when conducting environmental surveys that extend beyond the respondent’s place of residence. Since perception of the environment may be most accurate close to one’s home, researchers should also consider using even shorter distances, such as the respondent’s street or block, when assessing local environmental supports for physical activity. The procedures used in this study should be replicated for comparison of validity measures across different counties and cities. Lastly, exploration and use of other methods to characterize social issues, traffic volume, and unattended dogs should be investigated in future validation studies.
Conclusion This study assessed agreement between perceptions of environmental supports for physical activity and objective measures of the social and physical environments. GIS technology was successfully implemented to visually present environmental supports for physical activity and to examine the validity of individual perceptions of the environment. Individual differences in physical activity levels were also important in understanding agreement between environmental perceptions and objective environmental measures. Future studies should examine the effect of safety-related issues for people who do not meet the recommended guidelines for physical activity. The Cardiovascular Health Branch, Centers for Disease Control and Prevention (CDC) cooperative agreement U48/ CCU409664-06 (Prevention Research Centers Program), funded this study, and the CDC Division of Nutrition and Physical Activity provided administrative support. The Office of Minority Health of the Department of Health and Human Services provided additional support. The University of South Carolina (USC) Institutional Review Board approved this study. We are grateful to Marlo Cavnar and Martin Evans for assistance in data collection and to the USC Survey Research Laboratory for sampling. We are especially thankful to the Sumter County Active Lifestyles Committee for volunteering their time to this study. We also thank Dawn Wilson, PhD, Fran Wheeler, PhD, Dennis Shepard, and Regina Fields from the USC Prevention Research Center for their constructive comments in preparing the manuscript.
330
References 1. Sallis JF, Baumann A, Pratt M. Environmental and policy interventions to promote physical activity. Am J Prev Med 1998;15:379 –97. 2. King AC, Blair SN, Bild DE, et al. Determinants of physical activity and interventions in adults. Med Sci Sports Exerc 1992;24:221–36. 3. King AC, Jeffery RW, Fridinger F, et al. Environmental and policy approaches to cardiovascular disease prevention through physical activity: issues and opportunities. Health Educ Q 1995;22:499 –511. 4. Brownson RC, Baker EA, Houseman RA, Brennan LK, Bacak SJ. Environmental and policy determinants of physical activity in the United States. Am J Public Health 2001;91:1995–2003. 5. Booth ML, Owen N, Bauman A, Clavisi O, Leslie E. Social-cognitive and perceived environmental influences associated with physical activity in older Australians. Prev Med 2000;31:15–22. 6. Centers for Disease Control and Prevention. Neighborhood safety and the prevalence of physical inactivity selected states, 1996. MMWR Morb Mortal Wkly Rep 1999;48:143–6. 7. Brownson RC, Houseman RA, Brown DR, et al. Promoting physical activity in rural communities. Am J Prev Med 2000;18:235–41. 8. Giles-Corti B, Donovan RJ. The relative influence of individual, social and physical environment determinants of physical activity. Soc Sci Med 2002; 54:1793–812. 9. King AC, Castro C, Wilcox S, Eyler AA, Sallis JF, Brownson RC. Personal and environmental factors associated with physical inactivity among different racial-ethnic groups of U.S. middle-aged and older-aged women. Health Psychol 2000;19:354 –64. 10. Sallis JF, Johnson MF, Calfas KJ, Caparosa S, Nichols JF. Assessing perceived physical environmental variables that may influence physical activity. Res Q Exerc Sport 1997;68:345–51. 11. Sallis JF, Melbourne H, Hofstetter CR, et al. Distance between homes and exercise facilities related to frequency of exercise among San Diego residents. Public Health Rep 1990;105:179 –85. 12. Troped PJ, Saunders RP, Pate RP, Reininger B, Ureda JR, Thompson SJ. Associations between self-reported and objective physical environmental factors and use of a community rail-trail. Prev Med 2001;32:191–200. 13. Siegel C, Davidson A, Kafadar K, Norris JM, Todd J, Steiner J. Geographic analysis of pertussis infection in an urban area: a tool of health services planning. Am J Public Health 1997;87:2022–5. 14. Estabrooks PA, Lee RE, Gyurcsik NC. Resources for physical activity participation: does availability and accessibility differ by neighborhood SES? Ann Behav Med 2003;25:100 – 04. 15. Reissman DR, Staley F, Curtis GB, Kaufmann RB. Use of geographic information system technology to aid health department decision making about childhood lead poisoning prevention activities. Environ Health Perspect 2001;109:89 –94. 16. Chen FM, Breiman RF, Farley M, Plikaytis B, Deaver K, Cetron MS. Geocoding and linking data from population-based survey and the US Census to evaluate the impact of median household-income on the epidemiology of invasive Streptococcus pneumoniae infections. Am J Epidemiol 1998;148:1212–8. 17. Glass GE, Schwartz BS, Morgan JM, Johnson DT, Noy PM, Israel E. Environmental risk factors for lyme disease identified with geographic information systems. Am J Public Health 1995;85:944 –8. 18. Golledge RG, Stimson RJ. Spatial behavior: a geographic perspective. New York: Guilford Press, 1997. 19. Lloyd R. Spatial cognition: geographic environments. Norwell MA: Kluwer Academic Publishers, 1997. 20. Henderson KA, Neff LJ, Sharpe PA, Greaney ML, Royce SW, Ainsworth BE. It takes a village to promote physical activity: the potential for public park and recreation departments. J Park Rec Admin 2001;19:23–41. 21. Pate R, Pratt M, Blair S, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995;273:402–7. 22. Landis J, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159 –74. 23. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Measurement 1960;20:37–46. 24. Szklo M, Nieto FJ. Epidemiology: beyond the basics. Gaithersburg MD: Aspen Publishers, 2000. 25. St. John C. Racial differences in neighborhood evaluation standards. Urban Affairs Q 1987;22:377–98. 26. Chiricos T, McEntire R, Gertz M. Perceived racial and ethnic composition of neighborhood and perceived risk of crime. Social Problems 2001;48:322. 27. Pedersen DM. Effects of size of home town on environmental perception. Perceptual Motor Skills 1997;45:955–66.
American Journal of Preventive Medicine, Volume 24, Number 4
28. Mesch GS, Manor O. Social ties, environmental perception, and local attachment. Environ Behav 1998;30:504 –20. 29. St. John C, Bates NA. Racial composition and neighborhood evaluation. Soc Sci Res 1990;19:47–61. 30. St. John C, Cosby V. Life cycle differences in neighborhood satisfaction. Soc Spectrum 1995;15:147–60. 31. Wentworth W, Nager A. Bryant Park: a comprehensive evaluation of its
image and use. New York: City University of New York Environmental Psychology Program, 1976. 32. Yang MJ, Yang MS, Shih CH, Kawachi I. Development and validation of an instrument to measure perceived neighbourhood quality in Taiwan. J Epidemiol Community Health 2002;56:492–6. 33. Hughey JB, Bardo JW. The structure of community satisfaction in a southeastern American city. J Soc Psychol 1984;123:91–9.
Am J Prev Med 2003;24(4)
331
