Walking or cycling to school appears to contribute to higher levels of total physical activity and improved weight status in youth (1,14). Several studies have shown that children who walk to school are more active than those who do not (8,26). In countries where active commuting to school is the norm, walking to school accounts for nearly half of total physical activity (27). Transportation surveys, however, show the proportion of U.S. children walking to school has declined steeply. From 1977 to 1995, a 37% decrease was seen in walking or biking to school by American children (16). One study found that only 19% of children walked to or from school at least once a week during the preceding month, whereas 6% biked (6). A survey in Britain found that 50% of children between the ages of 4 and 11 yr were driven regularly to schools that were within a mile of their homes (23).
Many different reasons are thought to account for low levels of walking to school. Parental safety concerns such as travel distance, traffic, and crime have been associated with inactive commuting (3,6,25). One report calculated that perceived traffic danger presented barriers to walking and biking to school for 20 million U.S. children (7). In one study, a high proportion of those parents who reported no barriers to their child actively commuting to school, also reported that their children walked to school (7). Barriers to active commuting to school, including parental concerns, need to be better understood.
In addition to parental concerns, physical environmental factors may be associated with children actively commuting to school. Environmental correlates of children's overall physical activity have been documented (22), so it is reasonable to expect environmental variables to be related to specific types of physical activity such as walking and biking to destinations. Timperio et al. (25) found that having to cross several roads was associated with less walking or cycling to local destinations for children 10-12 yr of age. It is well established that "walkable" neighborhoods with destinations near homes, interconnected streets, and higher residential densities are associated with more walking and cycling for transportation among adults (19). In one study, walking to school was more common in neighborhoods with sidewalks (10). Land use density and mix variables, although related to adult travel behavior, however, were unrelated to children's active commuting (10). In another study, population density around the school was positively associated with biking and walking to school (5).
The primary purpose of the present study was to examine the association of a variety of objective and perceived neighborhood environmental characteristics and parent concerns with active commuting to school. A second purpose was to examine whether parental concerns varied by environmental characteristics. A third set of analyses investigated the combined relationship between the perceived environment and parental concerns and objective walkability on the outcome of active commuting to school. These analyses were conducted to examine whether perceptions of the environment explained behavior, independent of the objective built environment. This has implications for measurement (Are both self-report and objective measurements required?) and intervention design (Is awareness and education required in addition to environmental change?).
METHODS
The Neighborhood Quality of Life Study (NQLS) is an observational epidemiologic study designed to allow comparisons between neighborhoods stratified based on their "walkability" characteristics and median household income derived from Geographic Information System (GIS) and Census data. The primary aim of NQLS was to evaluate the relationships between objectively defined, high-walkable and low-walkable neighborhoods and physical activity levels in randomly selected adults. Study participants completed questionnaires about perceptions of their neighborhoods, to complement the objective measures obtained using GIS. Those participants in Seattle, King County, WA (N = 495) who indicated in an initial survey there were children in their household were sent a second survey concerning their children's activity patterns, if their child was between the ages of 4 and 18 yr (N = 323). The present analyses were based on the parental responses to the supplemental survey they completed about their children.
Sampling design.
A walkability index was used to (a) select study neighborhoods and (b) characterize the environment around each participant's home. The index has been described elsewhere (12) and its validity was supported by associations with adults" physical activity (13) and body mass index (11). Walkability was defined for each block group (subunit of census tract) in King County. The GIS-based walkability index combined indicators shown to be related to active transportation (19). The index was calculated using the following equation:
;)
Net residential density was the ratio of residential units to the land area devoted to residential use. Retail floor area ratio was the retail building square footage divided by retail land square footage. A low ratio indicated a retail development likely to have substantial parking, whereas a high ratio indicated a more pedestrian-oriented design. Intersection density, which indicated street connectivity, was measured by the ratio between the number of true intersections (three or more legs) to the land area of the block group in acres. Land use mix indicated the degree to which a diversity of land use types was present in a block group. The mix measure considered five land use types: residential, retail, entertainment, office, and institutional. Values were normalized between 0 and 1, with 0 being single use and 1 indicating an even distribution across the five uses. The four components of the walkability index were normalized for each block group using a Z score.
Neighborhood walkability was then computed for each block group. Block groups were ranked and divided into deciles based on the walkability index. The top four and bottom four deciles represented high- and low-walkability areas. Similarly, the median household income data for each block group were deciled and categorized into high and low income. Annual household income values less than $15,000 and greater than $150,000 were not included in the deciling process to avoid skewing the data with outliers. The second, third, and fourth deciles constituted the "low-income" category, and the seventh, eighth, and ninth deciles made up the "high-income" category. Thus, a 2 × 2 walkability-income matrix, was created. Based on the GIS and census data, clusters of contiguous block groups with similar characteristics were identified. Candidate clusters, or neighborhoods, were visited by investigators, and final selections were made based on the walkability and income scores and the availability of participants for recruitment. Sample size estimations indicated that four neighborhoods would be required for each quadrant in the design matrix.
Participant recruitment.
Adults between the ages of 20 and 65 yr with telephone numbers and living within study neighborhoods were targeted for recruitment. Names, addresses, and telephone numbers were provided by a commercial marketing company. Of adults who were contacted by telephone, 28% provided informed consent and completed a survey at time 1. In the first survey, 495 adults reported having a child in the household. Those participants whose child was between 4 and 18 yr of age were asked to complete a supplemental survey 6 months later about the child's physical activity and child-related environmental variables. The response rate to the second survey was 93%. The study was approved by the institutional review board at San Diego State University and Cincinnati Children's Hospital and Health Center.
Objective measures.
The walkability index was used both to select block groups for the sampling framework and to act as an independent variable at the neighborhood level, with participants coded as living in high- or low-walkable neighborhoods. A more proximal walkability index was also calculated for each participant. Home addresses were geocoded and entered into the GIS. An individual walkability index was then computed for the 1-km parcel of land around each participant's home. A 1-km radius was drawn from each participant's home, and a polygon was created using the street network to define the edge of the shape. This type of bounded area is called a "network buffer." Within the buffer, all components of the walkability index were then computed, and an individual-level walkability score was created for each participant.
Subjective measures.
After providing written informed consent, study participants were sent a self-report survey assessing sociodemographic variables and their perception of the local environment. The main questions of relevance to this paper were assessed by the validated Neighborhood Environment Walkability Scale (18). This scale assessed environmental characteristics believed to be associated with physical activity: (a) residential density; (b) proximity and ease of access to nonresidential land uses, such as restaurants and retail stores (land use mix-diversity and land use mix-access); (c) street connectivity; (d) walking or cycling facilities, such as sidewalks and pedestrian or bike trails; (e) aesthetics; (f) pedestrian traffic safety; and (g) crime safety. With the exception of the residential density and land use mix-diversity subscales, items were scaled from 1 (strongly disagree) to 4 (strongly agree), with higher scores indicating a more favorable value of the environmental characteristic.
Residential density items asked about the frequency of various types of neighborhood residences, from single-family detached homes to 13-story or higher apartments and condominiums, with a response range of 1 (none) to 5 (all). Each dwelling category was weighted to estimate the number of household units in the same land area, compared with one household for single-family detached homes. Land use mix-diversity was assessed by the walking proximity from home to various types of stores and facilities, with responses ranging from 1- to 5-min walking time (coded as 5) to 30-min or more walking time (coded as 1). Higher scores on land use mix-diversity indicated closer average proximity. With the exception of the residential density subscale, all subscale scores were calculated as the mean across the subscale items. Residential density was a sum of the six items weighted based on their estimated household density. An additional variable, number of stores and facilities within a 20-min walk, was also created from the land use mix questions. Scoring information is available at www.drjamessallis.sdsu.edu.
Parents answered supplemental questions with regard to the youngest or only child in the household between 4 and 18 yr of age. Questions were adapted from previous studies of children's play environments (21) and the Centers for Disease Control Kids Walk to School program (http://www.cdc.gov/nccdphp/dnpa/kidswalk/). Parents reported how many days per week, in an average school week, their child walked or biked, rode in a car or school bus, or took public transportation to and from school. Similar measures have shown good (intraclass correlations: 0.7-0.8) test-retest reliability (1,20). Data were collected throughout an entire year, balanced by quadrant, to allow for variations in activity because of weather. Eleven parental concerns (on a scale of 1 to 4, with higher scores indicating more concerns) were assessed and the mean calculated. Items included crime (stranger danger, gangs, bullying), too much traffic in my neighborhood, too much traffic at school, cars drive too fast through my neighborhood, no (or inadequate) sidewalks or bikeways on the route to school, school is too far away, not enough time, child's after-school schedule, easier to drop off child on the way to work, child would be walking or biking alone to school, child does not want to, or like to, walk or bicycle to school. Cronbach's alpha for the 11 items was 0.80.
Statistical analyses.
The present study concerns travel to school so only data for children aged 5-18 yr, without a disability, were considered. The first aim was to examine the association of objective and perceived environmental characteristics and parent concerns with active commuting to school. Logistic regression analyses were conducted and controlled for child age (5-11 vs 12-18 yr), child gender, and responding parent's education (completed college or not), a proxy for socioeconomic status. The outcome was actively commuting at least once a week or not. Odds ratios and 95% confidence intervals are provided for all analyses. In the first model, the objective and self-report environmental variables and parental concern scores were entered separately as independent variables. To allow meaningful odds ratios, the scores for the independent variables were converted to categoric variables either below the median, or equal to the median and above, or tertiles, depending on the score range. The odds ratios represent the likelihood of a child actively commuting at least once per week.
Results for the tertile categories are only presented as the highest tertile versus the lowest (the referent) to simplify the presentation of the numerous analyses. Results for the median split categories are presented for median and above scores, with below the median scores as the reference. The exception to this is parental concerns, where scores below the median (i.e., low concern) are presented with high concern as the reference. Thus, all odds ratios were expected to be greater than 1. The neighborhood-level walkability and income categories-already grouped as high and low-were also investigated. When neighborhood income was a variable in the model, parent education was excluded to avoid multicollinearity. Interactions between income × neighborhood-level walkability index and parent concerns × neighborhood-level walkability index were tested for the outcome commuting at least once a week. The interaction between income × neighborhood-level walkability was also investigated for the outcome parent concerns.
The second aim of the current analyses was to examine whether parental concerns varied by environmental characteristics. The relationship between the perceived and objective environment variables and parent concerns (dichotomized at the median) was investigated using logistic regression controlling for child age, child gender, and parent education. The same variables were entered and tested as for the active commuting outcome.
The third aim was to examine whether perceived environment and parental concerns explained the association of objective walkability and active commuting to school and to investigate the role of the perceptions compared with objective measures. Each significant subjective perception variable and the objective neighborhood-level walkability variable were entered into a series of "combined" models. The relationship between these perceptions and neighborhood-level walkability was also examined to check for collinearity. Each combined model adjusted for child age, child gender, and parental education. Separate models were constructed for each perceived environment variable because of expected multicollinearity that is well known in the urban planning field (19).
RESULTS
Complete data were available for 259 parents and children. The mean age of parents for this group was 44.0 yr (SD 7.4). The mean age of the children they reported on was 11.3 yr (SD 4.0). Half the responding parents were women (49.4%), 83.3% were white, and 66.0% had a college education or above. Half the children were girls (48.6%), and 45.9% were age 12 yr or older. Parents reported that 18.1% of the children walked or biked to and from schools 5 d·wk−1, 25.1% (N = 65) actively commuted at least once a week. No significant differences were found between child commuting behavior by child age group, child gender, parent education, or parent gender. Parents of children aged 12-18 yr had significantly fewer concerns (P = 0.004), but child gender and parent education or gender were not significantly related to parent concerns.
Table 1 presents the odds ratios and 95% confidence intervals for the two outcomes, active commuting at least once a week and parental concerns. Each variable presented was entered in a separate logistic regression model controlling for participant demographics. Parental concerns demonstrated the strongest relationship with active commuting. Of the perceived environmental variables, stores within a 20-min walk and neighborhood aesthetics showed the strongest associations with commuting behavior. Both individual- and neighborhood-level walkability were significantly related to walking or biking to and from school. Residential density was the component of the walkability index most related to active commuting.
TABLE 1: Logistic regression models controlling for age, gender and parent education for the outcomes active commuting and parent concerns.
Fewer environmental variables were related to parent concerns. Pedestrian safety, which was not related to commuting behavior, was related to parental concerns. Parent concerns about their child walking or biking to school were significantly inversely associated with residential density and neighborhood-level walkability.
The interactions between neighborhood walkability and income were significant for both commuting behavior and parent concerns. The interaction terms are presented at the base of Table 1. These results are displayed graphically with percentage of active commuters in Figure 1 and percent with low parental concerns in Figure 2. In low-walkability neighborhoods, income was not related to commuting behavior. In high-walkability neighborhoods, those children in high-income neighborhoods were more likely to actively commute. Likewise, in neighborhoods with high walkability, income played a greater role in explaining parental concerns. Those in high-walkability, low-income neighborhoods were less likely to have low concerns (i.e., they had the most concerns). The interaction between walkability and parent concerns was also significant (Fig. 3). In neighborhoods where it was easy to walk for transportation, parental concerns played a greater role in predicting child active commuting behavior.
FIGURE 1: Income × neighborhood walkability interaction for the outcome children active commuting to and from school.
FIGURE 2: Income × neighborhood walkability interaction for the outcome parent concerns.
FIGURE 3: Neighborhood walkability × parent concern interaction for the outcome children active commuting to and from school.
The results of the combined analyses are presented in Table 2, indicating the relationships between the independent variables (objective neighborhood walkability and the perceived variables that were significantly associated with active commuting) and the outcome, active commuting to school at least once a week. Three patterns of findings were documented. Parental concerns and aesthetics showed independent associations with the outcome, with both remaining significant. Stores within a 20-min walk and perceived walk and bike facilities remained significant, but the relationship between walkability and commuting behavior was no longer significant. When entered with walkability, perceived street connectivity and land use mix-access were no longer related to the outcome.
TABLE 2: Logistic regressions exploring the relationships between the perceived and objective variables and their combined associations with the outcome active commuting, controlling for child age, gender, and parental education.
DISCUSSION
Understanding the factors that explain children's walking and biking to school could contribute to the development of effective interventions to increase these behaviors, which in turn could improve the health of young people. These interventions may involve developing more pedestrian friendly neighborhoods around homes and schools and increasing awareness of such supportive environments. School site policies and school walk zone guidelines could also be informed by the identification of consistent correlates of children's commuting activity. In the present study, both the design of neighborhoods and parental concerns were significantly associated with children's active commuting to school. These findings suggest that both environmental changes and educational programs directed to parents may be needed to substantially increase active commuting to and from school.
Making environmental changes to facilitate active commuting to school could have widespread and relatively permanent effects, so it is important to identify the specific environmental factors with the strongest associations with behavior. Both objectively measured and perceived environmental variables were significant. Overall neighborhood walkability, objectively measured at either the neighborhood or individual level, was significantly related to active commuting. This suggests that in walkable neighborhoods more children may live within walking and biking distance of their schools. A similar walkability index was related to physical activity (13), active transportation, and body mass index in adults (11). Thus, designing neighborhoods to facilitate walking to destinations appears to have health-related benefits for both adults and youth. Present findings add to the rationale for changing policies to favor or require building neighborhoods that support physical activity. Policies that encourage the building or renovation of schools in the middle of existing neighborhoods rather than placing them on busy streets on the edge of communities could make active commuting accessible to more children.
Results with perceived environmental variables both confirmed and extended findings with objectively measured environments. Significant associations of active commuting with perceived land use mix and street connectivity confirmed the objective walkability findings. Additional variables, not included in the walkability index, that were related to children's active commuting were reported number of stores within a 20-min walk, walking and biking facilities, and aesthetics of the neighborhood. These findings are consistent with those of Ewing et al. (10) that children were more likely to walk to school when sidewalks were present.
Overall, however, a parental concerns scale was the strongest explanatory variable. When parents had few concerns, children were five times as likely to actively commute to school as when parents had many concerns. A simple interpretation, that this association implies parental education could increase children's active commuting, should be resisted. Parental concerns were related to important safety issues such as presence and quality of walking and biking facilities as well as traffic danger. Thus, interventions that tried to change parental perceptions about their children's active commuting without ensuring the safety of the commuting environment could imperil children's safety. Numerous Safe Routes to School Programs (e.g., those sponsored by the National Highway Traffic Safety Administration, CDC, or the Pedestrian and Bicycle Information Center) and initial evidence that improving safety alone (4), in combination with promotion activities, can be effective (24). Thus, the most responsible approach to reducing parental concerns about children's active commuting may be improvement in the walking and biking infrastructure, protection from traffic, and aesthetics of routes to schools. In addition, it is important to make the public aware of such improvements and promote active commuting.
The interactions among environmental variables, parental concerns, and neighborhood income suggest that solutions may need to be tailored for various situations. In low-walkable neighborhoods, income and parental concerns were less related to active commuting than in high-walkable neighborhoods. Presumably, this is because most children live too far from schools to walk or bike in low-density, disconnected neighborhoods. If the distance is too far, maybe nothing can be done to increase children's active commuting to school. The interaction between walkability and income indicated that walkability was related to active commuting only in high-income neighborhoods (Fig. 1). It is essential to understand the reasons for the relatively low rate of active commuting in high-walkability, low-income neighborhoods, and part of the answer may be found in the high rates of parental concerns in this group (Fig. 2). Parental concerns about safety, infrastructure, and aesthetics may negate the potential of children to walk or bike to school. Similarly, Figure 3 shows the lowest active commuting rates in neighborhoods with high-walkability and high parental concerns. These findings, that demographic and social factors can overcome favorable walkability characteristics, should be a priority for further research.
The interaction between walkability and concerns also suggests there may be parents who are overly concerned and are preventing their children from walking to school, despite supportive environments. Disproportionate media coverage of child abduction stories, for example, may be influencing parent concerns. Thus, in addition to improving low-walkability areas, education of parents in high-walkability areas may be required.
In her review (17), McMillan cautions against expectations of change from simplistic interventions that do not consider the multiple influences on commuting behavior. The combined analyses underscored the complexity of the variables under study as they relate to children's active commuting to and from school. The main purpose of these analyses was to identify perceived variables that may explain the relationship between objectively measured walkability and the active commuting outcome. Number of stores within a 20-min walk of home and walking and biking facilities appeared more strongly related to behavior than the objective environment. When each of these variables was in the model simultaneously with walkability, they remained significant. In contrast, the objective neighborhood walkability variable became nonsignificant. This pattern of findings may indicate that nearby stores and pedestrian or biking facilities are the "active ingredients" of neighborhood walkability and are particularly supportive of active commuting to school. Parental concerns and aesthetics were independently related to active commuting to school, but did not significantly reduce the impact of neighborhood walkability. This pattern suggests that active commuting could be increased by improving all these variables. A third pattern was found whereby perceived land use mix-access and street connectivity became nonsignificant when included in the model with neighborhood walkability. This is likely because these perceived variables being highly related to components of the objective walkability index, as demonstrated by the high odds ratios.
A key strength of the study was the use of both objective and perceived measures of the environment. Although only one region was used in this analysis, participants were drawn from 16 diverse neighborhoods that varied substantially in income and walkability. Additionally, a nationally representative study found that 25% of children walked or biked to school at least one time per week (8). This is identical to the percent of active commuters identified in the present sample, suggesting it may be representative of the broader population. The National Longitudinal Study of Adolescent Health also found a similar percentage (14). Limitations include a small sample size and cross-sectional data, which limit the ability to infer causal relationships. Parental reports of children's mode of travel to school have not been validated, but this salient and discrete behavior is expected to be well known to parents and relatively easy to report (1,20). Measures of parental concerns about the children commuting to school require further validation and reliability testing. The location of the schools, distance from home, and walkability of the school neighborhoods are also important elements that may explain commuting behavior, but which were not included in these analyses. For the older adolescents, driving behavior, which was not assessed, may also be relevant. Inability to include these variables likely had the effect of underestimating associations with environmental variables. Clustering by school was also not addressed in the present analyses.
One of the key recommendations stemming from a recent Institute of Medicine report calls for improvements of routes for walking and biking to school and other built environment factors (15). Safety-related improvements should be a high priority, particularly because 13% of traffic fatalities involve pedestrians and, of those, 16% involved a child under 18 yr of age (9). Often these deaths are caused by lack of street crossings. Children cannot be encouraged to walk to school until more is done to ensure the safety of their trips. Intervention studies have shown safety improvements are feasible and appear to be effective (4,24).
The present study demonstrates the importance of a pedestrian-friendly neighborhood for children's activity levels and parent perceptions, which in turn are strongly related to children's active commuting. Future studies should consider using an experimental design to compare commuting rates before and after an environmental intervention alone, and before and after a combined environmental and parent-education intervention. This would also allow for analyses of mediators and moderators. Collecting objective measures of commuting behavior, as well as children's perceptions of barriers, would greatly improve the findings of such a study. Interventions may need to be tailored to neighborhoods with different walkability characteristics and populations that vary in income.
The research presented in this paper was funded by grant HL67350 from the National Heart, Lung, and Blood Institute of the National Institutes of Health
Thanks to Dr. Greg Norman for his advice on presentation and interpretation of the mediator analyses
REFERENCES
1. Alexander, L. M., J. Inchley, J. Todd, D. Currie, A. R. Cooper, and C. Currie. The broader impact of walking to school among adolescents: seven day accelerometry based study.
BMJ. 331:1061-1062, 2005
3. Black, C., A. Collins, and M. Snell. Encouraging walking: the case of journey to school trips in compact urban areas.
Urban Studies 38:1121-1141, 2001.
4. Boarnet, M. G., C. L. Anderson, K. Day, T. McMillan, and M. Alfonzo. Evaluation of the California safe routes to school legislation: urban form changes and children's active transportation to school.
Am. J. Prev. Med. 28(2S2):134-140, 2005.
5. Braza, M., W. Shoemaker, and A. Seeley. Neighborhood design and rates of walking and biking to elementary school in 34 California communities.
American Journal of Health Promotion 19:128-136, 2004.
6. Centers for Disease Control and Prevention. Barriers to children walking and biking to school-United States, 1999.
JAMA. 288:1343-1344, 2002.
7. Centers for Disease Control and Prevention. Barriers to walking and biking to school-United States, 1999.
MMWR. 51:701-704, 2002.
8. Cooper, A. R., A. S. Page, L. J. Foster, and D. Qahwaji. Commuting to school: are children who walk more physically active?
Am. J. Prev. Med. 25:273-276, 2003.
9. Ernst, M., and B. McCann. In:
Mean Streets 2002, Washington, DC: Surface Transportation Policy Project, 2002, p. 15.
10. Ewing, R., W. Schroeer, and W. Greene. School location and student travel: Analysis of factors affecting mode choice.
Transportation Research Record 1895:55-63, 2004.
11. Frank, L. D., M. A. Andresen, and T. L. Schmid. Obesity relationships with community design, physical activity, and time spent in cars.
Am. J. Prev. Med. 27(2):87-96, 2004.
12. Frank, L. D., J. F. Sallis, B. E. Saelens and W. Baachman. Many pathways from land use to health: association neighborhood walkability and active transportation, body mass index, and air quality
J. Am. Plann. Assoc 72:75-87, 2006.
13. Frank, L. D., T. L. Schmid, J. F. Sallis, J. Chapman, and B. E. Saelens. Linking objectively measured physical activity with objectively measured urban form: findings from SMARTRAQ.
Am. J. Prev. Med. 28(2 (Suppl 2)):117-125, 2005.
14. Gordon-Larsen, P., M. C. Nelson, and K. Beam. Associations among active transportation, physical activity, and weight status in young adults.
Obes. Res. 13(5):868-875, 2005.
15. Preventing childhood obesity: health in the balance. In:
Institute of Medicine, J. P. Koplan, C. T. Liverman, and V. I. Kraak (Eds.). Washington, DC, 2004, pp. 5-19.
16. McCann, B., and B. DeLille. Mean streets 2000: pedestrian safety, health and federal transportation spending. In:
Surface Transportation Policy Project, 2000, p. 16.
17. McMillan, T. E. Urban form and a child's trip to school: the current literature and a framework for future research.
Journal of Planning Literature 19:440-456, 2005.
18. Saelens, B. E., J. F. Sallis, J. B. Black, and D. Chen. Neighborhood-based differences in physical activity: an environment scale evaluation.
Am. J. Public Health 93(9):1552-1558, 2003.
19. Saelens, B. E., J. F. Sallis, and L. D. Frank. Environmental correlates of walking and cycling: findings from the transportation, urban design, and planning literatures.
Ann. Behav. Med. 25(2):80-91, 2003.
20. Sallis, J. F., J. E. Alcaraz, T. L. McKenzie, and M. F. Hovell. Predictors of change in children's physical activity over 20 months: Variations by gender and level of adiposity.
Am. J. Prev. Med. 16:222-229, 1999.
21. Sallis, J. F., P. R. Nader, S. L. Broyles, et al. Correlates of physical activity at home in Mexican-American and Anglo-American preschool children.
Health Psychol. 12(5):390-398, 1993.
22. Sallis, J. F., J. J. Prochaska, and W. C. Taylor. A review of correlates of physical activity of children and adolescents.
Med. Sci. Sports Exerc. 32:963-975, 2000.
23. Sleap, M., and P. Warburton. Are primary school children gaining heart health benefits from their journeys to school?
Child Care Health Development 19:99-108, 1993.
24. Staunton, C.E., D. Hubsmith, and W. Kallins. Promoting safe walking and biking to school: the Marin County success story.
Am. J. Public Health 93:1431-1434, 2003.
25. Timperio, A., D. Crawford, A. Telford, and J. Salmon. Perceptions about the local neighborhood and walking and cycling among children.
Prev. Med. 38:39-47, 2004.
26. Tudor-Locke, C., B.E. Ainsworth, L.S. Adair, and B.M. Popkin. Objective physical activity of Filipino youth stratified for commuting mode to school.
Med. Sci. Sports Exerc. 35:465-471, 2003.
27. Tudor-Locke, C., B. E. Ainsworth, and B. M. Popkin. Active commuting to school.
Sports Med. 31:309-313.
