associations between environmental characteristics and active commuting to school among children: a...
TRANSCRIPT
Associations Between Environmental Characteristics and ActiveCommuting to School Among Children: a Cross-sectional Study
Marie-Jeanne Aarts & Jolanda J. P. Mathijssen &
Johannes A. M. van Oers & Albertine J. Schuit
# International Society of Behavioral Medicine 2012
AbstractBackground Active commuting to school can contribute toactive living among children, and environmental character-istics might be related to transportation mode to school.Purpose The purpose of this study is to explore the associ-ation between physical and social environmental character-istics in the home, neighborhood, and school environmentand walking and bicycling to school.Method Data were collected among parents (n05,963) ofchildren of primary schools in four Dutch cities. Parentsreported mode of transportation to school, and individual,home environmental, neighborhood, and school environ-mental characteristics. Social as well as physical character-istics were included for the home and neighborhoodenvironment. Multilevel multinomial logistic regressionanalyses were conducted to quantify the association be-tween environmental characteristics and walking and bi-cycling to school.
Results Three quarter of all children usually commute toschool by active transportation, but age and distance fromhome to school were important prerequisites. Besides homeenvironmental characteristics, lower neighborhood socio-economic status was negatively associated with walking[odds ratio (OR)00.51] and bicycling (OR00.86). Per-ceived social safety was positively related to walking andbicycling (OR01.04 for both), as was perceived social co-hesion (OR01.04 and 1.02 for walking and bicycling).Living in the city center was positively associated withwalking (OR01.91), whereas living in a city green neigh-borhood was negatively associated with walking and bi-cycling (OR00.48 and 0.76, respectively). Traffic safety asperceived by school boards was positively associated withbicycling (OR01.25).Conclusion This study shows that there is a relation betweenseveral characteristics in the home, neighborhood, and schoolenvironment and walking and bicycling to school amongDutch primary school children. Especially the social neigh-borhood characteristics were related to active commuting.Therefore, apart from providing a physical infrastructure thatfacilitates safe and convenient active commuting to school,policy makers should be aware of opportunities to facilitateactive commuting by social initiatives in local communities.
Keywords Children .Physical activity .Active commuting .
Environment
Introduction
As in many other western countries, the majority of primaryschool children in The Netherlands do not meet the recom-mended health guidelines for physical activity [1]. Becauseof the health risks related to this lack of physical activity [2,3], it is important to find appropriate ways to stimulate
M.-J. Aarts : J. J. P. Mathijssen : J. A. M. van OersTilburg School of Social and Behavioral Sciences, DepartmentTranzo, Scientific Center for Care and Welfare, Tilburg University,PO Box 90153, 5000 LE Tilburg, The Netherlands
M.-J. Aarts : J. A. M. van Oers :A. J. SchuitNational Institute for Public Health and the Environment,Public Health and Health Services Division,PO Box 1, 3720 BA Bilthoven, The Netherlands
M.-J. Aarts (*)CAPHRI School for Public Health and Primary Care,Department of Health Services Research, Maastricht University,PO Box 616, 6200 MD Maastricht, The Netherlandse-mail: [email protected]
A. J. SchuitDepartment of Health Sciences and EMGO Institute for Health andCare Research, VU University Amsterdam,De Boelelaan 1105,1081 HVAmsterdam, The Netherlands
Int.J. Behav. Med.DOI 10.1007/s12529-012-9271-0
physical activity among children. Active commuting to andfrom school is an opportunity for children to be physicallyactive on a regular basis. Accelerometer data have shownthat children that go to school by means of active commut-ing are more physically active when compared to childrenthat use motorized travel, both during the journey fromhome to school itself [4], as well as during other timeperiods [5]. Maximizing the number of children that activelycommute to school can therefore be seen as a promisingpublic health strategy, to which also policy sectors outsidethe public health domain may contribute [6, 7].
According to the social ecology theory of health fromDahlgren and Whitehead [8], in addition to individual char-acteristics, both social and physical environmental character-istic may influence people’s health behavior, such as physicalactivity. With regard to active commuting to school, researchindicates that apart from individual characteristics of the childand the parents, environmental characteristics such as real andperceived safety and walking and bicycling facilities are re-lated to the mode of transportation to school [9, 10], andhence, optimizing (the perception of) these environmentalcharacteristics may be a valuable policy approach to increasethe number of children involved in active commuting.
Because parents act as gatekeepers for their children’scommuting behavior [11], both objective environmentalcharacteristics such as urban density, mixed land use plan-ning, proximity, and connectivity as well as parents’ per-ception of their living environment such as road safety mayplay an important role in the choice for transportation modeto school [11–13]. Moreover, not only the physical environ-mental characteristics but also social environmental charac-teristics may be related to active commuting to school [13].Panter and colleagues further indicate that three componentsof the living environment should be considered in relation tochildren’s commuting behavior: the neighborhood aroundthe home, the route from home to school, and the schoolenvironment [14]. It is therefore important to consider allthese different types of environmental characteristics (ob-jectively measured vs subjectively measured, physical andsocial environmental characteristics, and home, neighbor-hood, and school setting) when studying the associationbetween environmental characteristics and active commut-ing behavior of children. Furthermore, specific environmen-tal characteristics may be related to different forms of activecommuting, i.e., walking or bicycling [15, 16].
For policy makers to optimize the revenues of theirenvironmental policies aimed at stimulating active commut-ing among children, it is important to have insight into thecomplex relation between environmental characteristics andactive commuting to school. Much research addressing theabovementioned themes has been traditionally conducted inthe USA [9, 10] and Australia [17–20]. European countriesand cities however have a different infrastructure e.g. in
terms of street connectivity pattern, and therefore, Europeanstudies addressing this theme are specifically important aswell [14]. A UK study by Page et al. for example showedthat shorter distance from home to school was related toactive commuting to school [21]. Bringolf-Isler et al.showed that major road crossings and distance were signif-icantly related to non-active transportation and that increas-ing age of the child, daycare attendance, parental safetyconcerns, number of cars in the household, and belongingto French-speaking population were significantly associatedwith increased regular car trips in a Swiss setting [22].Panter et al. have shown that in a UK setting, parentalattitudes and safety concerns, the presence of social supportfrom parents and friends, and parent-reported neighborhoodwalkability were predictors of active commuting, with chil-dren receiving peer and family support and living in sup-portive environments (e.g., neighborhoods with high socialcohesion) being more likely to walk or cycle [23]. Thesestudies show that the relation between the environment andactive commuting behavior may be even country specific,and in order to assist local policy makers in designingpolicies that stimulate active commuting among children,country-specific results are warranted.
Therefore, the aim of this study is to explore the correlationbetween (perceived) characteristics in the home, neighbor-hood, and school environment related to active commutingto school among Dutch children. This study includes social aswell as physical environmental characteristics and examinesthe association with both walking and bicycling.
Methods
Study Setting
Cross-sectional data were collected between September2007 and January 2008 from parents of children of 42primary schools in four medium-sized Dutch cities in thesouthern part of The Netherlands. The number of inhabitantsranges from 77,450 to 201,259, and the degree of urbaniza-tion ranges from 727 to 1,716 citizens per km2. Althoughone city was somewhat smaller and less urbanized com-pared to the other cities, the municipalities were comparableregarding the composition of their population such as thepercentage of non-Western immigrants (9.9–13.4 %) andpercentage of inhabitants aged 0–14 years (16.7–17.6 %).The selection procedures and characteristics of the partici-pating cities are described in more detail elsewhere [24].
Study Population
Data were collected among parents of children aged 4–12 years. In the Netherlands, children in this age group attend
Int.J. Behav. Med.
primary school, which, in most cases, is close to or within thearea of residence.We invited all regular primary schools in thefour cities that were not already participating in projects aimedat stimulating physical activity among children (n0115). Wefirst invited them by letter, followed by a telephone call. Of the115 invited schools, 42 agreed to participate.
As outlined elsewhere [24], the schools in our study wererepresentative of the total population of schools in the par-ticipating municipalities in terms of school size, socioeco-nomic status (SES), and the type of neighborhood. Becauseno medical or physical measurements were conducted andconsidering the negligible (psychological) burden to fill inthe questionnaire, no ethics approval was required accordingto the Dutch Central Committee on Research InvestigatingHuman Subjects. Via a take-home envelope handed to theirchildren, parents were given written information aboutthe study and by returning the questionnaire they gaveconsent for the inclusion of their data in the study. Intotal, 11,094 parents were provided with a question-naire. Response rate was 60 %, resulting in 6,624returned questionnaires. During data entry, 12 question-naires could not be read, and 11 questionnaires wereremoved because they were completely empty, leaving6,601 questionnaires for analysis.
Questionnaires and Measures
This study encompassed a questionnaire for parents and aquestionnaire for the school board. The questionnaire forparents was based on questionnaires used in previous Dutchresearch [25, 26] and included the following topics: usualmode of transportation to school, individual factors (genderand age of the child, and parental report of height andweight of their child), social home environmental factors(parental education, parental ethnic background, workingsituation of parent(s), number of siblings, number of daysper week that the child goes home after school time), phys-ical home environmental characteristics (distance fromhome to school and number of cars in the household),perceived social neighborhood characteristics (social safety,social cohesion, degree of unoccupied houses, presence oftrash and litter, and presence of dog dirt), perceived physicalneighborhood characteristics (degree of high- vs low-risebuildings, presence of green, presence of water, traffic situ-ation, quality of sidewalks and bike lanes, and diversity ofroutes), and perceived physical characteristics of the schoolenvironment (traffic safety around school). In addition, typeof neighborhood and neighborhood SES were determinedbased on the child’s postal code [27, 28] (postal codereported by parents), and additional physical school envi-ronmental characteristics (traffic safety around school andsufficiency of bicycle shed at school as perceived by schoolboards) were derived from a questionnaire provided to the
board of the participating schools. The exact formulation ofthe items in the questionnaire, the calculation of all variablesin the analysis, and the descriptive data for each variable(such as mean or frequencies) are summarized in theAppendix.
Conceptually related items were summed when internalconsistency was acceptable (Cronbach’s alpha>0.6); other-wise, items were treated separately. Missing values were notimputed, unless it concerned a missing value on one of theitems of a sum score consisting of more than four items. Inthat case, the missing value was replaced by the mean of theother values. If more than one item was missing within onesum score, the sum score was not calculated. Throughoutthe questionnaire for parents, “neighborhood” was definedas the area that could be reached by parents in 10 to 15 minby foot or in 5 to 8 min by bike from the respondent’sresidence (street network distance). This matches the gener-al perception of a typical Dutch neighborhood and, in com-parison with distances in meters, distances in minutes aremore easily interpreted by the respondents [29, 30]. Themultinomial outcome measure is usual mode of transporta-tion to school encompassing the following categories: (1)walking, (2) bicycling, and (3) the reference category inac-tive transportation (on the back of parent’s bike or in abuggy, on the back of parent’s moped/scooter, brought bycar, or by bus).
Statistical Analyses
Questionnaires were excluded from further analyses becauseof missing values on the outcome measure transport modal-ity (n0366), and three important potential confounders: ageof the child (n02), distance from home to school (n0113),and parental education (n0154). Furthermore, question-naires of children living more than 3 days per week onanother address than the address described in the question-naire (n035) and questionnaires of children with severedisabilities that could hamper active commuting (n060)were removed. Some questionnaires had to be removedbecause of more than one exclusion criterion, and thefinal data base thus encompassed 5,963 respondents.Based on our power analysis described elsewhere [24],our study provided adequate power to detect smalleffects (f 200.02).
Descriptive analyses were conducted with SPSS, version17.0. T tests and chi-square tests were performed for con-tinuous and categorical variables respectively to assess dif-ferences (p<0.05) in characteristics (age, gender, BMI of thechild, and transport modality of the child, and parentaleducation and net household income) between the respond-ents that were included in the final analyses (n05,963)versus the respondents that were excluded for analysis(n0638).
Int.J. Behav. Med.
Multilevel multinomial logistic regression analyses wereconducted with SAS version 9.2, using PROC GLIMMIX.Inactive transportation was regarded as the reference cate-gory in all analyses. Random intercepts were allowed tocorrect for the clustering of the data within schools in allanalyses. After crude bivariate analyses in which the asso-ciation of each individual independent variable with modeof transportation was calculated, the association betweeneach individual independent variable with mode of trans-portation was calculated adjusted for age of the child anddistance from home to school (which were considered asimportant preconditions for active commuting to school)and parental education (which is considered as a goodindicator for SES in the Netherlands [31, 32] and was seenas an important possible confounder). In all analyses, paren-tal education was indicated by highest completed level ofeducation of the parent who filled in the questionnaire; inthe majority of cases, this was either the biological mother(86.4 %) or biological father (12.5 %).
Finally, in order to explore the association between envi-ronmental characteristics and active commuting to schooladjusted for the other environmental characteristics, a mul-tilevel forward sequential multinomial logistic regressionanalysis was performed. In a sequential analysis, variablesenter the equation in a theory-based order [33]. According tothe theoretical framework of Dahlgren and Whitehead [8], itwas assumed that proximal variables (individual character-istics and home environmental characteristics) are moreclosely related to children’s active commuting behavior thandistal characteristics (neighborhood and school environ-ment). Based on previous results from this research project[34], it was assumed that social neighborhood character-istics were more closely related to active commuting thanphysical neighborhood characteristics. Hence, the first stepof the sequential analysis consisted of a block of proximalvariables (individual and both social and physical homeenvironmental variables), followed by the second stepwhich introduced a block of distal (neighborhood) socialvariables to the model. During the third step, a block ofdistal (neighborhood) physical variables was added to themodel. The last step comprised the introduction of a blockof physical school environmental variables to the model.All steps in the sequential analysis were adjusted for ageof the child, distance from home to school, and parentaleducation. In order to prevent important variables to beexcluded from the model in a forward analysis too easily,a more liberal probability level of p>0.15 was chosen todecide on deletion of variables from the model [33, 35].The sequential analyses ended when all variables in themodel reached significance. In the final multivariate mod-els, only those variables with a p value <0.05 are shown.Prior to entry into the multivariate models, correlationsbetween (continuous) independent variables were checked
for collinearity, but none of the correlations exceeded theexclusion criterion of r>0.5 [33]. Because there werehardly any differences in characteristics between boysand girls and because there was no significant associationbetween gender and mode of transportation to school, datafor boys and girls were combined in the regressionanalyses.
Results
With regard to selective dropout, there were no differencesin age, gender, BMI, and mode of transportation of the child(walking, bicycling, inactive transportation) of the respond-ents that were included in the analyses, compared to thoserespondents that were excluded for analysis. However, therespondents that were excluded for analyses had a signifi-cant lower parental education level and net household in-come. Table 1 shows the characteristics of the studypopulation in terms of age, gender, BMI, the percentage ofoverweight and obese children (as determined by age- andgender-specific cutoff points as provided by Cole et al.[36]), parental education level, and net household income.This table also shows that approximately three quarter of allchildren usually commute to school by means of activetransportation (walking or cycling). The descriptive data inFig. 1 show that with increasing age, fewer children aregoing to school by means of inactive transportation (broughtto school by car or on the back of parent’s bike, moped, or ina buggy), in favor of children commuting to school by bike.Figure 2 depicts that within a distance of 1 km betweenhome and school, the majority of children (approximately70 %) commutes to school by foot. With increasing distanceup to 5 km from home to school, fewer children go to schoolby foot, in favor of children going to school by bike or byinactive transportation.
The results from the bivariate analyses are not shown, butcan be obtained from the corresponding author on request.Tables 2 and 3 summarize the results of the adjusted andfinal multivariate multilevel analyses and show the associa-tion between several environmental characteristics and ac-tive commuting to school. Only statistically significantassociations (p value <0.05) are shown, but the nonsignifi-cant results can be obtained from the correspondingauthor on request. The results of the final multivariateanalyses will be addressed per block of variables below(Tables 2 and 3).
Individual and Social and Physical Home EnvironmentalCharacteristics
In the final multivariate analyses, age of the child (years)was positively related to walking [odds ratio (OR)01.31]
Int.J. Behav. Med.
and bicycling (OR01.71) to school, whereas parental edu-cation was positively related to bicycling only (OR01.10).
Living in a single-parent family with the parent working 12–36 h per week was negatively associated with walking toschool (OR00.61). The number of siblings was positivelyrelated to walking (OR01.44) and bicycling (OR01.24), aswell as the number of days per week the child goes directly tohome after school (OR01.18 and 1.13 for walking and bi-cycling, respectively). The distance from home to school(kilometers) was negatively related to walking (OR00.18)and bicycling (OR00.70) to school, as well as thenumber of cars in the household (OR00.58 and 0.49for walking and bicycling, respectively).
Social Neighborhood Characteristics
A lower neighborhood SES was negatively associated withwalking (OR00.51) and bicycling (OR00.86) to school.Perceived social safety was positively related to walkingand bicycling to school (OR01.04 for both walking andbicycling), as was perceived social cohesion (OR01.04and 1.02 for walking and bicycling, respectively). The per-ceived presence of dog dirt was positively associated withwalking to school (OR01.19).
Physical Neighborhood Characteristics
Living in a city center type of neighborhood was pos-itively associated with walking to school (OR01.91),whereas living in a city green type of neighborhoodwas negatively associated with walking (OR00.48) andbicycling (OR00.76) to school. The perceived presenceof green was negatively associated with walking toschool (OR00.89), whereas the perceived diversity ofroutes was positively associated with bicycling to school(OR01.12).
Table 1 Characteristics of the study population
Total(n05,963)
Boys(n03,001)
Girls(n02,950)
Age (years) 7.8 (2.4) 7.8 (2.4) 7.8 (2.4)
BMIa (kg/m2) 16.3 (2.7) 16.3 (2.7) 16.2 (2.6)
Overweightb (%) 9.9 8.5 11.3
Obesityb (%) 2.8 3.1 2.4
Parental education (%)
Lowc 28.3 28.3 28.2
Intermediated 35.3 35.0 35.7
Highe 36.4 36.7 35.1
Net household income(Euros per month)
2,788 (1,351) 2,797 (1,326) 2,781 (1,376)
Usual mode of transportation to school
% Walking 43.4 43.5 43.2
% Bicycling 31.8 32.2 31.4
% Inactive 24.8 24.3 25.4
Values are mean (SD), unless otherwise specified. Twelve respondentshad a missing value on the gender of their childa Based on parental self-report of height and weight of their childb Based on age- and gender-specific cutoff points as provided by Coleet al. [36]c No education, primary education, lower general secondary education,or lower vocational educationd Higher general secondary education, pre-university education, orintermediate vocational educatione Higher vocational education or university
Fig. 1 Usual mode of transportation to school by age. In The Nether-lands, children aged 4–12 years are educated together at the sameprimary school. In the current study sample, 3 children in the lowestgrade were aged 3 years and 12 children in the highest grade were aged13 years. For this figure, these children were included in the lowest(4 years) and highest (12 years) age groups, respectively
Fig. 2 Usual mode of transportation to school by distance from hometo school
Int.J. Behav. Med.
Table 2 Association between environmental characteristics and walking to school
Variable Range/coding Adjusteda analyses Final multivariateb analysesOR (CI) OR (CI)
Individual and social and physical home environmental characteristics
Gender 0–1 (00boy, 10girl)
Age of the child (years) 3–13c 1.33 (1.29–1.38) 1.31 (1.26–1.37)
Parental education 1–8 0.95 (0.90–1.00)
Parental ethnic background 0–1 (00Dutch,10non-Dutch)
1.63 (1.24–2.14)
Working situation of parentsd Dummy coded: two-parentfamily, one parentworks 36 h per weekor more, one parentworks 12–36 h perweek is reference
0.58 (0.39–0.85)(single-parent family,parent works 12–36 hper week)
0.61 (0.40–0.94)(single-parent family,parent works12–36 h per week)
1.44 (1.16–1.79)(two-parent family,one parent works36 h per week ormore, one parentworks less than12 h per week)
Number of siblings 0–5 1.47 (1.33–1.63) 1.44 (1.29–1.61)
Distance from home to school (km) 0.5–6.0 0.19 (0.17–0.21) 0.18 (0.16–0.21)
Number of days childgoes home after school
0–5 1.21 (1.13–1.30) 1.18 (1.09–0.27)
Number of cars in the household 0–2 0.69 (0.59–0.81) 0.58 (0.48–0.69)
Social neighborhood characteristics
Neighborhood SES −4−4 (higher scoresrepresent lower SES)
0.58 (0.51–0.67) 0.51 (0.44–0.60)
Social safety 5–25 1.04 (1.02–1.07) 1.04 (1.01–1.07)
Social cohesion 6–30 1.04 (1.02–1.06) 1.04 (1.02–1.07)
Degree of unoccupied houses 1–5
Presence of trash and litter 1–5
Presence of dog dirt 1–5 1.15 (1.09–1.22) 1.19 (1.12–1.27)
Physical neighborhood characteristics
Type of neighborhoodd Dummy coded: citynon-center is referencee
1.91 (1.06–3.47)(city center)
0.48 (0.34–0.68)(city green)
Degree of high- vs low-rise buildings 2–10 (higher scorerepresent morehigh-rise buildings)
Presence of green 1–5 0.89 (0.82–0.98)
Presence of water 1–5
Traffic situation 5–25 (higher scoresrepresent less favorabletraffic situation)
Quality of sidewalks and bike lanes 4–20
Diversity of routes 1–5 1.09 (1.01–1.17)
Physical school environmental characteristics
Traffic safety around school(as perceived by parents)
0–1 (0–unsafe, 10safe) 0.73 (0.61–0.88) 0.70 (0.58–0.85)
Traffic safety around school(as perceived by school board)
1–5 (higher scoresrepresent higherperceived safety)
Int.J. Behav. Med.
Physical School Environmental Characteristics
The traffic safety around school as perceived by parents wasnegatively associated with walking and bicycling to school(OR00.70 and 0.72, respectively) indicating that childrenfrom parents that perceive the school environment as safeare less likely to walk or bicycle to school. Conversely, thetraffic safety around school as perceived by the school boardwas positively associated with bicycling to school (OR0
1.25). The sufficiency of the school’s bicycle shed (as per-ceived by the school board) was positively associated withwalking to school (OR01.91) and negatively associatedwith bicycling to school (OR00.69).
Discussion
Main Findings of This Study
This study showed that apart from characteristics in thehome environment such as the number of siblings, thenumber of days a child goes home directly after school,and the number of cars in het household, also characteristicsof the physical and social neighborhood environment arerelated to active commuting among Dutch primary schoolchildren. Neighborhood SES, perceived social cohesion,and perceived social safety were positively associated withactive commuting, as well as physical neighborhood char-acteristics such as living in a city center type of neighbor-hood type and the diversity of routes showed significant
positive associations with active commuting. Perceivedpresence of green was negatively associated with walkingand bicycling to school.
Discussion of the Main Findings and Comparisonwith Previous Research
This study confirmed that short distance from home toschool and a higher age of the child are factors related toactive commuting to school, a finding already known fromprevious studies [4, 19, 22]. With regard to the home envi-ronmental characteristics, this study showed that the numberof siblings was positively related to walking and bicyclingto school. This might be explained by the fact that siblingswalk together to school (safety in numbers), but as data fromother studies are somewhat inconsistent [18, 19], this topicrequires further study. In contrary to the general idea thatparents who are working (nearly) full time are more likely todrive their children to school by car [12, 22, 37], this studydid not show a consistent relation between the workinghours of the parents and their children’s mode of transpor-tation to school. This may be partly explained by the lownumber of respondents in some of the working situationcategories. We did find a positive association between thenumber of days a child goes home after school and walkingand bicycling to school. This might indicate that not the totalnumber of working hours by parents but the opportunity tosupervise the child during the journey from school to homeor a parent being present at home after school time may bean important factor related to active commuting. The
Table 2 (continued)
Variable Range/coding Adjusteda analyses Final multivariateb analysesOR (CI) OR (CI)
Sufficiency of bicycle shed at school(as perceived by school board)
1–4 1.44 (1.12–1.84) 1.91 (1.28–2.87)
All analyses are multilevel (random intercepts were allowed) to correct for the clustering of the data within schools. Only statistically significantassociations (p value<0.05) are shown
OR odds ratio, CI confidence intervala Adjusted for age of the child, distance from home to school, and parental educationb Adjusted for age of the child, distance from home to school, and parental education. In addition, the final multivariate model included thefollowing independent variables: working situation of parents, number of siblings, number of days child goes home after school, number of cars inthe household, neighborhood SES, social safety, social cohesion, presence of dog dirt, type of neighborhood, presence of green, diversity of routes,traffic safety around school (as perceived by parents), traffic safety around school (as perceived by school board), and sufficiency of bicycle shed atschool (as perceived by school board). Only those variables that show a statistical significant association with walking to school are shown in thistablec In The Netherlands, children aged 4–12 years are educated together at the same primary school. In the current study sample, 3 children in thelowest grade were aged 3 years and 12 children in the highest grade were aged 13 yearsd Only the dummy categories that show significant associations (p value <0.05) with walking to school are displayed (dummy categories that didnot show a significant association with walking to school were omitted)e Results of for work area type of neighborhood are as not shown because of extreme low number of respondents living in this type of neighborhood (n09)
Int.J. Behav. Med.
Table 3 Association between environmental characteristics and bicycling to school
Variable Range/coding Adjusteda analyses Final multivariateb analysesOR (CI) OR (CI)
Individual and social and physical home environmental characteristics
Gender 0–1 (00boy, 10girl)
Age of the child (years) 3–13c 1.73 (1.67–1.80) 1.71 (1.64–1.78)
Parental education 1–8 1.07 (1.02–1.13) 1.10 (1.04–1.16)
Parental ethnic background 0–1 (00Dutch,10non-Dutch)
Working situation of parentsd Dummy coded: two-parent family, oneparent works 36 h perweek or more, oneparent works 12–36 hper week is reference
1.48 (1.07–2.05)(two-parent family,both parents work12–36 h per week)
1.33 (1.07–1.65)(two-parent family,one parent works 36 hper week or more,one parent works lessthan 12 h per week)
Number of siblings 0–5 1.26 (1.14–1.39) 1.24 (1.12–1.38)
Distance from hometo school (km)
0.5–6.0 0.68 (0.64–0.72) 0.70 (0.66–0.75)
Number of days child goeshome after school
0–5 1.14 (1.06–1.22) 1.13 (1.05–1.22)
Number of cars in the household 0–2 0.56 (0.48–0.65) 0.49 (0.41–0.58)
Social neighborhood characteristics
Neighborhood SES −4−4 (higher scoresrepresent lower SES)
0.91 (0.83–0.99) 0.86 (0.77–0.95)
Social safety 5–25 1.04 (1.03–1.08) 1.04 (1.01–1.07)
Social cohesion 6–30 1.03 (1.01–1.04) 1.02 (1.00–1.04)
Degree of unoccupied houses 1–5
Presence of trash and litter 1–5
Presence of dog dirt 1–5
Physical neighborhood characteristics
Type of neighborhoodd Dummy coded:city non-centeris referencee
0.76 (0.58–0.99)(city green)
Degree of high- vslow-rise buildings
2–10 (higher scorerepresent morehigh-rise buildings)
0.95 (0.91–0.99)
Presence of green 1–5
Presence of water 1–5
Traffic situation 5–25 (higher scoresrepresent lessfavorable trafficsituation)
0.98 (0.96–0.99)
Quality of sidewalks and bike lanes 4–20
Diversity of routes 1–5 1.16 (1.09–1.25) 1.12 (1.04–1.21)
Physical school environmental characteristics
Traffic safety around school(as perceived by parents)
0–1 (0–unsafe, 10safe) 0.76 (0.64–0.91) 0.72 (0.60–0.87)
Traffic safety around school(as perceived by school board)
1–5 (higher scoresrepresent higherperceived safety)
1.22 (0.99–1.50) 1.25 (1.03–1.53)
Int.J. Behav. Med.
importance of social support from parents in stimulatingactive school transportation was shown in other studies aswell [23], and likewise, data from Switzerland show apositive association between daycare attendance and regularcar trips to school [22]. A study among Spanish adolescentsfurther showed that especially higher maternal professionallevel might be inversely associated with active commutingto school [38]. Comparable with other studies [15, 18, 22],the number of cars in the household was negatively associ-ated with active transportation to school.
Regarding the social neighborhood characteristics, chil-dren in lower SES neighborhoods were less likely to go toschool by means of active transportation. Previous researchhas shown a negative association between the likelihood ofwalking/bicycling home from school in deprived neighbor-hoods as well [14, 21]. Moreover, the current study under-lined the importance of perceived social neighborhoodcharacteristics such as perceived social safety and perceivedsocial cohesion, which were consistently related to walkingand bicycling to school. Social contacts that facilitate col-lectively commuting to school and parents’ perceptions ofsocial neighborhood characteristics were shown to be par-ticularly important for primary school children and adoles-cents in other studies as well [39, 40]. Together, thesefindings suggest an important role for social neighborhoodcharacteristics in relation to active commuting to school.The rather counterintuitive finding that the perceived pres-ence of dog dirt was positively related to walking to schoolmight reflect the walkability of those areas, which attracts
both dog walkers and active commuters to school. It is alsopossible that active commuters are actually out in theirneighborhood noticing these factors.
With regard to the physical neighborhood character-istics, living in a city center neighborhood was positive-ly related to walking to school, whereas living in a citygreen neighborhood was negatively associated withwalking and bicycling to school. In general, city centerneighborhoods are considered more walkable, due to theproximity of facilities. Together with a discouragingparking environment for cars in city center neighbor-hoods, this might be an explanation for the abovemen-tioned findings. Moreover, the perceived presence ofgreen was negatively associated with walking to schoolin the present study, which indicates that although livingin a green environment may stimulate active commutingamong adults [41], for children this might not be thecase. Although we did include items on perceived trafficsituation and quality of sidewalks and bike lanes in ourstudy, these were not significantly related to eitherwalking or cycling to school, which is in contrast withmany other studies showing the possible associations offor example major road crossings [22], road safety [13],road density [14], and the presence of walk and bikepaths [18] with active commuting among children. Inthe current study, quality of foot paths and bike laneswere pooled to one variable to prevent multicolinearity,while it may be more realistic to assess the associationbetween foot paths and walking and bike lanes and
Table 3 (continued)
Variable Range/coding Adjusteda analyses Final multivariateb analysesOR (CI) OR (CI)
Sufficiency of bicycle shed at school(as perceived by school board)
1–4 0.73 (0.57–0.93) 0.69 (0.54–0.90)
All analyses are multilevel (random intercepts were allowed) to correct for the clustering of the data within schools. Only statistically significantassociations (p value <0.05) are shown
OR odds ratio, CI confidence intervala Adjusted for age of the child, distance from home to school, and parental educationb Adjusted for age of the child, distance from home to school, and parental education. In addition, the final multivariate model included thefollowing independent variables: working situation of parents, number of siblings, number of days the child goes home after school, number of carsin the household, neighborhood SES, social safety, social cohesion, presence of dog dirt, type of neighborhood, presence of green, diversity ofroutes, traffic safety around school (as perceived by parents), traffic safety around school (as perceived by school board), and sufficiency of bicycleshed at school (as perceived by school board). Only those variables that show a statistical significant association with bicycling to school are shownin this tablec In The Netherlands, children aged 4–12 years are educated together at the same primary school. In the current study sample, 3 children in thelowest grade were aged 3 years and 12 children in the highest grade were aged 13 yearsd Only the dummy categories that show significant associations (p value <0.05) with bicycling to school are displayed (dummy categories that didnot show a significant association with bicycling to school were omitted)e Results of for work area type of neighborhood are as not shown because of extreme low number of respondents living in this type of neighborhood (n09)
Int.J. Behav. Med.
bicycles separately. In the UK study from Panter et al.,it was concluded that both attitudinal and environmentalperceptions of parents were associated with children’sactive commuting behavior [23]. Possibly, the overallneighborhood type included as a variable in the presentstudy already accounted for much of the differences inthe physical environmental characteristics. An alternativeexplanation for the lack of a clear association betweenperceived physical neighborhood characteristics and ac-tive commuting might be found in the specific Dutchinfrastructure, which possibly already provides childrenwith a facilitating physical environment with regard towalking and bicycling.
While the traffic safety around school as perceivedby the school board showed a positive association withbicycling to school, children of parents reporting thatthey perceive the traffic situation around school as safewere less likely to walk or bicycle to school. As thisstudy has a cross-sectional design and causality cannotbe demonstrated, a possible explanation for this findingmight be that parents who do walk or bicycle with theirchildren to school have more experience with the(unsafe) traffic situation around school. The fact thatonly 23.8 % of the school boards in this study perceivethe traffic situation around their school as safe indicatedthat there is room for improvement of traffic situationaround primary schools. Furthermore, the capacity ofbicycle sheds at primary schools as perceived by schoolboards was positively related to walking to school, andnegatively related to bicycling to school. Although thisfinding may seem awkward at first, this might reflectreverse causality as schools where many children walkand few children bicycle to school perceive to haveenough bicycle shed capacity and vise versa.
Strengths and Limitations of This Study
This large-scale explorative study addresses a broad spec-trum of social as well as physical environmental associa-tions with walking and bicycling to school, which is in linewith the social ecological perspective on health as presentedby Dahlgren and Whitehead [8]. However, because of thecross-sectional design, no causal relationships can be dem-onstrated, and this hampers the interpretation of some of thefindings. Moreover, this study relied mostly on parentalperceptions of the neighborhood characteristics. Althoughthe neighborhood characteristics as perceived by parentsmay be of crucial importance [11, 42], and social neighbor-hood characteristics are also difficult to measure objectively,measurement of neighborhood characteristics by means ofneighborhood audits or geographical information systemsmay be a valuable tool in future research. This study alsorelied on parental report of their children’s usual
commuting behavior to school because more objectivemethods such as accelerometer data are often unable toadequately measure bicycling, and therefore are notsuitable for measuring active commuting. Further, thisstudy only asked parents to report the usual mode oftransportation to school, and we implicitly assumed thatthis was also the usual mode of transportation backfrom school to home in the afternoon. Although it ispossible that there are temporal differences in transportmode (i.e., between morning and afternoon trips) [37],UK data show that travel mode to and from school ishighly correlated [4]. Due to the abovementioned diffi-culties, there are no data available on the validity andreliability of the measures used in this study at thismoment.
With regard to the analyses of this study, a sequentialregression analysis was applied, so that variables enter themodel in a theory-driven manner. Additional analyses(which can be obtained from the corresponding author onrequest) have shown that altering the sequence of entry ofblocks of variables (i.e., reverse the order of entry of socialand physical neighborhood variables into the model) did notmodify the results of the study. Lastly, because the studywas conducted in four medium-sized Dutch cities, caution iswarranted in generalizing the findings of this study to otherareas.
Conclusions
This study shows the relation between several character-istics in the home, neighborhood, and school environmentand walking and bicycling to school. This study suggests apotentially important role for social characteristics at theneighborhood level in relation to walking and bicycling toschool. Therefore, apart from providing a physical infra-structure that facilitates safe and convenient active commutingto school, policy makers should be aware of opportunities tofacilitate active commuting by social initiatives in local com-munities. Future research should focus on the effectiveness ofsuch initiatives to further underpin the possible role of socialenvironmental characteristics in relation to children’s com-muting behavior.
Acknowledgments This project was supported by a grant fromZonMw, The Netherlands Organization for Health Research andDevelopment (grant number 71600003). The authors thank TNOQuality of Life and the ChecKid research team for providing thequestionnaires of their research projects. We thank Denise Hung,Eva Laan, Anne Snijders, and Coryke van Vulpen for theirassistance in the data collection, and Karin van Beek, TimGotjé, and Nienke Raaijmakers for their assistance in the datacleaning.
Int.J. Behav. Med.
Appendix
AOverview
ofthedepend
entandindepend
entvariablesinclud
edin
theanalyses
Con
cept
Num
berof
items
Exact
form
ulationof
items
Cronb
ach’s
alph
aAnswer
catego
ries
Recod
e/sum
scorecalculation
andrang
eDescriptiv
edata:mean(SD)or
prop
ortio
ns
Dependent
variable
(questionn
aire
forparents)
Mod
eof
transportatio
nto
scho
ol
1Whatistheusualmod
eof
transportatio
nto
scho
olforyo
urchild
?
NA
Walking
,bicycling;
ontheback
ofparent’sbike/in
abu
ggy,on
the
back
ofparent’smop
ed/scooter,
brou
ghtby
car,by
bus,other
(with
possible
specificationin
open
questio
n)
Ontheback
ofparent’sbike/in
abu
ggy,on
theback
ofparent’s
mop
ed/scooter,brou
ghtby
car
andby
buswererecodedinto
inactiv
etransportatio
n(reference
catego
ry)
43.4
%walking
,31
.8%
bicycling,
24.8
%inactiv
e
Individu
alfactors(questionn
aire
forparents)
Genderof
the
child
1Whatisthegend
erof
your
child
?NA
Boy,girl
NA
50.3
%bo
y,49
.5%
girl,0.2%
missing
value
Age
ofthechild
1How
oldisyo
urchild
?NA
Openqu
estio
nRange
from
3to13
yearsa
7.82
(2.39)
Socialandph
ysical
homeenvironm
entalfactors(questionn
aire
forparents)
Parental
education
1Whatisyo
urhigh
est
completed
education?
NA
Noeducation,
prim
aryeducation,
lowervo
catio
naledu
catio
n,lower
generalsecond
aryeducation,
interm
ediate
vocatio
nal
education,
high
ergeneral
second
aryeducationor
pre-un
iversity
education,
high
ervo
catio
naleducation,
university
Parentaleducationwas
treatedas
anordinalvariable
rang
ingfrom
1to
8,high
erscores
represent
high
ereducation
4.1%
noeducation,
2.5%
prim
aryeducation,
10.8
%lower
vocatio
naleducation,
10.8
%lower
general
second
aryeducation,
24.8
%interm
ediate
vocatio
nal
education,
10.6
%high
ergeneralsecond
aryeducation
orpre-un
iversity
education,
24.8
%high
ervo
catio
nal
education,
11.5
%un
iversity
Parentalethn
icbackgrou
nd1
Whatisyo
urcoun
try
ofbirth?
NA
The
Netherlands,Surinam
,The
Netherlands
Antilles,Turkey,
Morocco,othercoun
try
Surinam
,The
Netherlands
Antilles,
Turkey,Morocco,andother
coun
trywererecodedinto
“other
coun
try”
84.0
%bo
rnin
theNetherlands,
15.5
%bo
rnin
another
coun
try,0.5%
missing
value
Working
situationof
parent(s)
2Which
situationismost
applicable
toyo
uand
your
partner?
(respo
ndents
wereaskedto
indicate
theircivilstatus
and
working
situation)
NA
For
civilstatus:married/official
registratio
nof
cohabitatio
nand
livingtogether;liv
ingtogether,
notmarried;un
married,never
been
married;divo
rced,
separated;
widow
(er);endu
ring
relatio
nship,
livingapart;liv
ing
with
myparents;other(open
answ
ers)
Based
ontheircivilstatus,
respon
dentswerecatego
rizedas
either
sing
le-parentfamily
ortwo-parent
family.Num
berof
working
hoursof
therespon
dent
and(ifapplicable)thepartnerwas
calculated.Full-tim
ejobwas
recodedinto
38hperweek.
Part
timejobworking
hourswere
3.1%
sing
le-parentfamily,
parent
works
less
than
12h
perweek;
5.5%
sing
le-parent
family,parent
works
12–36
hperweek;
2.3%
sing
le-parent
family,parent
works
36hper
weekor
more;
2.2%
two-
parent
family,bo
thparents
workless
than
12hperweek;
Appendix
Int.J. Behav. Med.
Appendix
A(con
tinued)
Con
cept
Num
berof
items
Exact
form
ulationof
items
Cronb
ach’s
alph
aAnswer
catego
ries
Recod
e/sum
scorecalculation
andrang
eDescriptiv
edata:mean(SD)or
prop
ortio
ns
filledin
bytherespon
dents(open
questio
n).Other
answ
ercatego
ries
wererecodedin
0h
perweek,
except
forrespon
dents
who
indicatedthat
they
areself-
employ
ed/entrepreneurwhich
was
recodedinto
38h.
Working
hoursof
therespon
dentsand(if
applicable)thepartnerwere
collapsed
into
threecatego
ries:
less
than
12hperweek,
12–36
hperweek,
and36
hperweekor
more
1.4%
two-parent
family,on
eparent
works
12–36
hper
week,
oneparent
works
less
than
12hperweek;
6.9%
two-parent
family,bo
thparentswork12–36
hper
week;
23.2
%two-parent
family,on
eparent
works
36h
perweekor
more,on
eparent
works
less
than
12hperweek;
47.5
%two-parent
family,o
neparent
works
36hperweekor
more,on
eparent
works
12–
36hperweek;
5.6%
two-
parent
family,bo
thparents
work36
hperweekor
more;
2.3%
missing
value
For
working
situationof
the
respon
dent
and(ifapplicable)
theirpartner:full-tim
ejob;
part-
timejob(…
hours);ho
usew
ife/
househusband
/volun
tary
work;
outof
work/un
able
towork;
stud
enton
adaytim
ecourse;
other(openansw
ers)
Finally,respo
ndentswereclassified
into
nine
catego
ries
(see
next
column).These
nine
catego
ries
werecodedinto
eigh
tdu
mmy
variables,with
two-parent
family,
oneparent
works
36hperweek
ormore,on
eparent
works
12–
36hperweekas
reference
catego
ryNum
berof
siblings
1How
manybrothers
and
sistersdo
esyo
urchild
have?
NA
Openqu
estio
n.Range
from
0to
8siblings.5,
6,7,
or8siblings
werecombinedinto
onecatego
ry“5
siblings
ormore”
10.6
%0siblings;54
.5%
1sibling;
24.7
%2siblings;
6.8%
3siblings;1.9%
4siblings;1.5%
5siblings
ormore
Distancefrom
hometo
scho
ol
1Whatisthedistance
(in
kilometers)
betweenyo
urho
useandyo
urchild
’sscho
ol?
NA
Lessthan
1km
,1–
2km
,2–
5km
,morethan
5km
Lessthan
1km
was
recodedinto
0.5km
,1–2km
was
recodedinto
1.5km
,2–5km
was
recodedinto
3.5km
,morethan
5km
was
recodedinto
6.0km
50.0
%less
than
1km
;28
.3%
1–2km
;18
.7%
2–5km
;3.0%
morethan
5km
Num
berof
days
per
weekthat
thechild
goes
home
afterscho
oltim
e
1How
manydays
perweek
does
your
child
goho
me
afterscho
oltim
e?
NA
0,1,
2,3,
4,5days
Range
from
0to
5days.
1.7%
0days;1.6%
1day;
7.6%
2days;19
.2%
3days;
16.5
%4days;51
.9%
5days;1.4%
missing
value.
Int.J. Behav. Med.
Appendix
A(con
tinued)
Con
cept
Num
berof
items
Exact
form
ulationof
items
Cronb
ach’s
alph
aAnswer
catego
ries
Recod
e/sum
scorecalculation
andrang
eDescriptiv
edata:mean(SD)or
prop
ortio
ns
Num
berof
cars
inho
usehold
1How
manycars
does
your
householdhave?
NA
Nocar,1car,2or
morecars
2or
morecars
was
recodedinto
2cars
5.3%
nocar;49
.3%
1car;
44.9
%2or
morecars;0.5%
missing
value
Socialneighb
orho
odcharacteristics(questionn
aire
forparents)
Neigh
borhoo
dSES
1Whatisyo
urpo
stal
code?
NA
NA
Statusscorebasedon
thepo
stal
code
oftherespon
dent’saddress.
The
status
scoreisbasedon
percentage
unem
ploy
edpeop
le,
percentage
peop
lewith
low
education,andpercentage
oflow-
incomeho
useholds.Scorescould
rang
efrom
−4to
4,high
erscores
representlower
neighb
orho
odSES
−0.18
(1.16)
0.4%
missing
value
Socialsafety
5The
streetsin
the
neighb
orho
odarewelllit
atnigh
t.
0.60
9Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Negativelywordeditemswere
reversed.Sum
scorewas
compu
ted,
rang
efrom
5to
25,
high
erscores
representgreater
social
safety
19.19(3.79)
0.9%
missing
value
Pedestrians
andcyclistscan
beeasily
seen
bythe
residentsof
theho
uses
intheneighb
orho
od.
The
risk
ontrou
bleandpetty
crim
ein
theneighb
orho
odmakes
itun
safe
towalk
throug
htheneighb
orho
odwith
mychild
during
daytim
e.The
risk
ontrou
bleandpetty
crim
ein
theneighb
orho
odmakes
itun
safe
towalk
throug
htheneighb
orho
odwith
mychild
atnigh
t.The
neighb
orho
odissafe
enou
ghformychild
towalkor
play
onthestreets
with
outsupervision(during
daytim
e).
Socialcohesion
6Peoplein
theneighb
orho
odarewillingto
help
each
other.
0.84
1Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Sum
scorewas
compu
ted,
rang
efrom
6to
30,high
erscores
representgreatersocial
cohesion
23.20(4.76)
0.9%
missing
value
Int.J. Behav. Med.
Appendix
A(con
tinued)
Con
cept
Num
berof
items
Exact
form
ulationof
items
Cronb
ach’s
alph
aAnswer
catego
ries
Recod
e/sum
scorecalculation
andrang
eDescriptiv
edata:mean(SD)or
prop
ortio
ns
The
neighb
orho
odisatig
htcommun
ityThe
peop
lein
the
neighb
orho
odcanbe
trusted.
Ingeneral,thepeop
lein
the
neighb
orho
odgetalon
gwell.
Peoplein
theneighb
orho
odsharethesameno
rmsand
values.
There
aremanychild
ren
livingin
the
neighb
orho
od.
Degreeof
unoccupied
houses
1There
areno
tmany
unoccupied
build
ings
orho
uses
inthedirect
environm
ent.
NA
Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Scoresrang
efrom
1to
5;high
erscores
representm
oreun
occupied
houses
1.56
(1.12)
1.5%
missing
value
Presenceof
trashandlitter
1There
isno
tmuchtrash
orlitterin
thedirect
environm
ent.
NA
Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Scoresrang
efrom
1to
5;high
erscores
representmoretrashor
litter
1.89
(1.18)
1.2%
missing
value
Presenceof
dog
dirt
1There
isno
tmuchdo
gdirtin
thedirect
environm
ent.
NA
Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Scoresrang
efrom
1to
5;high
erscores
representmoredo
gdirt
3.38
(1.46)
1.1%
missing
value
Phy
sicalneighb
orho
odcharacteristics(questionn
aire
forparents)
Typ
eof
neighb
orho
od1
Whatisyo
urpo
stal
code?
NA
NA
Based
onpo
stal
code
ofresidence
address:city
center,city
non-
center,city
green,
towncenter,
ruralarea,workarea.Dum
my
coded,
city
non-center
isreference
4.9%
city
center;50
.6%
city
non-center;40
.3%
city
green;
0.5%
towncenter;2.5
%rural
area;0.2%
workarea;1.1%
missing
value
Degreeof
high
-vs.
low-rise
build
ings
2There
areno
tmanyhigh
-rise
build
ings
inou
rneighb
orho
od.
0.60
6Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Negativelywordeditemswere
reversed.Sum
scorewas
compu
ted,
rang
efrom
2to10
;high
erscores
representmore
high
-risebu
ildings.
3.43
(1.92)
1.7%
missing
value
There
aremanylow-rise
build
ings
inou
rneighb
orho
od.
Presenceof
greenin
the
neighb
orho
od
1There
ismuchgreenin
the
direct
environm
ent.
NA
Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Scoresrang
efrom
1to
5;high
erscores
represent
moregreen
4.21
(1.11)
0.8%
missing
value
Int.J. Behav. Med.
Appendix
A(con
tinued)
Con
cept
Num
berof
items
Exact
form
ulationof
items
Cronb
ach’s
alph
aAnswer
catego
ries
Recod
e/sum
scorecalculation
andrang
eDescriptiv
edata:mean(SD)or
prop
ortio
ns
Presenceof
water
inthe
neighb
orho
od
1There
ismuchwater
inthe
direct
environm
ent.
NA
Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree).
Scoresrang
efrom
1to
5;high
erscores
representmorewater
3.63
(1.48)
1.2%
missing
value
Traffic
situation
5There
aresufficient
pedestrian
crossing
sand
trafficlig
htsto
help
pedestrians(children)
crossing
busy
trafficpo
ints
intheneighb
orho
od.
0.65
5Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Negativelywordeditemswere
reversed.Sum
scorewas
compu
ted,
rang
efrom
5to
25;
high
erscores
representless
favo
rabletrafficsituation
16.78(4.51)
0.5%
missing
value
The
speedof
thetrafficin
the
neighb
orho
odisusually
low
(maxim
um30
km/h).
Mostdriversin
the
neighb
orho
oddrivetoo
fast.
There
isso
muchtrafficin
ourneighb
orho
odthat
itis
difficultor
inconv
enient
formychild
towalk
there.
There
arealotof
exhaust
fumes
(from
cars,bu
ses)
intheneighb
orho
od.
Qualityof
sidewalks
andbike
lanes
4Mostsidewalks
inthe
neighb
orho
odarewell
maintained.
0.60
3Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Sum
scorewas
compu
ted,
rang
efrom
4to
20;high
erscores
representbetterqu
ality
12.47(3.76)
3.2%
missing
value
Mostsidewalks
inthe
neighb
orho
odareseparated
from
thestreet
(byparked
cars,plants,concrete
obstacles).
Moststreetsin
the
neighb
orho
odhave
bike
lanes.
Mostbike
lanesin
the
neighb
orho
odareseparated
from
thestreet
(byparked
cars,plants,concrete
obstacles).
Diversity
ofroutes
1There
aremanydifferent
routes
intheneighb
orho
odmychild
cantake
toget
somew
here.
NA
Five-po
intLikert-type
scale
(stron
glydisagree
tostrong
lyagree)
Scoresrang
efrom
1to
5;high
erscores
representgreaterdiversity
3.84
(1.18)
0.8%
missing
value
Int.J. Behav. Med.
Appendix
A(con
tinued)
Con
cept
Num
berof
items
Exact
form
ulationof
items
Cronb
ach’s
alph
aAnswer
catego
ries
Recod
e/sum
scorecalculation
andrang
eDescriptiv
edata:mean(SD)or
prop
ortio
ns
Phy
sicalscho
olenvironm
entalcharacteristics(questionn
aire
forparentsandqu
estio
nnaire
forscho
olbo
ards)
Traffic
safety
arou
ndscho
olas
perceived
byparents
1Doyo
uthinkthetraffic
situationarou
ndyo
urchild
’sscho
olissafe?
NA
Yes;yes,prov
ided
that
Iwatch
my
child
carefully
;no
(reasoncould
bespecified)
“Yes”and“yes,prov
ided
that
Iwatch
mychild
carefully
”were
recodedinto“safe”
(cod
e1),“no”
was
recodedinto
“unsafe”
(cod
e0)
61.6
%un
safe;38
.4%
safe;
0.4%
missing
value
Traffic
safety
arou
ndscho
olas
perceived
byscho
oldirectionb
1How
doyo
ujudg
ethetraffic
safety
arou
ndyo
urscho
ol?
NA
Verysafe;safe;average;
unsafe;
very
unsafe
Scoresrang
efrom
1to
5,high
erscores
representgreatersafety.
2.84
(0.91)
Sufficiency
ofbicycleshed
atscho
olas
perceivedby
scho
oldirectionb
1Doesthecapacity
ofthe
bicycleshed
satisfy
the
needsof
your
scho
ol?
NA
Yes,morethan
sufficient;yes,
sufficient;no
,insufficient;no
,very
insufficient
Scoresrang
efrom
1to
4;high
erscores
representgreater
sufficiency
2.90
(0.70)
SDstandard
deviation,
NAno
tapplicable
aIn
The
Netherlands,childrenaged
4–12
yearsareeducated
together
atthesameprim
aryscho
ol.Inthecurrentstudy
sample,3child
renin
thelowestg
rade
wereaged
3yearsand12
child
renin
the
high
estgradewereaged
13years
bDerived
from
thequ
estio
nnaire
forscho
olbo
ard
Int.J. Behav. Med.
References
1. CBS Statline. http://www.cbs.nl. Accessed 21 Oct 2010.2. Boreham C, Riddoch C. The physical activity, fitness and health of
children. J Sports Sci. 2001;19(12):915–29.3. Dencker M, Andersen LB. Health-related aspects of objectively
measured daily physical activity in children. Clin Physiol FunctImaging. 2008;28(3):133–44.
4. Van Sluijs EM, Fearne VA, Mattocks C, Riddoch C, Griffin SJ,Ness A. The contribution of active travel to children’s physicalactivity levels: cross-sectional results from the ALSPAC study.Prev Med. 2009;48(6):519–24.
5. Cooper AR, Andersen LB, Wedderkopp N, Page AS, Froberg K.Physical activity levels of children who walk, cycle, or are drivento school. Am J Prev Med. 2005;29(3):179–84.
6. Richards R, Murdoch L, Reeder AI, Rosenby M. Advocacy foractive transport: advocate and city council perspectives. Int JBehav Nutr Phys Act. 2010;7:5.
7. Cole R, Burke M, Leslie E, Donald M, Owen N. Perceptions ofrepresentatives of public, private, and community sector institu-tions of the barriers and enablers for physically active transport.Transport policy. 2010;17(6):496–504.
8. Dahlgren G, Whitehead M. Policies and strategies to promote socialequity in health. Stockholm: Institute of Futures Studies; 1991.
9. Tudor-Locke C, Ainsworth BE, Popkin BM. Active commuting toschool: an overlooked source of childrens’ physical activity?Sports Med. 2001;31(5):309–13.
10. Kerr J, Rosenberg D, Sallis JF, Saelens BE, Frank LD, ConwayTL. Active commuting to school: associations with environmentand parental concerns. Med Sci Sports Exerc. 2006;38(4):787–94.
11. Faulkner GE, Richichi V, Buliung RN, Fusco C, Moola F. What’s“quickest and easiest?”: parental decision making about school tripmode. Int J Behav Nutr Phys Act. 2010;7:62.
12. Giles-Corti B, Kelty SF, Zubrick SR, Villanueva KP. Encouragingwalking for transport and physical activity in children and adoles-cents: how important is the built environment? Sports Med.2009;39(12):995–1009.
13. Panter JR, Jones AP, van Sluijs EM. Environmental determinantsof active travel in youth: a review and framework for futureresearch. Int J Behav Nutr Phys Act. 2008;5:34.
14. Panter JR, Jones AP, Van Sluijs EM, Griffin SJ. Neighborhood,route, and school environments and children’s active commuting.Am J Prev Med. 2010;38(3):268–78.
15. Grize L, Bringolf-Isler B, Martin E, Braun-Fahrlander C. Trend inactive transportation to school among Swiss school children and itsassociated factors: three cross-sectional surveys 1994, 2000 and2005. Int J Behav Nutr Phys Act. 2010;7:28.
16. De Vries SI, Hopman-Rock M, Bakker I, Hirasing RA, vanMechelen W. Built environmental correlates of walking and cy-cling in Dutch urban children: results from the SPACE study. Int JEnviron Res Public Health. 2010;7(5):2309–24.
17. Pikora T, Giles-Corti B, Bull F, Jamrozik K, Donovan R. Devel-oping a framework for assessment of the environmental determi-nants of walking and cycling. Soc Sci Med. 2003;56(8):1693–703.
18. Pont K, Ziviani J, Wadley D, Bennett S, Abbott R. Environmentalcorrelates of children’s active transportation: a systematic literaturereview. Health Place. 2009;15(3):827–40.
19. Timperio A, Ball K, Salmon J, Roberts R, Giles-Corti B, SimmonsD, et al. Personal, family, social, and environmental correlates ofactive commuting to school. Am J Prev Med. 2006;30(1):45–51.
20. Timperio A, Crawford D, Telford A, Salmon J. Perceptions aboutthe local neighborhood and walking and cycling among children.Prev Med. 2004;38(1):39–47.
21. Page AS, Cooper AR, Griew P, Jago R. Independent mobility,perceptions of the built environment and children’s participation
in play, active travel and structured exercise and sport: the PEACHProject. Int J Behav Nutr Phys Act. 2010;7:17.
22. Bringolf-Isler B, Grize L, Mader U, Ruch N, Sennhauser FH,Braun-Fahrlander C. Personal and environmental factors associat-ed with active commuting to school in Switzerland. Prev Med.2008;46(1):67–73.
23. Panter JR, Jones AP, van Sluijs EM, Griffin SJ. Attitudes, socialsupport and environmental perceptions as predictors of activecommuting behaviour in school children. J Epidemiol CommunityHealth. 2010;64(1):41–8.
24. Aarts MJ, Van de Goor IA, Van Oers HA, Schuit AJ. Towardstranslation of environmental determinants of physical activity inchildren into multi-sector policy measures: study design of a Dutchproject. BMC Publ Health. 2009;9(1):396.
25. Kruizinga AG, Bakker I, Stafleu A, De Vries SI. KOALA deel-project “Leefstijl en gewicht”. Ontwikkeling van de vragenlijst“Uw mening over eten en bewegen”. [KOALA project “Life styleand weight”. Development of the questionnaire “Your opinionabout food and exercise”.]. Zeist: TNO Quality of Life 2007.
26. De Jong E, Schokker DF, Visscher TL, Seidell JC, Renders CM.Behavioural and socio-demographic characteristics of Dutchneighbourhoods with high prevalence of childhood obesity. Int JPediatr Obes. 2011;6(3–4):298–305.
27. VROM. Nota Mensen, Wensen, Wonen. Wonen in de 21ste eeuw.[Memorandum People, Wishes, Living. Living in the 21st centu-ry.]. Den Haag 2000.
28. SCP. http://www.scp.nl. Accessed 21 Oct 2010.29. Colabianchi N, Dowda M, Pfeiffer KA, Porter DE, Almeida MJ,
Pate RR. Towards an understanding of salient neighborhoodboundaries: adolescent reports of an easy walking distance andconvenient driving distance. Int J Behav Nutr Phys Act. 2007;4:66.
30. Coulton CJ, Korbin J, Chan T, Su M. Mapping residents’ percep-tions of neighborhood boundaries: a methodological note. Am JCommunity Psychol. 2001;29(2):371–83.
31. Oakes JM, Rossi PH. The measurement of SES in health research:current practice and steps toward a new approach. Soc Sci Med.2003;56(4):769–84.
32. Van Berkel-Van Schaik AB, Tax B. Naar een standaard oper-ationalisatie van sociaal-economische status voor epidemiolo-gisch en sociaal medisch onderzoek. Reeks sociaal-economische gezondheidsverschillen, nr. 6 [Towards a standardoperationalisation of socio-economic status in epidemiologicaland social medical research. Series socio-economic healthinequalities, nr 6.] Ministry of Welfare, Health and Culture.Rijswijk, The Netherlands 1990.
33. Tabachnick BG, Fidell LS. Using multivariate statistics. 5th ed.Boston: Pearson Education Inc.; 2007.
34. Aarts MJ, Wendel-Vos W, van Oers HA, van de Goor IA, Schuit AJ.Environmental determinants of outdoor play in children: a large-scalecross-sectional study. Am J Prev Med. 2010;39(3):212–9.
35. Bendel RB, Afifi AA. Comparison of stopping rules in forwardregression. J Am Stat Assoc. 1977;72:46–53.
36. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH. Establishing astandard definition for child overweight and obesity worldwide:international survey. BMJ. 2000;320(7244):1240–3.
37. Mitra R, Buliung RN, Faulkner GE. Spatial clustering and thetemporal mobility of walking school trips in the Greater TorontoArea, Canada. Health Place. 2010;16(4):646–55.
38. Chillón P, Ortega FB, Ruiz JR, Pérez IJ, Martín-Matillas M,Valtueña J, Gómez-Martínez S, Redondo C, Rey-López JP,Castillo M, Tercedor P, Delgado M. Socioeconomic factorsand active commuting to school in Spanish adolescents; theAVENA study. European Journal of Public Health. 2009;19(5):470–6.
39. Hume C, Timperio A, Salmon J, Carver A, Giles-Corti B,Crawford D. Walking and cycling to school: predictors of
Int.J. Behav. Med.
increases among children and adolescents. Am J Prev Med.2009;36(3):195–200.
40. McDonald NC, Deakin E, Aalborg AE. Influence of the socialenvironment on children’s school travel. Prev Med. 2010;50 Suppl1:S65–8.
41. Wendel-Vos GC, Schuit AJ, de Niet R, Boshuizen HC, SarisWH, Kromhout D. Factors of the physical environment
associated with walking and bicycling. Med Sci Sports Exerc.2004;36(4):725–30.
42. Evenson KR, Birnbaum AS, Bedimo-Rung AL, Sallis JF,Voorhees CC, Ring K, et al. Girls’ perception of physical envi-ronmental factors and transportation: reliability and associationwith physical activity and active transport to school. Int J BehavNutr Phys Act. 2006;3:28.
Int.J. Behav. Med.