school physical activity environment related to student obesity and activity: a national study of...

Journal of Adolescent Health 45 (2009) S71–S81

Original article

School Physical Activity Environment Related to Student Obesity

and Activity: A National Study of Schools and Students

Patrick M. O’Malley, Ph.D.*, Lloyd D. Johnston, Ph.D., Jorge Delva, Ph.D.,and Yvonne M. Terry-McElrath, M.S.A.

Institute for Social Research, University of Michigan, Ann Arbor, Michigan

Manuscript received November 13, 2008; manuscript accepted April 20, 2009

Abstract Purpose: To explore whether characteristics of the U.S. secondary school physical activity environ-

*Address correspo

Research, University

Ann Arbor, MI 48106

E-mail address: po

1054-139X/09/$ – see

doi:10.1016/j.jadoheal

ment are associated with student body mass index (BMI) and physical activity.

Methods: This report uses data from two studies: Monitoring the Future (MTF; an annual nationally

representative survey of 8th-, 10th-, and 12th-grade public and private school students) and Youth,

Education, and Society (a survey of administrators in schools that have completed their 2-year partic-

ipation in the MTF study). School policies and programs related to various health issues, including

physical education (PE) and sports activity, were examined for relationships with student self-reported

height, weight, being active in sports, exercising vigorously, and participating in school athletics.

Results: The results show that in 2004–2007, the percentage of students who attended schools that

required PE in their grade differed sharply by grade level: 88% of 8th graders, 48% of 10th graders,

and 20% of 12th graders. There were few statistically significant associations between school PE

requirements and student BMI. The average percentage of students who participated in interscholastic

or varsity sports was associated at the bivariate level with a lower percentage of students being over-

weight in all three grades. Other measures of PE and sports activity showed varying associations with

BMI and physical activity measures.

Conclusions: Relationships between the school physical activity environment and student BMI and

physical activity were not uniformly strong. We conclude that, as currently practiced in schools, ex-

isting variations in physical activity policies may not be sufficient to produce discernible school-

wide differences; thus, there is a need for more vigorous PE programming than is typically provided.

� 2009 Society for Adolescent Medicine. All rights reserved.

Keywords: Body mass index; Obesity; Overweight; Physical education; Schools; Physical activity; Students

Two factors are generally considered important contribu-

tors to the recent increases in obesity among American youth:

insufficient physical activity and high caloric intake [1–4].

Because young people spend a great deal of time in school,

the school environment is a natural place to consider prac-

tices and policies that may affect students’ physical activity

and food and beverage intake. In this paper, we examine

secondary schools’ practices regarding physical education

ndence to Patrick M. O’Malley, Institute for Social

of Michigan, 426 Thompson Street, Room 2341,

-1248.

[email protected]

front matter � 2009 Society for Adolescent Medicine. All

th.2009.04.008

and sports participation, and we explore the extent to which

those practices are associated with students’ self-reported

overweight status and their exercise and sports habits [5].

Physical activity and the school environment

Several objectives set in Healthy People 2010 focus on

increasing physical activity (PA) among both children and

adolescents [6]. The stated objectives include school-specific

issues such as daily physical education (PE) and increased

walking and bicycling to school. Additional recommenda-

tions from the U.S. Department of Health and Human

Services [7] state that children should have a minimum of

60 minutes of PA daily, most of which should be moderate

rights reserved.

mailto:[email protected]

P.M. O’Malley et al. / Journal of Adolescent Health 45 (2009) S71–S81S72

to vigorous physical activity (MVPA). Such objectives could

be addressed within the school setting via PE and other activ-

ities such as intramural and varsity sports. Given that the

Institute of Medicine recommends that children expend

approximately 50% of daily energy at school [2], school-

based PA provides an important opportunity to shape an indi-

vidual’s PA habits through completion of secondary school.

Although school-based PE is required in virtually all

states, participation requirements are generally low [8–10].

A nationally representative longitudinal study in Australia

of both primary and secondary schools found that median

total PA, fitness, and body mass index (BMI) were generally

similar across schools regardless of the levels of compulsory

PE and school sports provided, and concluded that current

policies regarding school PA may not be of an intensity suffi-

cient to relate to individual student PA or obesity [11]. Mixed

results have been found regarding the relationship between

walking or bicycling to school and student health. A national

survey of U.S. youth in grades 4–12 found no relationship

between walking or biking to school and BMI, but did find

that such active transportation was related to a higher likeli-

hood of frequent moderate physical activity [12]. Two small

studies of primary school students (one of urban students in

England, and one of suburban students in southern Califor-

nia) found that for boys, active commuting (walking in

England; walking, biking, or skateboarding in California)

was related to increased physical activity [13] and lower

BMI [14]. A small study of rural Nebraska elementary school

children showed surprising results, with active commuting

associated with increased physical activity, but also associ-

ated with increased BMI among overweight children (no

significant relationship was found with BMI among

normal-weight children) [15].

Interventions targeting existing school PE curricula (e.g.,

adding additional PE classes, extending the length of PE

classes, increasing time spent in PE classes in MVPA) have

been found to relate positively to time spent in PA at school,

as well as increased MVPA, energy expenditure, aerobic

capacity and muscular endurance, flexibility, and PA knowl-

edge and self-efficacy [8]. However, few relationships have

been found between PE interventions and student BMI or

body fat [8]. Some researchers posit that the lack of observed

relationships between interventions that mandate dietary and

PA changes in the school setting and student BMI may be due

to students compensating by altering their behaviors outside

of school [16]. Another possibility is that PE may be effective

in increasing energy expenditures but may not be enough to

affect weight.

As noted above, the literature provides several examples

of intervention studies to increase youth PA. However, no

nationally representative study involving U.S. students has

examined overall school PA environments and their relation-

ships with student self-reported BMI and PA levels. This

article is intended to provide that examination. In 2007, John-

ston et al reported data on the levels of PE and sports partic-

ipation among American secondary school students in 2003

through 2005 and the extent to which these levels varied by

student grade level, racial/ethnic background, and socioeco-

nomic status (SES) [17]. A principal finding was that PE

was noticeably lacking in American high schools for all

groups. In the present study, we again report data on the

levels of secondary school student PE and sports participa-

tion (for 2004–2007) and extend the analysis to incorporate

both (a) school-level variables as reported by school admin-

istrators and (b) student self-reports of BMI, exercise, and

sports participation. The key research question to be ad-

dressed is: Are characteristics of the U.S. secondary school

PA environment associated with relevant student outcomes,

both bivariately and controlling for individual variables and

other school-level variables? The conceptual model assumes

that the school PA environment can have a direct effect on

student outcomes, including BMI and PA levels. Thus, we

hypothesize that schools with policies and practices that

enhance student activity will have ‘‘better’’ outcomes,

compared to schools that do not have such policies or prac-

tices. Other school-level factors may be confounded with

PA variables, so we control for some key related measures.

Of course, individual-level variables will also be associated

with outcomes, so we also control for important factors at

the individual level.

Methods

Samples and survey methods

This report uses data from two studies, Monitoring the

Future (MTF) and Youth, Education, and Society (YES).

The MTF study annually surveys nationally representative

samples of 8th-, 10th-, and 12th-grade students located in

approximately 410 public and private schools in the 48

contiguous states; each school participates for 2 consecutive

years. Thus, half of the sample of schools are ‘‘cycling out’’

of the study each year. In the YES study, administrators in

schools that have just completed their second and final year

of participation in the MTF study are requested to complete

a questionnaire describing school policies and programs

related to various health issues, including PE, food services,

and substance use.

MTF design and methods

The design and methods for the MTF project are summa-

rized briefly here; a detailed description is available else-

where [18]. At each of three grade levels (8th, 10th, 12th),

a stratified random sampling procedure is used that involves

three stages: (1) geographic regions are selected; (2) schools

are selected within regions with probability proportional to

the estimated number of students in the target grade; and

(3) students are selected within schools, usually by means

of randomly selecting whole classrooms. A total of about

45,000 students are surveyed each year. Sample weights

are assigned to each student to take into account variations

in selection probabilities that occurred at all stages. Ethical

P.M. O’Malley et al. / Journal of Adolescent Health 45 (2009) S71–S81 S73

approval for the study was obtained from the University of

Michigan Behavioral Sciences Institutional Review Board

[18], and informed consent was given for survey participa-

tion. Surveys were administered in group settings (that is,

classrooms) by University of Michigan personnel; students

self-completed the survey questionnaires during a normal

class period.

MTF student measures. Gender, race/ethnicity, and parental

education were reported. Students were coded as being of

White, African American, Hispanic, or Other background;

the numbers of cases in other specific race/ethnic groups

were too small to allow separate estimates. Parental educa-

tion, an average of father’s and mother’s educational attain-

ment (with one missing data case permitted), is used as

a proxy for SES, because students are generally unable to

provide accurate information on family income and parental

occupation. Students self-reported their height (in feet and in-

ches) and weight (in pounds), using pre-coded close-ended

response alternatives. BMI was calculated by dividing weight

(in kilograms) by height (in meters) squared. Age- and

gender-specific growth curves produced by the Centers for

Disease Control and Prevention were used to determine

whether each student’s BMI was categorized as overweight

(85th to less than the 95th percentile) or obese (equal to or

greater than the 95th percentile) [19,20]. Measures of student

PA were obtained from responses to three questions: (1) How

often do you actively participate in sports, athletics or exer-

cising? (2) How often do you exercise vigorously (jogging,

swimming, calisthenics, or any other active sports)? (3) To

what extent did you participate in school athletic teams this

school year?

YES design and methods

The half samples of nationally representative schools that

were cycling out of the MTF study in 2004 through 2007

comprised the YES target sample. School administrators

were asked to complete a self-administered questionnaire

on a variety of school policies and relevant student participa-

tion rates. More than 85% of respondents were school admin-

istrators; teachers, and other school personnel were other

respondents. The combined number of YES secondary

schools that participated in the 2004–2007 surveys is 693, re-

flecting a response rate of 84%. A comparison of all schools

responding to the YES questionnaire with all schools eligible

for participation in the YES study showed minimal differ-

ences in student outcomes. For example, mean BMI was

21.6 in both groups for 8th grade, 22.9 for both in 10th grade,

and 23.6 for both in 12th grade. Other student outcomes were

similarly virtually identical. There were some very minor

differences in school characteristics, but these were not suffi-

cient to produce dissimilar student outcomes.

YES measures. Questionnaires were completed in late spring

each year from 2004 through 2007, and administrators were

instructed to answer for the current (just ending) school year.

Respondents were asked whether students in the target grade

(8th, 10th, or 12th) are required to take PE, what percentage

of students actually take a PE class, the number of days per

week they have PE, and the duration of a PE class (in

minutes). These questions permit a calculation of the total

number of minutes per week a student who takes PE would

spend in PE. Questions were also asked about the rates of

student participation, separately for boys and girls, in (a)

interscholastic or varsity sports and (b) intramural sports or

PA clubs. An additional question asked what percentage of

students walk or bike from home to school on an average

school day. Finally, a yes–no question was asked about

whether the school gives physical fitness tests, and if so,

whether parents are provided with the results.

School-level control variables

Type of school (public or private), size of school (number

of students in target grade), percentage of students eligible for

a free or reduced-cost lunch (an indicator of school SES),

region of the country, and urbanicity were used as control

variables in multivariate analyses. Region and urbanicity

were derived from the sampling data from which the school

was drawn. Four regions of the country were distinguished

(Northeast, Midwest, South, and West) based on U.S. Census

Bureau classifications. Three levels of population density

were coded at the school level using locale code data from

the National Center for Education Statistics: urban (large or

midsize central city), suburban (urban fringe of either large

or midsize city, as well as large or small town), and rural.

Analyses

All analyses were conducted using SAS version 9.1.3.

Multivariate models took account of the complex sample

design in estimation of parameters by using the survey

logistic procedure, as well as using sampling weights [21].

The sample weights are designed to provide accurate infer-

ence at the student level. Each of the 11 relevant variables

from the school administrator questionnaire was used as

a predictor of each of the four student outcome variables, first

bivariately and then multivariately, adding controls for other

school- and student-level variables. All analyses were con-

ducted separately by grade level, with data pooled across

the four years (2004–2007).

Results

School and student data were available on the variables

relevant to the present analyses from over 70,000 MTF

student respondents in those schools. Female students

comprised 50.6% of the combined samples. The racial/ethnic

composition of the resulting student samples was 65.0%

white, 10.6% African-American, 10.8% Hispanic, and

13.6% from other racial/ethnic backgrounds.


Table 1 provides information on school administrators’

responses to questions on PA. The data are weighted to reflect

the proportions of students who attend schools with various

characteristics.

For the years 2004 through 2007 combined, 88% of 8th-

graders attended schools that required PE in their grade,

compared to 48% for 10th graders, and 20% for 12th graders.

The mean percentage of students reported to take PE showed

similar sharp differences by grade at 92%, 64%, and 34%.

Other results from the school administrator surveys can be

seen in Table 1.

Table 2 provides descriptive data from student self-report

information on the outcomes of interest: BMI and overweight

status, and measures of PA for the total sample by grade.

Slightly more than 10% of students in each grade were

obese, and approximately 25% were overweight (or obese).

Analyses by gender within grade (not shown) indicated that

males averaged slightly higher in BMI than females but

were distinctly more likely to exceed the 85th and 95th

percentile cutoffs. Male students also were generally more

likely than female students to report actively participating

in sports, exercising vigorously, or having much participation

in school athletic teams.

Tables 3–6 provide results of logistic regression analyses

addressing the key research question. We first report the asso-

ciation between each environmental variable and the

percentage of students who were at or above the 85th percen-

tile, and then between each of the environmental variables

and the three measures of being active in sports, exercising

vigorously, and participating in school athletics.

In general, the school PE variables did not show statisti-

cally significant associations with students’ overweight

Table 1

School physical education (PE) participation by grade level, 2004–2007 combined

School PE participation

% of Students required to take PE

Mean % of students who take PE

Mean number of days/week students have PE among those who take PE

Mean number of days/week that students have PE in all schools

Mean length of PE classes in minutes

Mean time students in all schools spend in PE classes (in minutes per week)

School varsity and intramural sports participation

Mean % of boys who participate in interscholastic or varsity sports

Mean % of girls who participate in interscholastic or varsity sports

Average % of both boys and girls who participate in interscholastic or varsity sports

Mean % of boys who participate in intramural sports or physical activity clubs

Mean % of girls who participate in intramural sports or physical activity clubs

Average % of both boys and girls who participate in intramural sports or physical

activity clubs

Walking or biking to school

Mean % of students who walk or bike from home to school on an average school day

School physical fitness tests

% of Students who attend schools that provide physical fitness tests to at least some

students

% of Students in schools in which at least some parents are provided with the results

of physical fitness tests

status, particularly after controls for other school- and

student-level variables were included in the regressions.

Similarly, the percentage of students walking or biking to

school, along with school fitness test measures, were usually

not statistically significant.

The average percentage of students (both boys and girls)

who participated in interscholastic or varsity sports was asso-

ciated at the bivariate level, with a lower percentage of

students being overweight or obese in all three grades. This

association remained significant after controls for indi-

vidual-level variables for 10th and 12th graders. The average

percentage of students (both boys and girls) who participated

in intramural sports or PA clubs was associated bivariately

with a lower percentage of students being overweight or

obese in all three grades, but significantly so in 8th grade

only. This association remained significant after controls

for individual-level variables, but not after school-level vari-

ables were added.

The PE measures were, for the most part, not significantly

systematically related to the students’ self-reported exercise

variables (Tables 4–6). One exception was that the mean

length of PE classes was negatively associated with being

active in sports and participating in school athletics in all

three grades; that is, longer PE classes were associated with

students reporting being less active and participating less in

school athletics. Only one of the nine associations with exer-

cising vigorously was significant (at p < .05), suggesting no

important relationships.

The average percentage of students (both boys and girls)

estimated by school administrators to have participated in

interscholastic or varsity sports was generally associated at

the bivariate level with significantly higher percentages of

data

Grade

8

Standard

error

Grade

10

Standard

error

Grade

12

Standard

error

87.6 2.4 48.4 3.9 20.4 3.1

91.8 1.5 64.1 2.9 34.0 2.8

3.8 .1 4.3 .1 3.9 .1

3.5 .1 2.8 .1 1.4 .1

52.9 1.2 61.8 1.7 59.8 2.1

179.9 5.3 173.0 9.2 82.0 7.2

36.2 1.6 38.8 1.7 35.9 1.6

32.6 1.6 34.5 1.6 32.3 1.6

34.4 1.6 37.3 1.7 34.4 1.6

27.4 1.8 16.5 1.5 15.8 1.9

22.9 1.6 13.8 1.4 13.8 1.8

25.2 1.7 15.6 1.5 15.0 1.8

19.3 1.6 12.2 1.4 8.6 1.3

64.8 3.5 41.6 3.9 23.4 3.2

39.2 3.5 20.6 3.3 10.2 2.3

Table 2

Body mass index, overweight, obesity, and physical activity by grade level, 2004–2007 combined data

Grade

8

Standard

error

Grade

10

Standard

error

Grade

12

Standard

error

Body mass index:

Mean body mass index 21.6 .1 22.9 .1 23.6 .1

% of Students obese (equal to or greater than the 95th percentile) 12.3 .4 12.0 .4 11.0 .4

% of Students overweight (equal to or greater than the 85th percentile) 26.9 .6 26.9 .6 24.5 .7

Individual sports participation:

% of Students who actively participate in sports, athletics or exercisingy

0. Less than almost every day 44.8 .6 47.4 .7 56.3 .8

1. Almost every day 55.2 .6 52.6 .7 43.7 .8

% of Students who exercise vigorously: e.g., jogging,

swimming, calisthenics, or any other active sportsz

0. Less than nearly every day or every day 45.1 .7 51.3 .9 64.1 .7

1. Nearly every day or every day 54.9 .7 48.7 .9 35.9 .7

Extent of participation in school athletic teams (%)þ

0. Less than great participation 61.4 .9 63.8 .8 71.1 .8

1. Great participation 38.6 .9 36.2 .8 28.9 .8

y Original scale: Never; a few times a year; once or twice a month; at least once a week; almost every day.z Original scale: Never; seldom; sometimes; most days; nearly every day; every day.þ Original scale: Not at all; slight; moderate; considerable; great.


students self-reporting being active in sports, exercising

vigorously, and participating in school athletics; and most

of the associations remained significant after controlling for

individual- and school-level variables. The average

percentage of students (both boys and girls) who participated

in intramural sports or PA clubs was generally not associated

with the percentages of students being active in sports, exer-

cising vigorously, or participating in school athletics. (Bivar-

iate associations were significant among 8th graders—who

have higher rates of participation—for being active in sports

and exercising vigorously, but these did not remain signifi-

cant in multivariate models.)

Discussion

The results show that from 2004 through 2007, the

percentage of students who attended schools that required

PE in their grade differed sharply by grade level: 88% of

8th graders, 48% of 10th graders, and 20% of 12th graders.

This finding is fully consistent with previous reports [17].

Clearly, in recent years, required PE has become a rarity

for American high school seniors. However, it is difficult to

say whether this has major implications for obesity rates

among students, given the lack of significant association

between the requirement and the percentage of students

who are at or above the 85th percentile on BMI after control-

ling for both individual- and school-level variables. In bivar-

iate models, the percentage of overweight students was

higher in schools that did not require PE than in schools

that did: 29.7% versus 26.5% in 8th grade, 28.1% versus

25.5% in 10th grade, and 24.9% versus 23.0% in 12th grade.

These are modest differences, and become nonsignificant for

8th and 12th grade when individual variables are controlled.

The differences become nonsignificant in all grades when

school-level variables are added to the individual-level

controls.

The percentage of students who actually take PE showed

a similar pattern. Another indicator of extent of PE (number

of days per week that students take PE, averaged across all

students) also showed a negative association with percent

at or above the 85th percentile, significantly so for 10th

graders. Other indicators (length of PE classes, amount of

time spent in PE) generally showed no association with the

percentage of students at or above the 85th percentile. It

should be noted that analyses examining the associations of

the various school PE-related measures and both continuous

BMI score and classification as obese produced similar

results.

Thus, it appears that, as measured in the present study,

variations in PE in American secondary schools have at

best only modest association with student BMI. These find-

ings are consistent with those noted in the Introduction ob-

tained from nationally representative work in Australia [11].

Why do we see so little association? A number of factors

should be considered. One might argue that controlling for

school-level variables such as percentage of students eligible

for a free or reduced-cost lunch is in some sense ‘‘overcon-

trolling’’ because any effect of a PE requirement might over-

lap or be shared with school SES. However it does seem

entirely reasonable to control for individual-level variables,

and such controls make the PE requirement effect nonsignif-

icant for 8th and 12th graders.

It may be useful to note that school-level variables in

general cannot explain much variance in student outcomes

when much of the variance in the outcomes is within schools.

This is certainly the situation with the percent of overweight

students: O’Malley et al [22] reported that less than 4% of the

variation in BMI or in the percentage of students at or above

Table 3

School physical education (PE) participation association with percentage of

students at or above the 85th percentile on BMI by grade level, 2004–2007

combined data

Environment

Predictor

Bivariatey Multivariate

model 1zMultivariate

model 2þ

OR OR OR

School requires PE in specified grade

8th Grade .86* .96 1.08

10th Grade .88* .91* .96

12th Grade .90 .95 .95

% of Students in specified grade

that take PEþþ

8th Grade .81 .95 1.12

10th Grade .79** .86* .95

12th Grade .88 .94 .96

Days/week students have PE among

those who take PE in specified grade

8th Grade 1.00 .99 .98

10th Grade 1.03 1.00 .99

12th Grade 1.00 .99 .99

Days/week that students in specified

grade have PE in all schools

8th Grade .98 .99 .99

10th Grade .96* .97* .99

12th Grade .98 .99 .99

Mean length of PE classes (in minutes) in specified gradeyy

8th Grade 1.25* 1.07 1.08

10th Grade 1.02 1.01 .98

12th Grade 1.13 1.12 1.09

Time students in all schools spend in

PE classes (in minutes) per weekyy

8th Grade 1.01 1.00 1.00

10th Grade .98 .99 1.00

12th Grade 1.01 1.01 1.02

Average % of both boys and girls who

participate in interscholastic or varsity sportsþþ

8th Grade .75* .95 .88

10th Grade .57*** .76* .89

12th Grade .63* .72* .70

Average % of both boys and girls who participate

in intramural sports or physical activity clubsþþ

8th Grade .62*** .74** .85

10th Grade .87 .93 1.14

12th Grade .83 .88 .95

Percentage of students who walk or bike from home

to school on an average school dayþþ

8th Grade 1.18 .99 1.05

10th Grade 1.20 .91 1.06

12th Grade 1.50 1.05 .87

School provides physical fitness tests

to at least some students

8th Grade .95 .95 .96

10th Grade .97 1.00 1.08*

12th Grade 1.01 .99 .99

Schools in which at least some parents are

provided with the results of physical fitness tests

8th Grade .97 1.01 1.02

10th Grade .92 .95 1.06

12th Grade 1.00 .91 .89

OR ¼ odds ratio.

* p < .05.

** p < .01.

*** p < .001.y OR from model with each environment variable entered separately.

z OR from model with each environment variable entered separately, plus

year and individual variables gender, race/ethnicity, and parental education.þ OR from model with each environment variable entered separately, plus

year and individual variables noted in Multivariate Model 1, plus school-

level variables type, size, percentage of students eligible for reduced-cost

lunch, region, and population density.þþ Environment predictor multiplied by 10 to obtain more meaningful

ORs.yy Environment predictor multiplied by 60 to obtain more meaningful

ORs.


the 85th percentile lies between American secondary schools,

while more than 96% is within schools. Similarly, in the

present study only 3–4% of the variance in being active in

sports and in exercising frequently is between schools at

each grade. Participation in school athletic teams varied

somewhat more, at 5–9% between schools.

This study has other limitations that could reduce our

ability to detect associations. The data are cross-sectional,

making it difficult to discern even the directionality of any

associations. An additional important limitation is reliance

on one or two observers to provide information that may be

difficult for them to know with high accuracy. We have tried

to limit our requests for information to data that should be

fairly readily known. However, information such as what

percentage of students walk or bike to school, or what percent

participate in intramural sports will clearly have some degree

of reporting error. However, the percentage of students that

administrators reported as participating in interscholastic or

varsity sports was positively associated with the percentage

of students who self-reported being active in sports, exer-

cising vigorously, and participating in school athletic teams.

This suggests that the administrator reports are indeed

reasonably valid indicators of student participation. We

were also limited in the extent of information that was ob-

tained. For example, details about extent of teacher training

or accreditation for PE, length of time students participated

in PE (full year, half year), and other factors were not ob-

tained. Although we attempted to assess the most important

potential factors, the range of policies addressed in this study

is relatively narrow. Still another limitation is that we are to

some extent analyzing reports of policies that may not be

fully realized in practice.

Furthermore, although this study is large by most stan-

dards, the number of schools per grade combined across all

4 years is less than 250; thus, small effects may not rise to

the level of statistical significance.

Another possible contributing factor to our lack of find-

ings is that PE as practiced in schools is simply not physically

demanding enough to have a discernible effect on BMI.

Several studies indicate that in order to find significant rela-

tionships between PE and body fat, PE programs need both

significant daily class commitment (e.g., 1.25 hours per day

[23]) and purposeful increases in MVPA [24,25]. Studies

that increased the percentage of MVPA per PE class, but

did not increase the total number of PE classes required,

Table 4


students active in sports daily or almost every day vs. other by grade level,

2004–2007 combined data

Environment

predictor



model 2þ

OR OR OR


8th Grade 1.04 .95 .95

10th Grade 1.15* 1.11* 1.07

12th Grade 1.03 1.01 1.11

% of Students in specified grade that take PEþþ

8th Grade 1.08 .96 .96

10th Grade 1.21* 1.15* 1.12*

12th Grade 1.04 1.01 1.02

Days/week students have PE among those

who take PE in specified grade

8th Grade 1.05** 1.06*** 1.08***

10th Grade .94* .98 1.01

12th Grade 1.00 1.02 1.02

Days/week that students in specified

grade have PE in all schools

8th Grade 1.05** 1.05** 1.05*

10th Grade 1.02 1.02 1.03*

12th Grade 1.01 1.01 1.01


8th Grade .79** .86* .81*

10th Grade .79** .81** .87*

12th Grade .85 .85* .92

Time students in all schools spend in

PE classes (in minutes) per weekyy

8th Grade 1.02 1.03 1.03

10th Grade 1.00 1.00 1.01

12th Grade .99 1.00 1.00

Average % of both boys and girls who participate

in interscholastic or varsity sportsþþ

8th Grade 1.80*** 1.55*** 1.43**

10th Grade 2.47*** 1.72*** 1.41**

12th Grade 1.91** 1.74** 1.26

Average % of both boys and girls who participate in

intramural sports or physical activity clubsþþ

8th Grade 1.31** 1.13 1.12

10th Grade 1.22 .95 .78

12th Grade 1.04 .97 .88

Percentage of students who walk or bike from home to

school on an average school dayþþ

8th Grade .82* 1.00 1.20

10th Grade .80 1.07 1.09

12th Grade .81 1.23 1.32

School provides physical fitness tests to at least some students

8th Grade .91 .90 .90

10th Grade 1.03 1.04 1.00

12th Grade .93 .95 1.02

Schools in which at least some parents are provided with

the results of physical fitness tests

8th Grade 1.04 1.01 .99

10th Grade 1.07 1.06 1.04

12th Grade .97 1.02 1.05

OR ¼ odds ratio.

* p < .05.

** p < .01.

*** p < .001.y OR from model with each environment variable entered separately.z OR from model with each environment variable entered separately, plus

year and individual variables gender, race/ethnicity, and parental education.

þ OR from model with each environment variable entered separately, plus


level variables type, size, percent students eligible for reduced-cost lunch,

region, and population density.þþ Environment predictor multiplied by 10 to obtain more meaningful


ORs.


showed no significant differences in body size [26,27]. In

fact, the Institute of Medicine Committee on Prevention of

Obesity in Children and Youth recommends that half of chil-

dren’s daily PA should be accrued during the school day [2].

Many school districts may be concerned that requiring daily

PE would result in decreased academic performance due to

time limitations; however, research indicates that significant

allocation of curricular time to PA may actually enhance

learning rates [28,29].

There are also limitations with respect to the students’

self-report measures. Regarding height and weight, a number

of studies have investigated use of self-reports, and have

generally reported that, although there may be biases in abso-

lute levels associated with self-reports, the data are certainly

adequate for research purposes. Brener et al [30] obtained

both objective and self-reported data on height and weight

for more than 2000 students in grades 9 through 12, and

found that ‘‘.self-reported values of height, weight, and

BMI were highly correlated with their measured values’’

(p. 281). Brener et al [30] also noted that surveillance systems

can yield ‘‘.valuable results by using self-reported height

and weight to assess trends in the prevalence of obesity’’

(p. 287). Goodman et al [31] analyzed data from more than

10,000 respondents in the National Longitudinal Study of

Adolescent Health, with both self-reported and objectively

measured height and weight. They report that ‘‘correlations

between measured and self-reported anthropomorphic

indices [height and weight] were very strong’’ (p. 54).

They conclude that ‘‘findings from other studies that have

used self-reported BMI should be considered valid, and

future studies can use self-reported data to understand adoles-

cent obesity, its correlates, antecedents, and sequelae’’ [31 (p.

57)]. Although the findings are supportive of use of self-

reports for research, there is also evidence of bias. Ogden

et al [32] reported that about 34% of 12–19-year-olds were

overweight or at risk of overweight based on actual measure-

ments, compared to the 25% found here, based on self-

reports.

The students self-reported their levels of physical activity,

exercise, and sports participation in broad, general terms. It

would of course be preferable to have more specific informa-

tion, but we believe that the measures likely capture much of

the valid interindividual variation. (Using procedures devel-

oped by Heise [33] and Wiley and Wiley [34], we estimate

the reliabilities of the single-item measures of exercise and

sports at .60 for each, based on longitudinal data gathered

at 2-year intervals; we do not have an estimate of the reli-

ability of the participation in school athletic teams item,

Table 5


students exercising vigorously nearly every day/every day vs. other by grade,

2004–2007 combined data

Environment

predictor



model 2þ

OR OR OR


8th Grade .90 .81* .82*

10th Grade 1.11 1.07 1.04

12th Grade .94 .95 1.04

% of Students in specified grade that take PEþþ

8th Grade .91 .78 .80

10th Grade 1.26* 1.19* 1.16

12th Grade 1.01 1.01 1.07



8th Grade 1.05* 1.07** 1.06**

10th Grade .96 1.00 1.01

12th Grade 1.00 1.02 1.00

Days/week that students in specified grade have

PE in all schools

8th Grade 1.03 1.02 1.01

10th Grade 1.03 1.03 1.03

12th Grade 1.01 1.02 1.02


8th Grade .77* .84 .84

10th Grade .95 .98 1.03

12th Grade .89 .90 .92

Time students in all schools spend in PE classes

(in minutes) per weekyy

8th Grade 1.00 1.01 1.00

10th Grade 1.03 1.03 1.03

12th Grade 1.01 1.01 1.01


interscholastic or varsity sportsþþ

8th Grade 1.59*** 1.39* 1.36*

10th Grade 1.61* 1.14 1.01

12th Grade 1.65** 1.50** 1.39*



8th Grade 1.28* 1.10 1.06

10th Grade 1.15 .93 .83

12th Grade 1.03 .95 .94



8th Grade .86 .96 1.06

10th Grade .98 1.26 1.20

12th Grade .87 1.15 1.08


8th Grade .92 .92 .94

10th Grade 1.13 1.15* 1.11

12th Grade 1.01 1.03 1.11

Schools in which at least some parents are provided

with the results of physical fitness tests

8th Grade .97 .95 .97

10th Grade 1.11 1.12 1.09

12th Grade 1.07 1.13 1.14

OR ¼ odds ratio.

* p < .05.

** p < .01.





level variables type, size, percent students eligible for reduced-cost lunch,

region, and population density.þþ Environment predictor multiplied by 10 to obtain more meaningful


ORs.


because we do not have longitudinal data for that question). It

is possible that increased sensitivity in the self-reported PA

measures might improve the ability to detect relationships

with school-level PA environment measures.

Although the physical education measures showed limited

associations, the average percentage of students who partici-

pated in interscholastic or varsity sports showed the expected

relationship, with a lower percentage of students at or above

the 85th percentile in all three grades, and remained signifi-

cant after controls for individual-level variables for 10th

and 12th graders. The average percentage of students (both

boys and girls) who participated in intramural sports or PA

clubs was associated with a lower percentage of students

being overweight; this was significant only in 8th grade, bi-

variately and with individual-level variables controlled.

None of these associations remained significant after control-

ling for both individual- and school-level variables, but as

noted above, controlling for the school variables may be in

some sense ‘‘over-controlling.’’ The overall pattern suggests

that, in schools with high percentages of students partici-

pating in sports (based on administrators’ reports), there are

likely to be somewhat lower levels of student overweight,

though the associations were not always statistically signifi-

cant.

The percentage of students walking or biking to school

was not associated with the percentage at or above the 85th

percentile for any grade. Low rates of walking or bicycling

to school reflect previously published rates [35], and likely

limit the possible overall effects of these activities. There

may be salutary effects of an individual student walking or

biking to school, especially if the distance is substantial,

but the effects (particularly because the effects are likely to

be small) may be obscured by our use of a school-level

measure of percentage of all students walking or biking. If

rates of walking and bicycling reached the Healthy People2010 goal of 50% (rates were at 48% in the late 1960s

[36]), direct relationships might be possible to detect.

Research among youth in the Philippines has shown that

commuting to school via motorized transportation versus

walking may result in an expected weight gain of 2 to 3 lb

yearly [37].

Whether a school provided physical fitness tests and

whether a school provided parents with results of such tests

did not appear to have any meaningful association with

percentage of students being at or above the 85th percentile.

Requirement of PE, the percentage of students who take

PE, mean PE class length, and total time spent in PE classes

were for the most part not systematically associated with the

Table 6

School physical activity environment association with percent participating

in school athletic teams to a great extent vs. other by grade level, 2004–2007

combined data

Environment

predictor



model 2þ

OR OR OR


8th Grade 1.07 .98 1.04

10th Grade 1.12 1.07 1.04

12th Grade .90 .89 1.11

Percentage of students in specified grade that take PEþþ

8th Grade 1.40 1.23 1.25

10th Grade 1.07 1.01 1.03

12th Grade .91 .90 .96



8th Grade .98 .98 1.04

10th Grade .93** .96 1.01

12th Grade 1.01 1.02 1.03

Days/week that students in specified grade have

PE in all schools

8th Grade 1.00 .99 1.04

10th Grade .99 .99 1.01

12th Grade .98 .98 1.00


8th Grade .62*** .72** .93

10th Grade .81** .83** .95

12th Grade .78* .77** .83*

Time students in all schools spend in PE classes

(in minutes) per weekyy

8th Grade .95 .96 1.03

10th Grade .99 .99 1.01

12th Grade .98 .99 .99


interscholastic or varsity sportsþþ

8th Grade 4.46*** 3.90*** 2.59***

10th Grade 3.28*** 2.45*** 2.01***

12th Grade 2.24*** 2.21*** 1.39



8th Grade 1.32 1.19 1.26

10th Grade 1.32 1.06 .91

12th Grade .86 .80 .74*



8th Grade .63** .79 1.10

10th Grade .70 .94 1.16

12th Grade .96 1.50 1.89**


8th Grade .88 .87 .87*

10th Grade 1.02 1.02 1.01

12th Grade .86 .88 1.00

Schools in which at least some parents are provided

with the results of physical fitness tests

8th Grade .98 .96 .94

10th Grade 1.03 1.02 1.06

12th Grade 1.01 1.06 1.09

OR ¼ odds ratio.

* p < .05.

** p < .01.





level variables type, size, percentage of students eligible for reduced-cost

lunch, region, and population density.þþ Environment predictor multiplied by 10 to obtain more meaningful


ORs.


percentage of students reporting being active in sports, exer-

cising vigorously, or participating in school athletics, particu-

larly after controls for individual- and school-level variables.

The number of days per week that students take PE did show

significant associations, before and after controls, with the

percentage of students reporting being active in sports and

exercising vigorously, but only among 8th graders.

One possibility is that although the mean number of days

per week that students have PE is generally similar across

grades among students who take PE, 8th-grade students are

much more likely both to be required to take PE and to actu-

ally take PE. Thus, the lack of findings for 10th and 12th

graders between days per week PE is taken and participation

in sports or vigorous exercise may be related to the lower

number of students exposed to higher numbers of days of

PE. This finding may indicate that if the Healthy People2010 recommendation of daily PE participation were to be

broadly implemented, students might have higher levels of

overall PA, as well as MVPA.

Another possibility relates to the findings from prior

research on PA compensation. In an intervention study

with 3rd- to 5th-grade youth in two school districts in Ne-

braska, Donnelly et al [38] found that intervention schools

were able to improve lunch nutrition (lowered food energy,

fat, and sodium; increased fiber) significantly, and to improve

class-based PA slightly. However, measures of 24-hour

energy intake showed no differences for food energy, fat,

or fiber; further, PA outside of school was significantly lower

for intervention students than control. The authors concluded

that students may downwardly adjust their spontaneous PA

outside of the school environment in response to increases

in school-related structured PA. However, a California-based

intervention to increase PA both in PE classes as well as

outside of school with elementary school children in seven

small schools indicated no differences between intervention

and control subjects on PA outside of school over a 2-year

period [39]. Thus, the limited evidence for a phenomenon

of compensation is mixed. Findings from the present anal-

yses may reflect such compensation, especially for high

school students with increased academic demands.

Participation in intramural sports or PA clubs was gener-

ally not related to student self-reports of being active in

sports, exercising vigorously, or participating in school

athletics for the high school students. There was some asso-

ciation at 8th grade, where the participation is greatest. It

seems likely that intramural sports have a substantially lower

level of PA compared to varsity sports, which would atten-

uate the relationships.


The percentage of students walking or biking to school,

and whether a school conducted or reported results of phys-

ical fitness tests did not appear to be systematically associated

with any of the three student self-report measures. This is not

surprising; these self-report measures may not measure with

great accuracy actual levels of walking or biking to school.

Conclusion

In summary, rates of PE participation remain low, and

decline sharply with grade level. We have not been able to

demonstrate clear effects of a number of aspects of school

environment on student overweight, particularly those

having to do with PE and walking or biking to school (now

a rare behavior). There are, however, demonstrable associa-

tions with student participation in varsity and intramural

sports. These associations suggest that this might be a fruitful

area for intervention studies.

We remain convinced that PA is an important part of the

effort to reduce child obesity, and that schools can play an

important role in that effort. We have shown that, as currently

practiced in schools, existing variations in PA policies may

not be sufficient to produce discernible school-wide differ-

ences. We suspect that schools need to have policies that

support more vigorous PA than do current policies.

A number of studies support the value of enhanced quan-

tity and quality of physical education to increase physical

activity. In addition, researchers such as Leupker et al [26]

have noted that improving lifetime PA habits among youth

may have an important and significant cumulative impact

over and above any reduction in immediate physiological

risk levels. Healthy People 2010 calls for increasing the

proportion of adolescents who spend at least 50 percent of

school physical education class time being physically active

[40]. Schools should be encouraged to provide their students

opportunities for more moderate to vigorous PA in PE, and

the motivation to incorporate PA into their lives, both in

and out of schools.

Acknowledgments

This research was conducted as part of the Youth, Educa-

tion, and Society (YES) project (part of a larger research

initiative called Bridging the Gap: Research Informing Policy

and Practice for Healthy Youth Behavior) funded by the Rob-

ert Wood Johnson Foundation. The Monitoring the Future

study is funded by the National Institute on Drug Abuse

(DA01411). The views expressed in this article are those of

the authors and do not necessarily reflect the views of the

sponsors. We thank Kathryn Johnson for editorial assistance.

References

[1] Johnston LD, O’Malley PM. Obesity among American Adolescents:

Tracking the Problem and Searching for Causes. Youth, Education,

& Society Occasional Paper No. 3. Ann Arbor, MI: Institute for Social

Research, 2003. Available at: http://www.yesresearch.org/pubs.html#

reports. Accessed October 30, 2008.

[2] Koplan JP, Liverman CT, Kraak VI, eds. Preventing Childhood

Obesity: Health in the Balance. Washington, DC: National Academies

Press, 2005.

[3] Patrick K, Norman GJ, Calfas KJ, et al. Diet, physical activity, and

sedentary behaviors as risk factors for overweight in adolescence.

Arch Pediatr Adolesc Med 2004;158:385–90.

[4] U.S. Department of Health and Human Services. Physical Activity and

Health: A Report of the Surgeon General. Atlanta, GA: Centers for

Disease Control and Prevention, 1996.

[5] Lowry R, Brener N, Lee S, et al. Participation in high school physical

education—United States, 1991–2003. MMWR 2004;53:844–7.

[6] U.S. Department of Health and Human Services. Healthy People 2010:

Conference Edition. Washington, DC: U.S. Department of Health and

Human Services, 2000.

[7] U.S. Department of Health and Human Services. 2008 Physical

Activity Guidelines for Americans. ODPHP Publication No. U0036.

Washington, DC: U.S. Department of Health and Human Services.

Available at: http://www.health.gov/paguidelines/guidelines/default.

aspx; 2008. Accessed February 11, 2009.

[8] Kahn EB, Ramsey LT, Brownson RC, et al. The effectiveness of

interventions to increase physical activity. A systematic review. Am

J Prev Med 2002;22:73–107.

[9] Lee SM, Burgeson CR, Fulton JE, et al. Physical education physical

activity: Results from the School Health Policies and Programs Study

2006. J Sch Health 2007;77:435–63.

[10] National Association for Sport and Physical Education & American

Heart AssociationShape of the Nation Report. Status of Physical Educa-

tion in the USA. Reston, VA: National Association for Sport and Phys-

ical Education. Available at: http://www.aahperd.org/Naspe/Shape

OfTheNation/PDF/ShapeOfTheNation.pdf; 2006. Accessed February

11, 2009.

[11] Cleland V, Dwyer T, Blizzard L, et al. The provision of compulsory

school physical activity: Associations with physical activity, fitness

and overweight in childhood and twenty years later. Int J Behav

Nutrition Phys Activity 2008;5:14–22.

[12] Fulton JE, Shisler JL, Yore MM, et al. Active transportation to school:

Findings from a national survey. Res Q Exerc Sport 2005;76:352–7.

[13] Cooper AR, Page AS, Foster LJ, et al. Commuting to school: Are chil-

dren who walk more physically active? Am J Prev Med 2003;25:273–6.

[14] Rosenberg DE, Sallis JF, Conway TL, et al. Active transportation to

school over 2 years in relation to weight status and physical activity.

Obesity 2006;14:1771–6.

[15] Heelan KA, Donnelly JE, Jacobsen DJ, et al. Active commuting to and

from school and BMI in elementary school children—preliminary data.

Child Care Health Dev 2005;31:341–9.

[16] Stice E, Shaw H, Marti N. A meta-analytic review of obesity prevention

programs for children and adolescents: The skinny on interventions that

work. Psychol Bull 2006;132:667–91.

[17] Johnston LD, Delva J, O’Malley PM. Sports participation and physical

education in American secondary schools: Current levels and racial/

ethnic and socioeconomic disparities. Am J Prev Med 2007;33(Suppl

1):S195–208.

[18] Johnston LD, O’Malley PM, Bachman JG, et al. Monitoring the Future

National Survey Results on Drug Use, Secondary School Students,

Volume I. 1975–2006. NIH Publication No. 07–6205. Bethesda,

MD: National Institute on Drug Abuse, 2007.

[19] Hammer LD, Kraemer HC, Wilson DM, et al. Standardized percentile

curves of body-mass index for children and adolescents. Am J Dis

Child 1991;145:259–63.

[20] Pietrobelli A, Faith MS, Allison DB, et al. Body mass index as

a measure of adiposity among children and adolescents: A validation

study. J Pediatr 1998;132:204–10.

[21] SAS Institute Inc. SAS OnlineDoc 9.1.3 [Online]. Available at: http://

support.sas.com/onlinedoc/913/docMainpage.jsp. Accessed November

5, 2008.

http://www.yesresearch.org/pubs.html#reports

http://www.yesresearch.org/pubs.html#reports

http://www.health.gov/paguidelines/guidelines/default.aspx

http://www.health.gov/paguidelines/guidelines/default.aspx

http://www.aahperd.org/Naspe/ShapeOfTheNation/PDF/ShapeOfTheNation.pdf

http://www.aahperd.org/Naspe/ShapeOfTheNation/PDF/ShapeOfTheNation.pdf

http://support.sas.com/onlinedoc/913/docMainpage.jsp

http://support.sas.com/onlinedoc/913/docMainpage.jsp


[22] O’Malley PM, Johnston LD, Delva J, et al. Variation in obesity among

American secondary school students by school and school characteris-

tics. Am J Prev Med 2007;33:187–94.

[23] Dwyer T, Coonan WE, Leitch DR, et al. An investigation of the effects

of daily physical activity on the health of primary school students in

south Australia. Int J Epidemiol 1983;12:308–13.

[24] Flores R. Dance for health: Improving fitness in African American and

Hispanic adolescents. Public Health Rep 1995;110:189–93.

[25] Sallis JF, McKenzie TL, Conway TL, et al. Environmental interven-

tions for eating and physical activity. A randomized controlled trial

in middle schools. Am J Prev Med 2003;24:209–17.

[26] Leupker RV, Perry CL, McKinlay SM, et al. Outcomes of a field trial to

improve children’s dietary patterns and physical activity. The Child and

Adolescent Trial for Cardiovascular Health (CATCH) Collaborative

Group. J Am Med Assoc 1996;275:768–76.

[27] Webber LS, Catellier DJ, Lytle LA, et al. Promoting physical activity in

middle school girls: Trial of activity for adolescent girls. Am J Prev

Med 2008;3:173–84.

[28] Shephard RJ. Curricular physical activity and academic performance.

Pediatr Exerc Sci 1997;9:113–26.

[29] Trudeau F, Shephard RJ. Physical education, school physical activity,

school sports and academic performance. Int J Behav Nutrition Phys

Activity 2008;5:10–21.

[30] Brener ND, McManus T, Galuska DA, et al. Reliability and validity of

self-reported height and weight among high school students. J Adolesc

Health 2003;32:281–7.

[31] Goodman E, Hinden BR, Khandelwal S. Accuracy of teen and parental

reports of obesity and body mass index. Pediatr 2000;106:52–8.

[32] Ogden CL, Carroll MD, Curtin LR, et al. Prevalence of overweight and

obesity in the United States, 1994–2004. J Am Med Assoc 2006;

295:1549–55.

[33] Heise DR. Separating reliability and stability in test-retest correlation.

Am Sociol Rev 1969;34:93–101.

[34] Wiley DE, Wiley JA. The estimation of measurement error in panel

data. Am Sociol Rev 1970;35:112–7.

[35] Dellinger AM, Staunton CE. Barriers to children walking and biking to

school—United States. MMWR 1999;2002(51):701–4.

[36] U.S. Environmental Protection Agency. Travel and Environmental

Implications of School Siting. EPA 231-R-03–004. Washington, DC:

U.S. Environmental Protection Agency, 2003.

[37] Tudor-Locke C, Ainsworth BE, Adair LS, et al. Objective physical

activity of Filipino youth stratified for commuting mode to school.

Med Sci Sports Exerc 2003;35:465–71.

[38] Donnelly JE, Jacobsen DJ, Whatley JE, et al. Nutrition and physical

activity program to attenuate obesity and promote physical and

metabolic fitness in elementary school children. Obes Res 1996;

4:229–43.

[39] Sallis JF, McKenzie TL, Alcaraz JE, et al. The effects of a 2-year phys-

ical education program (SPARK) on physical activity and fitness in

elementary school students. Am J Public Health 1997;87:1328–34.

[40] U.S. Department of Health and Human Services. Healthy People 2010,

2nd edition. With Understanding and Improving Health and Objectives

for Improving Health. 2 vols. Washington, DC: U.S. Government

Printing Office, November 2000. Available at: http://www.

healthypeople.gov/Document/HTML/Volume2/22Physical.htm#_Toc

490380803. Accessed April 2, 2009.

http://www.healthypeople.gov/Document/HTML/Volume2/22Physical.htm#_Toc490380803



school physical activity environment related to student obesity and activity: a national study of...

Documents