April 2001, Vol. 91, No. 4
James F. Sallis, PhD, Terry L. Conway, PhD, Judith J. Prochaska, MS,
Thomas L. McKenzie, PhD, Simon J. Marshall, MS, and Marianne Brown, MPH
A B S T R A C T
Objectives. This study assessed the
association of school environmental char-
acteristics with student physical activity
Methods. Physical activity areas (n=
137) at 24 public middle schools were
assessed for area type, size, and im-
provements (e.g., basketball courts). Stu-
dent physical activity and the presence
of equipment and supervision were di-
rectly observed before school, after
lunch, and after school.
Results. Environmental character-
istics explained 42% of the variance in
the proportion of girls who were physi-
cally active and 59% of the variance for
Conclusions. School environments
with high levels of supervision and im-
provements stimulated girls and boys to
be more physically active. (Am J Public
James F. Sallis is with the Department of Psychol-
ogy, Terry L. Conway is with the Graduate School
of Public Health, and Thomas L. McKenzie is with
the Department of Exercise and Nutritional Sciences,
San Diego State University, San Diego, Calif. Judith
J. Prochaska is with the SDSU/UCSD Joint Doctoral
Program in Clinical Psychology, San Diego, Calif.
At the time of the study, Simon P. Marshall and Mar-
ianne Brown were with Project M-SPAN, San Diego
Requests for reprints should be sent to James
F. Sallis, PhD, Department of Psychology, San Diego
State University, 6363 Alvarado Court, #103, San
Diego, CA 92120 (e-mail: email@example.com).
This brief was accepted June 21, 2000.
Ecologic models of behavior lead to hy-
potheses that environmental characteristics
can influence physical activity.1Schools are
particularly important environments for chil-
dren. Identifying environmental correlates of
student physical activity could facilitate in-
terventions that benefit all children at school.
The present study tested the hypothesis that
schools with adequate space, facilities, equip-
ment, and supervision stimulate students to
be physically active at school during free time.
Twenty-four public middle schools
(grades 6–8) in San Diego County, Calif, were
studied during spring 1997. Mean enrollment
was 1081 students (SD=352); 43% were non-
White, 39% received subsidized meals, and
38% were bused.
All potential physical activity areas at
schools were assessed by observation. Envi-
ronmental variables included the following:
(1) area type—court space with permanent
marks or improvements, open field space with
no markings, and indoor activity space in-
cluding multipurpose rooms and gymnasiums;
(2) area size in square meters; (3) permanent
improvements, including number of basket-
ball hoops, tennis courts, baseball diamonds,
and football or soccer goals. Two assessors
agreed on the coding. Schools had from 2 to
8 activity areas (mean=6.3; total=151).
SOPLAY (System for Observing Play
and Leisure Activity in Youth) was used by
trained assessors to code the number of par-
ticipants and their activity levels.2Activity
areas were “scanned” for girls and boys, and
they were observed separately on fixed sched-
ules during 3 time periods: before school, dur-
ing lunch, and after school. The physical ac-
tivity of each student was coded as sedentary,
walking, or very active (more active than walk-
ing) by validated codes.3,4Assessors recorded
temperature, accessibility of the area, and the
presence of supervision, organized activities,
and equipment. Each school was observed on
3 randomly scheduled days (total=72 days)
with fair weather. Interobserver reliabilities
Each accessible area was a “case.” The
dependent variable was the number of chil-
dren in an area who were engaged in moder-
ate to vigorous physical activity, averaged
across 3 time periods and 3 assessment days
sex-specific linear regressions included 6
sidized lunch, percentage bused, school start
ucation (derived from a parent survey [n=
1609; response rate=72%]). Environmental
observation; area type; area size; total im-
provements; and proportion of observations
with equipment, supervision, and organized
cluded. Variables with P>.20 were deleted
from final models.
Of the 151 areas, 137 were accessible to
students. Forty-four percent of areas were
The Association of School Environments
With Youth Physical Activity
618 American Journal of Public Health
April 2001, Vol. 91, No. 4 American Journal of Public Health619
TABLE 1—Multiple Regression Results Explaining Variance in Students’
Participation in Moderate to Vigorous Physical Activity,With School
F scoreP Eta2
F scoreP Eta2
% students bused
Note. NS=variable not entered in model. Eta2indicates the proportion of variance
explained by the variable.
bR2=.630 (adjusted R2=.594)
Note. “% of students in MVPA” is the percentage of boys and girls attending school
who were observed to engage in moderate to vigorous physical activity (MVPA)
during free time throughout the school day. High and low levels of variables were
determined by median splits. Data are based on observations in 137 activity areas.
FIGURE 1—Area type×supervision: interactions for boys and girls.
outdoor courts, 43% were outdoor fields,
and 13% were indoors. The mean number of
permanent improvements was 66.6 per
school (SD=20.5; range=10–97). The most
common improvements were basketball
hoops and courts (100% of schools), baseball
backstops (87%), volleyball nets (87%), ten-
nis courts (79%), and racquetball courts
To obtain a schoolwide estimate of the
proportion of physically active students, we
summed average rates of moderate to vigor-
ous physical activity across activity areas. The
proportion of girls observed being physically
active ranged from 0% to 5% across schools,
with a mean of 1.6% (SD=1.2%). The pro-
portion of physically active boys ranged from
1% to 11%, with a mean of 5.5% (SD=2.7%).
The final multiple regression model for
girls explained 42% of the variance in ob-
served physical activity (Table 1).The inter-
action “area type×supervision” explained
11% of the variance and revealed that su-
ment” indicated that more girls were active
when equipment was not present in indoor
areas, but equipment enhanced activity lev-
els in outdoor areas (Figure 2). The interac-
tion “supervision×area improvements” in-
dicated that girls were most active when
school environments had high levels of both
improvements and supervision (Figure 3).
of the variance in physical activity (Table 1).
Figure 1 shows that boys were most likely to
ure 2 indicates that the largest proportion of
boys were active on courts when equipment
was available. The interaction “area type×
equipment” explained 16% of the variance.
boys were active when areas had high levels
of both improvements and supervision.
The main findings of this study can be
summarized by paraphrasing a line from the
movie Field of Dreams: If we build it, they will
come—and be active.
ical activity and 59% of the variance for boys.
alistic improvements to school environments
and 5-fold higher, respectively, than when the
of girls and 6% of boys.An absence of envi-
ronmental support was associated with near-
students would be stimulated to be physically
environment. It is not clear how much further
improvements in school environments could
April 2001, Vol. 91, No. 4 620American Journal of Public Health Download full-text
Note. See note for Figure 1.
FIGURE 2—Area type×equipment: interactions for boys and girls.
Note. See note for Figure 1.
FIGURE 3—Supervision×improvements: interactions for boys and girls.
Strengths of the study included objec-
tive measurement of all variables, the de-
pendent variable that reflected population-
wide physical activity, and multiple obser-
vation days at schools. Because the study was
conducted in one region with favorable
weather, generalization to other regions needs
to be tested.
Girls and boys were more likely to
choose to be physically active when there
were many improvements to activity areas
and when adults supervised activities. Inter-
ventions that enhance school physical envi-
ronments and social resources are expected
to be effective in attracting students to ac-
tivity areas and stimulating student physical
J.F. Sallis was the lead author and made most of
the revisions. T.L. Conway, J.J. Prochaska, and S.J.
Marshall contributed to conceptualizing and running
the statistical analyses. Each author wrote a portion
of the original draft and edited all versions: J.F. Sal-
lis drafted the introduction and part of the Discus-
sion section; T.L. Conway drafted part of the Dis-
cussion section and contributed to the Methods
section; J.J. Prochaska wrote the analysis methods
and the Results section; T.L. McKenzie drafted the
measurement methods; S.J. Marshall contributed
to the Methods section; and M. Brown drafted the
part of the Methods section dealing with subjects
The study was supported by NIH grant HL54564.
Diane Ward contributed to the development of
measures used in the study.
The study was approved by the Committee for
the Protection of Human Services, San Diego State
1. Sallis JF, Bauman A, Pratt M. Environmental
and policy interventions to promote physical ac-
tivity. Am J Prev Med. 1998;15:379–397.
2. McKenzie TL, Marshall SJ, Sallis JF, Conway
TL. Leisure-time physical activity in school en-
vironments: an observational study using SO-
PLAY. Prev Med. 2000;30:70–77.
3. McKenzie TL, Sallis JF, Patterson TL, et al.
BEACHES: an observational system for as-
sessing children’s eating and physical activity
behaviors and associated events. J Appl Behav
4. Rowe PJ, Schuldheisz JM, van der Mars H.
Measuring physical activity in physical educa-
tion: validation of the SOFIT direct observa-
tion instrument for use with first to eighth
grade students. Pediatr Exerc Sci. 1997;9: