ArticlePDF Available

A Meta-analysis of the Relationship Between Children's Physical Activity and Mental Health



The present study was a comprehensive, quantitative synthesis of the literature examining the effects of physical activity on children’s mental health outcomes. The final analysis included 73 published and unpublished studies, totaling 246 effect sizes. Various study and participant characteristics were coded to assess moderator effects, including type of physical activity, mental health outcome, gender, cognitive ability, mental status, and implementer of the physical activity, etc. Results demonstrated varying effects depending on the methodology of the examined study [i.e., correlational vs. randomized controlled trial (RCT)/non-RCT] and characteristics of the participants, although overall effects of physical activity on children’s mental health were small but significant, indicating that on average physical activity led to improved mental health outcomes for all children.
A Meta-analysis of the Relationship Between Children’s Physical
Activity and Mental Health
Soyeon Ahn,
PHD, and Alicia L. Fedewa,
Department of Educational and Psychological Studies, University of Miami, and
Department of Educational,
School, and Counseling Psychology, University of Kentucky
All correspondence concerning this article should be addressed to Soyeon Ahn, PHD, 5202 University Dr.,
Merrick Building #305-9, Coral Gables, FL, 33146, USA. E-mail:
Received May 28, 2009; revisions received October 24, 2010; accepted October 31, 2010
The present study was a comprehensive, quantitative synthesis of the literature examining the effects of physi-
cal activity on children’s mental health outcomes. The final analysis included 73 published and unpublished
studies, totaling 246 effect sizes. Various study and participant characteristics were coded to assess moderator
effects, including type of physical activity, mental health outcome, gender, cognitive ability, mental status, and
implementer of the physical activity, etc. Results demonstrated varying effects depending on the methodology
of the examined study [i.e., correlational vs. randomized controlled trial (RCT)/non-RCT] and characteristics
of the participants, although overall effects of physical activity on children’s mental health were small but
significant, indicating that on average physical activity led to improved mental health outcomes for all children.
Key words interventions; mental health; overweight; physical activity.
Within the past decade, the US has seen a steady decline in
the numbers of physically active children (Centers for
Disease Control [CDC], 2008). Although children spend
the majority of the day in classrooms, schools are increas-
ingly under more pressure to meet high stakes testing
standards. This pressure has created the push for more
instructional time and less time devoted to physical activity
(i.e., physical education or recess breaks; Burgeson,
Weschler, Brener, Young, & Spain, 2001; Hardman,
2008). Moreover, technology has afforded children more
opportunities to play video games, watch TV, or browse the
Internet, activities that contribute to sedentary behaviors
(Stevens, To, Stevenson, & Lochbaum, 2008). Parents
also report having more concern about their children’s
safety in playing outside or using active means of trans-
portation (i.e., biking, walking) on their way to school,
further limiting the amount of children’s physical activity
(Stevens et al., 2008; World Health Organization [WHO],
When physical activity is restricted during school
hours, children do not compensate for loss of physical
activity after school, resulting in children who are incredibly
sedentary throughout the majority of the day (Dale, Corbin,
& Dale, 2000). The relationship between sedentary behav-
iors and prevalence of obesity has been well documented
(CDC, 2008; Pate et al., 2002). For the first time in history,
children have a shorter lifespan than their parents due
to obesity-related diseases (Wang & Veugelers, 2008;
WHO, 2009). Although, but one factor in a myriad of influ-
ences, the amount of physical activity children engage in is
linked to their status of being overweight or obese (National
Center for Health Statistics, 2009).
One critical relationship that has been examined in the
literature has been the link between physical activity and
mental health. Although the research in this area is scarce
compared to studies examining the effects of physical ac-
tivity on mental health in adults, there is a considerable
need for this body of research. Approximately 20% of
school-age children have a diagnosable mental health
Journal of Pediatric Psychology pp. 113,2011
Journal of Pediatric Psychology ßThe Author 2011. Published by Oxford University Press on behalf of the Society of Pediatric Psychology.
All rights reserved. For permissions, please e-mail:
Journal of Pediatric Psychology Advance Access published January 11, 2011
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
disorder and require psychological treatment (U.S. Public
Health Service, 2000). However, the majority of children
do not receive services to remediate their concerns
(Thompson, 2005; U.S. Public Health Service, 2000).
One subset of the childhood population that is particularly
at risk for mental health disorders are those children clas-
sified as overweight or obese. It is well documented in the
literature that children who are classified as overweight or
obese have significantly more psychosocial problems than
do children who are of typical weight (Farhat, Iannotti, &
Simons-Morton, 2010). Yet, when studies examine the
effects of physical activity on children’s mental health,
differential effects for children who are either overweight
or obese are commonly not considered. Although physical
activity interventions that have been used to build
self-esteem and physical fitness in youth have exhibited
promising results (Ekeland, Heian, & Hagen, 2005), it is
unclear what the impact of physical activity has on the
mental health of all children, including those who are
considered overweight or obese.
In a meta-analysis of 16 randomized controlled stud-
ies, Larun and colleagues (Larun, Nordheim, Ekeland,
Hagen, & Heian, 2006) investigated the effects of vigorous
exercise interventions in preventing or reducing anxiety or
depression in children and youth. Although depression
and anxiety were the only outcome variables in this
meta-analysis, results were in favor of exercise interven-
tions in alleviating or preventing negative symptoms in
children and youth (Larun et al., 2006). Although the
2006 meta-analysis by Larun and colleagues was a compre-
hensive synthesis of the literature with respect to the out-
comes of anxiety and depression symptomology in youth,
there are a number of other mental health concerns that
affect school-aged populations. Attention-deficit hyperac-
tivity disorder (ADHD) is typically a comorbid condition
in children with such diagnoses as anxiety or depression,
and there have been a handful of studies investigating the
relationship between physical activity and mental health
outcomes in children diagnosed with ADHD. In addition,
self-esteem has been found to be an important buffer in the
onset of childhood mental disorders (Ekeland et al., 2005);
it is therefore imperative to consider the role of self-esteem
in relation to physical activity and children’s mental health.
Including these studies in the meta-analysis would
have provided researchers a more comprehensive picture
of the relationship between physical fitness and mental
health in children. Further, the 2006 meta-analysis
excluded studies that were not randomized controlled clin-
ical trials, leaving open the question as to whether effects
have been found in other studies, including quasi-
experimental and correlational designs. Last, Larun and
colleagues (2006) excluded children who were classified
as overweight or obese and therefore did not take into
account children’s health status. It is important to consider
the differential effects physical activity may have on chil-
dren’s health status, especially considering the docu-
mented risk of overweight/obese children with increased
psychosocial difficulties (Farhat et al., 2010).
Given the number of studies that were not included in
the 2006 meta-analysis, as well as the multitude of other
mental health concerns pervasive in youth not taken into
account, this study attempted to fill the gap in examining
the relationship between physical activity and children’s
mental health. Moreover, a comprehensive set of modera-
tor variables were also examined. No meta-analysis to
date has examined differential effects of moderator vari-
ables in the relationship of physical activity and mental
health in children, although there is reason to believe
that intervention effects of physical activity may differ
depending on such moderators as gender (Kremers,
Droomers, Van Lenthe, & Brug, 2007; Simen-Kapeu &
Veugelers, 2010), age (Fedewa & Ahn, in press; Kremers
et al., 2007), methodological design (Conn, 2010), and
implementer (Stice, Shaw, & Marti, 2006), to name a
few. Researchers have argued for the increased use of mod-
erator analyses in examining outcomes of physical inter-
ventions in children given the multitude of differential
relationships and mechanisms of behavior change in
youth (Kremers et al., 2007). Thus, a number of moderator
analyses were conducted to determine if child mental
health outcomes were associated with various charac-
teristics of samples, research methodology, or interven-
tions. Therefore, the present meta-analysis addressed the
following two questions:
1. What are the overall effects of physical activity on
children’s mental health?
2. Do the effects of physical activity on children’s
mental health vary depending on the intervention,
sample, and study design characteristics? In partic-
ular, is physical activity more important for chil-
dren who are classified as obese or overweight?.
The Search Process
The location of relevant studies in this research synthesis
was as exhaustive as possible, and included both published
and unpublished literature based on a manual as well as a
computerized search of pertinent databases including
PsychLit, PsychInfo, Dissertation Abstracts, MedLine, and
ERIC. Key terms for literature searches included the words
2Ahn and Fedewa
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
‘‘physical activity,’’ ‘‘physical fitness,’’ ‘‘physical exercise,’’
‘‘curricular activity,’’ ‘‘exercise,’’ ‘‘mental health,’’ ‘‘anxiety,’’
‘‘depression,’’ ‘‘dysthymia,’’ ‘‘bipolar,’’ ‘‘post-traumatic stress
disorder,’’ ‘‘attention deficit hyperactivity disorder,’’ ‘‘eating
disorder,’’ ‘‘anorexia,’’ ‘‘bulimia,’’ ‘‘youth,’’ ‘‘adolescents,’’
and ‘‘children’’. As well as database resources, general
search engines (e.g., Google) were employed with the
above key terms to capture those studies that had not
been included in the databases. Lastly, literature reviews,
ancestry searches, and comprehensive analyses conducted
in the area (i.e., Jorm, Allen, O’Donnell, Parslow, Purcell,
& Morgan, 2006; Larun et al., 2006; Ortega, Ruiz, Castillo,
& Sjostrom, 2008) were searched to include any additional
bibliographic information. Results yielded over 150 refer-
ences between 1960 and 2010.
Studies retrieved from the initial searches were
screened using specific criteria: (a) studies had to investi-
gate the effect or relationship of some type of physical
activity and children’s mental health (i.e., the dependent
variable was a mental health outcome of some form);
(b) target populations had to range from pre-school to
high-school age (3–18 years); (c) no qualitative or concep-
tual studies were included; (d) data that have only been
used once in a manuscript to avoid replication (i.e., studies
that had published more than one article on the same
participants were not included, as were studies that were
done as unpublished theses and subsequently published);
and (e) studies must have been reported in English. This
process identified a total of 95 studies.
Out of 95 studies, 22 studies (a reference list of 22
excluded studies is available online as Supplementary
Data) were excluded due to the following reasons: (a) 20
studies did not provide sufficient information (i.e., mean,
standard deviation) for calculating effect size and (b) two
studies used advanced data analysis techniques such as
regression. Therefore, a total of 73 studies (a reference
list of 73 included studies is available online as
Supplementary Data) were included in the current research
Coding of Studies
Based on a literature review, a systematic coding scheme was
developed to identify salient features of each study.
Specifically, variables with regard to (a) study design,
(b) participant, (c) physical activity/exercise, and
(d) mental health characteristics were independently
coded and entered into the computer database for statistical
analyses. Coding of these variables was mainly based on
author’s report. When no information was given by
author(s), variables were coded as ‘‘not informed.’’ The
author and a graduate student independently coded and
entered variables described above. All discrepancies were
resolved upon discussion.
Study Design Characteristics
Study design was coded as (1) between-subject design
(i.e., posttest-only-control group design), (2) within-subject
design (i.e., pretest–posttest design), (3) mixed-design
(i.e., pretest–posttest control group design), and (4) cross-
sectional or correlational design. Based on the research
questions being asked, the included studies were catego-
rized into either (a) group comparison study examining
the effect of physical activity interventions on mental
health outcomes or (b) cross-sectional/correlational study
examining the relationship between physical activity and
mental health outcomes.
When comparison groups were used, the assignment
methods that allocate subjects into comparison groups
were categorized into the following areas: (a) random,
(b) not random, and (c) not informed. Then, studies
using random assignment were next categorized as ran-
domized controlled trials (RCTs), which were compared
to the rest of studies (i.e., non-RCTs). Study setting was
coded as (1) school, (2) clinic, (3) after-school program,
(4) research center, and (5) other. Last, other study char-
acteristics such as publication type (i.e., published vs.
unpublished) and study location (i.e., US vs. non-US)
were coded.
Participant Characteristics
Participants were coded as (1) typical/typical inferred, (2)
cognitively impaired, (3) learning disabled or children with
academic delays, (4) children with ADHD, (5) children
with Post-Traumatic Stress Disorder (PTSD), (6) children
with emotional problems (e.g., anxiety, depression), (7) chil-
dren with behavior problems (including children with con-
duct disorder), (8) children who had undergone cancer
treatment, and (9) not informed. Participants were also
coded as (1) typical, (2) fit, (3) mixed, and (4) not informed,
depending on their physical fitness status. Other informa-
tion including whether participants were diagnosed or not
(i.e., Yes or No), mean age and gender (i.e., male, female, or
mixed) were also coded.
Physical Activity Characteristics
Specific characteristics of physical activity were coded.
First, the focus of physical activity was qualitatively collect-
ed and then categorized into (1) aerobic training, (2) resis-
tance/strength/circuit training, (3) flexibility training, (4)
regular PE program, (5) sport participation such as ski,
football, and volleyball, (6) movement/motor skill training,
(7) yoga (including meditation), (8) combined, and (9) not
Meta-analysis of Physical Activity 3
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
informed. These categories were constructed based on the
description provided by the authors of the study, as well as
the features of the activity children were receiving. If two
categories overlapped (i.e., if aerobic training and resis-
tance training were both targeted for the intervention),
then the study features were coded as a ‘‘combined’’ inter-
vention. The exception to this coding scheme was for the
category of PE intervention, as it was often unclear what
the focus of the PE intervention focus was with respect to
the type of physical activity. Thus, studies using PE pro-
gramming as their physical activity intervention received a
coding of ‘‘PE intervention,’’ even if the PE intervention
may have included aerobic training. The authors chose this
coding framework to maintain consistency across studies,
as the majority of PE programming interventions did not
explicitly state the target of their physical activity. Second,
total hours, frequency per week, intensity (i.e., light, mod-
erate, intense, and mixed when different levels of intensity
were involved), and unit (i.e., individual-based, small
group with less than 10 subjects, medium group with sub-
jects between 10 and 30, large group with more than
30 subjects, and whole class) of the physical activity
were collected. Last, the administrator who led the physical
activity (i.e., teacher/instructor, researcher, PE specialist,
external instructor including after-school counselors, vol-
unteers, after-school counselors, and recreational therapist)
was also coded.
Mental Health Outcome Characteristics
Mental health outcome measures were categorized into de-
pression/dejection, anxiety, global self-esteem, self-concept
(including exercise self-concept, physical self-concept, ac-
ademic self-concept, social self-concept, and family/home
self-concept), distress/PTSD and emotional distress, psy-
chological distress or a combination of multiple symptoms
(e.g., depression and anxiety; emotional disturbance),
suicidal ideation, ADHD, life satisfaction, somatic symp-
toms, problems in social functioning, conduct/behavioral
problems, cognitive impairment, and psychosis.
Effect Size
Depending on the research questions and/or designs of the
included study, the following two types of effect sizes (ESs)
were computed in the current meta-analysis: one for com-
parison studies (d) and the other for correlational/
cross-sectional studies (r). From the comparison studies,
the standardized mean difference between treatment/inter-
vention and control groups was computed. The group with
no physical activity intervention was treated as a control
group. When studies used a pretest–posttest control group
design, dwas computed using the formulas in Morris
(2008). If no pretest was used, dwas computed using
the formula in Lipsey and Wilson (2001). When no suffi-
cient statistics were reported, dwas computed from the
reported tor Fstatistics using the formulas outlined in
Rosenthal (1994). Also, the odds ratio was converted to
dusing the formula presented in Borenstein (2009). From
the correlational studies examining the association between
physical activity and mental health outcome, the reported r
was obtained. In addition, if comparison studies examined
the relationship of physical activity level with mental health
outcome, d-ESs were converted to rby the formulas shown
in Rosenthal (1994).
Statistical Analyses
The statistical analyses were based on the methods pro-
posed by Hedges and Olkin (1985) and also described in
Cooper, Hedges, and Valentine (2009). Under the
fixed-effect model, the computed effect-sizes were weighted
by the inverse of its variance, and an overall homogeneity
test of these effects (Qtotal) was initially performed. When
the fixed-effects model did not hold (or Qtotal was signifi-
cant), the random-effects model or mixed-effects model
with predictors were applied. The random-effects model
incorporated the additional uncertainty to the effect vari-
ances, which was estimated using the methods of mo-
ments. Further, the mixed-effects model with moderators
(i.e., children’s mental health outcomes) incorporated ad-
ditional uncertainty within each level of categorical mod-
erators, whose weights were computed for effect in each
level of moderators. More details about random-effects or
mixed-effects models with categorical moderators can be
found in Raudenbush (2009).
Studies often provided dependent effect sizes by using
multiple measures of variables, which in turn violates the
assumption of independence (Glesser & Olkin, 2009).
For instance, Allison et al. (2005) used three mental
health outcomes including psychological distress, prob-
lems in social functioning, and depression/anxiety. Such
dependency issues can be handled in various ways
(Becker, 2000).
In this meta-analysis, the issue of dependency was first
handled by choosing effect sizes from the total score or
averaging effect sizes from subtest scores if no total score
was presented, rather than using a subtest score. Effect
sizes were then grouped into subcategories of physical ac-
tivity and mental health measures described above and
thus they were no longer dependent within each subcate-
gory for the computation of the overall effect sizes. The
authors chose this method due to its simplicity and
4Ahn and Fedewa
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
feasibility as compared to multivariate methods, which re-
quire a full variance-covariance matrix of dependent effect
Description of Studies
The included 73 studies yielded a total of 246 ESs. These
included 182 d- and 64 r-ESs. Studies were published
between 1974 and 2009 and most of the studies (s¼55)
were conducted in the US, while the rest of them were
from various countries including Brazil (s¼1), German
(s¼1), Hungary (s¼1), South Africa (s¼3), Switzerland
(s¼1), the UK (s¼6), China (s¼2), Canada (s¼4), and
Australia (s¼2). Sample sizes used in the 73 studies
ranged from 9 to 14,594 (M¼484.13, SD ¼1714.82),
including 14 to 502 (M¼55.67, SD ¼86.86) from RCT
(s¼30), 9 to 2,444 (M¼192.85, SD ¼502.27) from
non-RCT (s¼24), and 35 to 14,594 (M¼1398.58,
SD ¼3441.39) from correlational studies (s¼19).
Participants’ age ranged from 3.67 to 17.66 years
(M¼12.67, SD ¼2.94). In the majority of studies, chil-
dren were typical in their mental, fitness, and diagnostic
status (i.e., represented non-clinical populations).
Publication Bias
The current review included both unpublished and pub-
lished studies. However, the Egger’s regression tests
(Sutton, 2009) were found to be significant for both
effect sizes (t(180) ¼1.34, p< .01 for d-ESs;
t(62) ¼1.60, p¼.02 for r-ESs), indicating the presence
of potential publication bias.
Comparison Studies
A total 182 d-effect sizes examining the intervention
effect of the physical activity on children’s mental out-
comes were first analyzed. The significant Q-statistic
of 1395.11 indicates that the included d-ES were hetero-
geneous. Further, the estimated d-ES under the random-
effects model was 0.38 with a SE of 0.11, which was
statistically significant. Such a significant but negative
result indicates that physical activity has a moderate
effect on alleviating children’s negative mental health
However, a statistically significant mean d-ES differ-
ence was found between RCT and non-RCT studies
(Qð1Þ¼65:58,p<:01), showing significantly lower
mean d-ES from RCT studies (
d¼:30,SE ¼0:06)
when compared to non-RCT studies (
SE ¼0:24). Also, the statistically significant Qstatis-
tics suggested that d-ESs for both studies were
statistically different (Qð101Þ¼298:68,p<:01for RCT;
Qð78Þ¼1030:80,p<:01 for non-RCT). Thus, the follow-
ing moderator analyses were performed separately for
RCTs and non-RCT studies. Table I shows the results
from the RCT and non-RCT studies.
Type of Mental Health Outcome
The intervention effect of the physical activity program
significantly differed by type of mental health outcome
for both RCT (Qð11Þ¼133:67,p<:01) and non-RCT
studies (Qð11Þ¼181:23,p<:01). From RCT studies,
the physical activity intervention was found to be effec-
tive for reducing depression (
d¼:41,SE ¼0:13),
anxiety (
d¼:35,SE ¼0:18), psychological distress/
d¼:61,SE ¼0:30), and emotional disturbance
d¼:33,SE ¼0:17). It was also found that physical
activity significantly enhanced children’s self-esteem
d¼:29,SE ¼0:08) and their self-concept (
SE ¼0:10). However, the treatment effect from non-RCT
studies was significant only for increasing children’s
self-esteem (
d¼:78,SE ¼0:28).
Physical Activity Intervention
From both RCT and non-RCT studies, the estimated
mean ESs were significantly different depending on the
characteristics of the physical activity programs, including
focus, intensity, intervention unit, total hours, frequency
per week, and administrator of the physical activity
First, the mean d-ESs from RCT studies were statis-
tically significant and largest when the intervention
was focused exclusively on circuit training (
SE ¼0:29), followed closely by the intervention with a
combined physical activity focus (
d¼:57,SE ¼0:11).
These results indicate that interventions with a focus on
circuit training or a combination of aerobic and resistance
training significantly lowered children’s mental health dis-
turbances when compared to a control group with no phys-
ical activity. From non-RCT studies, the intervention with
sport participation was the sole program that showed a
statistically significant reduction on children’s mental
disturbance, compared to control groups with no physical
activity (
d¼1:06,SE ¼0:27).
Second, an intervention with an intense level of
physical activity was found to be significant for reducing
children’s mental health disturbance from RCT studies
d¼:27,SE ¼0:08), although the largest and most
significant intervention effect was found from RCT studies
that did not indicate the intensity level of their physical
activity intervention (
d¼:41,SE ¼0:10). From
non-RCT studies, a moderate intensity of physical activity
Meta-analysis of Physical Activity 5
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
Tab le I . Moderator Analyses for d-Effect Sizes
RCT studies Non-RCT studies
k M SE Qwithin K M SE Qwithin
Outcome Q(11) ¼133.67** Q(11) ¼182.23**
Depression 14 0.41** 0.13 32.19** 16 1.14 0.71 94.19**
Anxiety 16 0.35* 0.18 69.96** 9 1.51 0.85 27.16**
Self-esteem 26 0.29** 0.08 60.71** 16 0.78** 0.28 163.25**
Distress/PTSD 5 0.61* 0.30 13.83** 7 1.42 1.15 36.01**
Emotional disturbance 4 0.33* 0.17 7.10 5 0.42 0.41 4.53
ADHD 5 0.92 0.64 17.17** 2 0.31** 0.10 6.09*
Somatic symptom 3 0.35 0.21 1.14 4 0.43 0.40 6.26
Social function problem 11 0.26 0.15 27.48** 6 1.15 0.93 439.63**
Conduct problem 6 0.00 0.46 2.20 6 0.12 0.63 48.99**
Cognitive problem 2 0.50 0.66 6.54* 1 0.82 2.77 -
Self-concept 9 0.16** 0.10 11.59 6 0.12 0.31 16.22**
Quality of life 2 0.15 0.09 0.04 1 0.79 0.9 -
PA focus Q(4) ¼44.21** Q(6) ¼145.22**
Aerobic 30 0.14 0.09 48.51* 48 0.75 0.38 226.17**
Circuit/Strength 9 0.72* 0.29 9.57 3 0.32 0.33 1.16
Flexibility 34 0.13 0.11 95.53** 10 0.69 0.39 111.72**
Combined focus 29 0.57** 0.11 100.86** 3 0.27 0.37 46.55**
Sport participation 1 0.11  51.06** 0.27 17.57**
Yoga/Meditation    10.99 –
Not indicated 9 0.32 0.27 482.46**
Intensity Q(4) ¼32.96** Q(4) ¼35.21**
Light 5 0.10 0.29 0.111 3 0.34 0.40 5.82
Moderate 7 0.18 0.17 3.018 3 1.89* 0.90 7.97
Intense 26 0.27** 0.08 36.35 23 0.22 0.41 138.77**
Mixed 13 0.06 0.18 5.31 12 1.44 0.87 36.61**
Not indicated 52 0.41** 0.10 220.93** 38 0.49 0.31 806.46**
Unit Q(4) ¼22.69** Q(4) ¼57.35**
Individualized 7 1.82** 0.65 17.01**
Small group (<10) 8 0.10 0.21 1.24 3 0.50** 0.14 1.16
Medium group (10–30) 26 0.13 0.08 33.69 6 1.73 1.67 28.59**
Large group (>30) 14 0.38 0.21 34.82**
Total class 16 0.45** 0.11 41.32** 7 0.25 1.39 33.01**
Not indicated 39 0.39** 0.10 164.91** 56 0.24 0.17 893.73**
Hours Q(2)¼24.46** Q(2)¼41.11**
Less than 20 hr 53 0.16** 0.04 53.69 25 1.84** 0.59 117.20**
20–33 hr 31 0.42** 0.13 125.61** 11 0.28** 0.08 10.27
More than 33 hr 10 0.55** 0.19 74.07** 17 0.09 0.43 109.27**
Frequency Q(6) ¼60.73** Q(4) ¼51.39**
1210.57** 0.18 108.38** 10 0.81* 0.48 103.85**
2180.37** 0.09 39.13** 1 0.99 –
3350.09 0.08 58.00** 39 0.60 0.46 197.99**
470.61 0.43 7.8 3 0.32 0.55 1.16
5140.30 0.16 5.58 – –
711.19 – –
Not indicated 7 0.39* 0.45 20.98** 26 0.37 0.25 676.46**
Administrator Q(5) ¼16.28** Q(5) ¼47.19**
Teacher 12 0.36* 0.15 60.66** 48 0.76* 0.35 226.17**
Researcher 37 0.20** 0.08 45.71 3 1.71** 0.65 1.16
PE specialist 9 1.02** 0.33 15.8 10 0.09 0.18 111.72**
6Ahn and Fedewa
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
showed a significant decrease on negative mental health
outcomes (
d¼1:89,SE ¼0:90), compared to control
groups with no physical activity.
Third, the mean d-ES from RCT studies was negative
and statistically significant for physical activity done in
school with the entire classroom (
d¼:45,SE ¼0:11).
However, the mean d-ESs from non-RCT studies were sta-
tistically significant for individualized physical activity
d¼1:82,SE ¼0:65) or small group physical activity
d¼:50,SE ¼0:14). Post hoc tests comparing mean
d-ESs from non-RCT studies indicated that the individu-
alized physical activity program appeared to be the most
effective for alleviating negative mental health outcomes
¼1.32, p¼.047).
Fourth, the total hours of the physical activity inter-
vention was categorized into three groups: (a) less than
20 hr, (b) 20–33 hr, and (c) more than 33 hr. These three
categories were created based on the mean hours of
the physical activity or physical education program
(20 hr) and its standard deviation (13 hr). These hours
were distributed over the length of the intervention,
which varied by study design. For RCT studies, the
average length of time for physical activity interventions
was 11.1 weeks (SD ¼3.6), while for non-RCT
studies the mean length was 8.95 weeks (SD ¼4.5).
From RCT-studies, physical activity programs with more
than 33 hr showed a statistically significant reduction
in mental health disturbance (
d¼:55,SE ¼0:21),
followed by 20–33 hr of physical training (
SE ¼0:21) and less than 20 hr physical activity
d¼:55,SE ¼0:21). From non-RCT studies, mean
d-ESs showing an intervention effect with less than 20 hr
d¼1:84,SE ¼0:59) or 20–33 hr (
SE ¼0:08) were significant.
Fifth, the mean d-ES from RCT studies was statistically
significant when the physical activity program was provid-
ed once per week (
d¼:57,SE ¼0:18) and twice per
week (
d¼:37,SE ¼0:21). From non-RCT studies, the
intervention with physical activity provided once per week
showed significantly more effect on decreasing negative
mental health outcomes when compared to the control
group (
d¼:81,SE ¼0:48).
Last, mean d-ESs from RCT studies were significant
when the physical activity program was administered by
Table I. Continued
RCT studies Non-RCT studies
k M SE Qwithin K M SE Qwithin
After-school counselor 7 0.11 0.46 1.35 3 0.08 0.53 46.55**
Not indicated 33 0.41** 0.12 158.77** 5 0.27 0.16 17.57**
Therapist 5 0.10 0.29 0.11 1 0.02
Clinician 9 0.82 0.78 482.46**
Gender Q(3) ¼33.80** Q(3) ¼49.33**
Female 6 0.41 0.43 7.60 23 1.60* 0.66 245.34**
Male 34 0.55** 0.11 119.79** 13 0.04 0.28 476.73**
Mixed 63 0.19** 0.07 137.48** 4 0.39 0.22 238.25**
Not informed 2 0.00 0.07
Diagnostic status Q(1) ¼59.19** Q(1) ¼25.23**
Yes 23 0.17 0.25 64.28** 30 1.00 0.52 197.13**
No 80 0.07* 0.12 218.03** 49 0.64* 0.27 808.49**
Mental status Q(6) ¼9.46 Q(6) ¼84.03**
Normal 53 0.29** 0.04 179.38** 48 0.27** 0.11 788.97**
Cognitively impaired 2 2.49* 1.42 18.63**
Learning disabled 20 0.36** 0.07 61.01**
ADHD 6 0.48 0.27 13.83* 1 0.99 –
Emotionally disturbed 19 0.11 0.13 17.61* 4 0.37 0.20 3.02
PTSD 1 1.27 11 3.42** 1.17 65.71**
Children w/behavior Problem 1 0.11 2 0.03 0.63 3.33
Not indicated 3 0.16 0.21 1.27 11 0.33 0.63 74.93**
Fitness status Q(1) ¼0.07 Q(1) ¼25.23**
Normal 67 0.29** 0.04 197.13** 56 0.23* 0.11 851.57**
Fit 36 0.27** 0.05 808.49** 4 0.06 0.26 0.71
Note. **p<.01; *p<.05.
Meta-analysis of Physical Activity 7
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
the classroom teacher (
d¼:36,SE ¼0:15), researcher
d¼:20,SE ¼0:08), or physical education specialist
d¼1:02,SE ¼0:33). From non-RCT studies, the
significant treatment effect was found when physical activ-
ity intervention was administered by the classroom teacher
d¼:76,SE ¼0:35) and researcher (
SE ¼0:65), whose mean d-ESs were not statistical
different (M
¼0.95, p¼.19).
Participant Characteristics
From both RCT and non-RCT studies, significant mean
differences were found by several participant characteristics
including gender and diagnostic status. However, the in-
tervention effect did not depend on age or whether partic-
ipants were on a prescribed regimen of medication. A
significant mean difference by children’s mental health
status and physical fitness status was found only from
non-RCT studies, not from RCT studies.
First, both male and mixed-gender groups from RCT
studies showed significant intervention effects, while only
females from non-RCT studies showed significant interven-
tion effects on alleviating negative mental outcomes.
Second, it was found from both RCT and non-RCT studies
that the physical activity programs were more effective
in reducing negative mental health outcomes for children
who were clinically diagnosed. Third, the overall means
from non-RCT studies varied depending on children’s cog-
nitive and mental health status. In particular, the overall
mean d-ES from the cognitively impaired and PTSD classi-
fied groups showed the largest effects. Post hoc analysis
indicates that the overall mean d-ES for PTSD was signifi-
cantly larger compared to typically developing children
¼2.68, p< .05). Fourth, the intervention effect
was equally effective for children classified as overweight/
obese and children who were of average weight.
Correlational Studies
A total of 64r-ESs represents the relationship between chil-
dren’s level of physical activity and mental health. The
significant Qstatistics of 1040.02 indicates that the 64r-
ESs were statistically different. Under the random-effects
model, the estimated average correlation was statistically
significant, having a weighted mean of 0.06 with a SE
of 0.02. The statistically significant and negative effect in-
dicates that greater physical activity was related to a lesser
likelihood of experiencing detrimental mental health out-
comes. Table II shows the results from correlational/
cross-sectional studies.
Type of Mental Outcome
The overall relationship between physical activity level
and mental health differed depending on the type of
mental health outcome (Qð11Þ¼272:20,p<:01). Of
the 11 types of mental health outcomes, the level of phys-
ical activity showed significant relations to depression
r¼:14,SE ¼0:04) and self-concept (
SE ¼0:05). Such results indicated that the level of phys-
ical activity had a significantly negative relationship
with depression and significantly positive relationship to
a child’s self-concept.
Participant Characteristics
The mean r’s varied by several participant characteristics
including gender, cognitive status, and physical fitness
Tab le I I. Moderator Analyses for r-Effect sizes
Study characteristics K M SE Qwithin Study characteristics k M SE Qwithin
Outcome, Q(10) ¼540.32** Fitness status, Q(3) ¼286.81**
Depression 12 0.14** 0.04 272.2** Typical 23 0 0.05 37.20*
Anxiety 7 0.09 0.06 22.85** Fit 4 0.02 0.02 0.29
Self-esteem 14 0.04 0.04 30.75** Obese 35 0.12** 0.04 174.61**
Distress/PTSD 5 0.04 0.06 73.03** Not indicated 2 0.03 0.15 541.19**
Emotional mood 3 0.09 0.08 18.15** Mental status, Q(2) ¼87.79**
Somatic symptom 3 0.01 0.01 1.65 Typical 60 0.07** 0.03 408.86**
Problem in social function 7 0.04 0.06 32.61** Learning disabled 2 0.04 0.05 2.27
Conduct problem 1 0.19 0.14 Not indicated 2 0.03 0.16 541.09**
Psychosis 2 0.01 0 0.01 Gender, Q(2) ¼43.46**
Self-concept 8 0.14** 0.05 52.6** Girl 13 0.13* 0.06 98.71**
Suicidal ideation 2 0.03 0.1 8.66** Boy 6 0.17* 0.09 31.72**
Diagnostic status, Q(1) ¼1.15 Mixed 45 0.03 0.03 866.13**
Yes 6 0.03 0.09 11.81*
No 58 0.07** 0.03 1027.07**
Note. **p<.01; *p<.05.
8Ahn and Fedewa
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
status, but they did not depend upon age or whether chil-
dren were clinically diagnosed. First, both girls and boys
showed significant mean rs between level of physical activ-
ity status and mental health. The magnitude of mean cor-
relations for boys and girls were almost identical. Second,
only children who were typically developing in their cog-
nitive abilities showed a significant but negative relation-
ship of physical activity to mental health. Third, the mean
correlation between physical activity and mental health
outcome for children classified as obese was found to be
statistically significant. Finally, the relationship between
physical activity and negative mental health outcomes
was negative and significant for children who were not
clinically diagnosed.
Other Study Characteristics
For both d- and r-ESs, statistical results from the moderator
analyses using other study characteristics such as setting of
the physical activity intervention and study location are
available online as Supplementary Data.
The current study examined the effect of children’s phys-
ical activity on their mental health outcomes. Through a
comprehensive synthesis of the literature, 73 studies yield-
ing 246 effect sizes revealed a number of critical findings.
Some of the findings varied depending on the methodolog-
ical design of the included studies. RCT, non-RCT, and
correlational results will be discussed, highlighting the sig-
nificance and practical implications of these findings.
As predicted, increased levels of physical activity had
significant effects in reducing depression, anxiety, psycho-
logical distress, and emotional disturbance in children.
Both RCT and non-RCT studies also showed that physical
activity increased children’s levels of self-esteem. An overall
effect size of 0.30 for RCT studies and 0.57 for
non-RCT studies is consistent with meta-analytic reviews
in adults, which have found effect sizes that range from
0.36 to 1.10 in both clinical and non-clinical populations
(see Stathopoulou et al., 2006). These results mirror stud-
ies done with adult populations, as physical activity has
shown significant benefits in lowering adult depression,
anxiety, and overall psychological distress (Dixon,
Mauzey, & Hall, 2003; Paluska & Schwenk, 2000). The
2006 meta-analysis using solely randomized controlled
studies with children found similar results with respect
to the small, but beneficial effects of physical activity on
depression and anxiety (Larun et al., 2006).
One of the main purposes of this analysis was to eval-
uate whether physical activity exerted a unique effect
for children who were classified as overweight or obese.
Given that prior research has identified increased psycho-
social distress among children classified as overweight or
obese (Farhat et al., 2010), it was expected that physical
activity may play an even more important role for this
group of children. Both RCT and non-RCT studies, how-
ever, demonstrated equal effects for children who were
obese/overweight and those who were of typical weight.
In other words, both groups of children showed statistical-
ly significant effects on improved mental health, regardless
of their weight/height ratio. This is a critical finding for
clinicians working with children from all physical fitness
backgrounds, as despite a child’s body mass index, chil-
dren appear to reap some clinical benefit from physical
Although correlational studies also found a significant
relationship between increased levels of physical activity
and decreased levels of depression (as well as an enhanced
self-concept), other mental health outcomes were not
found to relate significantly to heightened levels of physical
activity. Yet, one critique of correlational studies in this
area is that they leave open the question as to whether
the relationship between physical activity and mental
health is simply an effect of negative affect on the child’s
motivation to engage in physical activity. The experimental
studies have somewhat clarified this relationship and will
likely provide a more accurate indicator of how much phys-
ical activity influences mental health outcomes in children.
Given that depression, anxiety, psychological distress,
emotional disturbance, and self-esteem were all positively
affected through randomized-controlled designs in both
children and adults across multiple studies, these findings
can be interpreted as robust (Larun et al., 2006;
Stathopoulou, Powers, Berry, & Smits, 2006).
As demonstrated in this analysis, the type of physical
activity children received had varying effects on their
mental health. RCT studies demonstrated the greatest
effect with circuit training/strength training activities and
mixed activity interventions, meaning a combination of
aerobic and resistance training exercise. The more children
engaged in these types of activities, the less adverse mental
health issues they experienced. Results from aerobic exer-
cise and resistance training can be found in numerous
studies based on adult populations, as both types of activ-
ity have resulted in consistently beneficial effects for par-
ticipants’ mental health. Possible mechanisms include an
increase in serotonin or other neurotransmitters associated
with the ‘‘endorphin’’ effect in alleviating negative affect,
although additional clinical studies are needed to specify
Meta-analysis of Physical Activity 9
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
the precise neurological pathways that mediate the rela-
tionship between physical activity and mood on a physio-
logical level (Stathopoulou et al., 2006).
Although there is a dearth of research in this area using
child participants, the level of intensity of the intervention
played a significant moderating effect in both RCT and
non-RCT studies. Interventions designed with high levels
of intensity had the greatest effect on children’s mental
health in RCT studies, while moderate-level activities ex-
erted a greater effect for non-RCT studies. Again, there is a
sparse research base in which to make sense of these find-
ings, but in the only meta-analysis to date studying this
relationship, Larun and colleagues found no difference
between low and high intensity exercise when assessing
effects on depression and anxiety (2006). However, multi-
ple studies on adult populations suggest an overwhelming
consensus in favor of high intensity exercise (Stathopoulou
et al., 2006). Perhaps with the addition of RCT studies in
the present analysis, this relationship was better discerned.
As argued for adult populations, higher intensity exercise
may enhance neurological, physiological and cognitive
factors that mediate the relationship between activity and
mood. Although more research is needed to confirm this
hypothesis, perhaps similar mechanisms are at work in
children (Shephard, 1996; Stathopoulou et al., 2006;
Wiles, Jones, Haase, Lawlor, Macfarlane, & Lewis, 2008).
Curiously, interventions done approximately 1 to
2 days per week and for more than 33 hr (spanning the
length of the intervention) were most effective in RCT stud-
ies. For non-RCT studies, interventions done approximate-
ly 1 day per week for no more than 20 hr were the most
effective. That is, interventions that were done more than
3 days per week or for more than a total of 20 hr did not
alleviate children’s mental health ailments for non-RCT
studies. Although this finding from RCT studies appears
to contradict findings with the non-RCT studies, the seem-
ingly low amount of hours could be explained by the
relatively short duration of physical activity interventions
used in the included studies. In other words, the average
length of time for physical activity interventions was
8.95 weeks (SD ¼4.5), ranging from 2 to 20 weeks for
non-RCT studies. For RCT-studies, the length of the inter-
vention phase was longer (M¼11.1 weeks, SD ¼3.6).
Thus, with a mean total of 20 hr for the non-RCT interven-
tions, this would equate to approximately 2.2 hr per week
of additional physical activity for children—around
44 min, 3 times per week. For a mean total of 33 hr
for non-RCT studies, the average amount of physical
activity would come to approximately 2.9 hr per week,
or 58 min, 3 times per week. The differential mean
intervention length across study designs likely affects the
total of hours needed for significant results. A number
of possibilities might explain this finding of children’s
need for fewer hours of total physical activity. First, it
could be that since children already receive relatively
short and sporadic bouts of physical activity throughout
the week (i.e., recess or game play; see Dencker, Bugge,
Hermansen & Andersen, 2010; Rowland, 1996) they may
not require as much physical activity for an intervention to
be effective in reducing negative mood. Another possibility
concerns the dearth of studies that included these descrip-
tors (frequency and duration of intervention) in their meth-
odology. For example, there was only one effect size from
which to code daily (5 days/week) physical activity for chil-
dren. Without knowing this information, it was impossible
to code for these features of dosage, and thus it is likely
that an accurate assessment of the dose-response relation-
ship could not be obtained, just as in the prior 2006
meta-analysis (Larun et al., 2006).
When designing physical activity interventions for
children, this study showed that individualized- or class-
wide interventions had the greatest effect on children’s
mental health. When the intervention was led by teachers,
researchers, or PE specialists, children showed the signifi-
cant reduction in mental health problems through physical
activity in both RCT and non-RCT studies. This finding
has practical applications, as children spend the majority
of their waking hours in school. With many children
unable to access mental health treatment through outpa-
tient or clinical settings, schools are the one place where
services are both mandated and free for children with di-
agnosed mental health ailments (Hoagwood & Johnson,
2003). Thus, schools have the potential to be a vehicle
for improving children’s mental health outcomes.
Gender was found to be a moderator in both RCT and
non-RCT studies, although the findings were inconsistent.
In RCT samples, males and mixed-gender groups showed
the largest gains from physical activity with respect to their
mental health outcomes. However, non-RCT studies
showed that girls benefit more than boys do when it
comes to the effects of physical activity on their mental
health. Correlational studies demonstrated yet another
finding and showed no differential effect by gender. The
methodological design of each study likely plays a critical
role in whether gender and age are found to be moderating
factors in the relationship between physical activity and
mental health. In other words, the differential findings be-
tween RCT, non-RCT, and correlational studies may help
elucidate why the ‘‘gender difference’’ finding has been
markedly inconsistent in this body of research (Wiles
et al., 2008). Some authors have postulated that physical
activity exerts a greater effect on females’ mental health
10 Ahn and Fedewa
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
outcomes due to their ‘‘feeling better’’ (i.e., higher
self-esteem and self-concept) about their appearance and
overall lack of physical activity compared to same-age male
peers (Lagerburg, 2005). Other authors argue that girls and
boys in fact are not affected differently, and that gender
differences between activity levels and psychological well-
being can be ‘‘smoothed out’’ over time (Parfitt & Eston,
2005). The present analysis, however, does show that
when RCT studies are analyzed separately, males appear
to reap larger psychological benefits from physical activity
than do their female peers. Further research is needed
to investigate the mechanisms behind these variations, per-
haps using a mixed-methods design to assess both female
and male children’s perceptions of the benefits of physical
activity (Loman, 2008).
Additional participant characteristics were found to
moderate the relationship between physical activity and
mental health outcome. For children who were diagnosed
as cognitively impaired or emotionally disturbed, effect
sizes from RCT studies were significantly greater compared
to children who were typically developing and did not have
an emotional disorder. In fact, the RCT studies used for
analysis demonstrated an effect size that was five times as
large for children with cognitive impairments and almost
twice as large for students with emotional disturbance.
Perhaps the severity of problems for these two groups of
children enhanced the effect of physical activity on their
mental health, just as children who were clinically diag-
nosed with a disorder or disability displayed higher levels
of mental health benefit than those children in the general
population without a clinical diagnosis. These findings are
critical for clinicians and school-based practitioners, as it
demonstrates the increased effectiveness of physical activ-
ity for children who display severe problem behaviors or
clinical symptomology.
As with any study, there are limitations that must be
addressed. First, regardless of both searching and including
unpublished and published studies, slight publication bias
existed for the current analysis, which might threaten the
validity of research findings in the current meta-analysis.
However, it should be pointed out that such problem re-
flects the way individual studies in the field are conducted
and disseminated (Sutton, 2009). In spite of the potential
validity threat due to publication bias, the current review at
least can inform the possible presence of publication bias
in the area and estimate the likely effect of bias based on
the distribution of the effect sizes from the available stud-
ies. By taking into account this potential validity threat, the
overall effect size between physical activity and children’s
mental health remained significant. Second, the majority of
studies (69%) did not include children’s ethnicity or
socioeconomic status. Thus, these variables could not be
included as potential moderators for the relationship be-
tween physical activity and child mental health outcomes.
Yet researchers have demonstrated higher levels of obesity
in racially, ethnically, and socioeconomically disadvan-
taged populations (Burton & VanHeest, 2007). It is likely
that different relationships exist for these populations than
for middle-class, Anglo-American groups. Future research
in this area should include detailed descriptors of the in-
cluded sample so that these relationships can be examined.
In summary, the present quantitative synthesis of
the literature demonstrated a small to moderate effect of
physical activity on children’s mental health. Given that
findings varied by methodological design, additional RCT-
designed studies are warranted to replicate and confirm
the current findings. The evidence from this meta-analysis
adds to the current body of knowledge documenting the
positive mental health effects of exercise in children, parti-
cularly for those children who exhibit a higher severity
of symptomology. For practitioners, physical activity
can thus be considered an effective component to already
well-established treatments (e.g., cognitive-behavioral ther-
apy) in the field. Clinicians, school-based professionals,
and parents should encourage physical activity in children,
not only for the physical health benefits, but for the
positive mental health outcomes as well.
Supplementary Data
Supplementary data can be found at: http://www.jpepsy.
Conflicts of interest: None declared.
Becker, B. J. (2000). Multivariate meta-analysis.
In H. E. A. Tinsley, & S. Brown (Eds.), Handbook of
applied and multivariate statistics and mathematical
modeling (pp. 499–525). San Diego: Academic Press.
Borenstein, M. (2009). Effect sizes for continuous data.
In H. M. Cooper, L. V. Hedges, & J. C. Valentine
(Eds.), The handbook of research synthesis and
meta-analysis. 2nd edn. New York, NY: Russell Sage
Burgeson, C. R., Weschler, H., Brener, N. D.,
Young, J. C., & Spain, C. G. (2001). Physical
education and activity: Results from the School
Health Policies and Programs Study 2000.
Journal of School and Health, 71(7), 279–293.
Meta-analysis of Physical Activity 11
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
Burton, L. J., & VanHeest, J. L. (2007). The importance
of physical activity in closing the achievement gap.
Quest, 59, 212–218.
Centers for Disease Control and Prevention (2008).
Healthy People 2010 update. Retrieved from http://
Conn, V. S. (2010). Depressive symptom outcomes of
physical activity interventions: Meta-analysis findings.
Annals of behavioral Medicine, 29, 128–138.
Dale, D., Corbin, C. B., & Dale, K. S. (2000). Restricting
opportunities to be active during school time: Do
children compensate by increasing physical activity
levels after school? Research Quarterly for Exercise and
Sport, 71(3), 240–248.
Dencker, M., Bugge, A., Hermansen, B., & Andersen, L. B.
(2010). Objectively measured daily physical
activity related to aerobic fitness in young children.
Journal of Sports Sciences, 28(2), 139–145.
Dixon, W. A., Mauzey, E. D., & Hall, C. R. (2003).
Physical activity and exercise: Implications for
counselors. Journal of Counseling & Development, 82,
Ekeland, E., Heian, F., & Hagen, K. B. (2005). Can
exercise improve self-esteem in children and young
people? A systematic review of randomized controlled
trials. British Journal of Sports Medicine, 39, 792–798.
Farhat, T., Iannotti, R. J., & Simmons-Morton, B. G.
(2010). Overweight, obesity, youth, and health-risk
behaviors. American Journal of Preventive Medicine, 3,
Fedewa, A. L., & Ahn, S. (in press). Determining the
effects of physical activity on children’s cognitive
outcome: A meta-analysis. Research Quarterly for
Exercise and Sport.
Gleser, L. J., & Olkin, I. (2009). Stochastically dependent
effect sizes. In H. M. Cooper, L. V. Hedges, &
J. C. Valentine (Eds.), The handbook of research
synthesis and meta-analysis. 2nd edn. New York, NY:
Russell Sage Foundation.
Hardman, K. (2008). Physical education in schools: A
global perspective. Kinesiology, 40(1), 5–28.
Hedges, L. V., & Olkin, I. (1985). Statistical methods for
meta-analysis. Orlando, FL: Academic Press.
Hoagwood, K., & Johnson, J. (2003). School psychology:
A public health framework I. From evidence-based
practices to evidence-based policies. Journal of School
Psychology, 41, 3–21.
Kremers, S. P. J., De Bruijin, G. J., Droomers, M.,
Van Lenthe, F., & Brug, J. (2007). Moderators of
environmental intervention effects on diet and
activity in youth. American Journal of Preventive
Medicine, 32(2), 163–172.
Lagerberg, D. (2005). Physical activity and mental health
in schoolchildren: A complicated relationship. Acta
Pediatrica, 94, 1699–1705.
Larun, L., Nordheim, L. V., Ekeland, E., Hagen, K. B., &
Heian, F. (2006). Exercise in prevention and
treatment of anxiety and depression among children
and young people. Cochrane Database of Systematic
Reviews, 3, 1–47.
Lipsey, M. W., & Wilson, D. B. (2001). Practical
meta-analysis. Thousand Oaks, California: Sage
Publications, Inc.
Loman, D. G. (2008). Promoting physical activity
in teen girls: Insight from focus groups.
The American Journal of Maternal-Child Nursing,
33(5), 294–299.
Morris, S. B. (2008). Estimating effect sizes from
pretest-posttest-control group designs. Organizational
Research Methods, 11(2), 364–386.
National Center for Health Statistics (2009). Health,
United States, 2008. Hyattsville, MD: U.S.
Department of Health and Human Services.
Ortega, F. B., Ruiz, J. R., Castillo, M. J., & Sjostrom, M.
(2008). Physical fitness in childhood and
adolescence: A powerful marker of health.
Pediatric Review, 32, 1–11.
Paluska, S. A., & Schwenk, T. L. (2000). Physical activity
and mental health. Journal of Sports Medicine, 29(3),
Parfitt, G., & Eston, R. G. (2005). The relationship
between children’s habitual activity level and
psychological well-being. Acta Paediatrica, 94,
Pate, R. R., Freedson, P. S., Sallis, J. F., Taylor, W. C.,
Sirard, J. C., Trost, S. G., & Dowda, M. (2002).
Compliance with physical activity guidelines:
Prevalence in a population of children and youth.
Annals of Epidemiology, 12, 303–308.
Pica, R. (1997). Beyond physical movement:
Why young children need to move. Children, 52(6),
Raudenbush, S. W. (2009). Analyzing effect sizes:
Random-effects models. In B. S. Cooper, L.
V. Hedges, & J. C. Valentine (Eds.), The handbook of
research synthesis and meta-analysis. 2nd edn. New
York, NY: Russell Sage Foundation.
Rosenthal, R. (1994). Parametric measures of effect size.
In H. M. Cooper, & L. V. Hedges (Eds.), The
handbook of research synthesis. New York, NY:
Russell Sage Foundation.
12 Ahn and Fedewa
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
Rowland, T. W. (1996). Developmental exercise physiology.
Champaign, IL: Human Kinetics.
Simen-Kapeu, A., & Veugelers, P. J. (2010). Should
public health interventions aimed at reducing
childhood overweight and obesity be gender-focused?
BMC Public Health, 10, 1471–2458.
Stathopoulou, G., Powers, M. B., Berry, A. C.,
Smits, J. A., & Otto, M. W. (2006). Exercise interven-
tions for mental health: A quantitative and qualitative
review. Clinical Psychology and Practice, 13, 179–193.
Stevens, T. A., To, Y., Stevenson, S. J., &
Lochbaum, M. R. (2008). The importance of
physical activity and physical education in the
prediction of academic achievement. Journal of
Sport Behavior, 31, 368–388.
Stice, E., Shaw, H., & Marti, C.N. (2006). A
meta-analytic review of obesity prevention
programs for children and adolescents: The skinny
on interventions that work. Psychological Bulletin,
132(5), 667–691.
Sutton, A. J. (2009). Publication bias. In B. S. Cooper,
L. V. Hedges, & J. C. Valentine (Eds.), The Handbook
of Research Synthesis and Meta-analysis. New York,
NY: Russell Sage Foundation.
Thompson, R. (2005). The course and correlates
of mental health care received by young
children: Descriptive data from a longitudinal urban
high-risk sample. Children and Youth Services Review,
27, 39–50.
U.S. Public Health Service (2000). Report of the Surgeon
General’s Conference on Children’s Mental Health: A
national action agenda. Washington, DC: Department
of Health and Human Services.
Wang, F., & Veugelers, P. J. (2008). Self-esteem and
cognitive development in the era of the childhood
obesity epidemic. International Association for the
Study of Obesity, 9, 615–623.
Wiles, N. J., Jones, G. T., Haase, A. M., Lawlor, D. A.,
Macfarlane, G. J., & Lewis, G. (2008). Physical
activity and emotional problems amongst
adolescents. Society of Psychiatry and Epidemiology,
43, 765–772.
World Health Organization. (2009). Global strategy
on diet, physical activity and health. Retrieved
Meta-analysis of Physical Activity 13
by guest on January 12, 2011jpepsy.oxfordjournals.orgDownloaded from
... Regular physical activity participation is beneficial for enhancing muscular strength and bone health [3,4], improving cardiovascular fitness [5], and preventing multiple chronic diseases such as cancer [6], obesity [7,8], and type II diabetes [8]. In addition, physical activity appears to be effective for reducing depression/depressive symptoms and improving physical self-perceptions [9,10], while also providing opportunities for children to develop their social skills and improve cognitive performance (i.e., concentration) and academic achievement [11][12][13][14]. Research suggests sedentary behavior also has an independent association with increasing health problems [15,16]. ...
... Consistent with previous studies [32,45], no significant differences between boys and girls were found. However, other research has observed higher scores in the self-description domain of PLAYself among boys compared to girls when a wider age range [8][9][10][11][12][13][14] or youth were examined [70,71]. Previous studies suggest that maturation affects children's physical self-perceptions differentially by gender [72]. ...
Full-text available
Background Physical literacy (PL) is considered an important determinant of children's physical activity through which health benefits may be derived. The purpose of this study is to describe a sample of Canadian children’s baseline levels of PL and movement behaviors, and explore whether the associations between PL and their mental wellbeing, if any, are mediated by moderate-to-vigorous physical activity (MVPA). Methods All grade two children in 14 elementary schools in the West Vancouver School District, Canada were invited to participate in a two-year longitudinal project. PL was assessed through PLAYfun and PLAYself tools. Physical activity was measured by wrist-worn accelerometers (GT3X + BT) for seven days. Children's mental well-being was assessed using the Strengths and Difficulties Questionnaire (SDQ). A score of total difficulties was aggregated for internalizing and externalizing problems. Results A total of 355 children aged 7–9 (183 boys, 166 girls, 6 non-binary) participated with 258 children providing valid accelerometer data. Children exhibited an average of 111.1 min of MVPA per day, with 97.3% meeting the physical activity guidelines. Approximately 43% (108/250) of participants were meeting the Canadian 24-h movement guidelines. Children were at an ‘emerging’ level of overall physical competence (45.8 ± 5.6) and reported a mean score of 68.9 (SD = 12.3) for self-perceived PL, with no significant differences between boys and girls. PL was significantly associated with MVPA (r = .27) and all SDQ variables (rs = -.26—.13) except for externalizing problems. Mediation analyses showed PL was negatively associated with internalizing problems and total difficulties when the association with MVPA was considered. However, the mediating role of MVPA was found only between PL and internalizing problems, β = -.06, 95%CI [-.12, -.01]. Conclusions Although most of our sample was physically active and showed higher adherence to 24-H movement guidelines than comparable population data, the motor competence and self-perceived PL of our sample were similar to those of previous studies. PL has an independent association with children’s internalizing problems and total difficulties. Ongoing assessment will investigate the relationships between PL and children’s mental health from a longitudinal perspective.
... Research suggests that bodily changes that occur during puberty play a significant role in reducing the number of positive thoughts girls have about themselves during adolescence [27]. Low self-esteem is associated with poorer physical and mental health [24], and reduced physical activity engagement [28]. However, engaging in physical activity can also improve self-esteem [7], which is important because, recently, data suggests that self-esteem acts as a mechanism of resilience to adolescents exposed to stress [29]. ...
Full-text available
Adolescent girls report low physical activity levels and poor body image and self-esteem. This study evaluated the impact of a girls’ health intervention on body image, self-esteem, and physical activity enjoyment. The intervention was grounded in self-determination theory, resulting in lessons designed to promote autonomy, competence, and relatedness. The two primary components of the intervention included opportunities for girls to learn about resistance training as well as ways to improve their psychosocial health (i.e., body image, self-esteem, and physical activity enjoyment). Girls (n = 590), in the intervention (Mage = 12.79, SD = 0.69) and control group (Mage = 12.92, SD = 0.73), completed pre and post measures. A repeated measures MANOVA was conducted to assess changes in body image, self-esteem, and physical activity enjoyment. The intervention resulted in a significant increase in body esteem-appearance, F = 9.23, p = 0.003, and body esteem-weight, F = 4.77, p = 0.029, and a greater, non-significant, increase in self-esteem (3.22%), and physical activity enjoyment (4.01%) compared to the control group. This highlighted the use of the intervention for significant improvements in appearance and weight-related body image. The results support implementing psychosocial lessons, as well as physical activity, in health programming for girls.
... Meanwhile, low selfesteem in childhood is negatively associated with negative thoughts; avoidance of new experiences; difficulty in managing conflicts (Khullar & Tyagi, 2014); loneliness (Luo, Liu, & Zhang, 2020); suicide attempts (Barrera, Montoya-Castilla, Pérez-Albéniz, Lucas-Molina, & Fonseca-Pedrero, 2020); anxiety (Pyszczynski, Greenberg, Solomon, Arndt, & Schimel, 2004); and behavioral problems (Weels, Hunnikin, Ash, & Goozen, 2020). Results show that high self-esteem works as a buffer against mental disorders emerging from one's childhood (Ahn & Fedewa, 2011). Researchers suggest that high self-esteem protects against stressful situations and that it reduces the negative consequences of stress (McGee, Williams, & Nada-Raja, 2001). ...
Full-text available
This study aimed to investigate the effects of a self-esteem development program on middle-school students in the seventh grade. The study included 21 students with low self-esteem, of which 11 were in the experimental group and 10 in the control group. After identifying the participants, the researcher implemented an eight-session program, each session lasting 90 minutes on average, for the students in the experimental group. Non-parametric Mann Whitney-U test was used to determine whether there was a significant difference between the pre-test and post-test score averages of the experimental and control groups. Non-parametric Wilcoxon Signed Ranks test was used to determine whether there was a significant within-group difference between the pre-test and post-test scores of the experimental and control groups. The results indicate that the self-esteem scores of the students in the experimental group, who participated in group activities, increased. There was no change in the self-esteem scores of the students in the control group. The results of the study were discussed and interpreted in light of the relevant literature.
... Η ενασχόληση των ατόμων με ΦΔ κατά την παιδική ηλικία και την εφηβεία ωφελεί τη σωματική και ψυχική υγεία (Janssen & LeBlanc, 2010;Ahn & Fedewa, 2011;Andersen, Riddich, Kriemler & Hills, 2011). Επιπρόσθετα τα άτομα που ασχολούνται με ΦΔ εμφανίζουν μικρότερα ποσοστά καρδιαγγειακών προβλημάτων κατά την ενηλι- κίωση (Telama, 2009). ...
Full-text available
... Not surprisingly, other studies found that a high level of PA is related to positive outcomes, such as well-being, self-esteem, selfconcept, and resilience [16][17][18]. As early as 2011, a metaanalysis based on randomized controlled trials (RCTs) revealed that increased levels of PA are significantly associated with improved MH among children [19]. In 2019, another meta-analysis reported that the effects of PA on psychological ill-being (effect size = 0.130, p = 0.007) and psychological well-being (effect size = 0.189, p = 0.001) among children and youth are small but significant [16]. ...
Full-text available
Abstract Background The coronavirus disease (COVID‐19) and universal mitigation strategies have fundamentally affected peoples’ lives worldwide, particularly during the first two years of the pandemic. Reductions in physical activity (PA) and increased mental health (MH) problems among children and youth have been observed. This systematic review and meta-analysis investigated the relationship between physical activity (PA) and mental health (MH) among children and youth during the COVID‐19 pandemic. Methods Four electronic databases (EMBASE, PsycINFO, PubMed, and Web of Science) were systematically searched to identify studies that (1) examined the relationship between PA and MH among children and youth (aged 2–24 years old) and (2) were published in peer-reviewed journals in English between January 2020 and December 2021. Relationships between PA and two MH aspects (i.e., negative and positive psychological responses) among children and youth at different age ranges and those with disabilities or chronic conditions (DCC) were synthesized. Meta-analyses were also performed for eligible studies to determine the pooled effect size. Results A total of 58 studies were eventually included for variable categorization, with 32 eligible for meta-analyses. Our synthesis results showed that greater PA participation was strongly related to lower negative psychological responses (i.e., anxiety, depression, stress, insomnia, fatigue, and mental health problems) and higher positive psychological responses (i.e., general well-being and vigor) in children and youth during COVID-19. The pattern and strength of relations between PA and MH outcomes varied across age ranges and health conditions, with preschoolers and those with DCC receiving less attention in the existing research. Meta-analysis results showed that the magnitude of associations of PA with negative (Fisher’s z = − 0.198, p
Background: Hearing impairment (HI) is the most common global disabling condition. It is a considerable public health condition in childhood that is associated with long-term socio-emotional-academic, and communication difficulties. The current study explored the knowledge and awareness of HI among school-age children and its related factor in Taif, Saudi Arabia. Materials and methods: A cross-sectional study was done on 268 Saudi school-age children in the population of the Taif region of Saudi Arabia. A predesigned questionnaire was used to collect their demographic data, consanguinity, education level, and academic performance. Results: About 45.9% of parents had good awareness related to hearing loss and its impact on children's life. Only 19% (n = 51) of parents reported that their children encountered language problems in communicating with others. When we assessed the relationship between this language problem and academic performance, it was found that children with language problems had below-average academic performance (P < 0.001). Conclusion: The study showed that parents' awareness about children's hearing was not that satisfactory, and there was a considerable amount of difficulties faced by the children in their personal and social life due to these hearing problems. It is imperative to create awareness among the public regarding various modifiable risk factors of HI by conducting health awareness campaigns.
Full-text available
This is the protocol for a review and there is no abstract. The objectives are as follows: 1.To determine whether exercise interventions reduce and/or prevent anxiety and/or depression among children and young people compared to other treatments or no treatment. 2.If so, what are the characteristics of the most effective interventions?
Much of the literature on meta-analysis deals with analyzing effect sizes obtained from k independent studies in each of which a single treatment is compared with a control (or with a standard treatment). Because the studies are statistically independent, so are the effect sizes. Studies, however, are not always so simple. For example, some may compare multiple variants of a type of treatment against a common control. Thus, in a study of the beneficial effects of exercise on blood pressure, independent groups of subjects may each be assigned one of several types of exercise: running for twenty minutes daily, running for forty minutes daily, running every other day, brisk walking, and so on. Each of these exercise groups is to be compared with a common sedentary control group. In consequence, such a study will yield more than one exercise versus control effect size. Because the effect sizes share a common control group, the estimates of these effect sizes will be correlated. Studies of this kind are called multiple-treatment studies. In other studies, the single-treatment, single-control paradigm may be followed, but multiple measures will be used as endpoints for each subject. Thus, in comparing exercise and lack of exercise on subjects' health, measurements of systolic blood pressure, diastolic blood pressure, pulse rate, cholesterol concentration, and so on, may be taken for each subject. Similarly, studies of the use of carbon dioxide for storage of apples can include measures of flavor, appearance, firmness, and resistance to disease. A treatment versus control effect-size estimate may be calculated for each endpoint measure. Because measures on a common subject are likely to be correlated, corresponding estimated effect sizes for these measures will be correlated within studies. Studies of this type are called multiple-endpoint studies (for further discussions of multiple-endpoint studies, see Gleser and Olkin 1994; Raudenbush, Becker, and Kalaian 1988; Timm 1999). A special, but common, kind of multiple-endpoint study is that in which the measures (endpoints) used are sub-scales of a psychological test. For study-to-study comparisons, or to have a single effect size for treatment versus control, we may want to combine the effect sizes obtained from the subscales into an overall effect size. Because subscales have differing accuracies, it is well known that weighted averages of such effect sizes are required. Weighting by inverses of the variances of the estimated subscale effect sizes is appropriate when these effect sizes are independent, but may not produce the most precise estimates when the effect sizes are correlated. In each of these above situations, possible dependency among the estimated effect sizes needs to be accounted for in the analysis. To do so, additional information has to be obtained from the various studies. For example, in the multiple-endpoint studies, dependence among the end-point measures leads to dependence between the corresponding estimated effect sizes, and values for between-measures correlations will thus be needed for any analysis. Fortunately, as will be seen, in most cases this is all the extra information that will be needed. When the studies themselves fail to provide this information, the correlations can often be imputed from test manuals (when the measures are subscales of a test, for example) or from published literature on the measures used. When dealing with dependent estimated effect sizes, we need formulas for the covariances or correlations. Note that the dependency between estimated effect sizes in multiple-endpoint studies is intrinsic to such studies, arising from the relationships between the measures used, whereas the dependency between estimated effect sizes in multiple-treatment studies is an artifact of the design (the use of a common control). Consequently, formulas for the covariances between estimated effect sizes differ between the two types of studies, necessitating separate treatment of each type. On the other hand, the variances of the estimated effect sizes have the same form in both types of study - namely, that obtained from considering each effect size in isolation (see chapters 15 and 16, this volume). Recall that such variances depend on the true effect size, the sample sizes for treatment and control, and (possibly) the treatment-to-control variance ratio (when the variance of a given measurement is assumed to be affected by the treatment). As is often the case in analyses in other chapters in this volume, the results obtained are large sample (within studies) normality approximations based on use of the central limit theorem.
This volume considers the problem of quantitatively summarizing results from a stream of studies, each testing a common hypothesis. In the simplest case, each study yields a single estimate of the impact of some intervention. Such an estimate will deviate from the true effect size as a function of random error because each study uses a finite sample size. What is distinctive about this chapter is that the true effect size itself is regarded as a random variable taking on different values in different studies, based on the belief that differences between the studies generate differences in the true effect sizes. This approach is useful in quantifying the heterogeneity of effects across studies, incorporating such variation into confidence intervals, testing the adequacy of models that explain this variation, and producing accurate estimates of effect size in individual studies. After discussing the conceptual rationale for the random effects model, this chapter provides a general strategy for answering a series of questions that commonly arise in research synthesis: 1. Does a stream of research produce heterogeneous results? That is, do the true effect sizes vary? 2. If so, how large is this variation? 3. How can we make valid inferences about the average effect size when the true effect sizes vary? 4. Why do study effects vary? Specifically do observable differences between studies in their target populations, measurement approaches, definitions of the treatment, or historical contexts systematically predict the effect sizes? 5. How effective are such models in accounting for effect size variation? Specifically, how much variation in the true effect sizes does each model explain? 6. Given that the effect sizes do indeed vary, what is the best estimate of the effect in each study? I illustrate how to address these questions by re-analyzing data from a series of experiments on teacher expectancy effects on pupil's cognitive skill. My aim is to illustrate, in a comparatively simple setting, to a broad audience with a minimal background in applied statistics, the conceptual framework that guides analyses using random effects models and the practical steps typically needed to implement that framework. Although the conceptual framework guiding the analysis is straightforward, a number of technical issues must be addressed satisfactorily to ensure the validity the inferences. To review these issues and recent progress in solving them requires a somewhat more technical presentation. Appendix 16A considers alternative approaches to estimation theory, and appendix 16B considers alternative approaches to uncertainty estimation, that is, the estimation of standard errors, confidence intervals, and hypothesis tests. These appendices together provide re-analyses of the illustrative data under alternative approaches, knowledge of which is essential to those who give technical advice to analysts.
That the published scientific literature documents only a proportion of the results of all research carried out is a perpetual concern. Further, there is good direct and indirect evidence to suggest that the unpublished proportion may be systematically different from the published, since selectivity may exist in deciding what to publish. (Song et al. 2000; Dickersin 2005) For example, researchers may choose not to write up and submit studies with uninteresting findings (such as nonstatistically significant effect sizes), or such studies may not be accepted for publication. Even if a study is published, there may be selectivity in which aspects are written up. For example, significant outcomes (Chan et al. 2004) or subgroups may be given precedent over nonsignificant ones. There is also evidence to suggest that studies with significant outcomes are published more quickly than those with nonsignificant outcomes (Stern and Simes 1997). Additionally, there are concerns that research with positive and statistically significant results are published in more prestigious places and cited more times, making it more visible, and hence easier to find. Indeed, the publication process should be thought of as a continuum and not a dichotomy (Smith 1999). Inkeeping with the previous literature, these (whole) publication and related biases will be referred to simply as publication bias throughout the chapter, unless a more specific type of bias is being discussed, though dissemination bias is perhaps a more accurate name for the collection (Song al. 2000). In areas where any such selectivity exists, a synthesis based on only the published results will be biased. Publication bias is therefore a major threat to the validity of meta-analysis and other synthesis methodologies. However, it should be pointed out that this is not an argument against systematic synthesis of evidence, because publication bias affects the literature and hence any methodology used to draw conclusions from the literature (using one or more studies) will be similarly threatened. Thus, publication bias is fundamentally a problem with the way individual studies are conducted, disseminated, and interpreted. By conducting a meta-analysis, we can at least attempt to identify and estimate the likely effect of such bias by considering the information contained in the distribution of effect sizes from the available studies. This is the basis of the majority of statistical methods described later. There is agreement that prevention is the best solution to the problem of selectively reported research. Indeed, with advances in electronic publishing making the presentation of large amounts of information more economically viable than traditional paper-based publishing methods, there is some hope that the problem will diminish, if not disappear. No evidence yet exists to indicate this is the case, however, nor is this a solution to the suppression of information due to vested economic interests (Halpern and Berlin 2005). The use of prospective registries of prospectively studies for selecting studies to be included in systematic reviews has been suggested (Berlin and Ghersi 2005), because it provides an unbiased sampling frame guaranteeing the elimination of publication bias (relating to the suppression of whole studies at least). However, trial registration, per se, does not guarantee availability of data and an obligation to disclose results in a form that is accessible to the public is also required. Further, registries exist for randomized controlled trials in numerous medical areas. Such a solution will not be feasible, however, even in the long term, for some forms of research, including that relating to analysis of (routine) observational data, where the notion of a study that can be registered before analysis may be nonexistent. The notion of prospectively designing multiple studies with the intention of carrying out a meta-analysis in the future has also been put forward as a solution to the problem (Berlin and Ghersi 2005), but again, this may be difficult to orchestrate in some situations. Carrying out as comprehensive a search as possible when obtaining literature for a synthesis will help minimize the influence of publication bias. In particular, this may involve searching the grey literature for studies not formally published (see chapter 6 in this volume; Hopewell, Clarke, and Mallett 2005). Since the beginning of the Internet, the feasibility and accessibility of publication by means other than commercial publishing houses has greatly increased. This implies that the importance of searching the grey literature may increase in the future. Despite researchers' best efforts, at least in the current climate, alleviation of the problem of publication bias may not be possible in many areas of science. In such instances, graphical and statistical tools have been developed to address publication bias within a meta-analysis framework. The remainder of this chapter provides an overview of these methods, which aim to detect and adjust for such biases. It should be pointed out that if a research synthesis does not contain a quantitative synthesis (for example, if results are considered too heterogeneous, or the data being synthesized are not quantitative) then, though publication bias may still be a problem, methods to deal with it are still very limited.
In any meta-analysis, we start with summary data from each study and use it to compute an effect size for the study. An effect size is a number that reflects the magnitude of the relationship between two variables. For example, if a study reports the mean and standard deviation for the treated and control groups, we might compute the standardized mean difference between groups. Or, if a study reports events and nonevents in two groups we might compute an odds ratio. It is these effect sizes that are then compared and combined in the meta-analysis. Consider figure 12.1, the forest plot of a fictional metaanalysis to assess the impact of an intervention. In this plot, each study is represented by a square, bounded on either side by a confidence interval. The location of each square on the horizontal axis represents the effect size for that study. The confidence interval represents the precision with which the effect size has been estimated, and the size of each square is proportional to the weight that will be assigned to the study when computing the combined effect. This figure also serves as the outline for this chapter, in which I discuss what these items mean and how they are computed. This chapter addresses effect sizes for continuous outcomes such as means and correlations (for effect sizes for binary outcomes, see chapter 13, this volume).
The authors address current recommendations for physical activity and health, physical activity and mental well‐being, and implications for counselors and the counseling profession. Specifically, they review a recent article (P. M. Dubbert, 2002) published in the Journal of Consulting and Clinical Psychology and examine in detail the resulting implications for counselors and the counseling profession.