Content uploaded by Po Sen Chu
Author content
All content in this area was uploaded by Po Sen Chu on Jun 12, 2015
Content may be subject to copyright.
Journal of Social and Clinical Psychology, Vol. 29, No. 6, 2010, pp. 624-645
624
Address correspondence to Po Sen Chu or Donald Saucier at Kansas State University,
Department of Psychology, 492 Bluemont Hall, Manhattan, KS 66506-5302; E-mail:
chupo@k-state.edu or saucier@ksu.edu.
CHU ET AL.
SOCIAL SUPPORT AND WELL-BEING
META-ANALYSIS OF THE RELATIONSHIPS
BETWEEN SOCIAL SUPPORT AND WELL-BEING
IN CHILDREN AND ADOLESCENTS
PO SEN CHU, DONALD A. SAUCIER, AND ERIC HAFNER
Kansas State University
Research has started to explore the associations between social support and well-
being among children and adolescents, but the overall relationship is still un-
clear. This study explored: (1) the overall association between social support and
well-being, (2) the association differences among categories of well-being, (3)
the association differences among different types of social support measures, (4)
the association differences among different support sources, and (5) whether the
association between social support and well-being changed with participants’
age. Two hundred forty-six studies were collected and analyzed, and the results
indicated a positive but small association between social support and well-being.
Additionally, moderator analyses indicated that social support was more strongly
associated with self-concept, perceived support was more strongly associated
with well-being, support from teachers and school personnel was more strongly
associated with well-being, and the association between social support and well-
being increased with age. The implications and possible applications of the rela-
tionship between social support and well-being among children and adolescents
are discussed.
The associations between social support and well-being in adults
have been heavily researched, but the study of such relationships
among children did not receive much attention until the 1980s (Belle,
1989; Dubow & Ullman, 1989; Wolchik, Beals, & Sandler, 1989), de-
spite it having been suggested that children and adolescents need
social support more than adults do (Belle, 1989). Fortunately, the
SOCIAL SUPPORT AND WELL-BEING 625
quantity of research on children and adolescents’ social support has
since grown plentifully (Wolchik, Beals, & Sandler, 1989). However,
the overall relationships between social support and well-being in
children and adolescents remain unclear, and researchers have not
yet identified which aspects of social support are critical for children
and adolescents (Dubow & Ullman, 1989). Therefore, the main pur-
pose of this meta-analysis was to explore the associations between
social support and well-being in children and adolescents. We be-
lieved it would be beneficial to aggregate related studies in order
to determine the valence and magnitude of the relationships, pro-
viding an authoritative basis for describing the association between
social support and well-being among children and adolescents.
Social support is defined as the provision of both psychological
and material resources with the intention of helping the recipients
to cope with stress (Cohen, 2004). Cohen and Wills (1985) proposed
that social support is related to well-being because it offers posi-
tive emotions, a sense of self-worth, and predictability in life; it also
functions as a stress buffer by reinforcing self-esteem, self-efficacy,
and problem solving behaviors. They stated that though most of
the studies on social support are correlational and therefore do not
imply causal relationships, the evidence from studies using animal
research, social-psychological analogue experiments, and prospec-
tive surveys suggest that social support has positive effects on well-
being.
However, in studying the relationship between social support and
well-being, researchers and practitioners have faced a legitimate
problem: the diverse definitions, measurements, and outcome vari-
ables utilized across studies (Barrera, 1986; Smith, Fernengel, Hol-
croft, Gerald, & Marien, 1994). The definitions of social support for
children and adolescents by different researchers, though similar,
often have slightly different emphases. Researchers have noted that
social support is a multidimensional construct and therefore is ana-
lyzed and measured via many different approaches, such as one’s
social network size, how much one believes he or she has actually
received support from others, or how much one believes he or she
will receive support when help is needed (Barrera, 1986; Dubow &
Ullman, 1989). We believe it is important to synthesize related stud-
ies to detect the overall relationship of social support and children
and adolescents’ well-being, and it is equally important to learn
whether these different types of measures associate with well-being
differently. Furthermore, many studies also assess how much chil-
626 CHU ET AL.
dren and adolescents receive support from different sources such as
family, friends, and teachers. Which source provides support that is
in general more strongly associated with the well-being of children
and adolescents?
The purposes of this meta-analysis, therefore, were to estimate the
overall relationship between social support and well-being among
children and adolescents, and to examine moderators that would
affect this relationship. The three categories of moderators exam-
ined in this study were types of measures of social support, sources
of social support, and different outcomes of well-being.
THE OVERALL RELATIONSHIP BETWEEN
SOCIAL SUPPORT AND WELL-BEING, AND
DIFFERENT OUTCOMES OF WELL-BEING
Evidence of the benefits of social support for adults is bountiful (for
example, see Cohen & Wills, 1985; Schwarzer & Leppin, 1991; Uchi-
no, Cacioppo, & Kiecolt-Glaser, 1997). Similarly, studies generally
show a positive relationship between social support and well-being
in children and adolescents (for example, see Malecki & Demaray,
2006 for a brief review about social support as a buffer to stress for
students). But some studies have found trivial or even negative rela-
tionships. For instance, Berndt (1989) suggested that the friendships
among children and adolescents can have negative aspects such as
conflicts and/or competition that need to be examined. There is also
evidence suggesting the association between social support and
psychological adjustment among pregnant teenage mothers could
be trivial (Barrera, 1981). Borcherding, SmithBattle, and Schneider
(2005) further stated that many methods used to measure teenage
mothers’ social support only focus on its benefits, rather than the
potential difficulties and conflicts in their relationships, indicating
that supportive relationships may not always be positive. Thus,
though evidence suggests an overall positive relationship between
social support and well-being among children and adolescents,
many factors may influence its magnitude.
Further, because children and adolescents’ relationships with their
support providers are potentially quite different from adults’, the as-
sociation between social support and well-being among them may
also be different. Unlike adults, most children, especially younger
children, tend to have vertical relationships with other adults (e.g.,
parents, teachers); in other words, those adults can potentially ex-
SOCIAL SUPPORT AND WELL-BEING 627
ert control over them (Russell, Pettit, & Mize, 1998). Hartup (1989)
stated that vertical and horizontal relationships have somewhat
different impacts on children and adolescents, and thus the social
support they receive and perceive, and its effects, may be different
as well.
Because the overall magnitude of the relationship between social
support and well-being is not well-established for children and ad-
olescents, and because children tend to experience support differ-
ently than adults (e.g., vertical vs. horizontal), a synthesis of studies
to investigate these issues is beneficial. Furthermore, Barrera (1986)
showed that the outcome variables examined among children and
adolescents across studies were very diverse. This suggests that a
moderator analysis is necessary. In Amato and Keith’s (1991) meta-
analysis on parental divorce and the well-being of children, they
coded outcome measures into 8 categories (e.g., academic achieve-
ment, conduct problems of children and adolescents, psychologi-
cal adjustment) and performed a moderator analysis across these
categories. In our current meta-analysis, we adopted and adjusted
their coding schemes to examine and compare different categories
of outcomes across studies.
THE TYPES OF MEASURES
When studying the relationship between social support and well-
being, the diverse definitions and measures of social support have
posed a challenge (Barrera, 1986; Smith et al., 1994). Barrera (1986)
attempted to overcome the problem by categorizing the instru-
ments into three types of measures based on what they assess: first,
social network assesses children and adolescents’ social network size
and density; second, enacted support assesses the frequency of sup-
port that children and adolescents’ believe they have received; and
third, perceived support assesses children and adolescents’ percep-
tions of how much support is available if needed. Research from
adult literature recognized perceived support as a better measure
when evaluating the association between social support and well-
being. For example, Cohen and Wills (1985) discovered that social
support’s protective effect was only found in studies measuring
perceived social support, and that those studies that utilized instru-
ments for measuring social network and actual support received in
the past did not produce significant effects. Their review suggested
628 CHU ET AL.
that among the three types of measures, perceived support is the
type that is most related to well-being.
Indeed, based on past literature, it appears that the measures of
social network and enacted support did not relate to well-being as
strongly as perceived support did. For the measure of social net-
work, Rook (1984) indicated that supportive relationships might
also be the sources of conflicts, and therefore are not necessarily
beneficial. Dubow and Ullman (1989) also found that one’s measure
of network size was unrelated to enacted and perceived support.
They reasoned that one deep and meaningful relationship may be
sufficient to protect one from stress, and more relationships are not
necessarily more beneficial. Consequently, simply counting one’s
number of relationships seems not to be the best predictor of one’s
well-being.
Regarding enacted support, studies show that support that is ac-
tually received may not be as beneficial as perceived support. Tay-
lor et al. (2004) summarized some potential reasons: the provided
support may not be what is really needed, the help offered may be
perceived as intrusive, and help by others may reduce one’s self-es-
teem. In contrast, an individual may be comforted by knowing that
social support is available during times of stress. From this perspec-
tive, Taylor et al. (2004) suggested that social support does not need
to be activated to be beneficial; the perception of social support may
be more important than received support.
Thus, based on past research, perceived social support should be
best related to well-being, followed by enacted support and social
network size. The moderator analysis of this meta-analysis exam-
ined this prediction.
THE SOURCES OF SUPPORT
Cauce and Srebnik (1990) identified three main sources of support
for children and adolescents: family, friends, and school person-
nel. Though research recognizes the importance of both family and
friend supports for children and adolescents’ well-being (e.g., see
Bugental & Grusec, 2006 for a review), these relationships can also
be sources of conflicts (e.g., Barrera, Chassin, & Rogosch, 1993).
Rook (1984) stated that the interactions within individuals’ social
networks are often assumed by many researchers to be supportive,
but in fact may sometimes be negative, including such things as
SOCIAL SUPPORT AND WELL-BEING 629
arguments and invasion of privacy. Further, some studies found
negative associations between friends’ support and well-being. For
example, Borum (2000) found adolescents’ relationships with peers
sometimes are associated with negative outcomes of well-being
(e.g., drug abuse or delinquency); Kerr, Preuss, and King (2006) also
found a positive relationship between peer support and suicidal
ideation among male adolescents. These studies, along with many
others, discovered trivial or negative outcomes from peer support
among children and adolescents. Though at least one study sug-
gested that the conflicts may not necessarily neutralize the positive
outcomes (Barrera, Chassin, & Rogosch, 1993), it is still important to
investigate the magnitude of overall relationships of these sources
of support with well-being.
Regarding school personnel (e.g., teachers) support, though there
is relatively little research on this source compared to parent and
friend support (Benhorin & McMahon, 2008), studies suggest posi-
tive associations between teachers’ support and well-being out-
comes, including improved emotional adjustment, better academ-
ic-related achievement and behaviors, and fewer psychosomatic
symptoms (e.g., Bru, Murberg, & Stephens, 2001; Natvig, Albrek-
sten, Anderssen, & Qvarnstrom, 1999; Wentzel, 1998). Murdock and
Bolch (2005) also found a positive link between teacher support
and school adjustment (assessed by school belonging and academic
achievement) among gay, lesbian, and bisexual high school stu-
dents. The empirical evidence generally shows a positive relation-
ship between school and teacher support and well-being among
children and adolescents.
Based merely on the related studies on different sources of sup-
port, it is difficult to distinguish which source of support will be
associated with well-being most strongly. Indeed, until now, re-
search seems to be inconclusive on the relative importance of these
sources of support on children and adolescents’ well-being (e.g.,
Criss, Shaw, Moilanen, Hitchings, & Ingoldsby, 2009). Therefore,
this meta-analysis explored the relationships between these sources
and well-being in order to examine which source was associated
with well-being most strongly.
To summarize, we investigated the following research questions
in our meta-analysis:
1. What is the magnitude of the mean effect size between social
support and well-being among children and adolescents?
630 CHU ET AL.
2. Are there differences in the magnitude and direction of mean
effect sizes among the different categories of well-being (e.g.,
academic achievement, conduct problems of children and ad-
olescents, psychological adjustment)?
3. Are there differences in the magnitude and direction of mean
effect sizes between different types of social support measures
(e.g., size, enacted, and perceived) and well-being?
4. Are there differences in the magnitude and direction of mean
effect sizes between different support sources (e.g., family
members, peers) and well-being?
Finally, because younger children tend to have vertical relationships
with their support providers, we examined whether the relation-
ship between support and well-being changes with age. Therefore,
we also examined whether age served as a significant moderator of
the relationship:
5. Is age a significant moderator of the relationship between so-
cial support and well-being?
METHOD
LOCATING STUDIES
We located studies for potential inclusion using the PsycINFO
(1900-2008) and Eric databases (1966-2008). First, the key word “so-
cial support” was used to search for studies (articles, book chapters,
and dissertations) that specifically used participants of childhood,
preschool age, school age, or adolescence as samples. Studies had
to be written in English, be published in the United States,1 and be
quantitative studies. A total of nearly 1,400 articles, abstracts, or texts
were scanned or reviewed. Reference lists of studies we retrieved
were examined to identify other studies for potential inclusion.
1. We set this criterion in order to control the variable of culture. Evidence suggests
that the relationships between social support and well-being are influenced by culture.
Kim, Sherman, and Taylor (2008) stated that the study of social support should take
the social relationships patterns of different cultures into account, because how
people sense and utilize social support depends on how they see the self and their
relationships with others, and different cultures have different assumptions and
expectations on these aspects. In other words, people of different cultures have different
perceptions of social support, and inclusion of those people could cause potential
confounds. This was the same reason we excluded studies of immigrant samples.
SOCIAL SUPPORT AND WELL-BEING 631
INCLUSION CRITERIA
Correlational studies that met the following inclusion criteria were
included. First, participants must have been children and/or ado-
lescents, including preschoolers, kindergarteners, elementary stu-
dents, middle school students, and high school students. Second,
studies were conducted in the United States. Third, correlations,
means and standard deviations, p values, or other statistics suffi-
cient for the calculation of an effect size of the relationship between
social support and well-being were included. If a study did not
have information adequate for the calculation of effect sizes (e.g.,
only reported multivariate statistics), it was excluded. Fourth,
samples were not sojourners or immigrants. Finally, the measure
of social support was derived directly from self-reports of the sam-
ples instead of from parent-report or lab-observation. For example,
Wertlieb, Weigel, and Feldstein (1987) realized the limitation of ob-
taining children’s social support from their mothers and suggested
children’s self-report was a more valid index of their own social
support. This opinion is shared by other researchers (e.g., Cauce &
Srebnik, 1990; Compas, 1987; Dubow & Ullman, 1989).
SAMPLE OF STUDIES
Two hundred forty-six published (N = 199) and unpublished (N
= 47) studies met the inclusion criteria and were included in the
analyses.2 The sample sizes of individual studies ranged from 21
to 14,211, and samples’ ages ranged from 3 to 20 years.3 The total
number of participants across all studies was 121,432. The publica-
tion years ranged from 1980 to 2008, with a mean publication year
of 2000.94.
CODING PROCEDURES
Each study was coded for the following characteristics: name of the
first author, publication year, number of total participants, mean age
2. All the studies included in the meta-analysis are available from the authors.
3. Though our maximum age limit was high school students only, 8 studies (k =
26) reported students in high schools up to the age of 20. Because the number of high
school students over age 18 was very small, in order to preserve the information the
studies could provide we decided to retain those studies.
632 CHU ET AL.
or grade levels of participants, races of participants if reported, the
instrument used to measure social support, gender differences in
terms of effect sizes if reported, and the relevant statistics for calcu-
lation of effect sizes. If a study did not provide an r value to report
the effect size between social support and well-being, we extracted
the statistics reported that could be transformed into correlation
coefficients assessing the effect sizes of the relationships between
social support and well-being according to the formulas provided
by Rosenthal and Rosnow (1991).
CATEGORIES OF OUTCOMES OF WELL-BEINGS
When studying the associations between parental divorce and the
well-being of children, Amato and Keith (1991) coded the variables
related to well-being into the following eight categories:
(a) academic achievement (standardized achievement tests, grades,
teachers’ ratings, or intelligence); (b) conduct (misbehavior, aggres-
sion, or delinquency); (c) psychological adjustment (depression, anxi-
ety, or happiness); (d) self-concept (self-esteem, perceived compe-
tence, or internal locus of control); (e) social adjustment (popularity,
loneliness, or cooperativeness); (f) mother-child relations (affection,
help, or quality of interaction); (g) father-child relations; and (h) other.
(p. 28)
After scanning and reviewing the 246 articles, we found that the
outcome variables that were studied generally fit this categoriza-
tion, and therefore we adopted this categorization with some modi-
fications. The first 5 categories (a-e) were adopted, and 4 additional
categories were added: health (e.g., exercise frequency, eating hab-
its, Body Mass Index (BMI), healthy habits such as diet and avoiding
substance use); coping skills (e.g., with anger, burnout); career (e.g.,
career planning, career outcome expectation, self-efficacy in finding
a good career path after graduation); and other (included different
overall measurements, such as overall life satisfaction, overall adap-
tive/maladaptive behaviors, quality of life, self-concept combined
with academic achievement, or some variables that did not belong
to any category, such as capacity for reflection).
SOCIAL SUPPORT AND WELL-BEING 633
TYPES OF MEASURES
Social support measurements were coded into 5 categories: first,
size (those that measure the number of people in participants’ social
network, such as number of friends); second, enacted (those that ask
participants’ perception of how much help they actually received
for a certain period of time in the past, such as participants’ esti-
mation of the frequency of parents praising or listening to them in
the past month); third, perceived (those that measure participants’
perception of availability of help if they need it, such as how much
the participants believe they can rely on their parents when they
are in trouble); fourth, seeking (those that measure how much par-
ticipants seek social support when they are depressed or anxious);
and fifth, other (those measures that failed to differentiate differ-
ent aspects of social support such as The Network of Relationships
Inventory by Furman and Buhrmester (1985), or the measures that
access size and satisfaction of social support but the studies failed
to report them separately).
SOURCES OF SUPPORT
We also coded the sources of support into 4 categories: family/rela-
tive support, friend support, teacher and school support, and others
(e.g., counselors, other adults, and communities).
OTHER MODERATORS
Age. We coded participants’ mean age in order to examine the rela-
tionship between age and effect sizes using a continuous moderator
analysis. Some studies did not provide the mean age of the partici-
pants, but almost all provided participants’ grade levels in school,
therefore we used that information to estimate the mean age.
Publication Years. Finally, we were interested in examining wheth-
er the relationships remained stable over time. Therefore, the publi-
cation years of the studies were tested as moderators of the relation-
ship between social support and well-being.
A study may contribute more than one effect size because social
support was often used to correlate with more than one outcome.
634 CHU ET AL.
A total of 901 effect sizes were extracted from the 246 studies and
included in the analysis. Forty-four studies (17.89%) contributed 1
effect size, 72 studies (29.27%) contributed 2 effect sizes, 44 studies
(17.87%) contributed 3 effect sizes, 26 studies (10.57%) contributed
4 effect sizes, 5 studies (2.03%) contributed 5 effect sizes, 27 studies
(10.98%) contributed 6 effect sizes, 8 studies (3.25%) contributed 8
effect sizes, 12 studies (4.88%) contributed 9 effect sizes, 3 studies
(1.22%) contributed 12 effect sizes, 2 studies (0.81%) contributed 15
effect sizes, 1 study (0.41%) contributed 16 effect sizes, and 2 studies
(0.81%) contributed 18 effect sizes.
Most of the coding of the variables was straightforward, but the
classification of well-being variables and types of support into cat-
egories was somewhat more subjective. Therefore, we performed
reliability tests in which the first and third authors independently
coded 50 well-being variables and types of support. The Cohen’s
Kappa was .71 (p < .05) for well-being variables and .80 (p < .01) for
types of support. The disagreements were resolved by discussion.
RESULTS
DSTAT (Johnson, 1989), a computer software program for the meta-
analytic review of research literature, was utilized to analyze the
effect sizes. The overall effect size was calculated using two meth-
ods. First, when one study provided more than one effect size, the
study’s effect sizes were first synthesized and corrected based on
sample sizes, such that each study contributed only one effect size.
Second, all 901 effect sizes were analyzed for an overall effect size.
A histogram of the distribution of all 901 effect sizes is presented
in Figure 1. As Figure 1 illustrated, the distribution approximates a
normal distribution.
Based on the analysis of only one effect size per study, both the
overall unweighted and weighted effect sizes have the same value
when rounded to two decimal places, r = .18, p < .0001. Using all 901
effect sizes, both the overall unweighted and weighted effect sizes
have the same value when rounded to two decimal places, r = .17,
p < .0001. These results indicated that the relationship between per-
ceived social support and overall well-being was positive in direc-
tion but only small in magnitude. Because the results of one effect
size per study and all effect sizes were very close, we used all 901
effect sizes for all following analyses.4
SOCIAL SUPPORT AND WELL-BEING 635
Among the 246 analyzed studies, 47 of them were unpublished
(19%). Interestingly, the mean effect size of unpublished studies (k
= 203, r = .191, p < .00001) was significantly stronger than that of
published studies (k = 698, r = .167, p < .00001), QB (1) = 47.84, p <
.00001. We calculated the Fail-safe N using Orwin’s (1983) formula
(dc = .01) in order to address the concern of the “file-drawer prob-
lem:” how many unpublished studies that we did not retrieve that
would be needed to nullify the results using the effect size from the
246 effect sizes. The result showed that 3,936 unpublished studies
with nonsignificant results were needed to nullify the effect size to
a near-zero level, which suggests the results are robust.
A homogeneity analysis indicated significant heterogeneity among
the effect sizes of the studies, Qw (900) = 17,844.57, p < .00001. The
analysis revealed that 462 effect sizes (51.28%) would need to be
excluded in order to achieve homogeneity among the effect sizes,
Qw (438) = 501.42, p = .051. The results indicated that it would be
worthwhile to perform moderator analyses in order to potentially
account for some of the variance among the effect sizes.
4. We realized using all 901 effect sizes could potentially raise independence issues.
However, because the results of one effect size per study and all effect sizes were very
close, it suggested independence was not a serious threat. Therefore, we retained all
effect sizes to have more informative moderator analyses.
FIGURE 1. The distribution of the 901 effect sizes of the relationships
between social support and well-being for children and adolescents.
636 CHU ET AL.
EFFECT SIZES ACROSS CATEGORIES OF
OUTCOMES OF WELL-BEINGS
The overall weighted mean effect sizes of the 9 categories of well-
being were calculated and the results are shown in Table 1. The
mean effect sizes range from .098 to .265, which suggests somewhat
weak to medium magnitudes of relationships between social sup-
port and these well-being outcomes. Post hoc contrasts were per-
formed and the results are indicated in Table 1. The results of the
contrasts show that the mean effect sizes of self-concept and other
outcomes are significantly stronger than the rest of outcomes, and
academic achievement and coping skills are significantly weaker
than the rest.
EFFECT SIZES ACROSS TYPES OF MEASURES
The relationships between types of social support measures and
well-being are shown in Table 2. As the results indicated, the mean
effect sizes range from .01 to .201, which suggests the mean effect
sizes of social support measures vary from almost zero (social net-
work size) to moderate in magnitude (perceived support). Because
research suggests that perceived support is a better predictor to
adjustment relatively to other types of measures (e.g., Dubow &
Ullman, 1989; Taylor et al., 2004), a priori contrasts was conducted
among the support measures, and the results are shown in Table 2.
The results indicated that the measures of perceived support were
TABLE 1. Mean Weighted Effect Sizes by Categories of Outcomes of Well-Being
Categories k r 90% CI
Academic achievement 164 .105a.101 / .108
Conduct 141 .141b.137 / .145
Psychological adjustment 262 .199c.195 / .202
Self-concept 138 .265d.259 / .270
Social adjustment 68 .212c, e .204 / .220
Health 45 .238e, f .230 / .247
Coping skills 17 .098a, b .081 / .116
Career 14 .215c, e, f .199 / .231
Other 34 .252d, f .241 / .262
Note. Effect sizes with different subscripts differ signicantly.
SOCIAL SUPPORT AND WELL-BEING 637
significantly more strongly associated with well-being than the oth-
er measures of social support.
EFFECT SIZES ACROSS SOURCES OF SUPPORT
The mean weighted effect sizes for support sources were also cal-
culated (see Table 3). The mean effect sizes ranged from .098 (other)
to .209 (teacher and school personnel support), which suggests that
the relationships between different sources of support and well-
being vary from somewhat weak to moderate in magnitude. Post
hoc contrasts were also performed and the results are indicated in
Table 3. The results of the contrasts show that the mean effect size of
teacher and school personnel support is significantly stronger than
the rest of the effect sizes, and the support of “other” sources is sig-
nificantly weaker than the rest.
OTHER MODERATOR ANALYSES
Age. Average age was used as a continuous moderator. The mean
age was 13.75, and the moderator analysis revealed a positive sig-
nificant correlation between age and effect sizes, z = 3.93, p < .0001.
This indicated that the strength of the relationship between social
support and overall well-being increases with the age of partici-
pants across the studies.
Publication Year. The publication year was analyzed as a continu-
ous moderator. The result was significant, z = 2.55, p < .05, suggest-
ing that the effect sizes have increased significantly over the years.
Gender Difference. Finally, there were 18 studies measuring male
and female participants’ relationships between support and well-
TABLE 2. Mean Weighted Effect Sizes by Types of Social Support Measures
Categories k r 90% CI
Size 53 .01a.004 / .017
Enacted 147 .143b.139 / .147
Perceived 604 .201c.199 / .204
Seeking 21 .069d.056 / .082
Other 76 .191e.183 / .199
Note. Effect sizes with different subscripts differ signicantly.
638 CHU ET AL.
being separately, 23 studies used female participants only, and 1
study used males only. Though these were only 17% of the total
number of studies analyzed, we still conducted an analysis compar-
ing the gender difference on mean effect sizes among these stud-
ies. The results indicated a significant difference between genders:
female participants (n = 8,430, k = 76, r = .199, p < .00001) had a
significantly stronger mean effect size than did male participants (n
= 2,683, k = 22, r = .144, p < .00001), QB (1) = 17.11, p < .0001. This sug-
gests that female participants have a stronger relationship between
social support and well-being than do male participants.
DISCUSSION
This meta-analysis explored the relationships between social sup-
port and children’s and adolescents’ well-being. The results indicat-
ed a significantly positive but small association, and that was mod-
erated by different outcomes of well-being (e.g., academic achieve-
ment, conduct problems, psychological adjustment). Additional
moderator analyses revealed that measures of perceived social sup-
port are more strongly associated with well-being than other types
of measures. Different sources of support had different strengths
of relationships with well-being, with teacher and school person-
nel support being stronger than other sources of support. Age was
a significant moderator in the relationship between social support
and well-being, which indicated that the effect sizes increased as the
average age of the children and adolescents increased. Studies that
reported effect sizes across genders were analyzed and the results
suggested female children and adolescents have stronger relation-
ships between support and well-being than do male children and
adolescents. Finally, publication year was a significant moderator
between social support and well-being, suggesting that effect sizes
increased significantly over the years.
TABLE 3. Mean Weighted Effect Sizes by Sources of Support
Categories k r 90% CI
Family/Relative 297 .192a.189 / .195
Friend 213 .100b.096 / .103
Teacher/School 125 .209c.204 / .213
Other 42 .098b.087 / .109
Note. Effect sizes with different subscripts differ signicantly.
SOCIAL SUPPORT AND WELL-BEING 639
THE OVERALL RELATIONSHIP BETWEEN
SOCIAL SUPPORT AND WELL-BEING,
AND DIFFERENT OUTCOMES OF WELL-BEING
Though the overall association between social support and well-
being was significant, the magnitude was small. This result may
fall short of the expectations of some researchers; after all, social
support is considered one of the most effective ways for individu-
als to cope with stress (e.g., Cohen & Wills, 1985; Dubow & Ull-
man, 1989; Taylor et al., 2004). However, small relationships do not
necessarily indicate insignificant or unimportant effects (Lipsey &
Wilson, 1993). In fact, according to Rosenthal’s (1984) formula for
binomial effect sizes, a correlation of .17 between social support and
well-being indicates that 58.5% of participants who have social sup-
port experience well-being, which implies that nearly 60% of chil-
dren and adolescents benefit from social support. We believe that
the magnitude of this impact is substantial, and therefore deserves
further research efforts. Further, the moderator analyses revealed
weak to medium magnitudes of relationships between social sup-
port and different well-being outcomes. The mean effect sizes for
self-concept and other were the strongest, while the effect sizes for
academic achievement and coping skills were the weakest. Future
research should further explore the potential reasons for the differ-
ences in these effect sizes, and also the mediators of the relation-
ships between support and these outcomes of well-being among
children and adolescents.
TYPES OF MEASURES
The analysis also revealed that the measures of perceived social
support were more strongly associated with well-being than other
types of measures, confirming what most literature suggests (e.g.,
Cohen & Wills, 1985; Dubow & Ullman, 1989; Taylor et al., 2004;
Wethington & Kessler, 1986). On the other hand, the size of social
networks was nearly unrelated to well-being, supporting the view
that counting one’s number of supportive relationships is not a
good indicator of well-being (Cohen & Wills, 1985; Dubow & Ull-
man, 1989). Also, enacted support was less strongly related to well-
being than was perceived support, suggesting that the support one
actually receives is not as beneficial as one’s perception of support.
As Taylor et al. (2004) suggested, receiving actual support from oth-
640 CHU ET AL.
ers might reduce one’s self-esteem, received support might not be
what one really needs, or the support may be seen to be intrusive.
These factors may cause stress to individuals that receive support,
and may therefore explain why enacted support is not strongly re-
lated to well-being.
Furthermore, the association between seeking support and well-
being was also weak. Bolger, Zuckerman, and Kessler (2000) indi-
cated that seeking support from one’s social network may cause ad-
ditional stress because asking for help makes one see him or herself
as less capable, and feel embarrassed to trouble others. Thus, enact-
ed and seeking support may not be as strongly related to well-being
as perceived support because they may activate stress. However, an
alternative viewpoint was suggested by Barrera (1986), who found
that some studies reported negative relationships between mea-
sures of enacted support and psychological well-being (assessed
by psychopathological symptoms), and explained that the reason
could be a mobilization of support due to individuals’ increased
stress. In other words, seeking or receiving actual support does not
activate stress; it is when under stress that individuals activate these
types of support. We suggest that future research should further
investigate how much the relationship between enacted or seeking
support and well-being is explained by individuals activating these
types of support due to increased stress.
Among the five types of social support measures, the other cat-
egory had a relatively strong relationship with well-being when
compared to other types of measures. We further examined the in-
struments in this category and found that some measures were mul-
tidimensional scales and the authors did not report effect sizes inde-
pendently (e.g., failed to report measures of support network size
and perceived support satisfaction separately). Some other scales
were developed by the authors themselves and some contained
only one item. Therefore, the validity of these measures is question-
able. Speculatively, this could potentially contribute to spuriously
high correlations between these measures and well-being, but fur-
ther research investigating this issue is needed.
SOURCES OF SUPPORT
In terms of sources of support, support from teachers and school
personnel had the strongest association with children and adoles-
SOCIAL SUPPORT AND WELL-BEING 641
cents’ well-being among the four sources, followed by the support
from family members. Friend support and “other” support had the
weakest relationships with well-being. The findings that the mean
effect sizes of both family and friend support were weaker than
support from teachers and school personnel deserve further explo-
ration. Many studies suggest the importance of parental support
in children and adolescents’ development (e.g., Ainsworth, Blehar,
Waters, & Wall, 1978; Steinberg, 1990), but some studies found that
relationships with family and friends can also be sources of conflicts
(Barrera, Chassin, & Rogosch, 1993; Rook, 1984). We speculated that
these conflicts might lessen the effects of family and friend support
on well-being. Further, research has failed to definitively determine
the relative importance of family support and friend support (Criss
et al., 2009). The results of this moderator analysis should offer evi-
dence for this issue: friend support was significantly less strongly
related to well-being than was family support. We speculate that
one of the reasons may be that friends are the only source of support
that children and adolescents can choose by themselves. Therefore,
children of similar characteristics, whether or not those are well-
adjusted characteristics, tend to associate with each other (Erdley,
Nangle, Newrnan, & Carpenter, 2001), and thus are more likely to
encourage each other on the qualities they share. For instance, ag-
gressive children will be more likely to choose aggressive friends
(Earls, Cairns, & Mercy, 1993). As a consequence the aggressive be-
havior may be magnified, and well-being diminished because it is
not a well-adjusted behavior. Future research should study whether
this magnifying effect of characteristics associated with maladjust-
ment does account for the weaker association between friend sup-
port and well-being, or whether there are other important factors.
OTHER MODERATORS
Age was significantly associated with the effect sizes between sup-
port and well-being. The older the participants were, the stronger
the effect sizes became. This seems to indicate that teenagers po-
tentially benefited from social support more than did children. One
potential implication is that vertical relationships, which are typical
among younger children and the adults they relate to, may not be as
beneficial for well-being as horizontal relationships. More research
is needed to examine why age is a moderator for the relationship
between support and well-being.
642 CHU ET AL.
The effect sizes have increased significantly over the years. Specu-
latively, because the number of studies of children and adolescents’
social support has been growing over the years (Wolchik et al.,
1989), the increasing of the effect sizes may imply that better mea-
sures were developed later that evaluated the relationship between
support and well-being more effectively.
Regarding gender differences, though some studies of children
and adolescents showed no gender differences in the correlation
between support and well-being (e.g., Santos, Richards, & Bleckley,
2007), and some found the relationship to be significant only for
male participants (e.g., Cumsille & Epstein, 1994), our analysis of 42
studies (k = 98) suggested that the mean effect size for female par-
ticipants was stronger than was the mean effect size for male par-
ticipants. This may indicate that teenage girls rely more on interde-
pendent relationships for well-being (Caselman, Self, & Self, 2006).
This seems to be consistent with the adult literature. For example,
Belle (1987) indicated that women use and receive more support
than do men. Miller and Stiver (1997) stated that women’s health
development is associated with their interpersonal relationships.
Kendler, Myers, and Prescott’s (2005) longitudinal study involving
more than 1,000 opposite-sex dizygotic twin pairs also found that
supportive relationships are more protective against depression for
women. Thus, it appears that women may have learned to rely on
their relationships for well-being at a young age. However, given
the diverse outcome variables of well-being studied in social sup-
port literature, the same pattern of gender difference may not be
consistent across all outcomes of well-being. We believe more re-
search is needed on these gender differences.
CONCLUSION
We believe the results found in this meta-analysis have practical
implications. First, to our knowledge, this meta-analysis is the first
attempt to synthesize the literature assessing the relationship be-
tween social support and well-being among these age groups. Con-
sequently, the findings should offer new knowledge and direction
for future research on the well-being of children and adolescents.
Furthermore, Cohen and Wills (1985) suggested that social support
is a significant contributor to the well-being of individuals. There-
fore, the finding that social support is positively linked to the well-
SOCIAL SUPPORT AND WELL-BEING 643
being of children and adolescents may indicate possibilities for in-
tervention. We also have suggested potential research directions in
regard to moderators such as measures of social support, sources of
social support, different outcomes of well-being, and age and gen-
ders of children and adolescents. In conclusion, this meta-analysis
has provided a further understanding of social support and well-
being among children and adolescents, and may serve as a foun-
dation for future research on the benefits of social support for the
well-being of children and adolescents.
REFERENCES
Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attach-
ment: A psychological study of the strange situation. Hillsdale, NJ: Erlbaum.
Amato, P. R., & Keith, B. (1991). Parental divorce and the well-being of children: A
meta-analysis. Psychological Bulletin, 110, 26-46.
Barrera, M., Jr. (1981). Social support in the adjustment of pregnant adolescents: As-
sessment issues. In B. H. Gottlieb (Ed.), Social networks and social support (pp.
69-96). Beverly Hills: Sage Publications.
Barrera, M., Jr. (1986). Distinctions between social support concepts, measures, and
models. American Journal of Community Psychology, 14, 413-445.
Barrera, M., Jr., Chassin, L., & Rogosch, F. (1993). Effects of social support and con-
flict on adolescent children of alcoholic and nonalcoholic fathers. Journal of
Personality and Social Psychology, 64, 602-612.
Belle, D. (1987). Gender differences in the social moderators of stress. In R. C. Bar-
nett, L. Biener, & G. K. Baruch (Eds.), Gender and stress (pp. 257-277). New
York: Free Press.
Belle, D. (Ed.). (1989). Children’s social networks and social support. New York: Wiley.
Benhorin, S., & McMahon, S. D. (2008). Exposure to violence and aggression: Protec-
tive roles of social support among urban African American youth. Journal of
Community Psychology, 36, 723-743.
Berndt, T. J. (1989). Obtaining support from friends during childhood and ado-
lescence. In D. Belle (Ed.), Children’s social networks and social support (pp.
308-331). New York: Wiley.
Bolger, N., Zuckerman, A., & Kessler, R. C. (2000). Invisible support and adjustment
to stress. Journal of Personality and Social Psychology, 79, 953-961.
Borcherding, K., SmithBattle, L., & Schneider, J. K. (2005). A preliminary investiga-
tion of the grandparent support scale for teenage mothers. Journal of Family
Nursing, 11, 289-306.
Borum, R. (2000). Assessing violence risk among youth. Journal of Clinical Psychol-
ogy, 56, 1263-1288.
Bru, E., Murberg, T. A., & Stephens, P. (2001). Social support, negative life events
and pupil misbehaviour among youth Norweigian adolescents. Journal of
Adolescence, 24, 715-727.
Bugental, D. B., & Grusec, J. E. (2006). Socialization processes. In N. Eisenberg (Ed.),
W. Damon (Series Ed.), Handbook of children psychology: Vol. 3. Social, emotional,
and personality development, 6th ed. (pp. 366-428). New York: Wiley.
644 CHU ET AL.
Caselman, T. D., Self, P. A., & Self, A. L. (2006). Adolescent attributes contributing to
the imposter phenomenon. Journal of Adolescence, 29, 395-405.
Cauce, A. M., & Srebnik, D. S. (1990). Returning to social support systems: A mor-
phological analysis of social networks. American Journal of Community Psy-
chology, 18, 609-616.
Cohen, S. (2004). Social relationships and health. American Psychologist, 59, 676-684.
Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis.
Psychological Bulletin, 98, 310-357.
Compas, B. E. (1987). Coping with stress during childhood and adolescence. Psycho-
logical Bulletin, 101, 393-403.
Criss, M. M., Shaw, D. S., Moilanen, K. L., Hitchings, J. E., & Ingoldsby, E. M. (2009).
Family, neighborhood, and peer characteristics as predictors of child adjust-
ment: A longitudinal analysis of additive and mediation models. Social Devel-
opment, 18, 511-535.
Cumsille, P., & Epstein, N. (1994). Family cohesion, family adaptability, social sup-
port, and adolescent depressive symptoms in outpatient clinic families. Jour-
nal of Family Psychology, 8, 202-222.
Dubow, E. F., & Ullman, D. G. (1989). Assessing social support in elementary school
children: The survey of children's social support. Journal of Clinical Child Psy-
chology, 18, 52-64.
Earls, F., Cairns, R.B., & Mercy, J. (1993). The control of violence and the promotion
of nonviolence in adolescence. In S.G. Millstein, A.C. Petersen, & E.O. Night-
ingale (Eds.), Promoting the health of adolescents: New directions for the twenty-
first century (pp. 285-304). New York: Oxford University Press.
Erdley, C.A., Nangle, D.W., Newrnan, J.E., & Carpenter, E.M. (2001). Children’s
friendship experiences and psychological adjustment: Theory and research.
In D.W. Nangle & C.A. Erdley (Eds.), The role of friendship in psychological ad-
justment. San Francisco: Jossey-Bass.
Furman, W., & Buhrmester, D. (1985). Children’s perceptions of the personal rela-
tionships in their social networks. Developmental Psychology, 20, 277-290.
Hartup, W. W. (1989). Social relationships and their developmental significance.
American Psychologist, 44, 120-126.
Johnson, B. (1989). DSTAT: Software for the meta-analytic review of research literature.
Hillsdale, NJ: Erlbaum.
Kendler, K. S., Myers, J., & Prescott, C. A. (2005). Sex differences in the relationship
between social support and risk for major depression: A longitudinal study of
opposite-sex twin pairs. American Journal of Psychiatry, 162, 250-256.
Kerr, D. C. R., Preuss, L. J., & King, C. A. (2006). Suicidal adolescents’ social support
from family and peers: Gender-specific associations with psychopathology.
Journal of Abnormal Child Psychology, 34, 103-114.
Kim, H. S., Sherman, D. K., & Taylor, S. E. (2008). Culture and social support. Ameri-
can Psychologist, 63, 518-526.
Lipsey, M. W., & Wilson, D. B. (1993). The efficacy of psychological, educational, and
behavioral treatment: Confirmation from meta-analysis. American Psycholo-
gist, 48, 1181-1209.
Malecki, C. K., & Demaray, M. K. (2006). Social support as a buffer in the relation-
ship between socioeconomic status and academic performance. School Psy-
chology Quarterly, 21, 375-395.
Miller, J. B., & Stiver, I. P. (1997). The healing connection. Boston: Beacon Press.
SOCIAL SUPPORT AND WELL-BEING 645
Murdock, T. B., & Bolch, M. B. (2005). Risk and protective factors for poor school
adjustment in lesbian, gay, and bisexual (LGB) high school youth: Variable
and person-centered analyses. Psychology in the Schools, 42, 159-172.
Natvig, G. K., Albreksten, G., Anderssen, N., & Qvarnstrom, U. (1999). School-relat-
ed stress and psychosomatic symptoms among school adolescents. Journal of
School Health, 69, 362-368.
Orwin, R. G. (1983). A fail-safe n for effect size in meta-analysis. Journal of Educa-
tional Statistics, 8, 157-159.
Rook, K.S. (1984). The negative side of social interaction: Impact on psychological
well-being. Journal of Personality and Social Psychology, 46, 1097-1108.
Rosenthal, R. (1984). Meta-analytic procedures for social research. Beverly Hills: Sage.
Rosenthal, R., & Rosnow, R. L. (1991). Essentials of behavioral research: Methods and
data analysis (2nd ed.). McGraw Hill: Boston.
Russell, A., Pettit, G. S., & Mize, J. (1998). Horizontal qualities in parent-child rela-
tionships: Parallels with and possible consequences for children’s peer rela-
tionship. Developmental Review, 18, 313-352.
Santos, M., Richards, C. S., & Bleckley, M. K. (2007). Comorbidity between depres-
sion and disordered eating in adolescents. Eating Behaviors, 4, 440-449.
Schwarzer, R., & Leppin, A. (1991). Social support and health: A theoretical and em-
pirical overview. Journal of Social and Personal Relationships, 8, 99-127.
Smith, C., Fernengel, K., Holcroft, C., Gerald, K., & Marien, L. (1994). Meta-analysis
of the associations between social support and health outcomes. Annals of
Behavioral Medicine, 16, 352–362.
Steinberg, L. (1990). Autonomy, conflict, and harmony in the family relationship. In
S. S. Feldman & G. R. Elliod (Eds.), At the threshold: The developing adolescent
(pp. 255-276). Cambridge, MA: Harvard University Press.
Taylor, S. E., Sherman, D. K., Kim, H. S., Jarcho, J., Takagi, K., & Dunagan, M. S.
(2004). Culture and social support: Who seeks it and why? Journal of Personal-
ity and Social Psychology, 87, 354-362.
Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K. (1997). The relationship be-
tween social support and physiological processes: A review with emphasis
on underlying mechanisms and implications for health. Psychological Bulletin,
119, 488-531.
Wentzel, K. R. (1998). Social relationships and motivation in middle school: The role
of parents, teachers and peers. Journal of Educational Psychology, 90, 202-209.
Wertlieb, D., Weigel, C., & Feldstein, M. (1987). Stress, social support, and behav-
ior symptoms in middle childhood. Journal of Clinical Child Psychology, 16,
204-211.
Wethington, E., & Kessler, R. C. (1986). Perceived support, received support, and
adjustment to stressful life events. Journal of Health and Social Behavior, 27,
78-89.
Wolchik, S. A., Beals, J., & Sandler I. (1989). Mapping children’s support networks:
Conceptual and methodological issues. In D. Belle (Ed.), Children’s social net-
works and social support (pp. 191-220). New York: Wiley.
