Content uploaded by Oliver Stoll
Author content
All content in this area was uploaded by Oliver Stoll
Content may be subject to copyright.
Self-Concept and Well-Being 1
Running head: SELF-CONCEPT AND WELL-BEING
REVISION submitted: March 4, 1999
Effects of Physical Exercise on Self-Concept and Well-Being
Dorothee Alfermann and Oliver Stoll
University of Leipzig, Leipzig, Germany
Self-Concept and Well-Being 2
Abstract
In 2 field experiments with middle-aged adults the effects of
exercise on self-concept and well-being were investigated. In
both experiments participants were randomly assigned to either
experimental or control groups. In Experiment 1 total = 24
female and male participated in an exercise program for six
months. Physical self-concept, self-esteem, and subjective well-
being were assessed before and after the 6 months program.
Compared with a wait-list control group (n=13), exercisers
improved significantly in physical self-concept and decreased in
psychosomatic complaints. In Experiment 2, 57 female and male
participated in exercise programs for six months. Placebo
attention groups (n = 36) were the control groups. The placebo
attention group members took part either in relaxation- or back-
training. Self-concept and well-being measures were assessed
three times: before and after the 6 months program (running or
mixed-sports program), and 6 months after the program’s
completion. The main effects of time showed that not only
exercise but also other kinds of intervention were able to
influence the dependent variables. Motor performance tests
likewise indicated that participants of all groups improved over
time. The results point to the fact that exercise is one, but
not the only strategy to improve mental health.
Self-Concept and Well-Being 3
Self-Concept and Well-Being 4
Effects of Physical Exercise on Self-Concept and Well-Being
The impact of exercise on mental health is a long standing
issue which became more clearly identified in the eighties. The
review of Folkins and Sime (1981) was one of the first concerned
with the relationship between physical exercise and mental
health. Though the authors were rather critical as to the
methodological rigor and the experimental design of the
literature they were nevertheless confident that physical
fitness training was beneficial to mental health. Since then a
growing body of research aimed to demonstrate that exercise
improves mental health (ISSP, 1992; McDonald & Hodgdon, 1991).
Physical activity, especially aerobic exercise, is thought to be
negatively associated to trait anxiety (Kerr & Vlaswinkel, 1990;
Landers & Petruzzello, 1994; Petruzzello, Landers, Hatfield,
Kubitz, & Salazar, 1991), depression (Morgan, 1994), and other
indices of mental health (McAuley, 1994; McAuley & Rudolph,
1995), including self-concept (Sonstroem, 1997a; b). In a meta-
analysis, Gruber (1986) demonstrated that physically active
children showed higher self-esteem than inactive children. In
another meta-analysis McDonald and Hodgdon (1991) likewise found
significant differences in self-concept measures of exercisers
and nonexercisers regardless of age and gender. However, the
generalizability of the results is limited due to the fact that
the majority of these studies has been done with „young adult
Self-Concept and Well-Being 5
groups, generally close to age 20“ (McDonald & Hodgdon, 1991, p.
142). Middle-aged and older adults were only rarely considered.
The two studies presented herein compared exercise groups
with no exercise control groups using a pretest-posttest design.
The purpose of the studies was to investigate long-term effects
of exercise on self-concept and on psychological well-being. In
addition, with regard to self-concept, the studies used as a
framework the Exercise and Self-Esteem Model of Sonstroem and
Morgan (1989) which was later revised into an expanded EXSEM
(Sonstroem, Harlow, & Josephs, 1994; s. also Sonstroem, 1997b;
1998). Thus, an additional objective of the second study was to
examine the hypothesized relationships within the EXSEM. It
assumes that exercise first influences physical self-concept
such that people develop a higher degree of physical competence
and physical acceptance. This subsequently should lead to
heightened feelings of global self-esteem (SE). Thus this model
„represents another example of the skill development hypothesis“
of self-esteem (Sonstroem, 1997b, p. 7).
But at the same time SE could also have an influence on
physical activity. The pathway from SE to activity and
performance is represented by the fact that SE influences (our
perception of) performance, and shapes our behaviors within the
environment. This pathway corresponds to the self-enhancement
hypothesis (Sonstroem, 1998, p. 134). Both causal pathways (from
Self-Concept and Well-Being 6
SE to environment and vice versa) represent a bidirectional
influence of SE and behavior. SE is not only influenced by
performance and success/failure, but is a causal agent itself.
Thus, physical activity can not only improve perceived physical
competence and this in turn self-esteem (the skill development
hypothesis), SE can also influence physical activity (Sonstroem,
1998).
Both versions of the Exercise and Self-Esteem Model (cf.
Sonstroem, 1997b) were tested in several studies since their
publication. Sonstroem, Harlow, Gemma, and Osborne (1991) could
demonstrate the validity of the older model with data from a
sample of adults in their middle and later years. As suggested
self-efficacy was directly related to physical competence (PC),
but not to self-esteem (SE), whereas PC showed a significant
path to SE. Baldwin and Courneya (1997) successfully
demonstrated the usefulness of the model in a study with female
breast cancer patients. Physical competence proved to be a
mediator between strenuous exercise and self-esteem. Sonstroem,
Harlow, and Salisbury (1993) tested the model in a longitudinal
study with male swimmers. Participants were administered a test
battery with a self-esteem scale, a physical competence scale,
and a study-developed swimming skills scale. Significant
increases were found in SE at Time 2, in PC at Time 3, and in
perception of skills at Times 2 and 3. Unfortunately the
Self-Concept and Well-Being 7
increase in psychological scores could not be explained by a
correspondent increase in physical performance. This might be
attributable to the fact that the swimmers in this study were
highly experienced and participated in competitions since years.
The amount of change in physical performance was not large
enough, and perhaps the study duration was not sufficient to
influence trait variables.
The expanded EXSEM represents a hierarchical, four level
model with self-efficacy/physical activity at the lowest
(behavioral) level, physical competence as a multidimensional
construct at the subdomain level, physical self-worth at the
next highest (domain) level, and self-esteem at the highest
level (s. also Fox & Corbin, 1989). Sonstroem, Harlow, and
Josephs (1994) tested the structural relationships between the
different components of the expanded EXSEM using the Physical
Self-Perception Profile (PSPP) of Fox and Corbin (1989).
Participants were middle-aged female aerobic dancers. Self-
efficacy had significant paths to perceived sport competence,
perceived physical condition, and perceived strength, but not to
perceived bodily attractiveness. But the latter variable was
associated rather strongly with overall physical self-worth
(PSW), as was PSW with SE. The authors used the EQS program of
Bentler and tested several models, with the EXSEM model having
the best fit with data.
Self-Concept and Well-Being 8
Caruso and Gill (1992) conducted two intervention studies
with male and female students. The correlation patterns in both
studies between the various self-concept measures offered some
support for the hierarchical EXSEM. Sonstroem and Potts (1996)
analyzed the relationship between physical self-concept and life
adjustment indices in a sample of male and female university
students. Using regression analyses the authors could show that
self-perceptions of PC significantly explained associations with
life adjustment. These relationships were robust across gender.
Besides changes of self-concept the experiments reported
herein also addressed the effects of exercise on subjective
well-being. Self-concept and well-being are both central aspects
of mental health. Though the psychological benefits of exercise
seem to be well established (International Society of Sport
Psychology, 1992; Martinsen & Stephens, 1994; McAuley & Rudolph,
1995; McDonald & Hodgdon, 1991) questions remain.
Thus, past research has demonstrated higher effects of
exercise for clinical populations than for nonclinical, ‘normal’
individuals (Raglin, 1990), and larger effects for state
variables than for psychological traits, like trait anxiety,
where „results are less consistent than those involving
psychological states“ (Leith, 1994, p. 189).
An additional problem of past research is the experimental
design (Morgan, 1997, p. 13). The majority of investigations are
Self-Concept and Well-Being 9
cross-sectional or correlational and thus nonexperimental in
nature (Leith, 1994, p. 193). As the mean effect size (mES)of
correlational studies in meta-analyses (mES = .41 + .16) doubles
that of experimental investigations (Dishman, 1994, p. 12),
nonexperimental research may somehow overestimate the mental
health effects of sport and exercise, not to mention the problem
of causal inference. In addition, even studies using an
experimental paradigm focus on brief interventions and short-
term effects. Thus experiments using longer interventions of 12
weeks and more seem warranted (cf. Craft & Landers, 1998;
McAuley & Rudolph, 1995; Petruzzello et al., 1991).
In the current two studies, the focus was on the effects of
exercise programs of six months duration on relatively stable
psychological variables of mental health, namely self-concept
and well-being, in middle-aged adults. The experimental
procedure in both studies followed a pretest-posttest design,
and participants were randomly assigned to exercise and control
groups. We hypothesized that the exercise groups, compared to
control groups, would improve in physical self-concept, in self-
esteem, and in subjective well-being after half a year of
regular exercise. In addition, because of the larger sample
size, Experiment 2 allowed a test of the EXSEM using Structural
Equation Modeling (Bentler, 1997).
Experiment 1
Self-Concept and Well-Being 10
Method
Participants. Participants volunteered for an exercise
program offered by the senior author’s research lab. The program
was advertised in local newspapers and in the university
faculties. To meet the requirements of the program applicants
had to be healthy, middle-aged adults with a sedentary life-
style and no exercise in the past 12 months. Due to restricted
gym capacities no more than forty applicants could be accepted
in the exercise program. As there were more applicants than
places in the program, a waiting group was established. The
waiting group got no program offered, but participated in data
assessment. They were promised and offered an exercise program
six months later. Participants were assigned randomly to the
experimental or the control group. To accomodate participants’
needs more persons in the experimental than in the waiting group
were accepted. Before starting all participants got information
about the exercise sessions, about the research project, and
voluntary participation and confidentiality were assured.
Initially the experimental group consisted of 39 adults (11
males, 28 females), 25 to 50 years of age (M = 37.5 yrs). During
the six-month program 11 persons dropped, five of them due to
psychosomatic illnesses. Accordingly, the dropouts differed from
the remaining participants significantly in their level of
psychosomatic complaints. Four participants of the exercise
Self-Concept and Well-Being 11
group did not complete the Time 2 questionnaires. Thus, data of
the experimental group include 24 participants (17 females, 7
males, M = 36.7 yrs, SD = 8.6). The control group initially
consisted of 19 adults. Six persons dropped leaving 13 control
participants in the whole study (12 females, 1 male; M = 39.3
years, SD = 11.27).
Instruments. The questionnaire package focused on physical
self-concept, self-esteem, and subjective well-being (Table 1).
Physical self-concept was assessed by four subscales containing
31 items rated on a 6-point, Likert-type scale ranging from
strongly disagree (1) to strongly agree (6). The scales had been
developed using exploratory factor analysis and item analyses.
The scales are measuring positive (12 items, e.g. „I am proud of
my body“, Cronbach alpha = .86) and negative physical self-worth
(7 items, e.g. „I tend to conceal my body“, Cronbach alpha = .
69), concerns about physical attractiveness (6 items, e.g. „I am
not satisfied with my figure“, Cronbach alpha = .54), and
perceived physical fitness (6 items, e.g. „I am in a good
shape“, Cronbach alpha = .80). All scales had been developed for
middle-aged adults (cf. Alfermann & Stoll, 1996). Self-esteem
was assessed by two scales tapping social self-esteem (6 items,
Cronbach alpha = .78) and global self-esteem (10 items; Cronbach
alpha = .57, Deusinger, 1986).
Self-Concept and Well-Being 12
Subjective well-being included two variables, trait anxiety
and psychosomatic complaints. Trait-anxiety was assessed by a
German version of the Spielberger scale (20 items; Cronbach
alpha = .90, Laux, Glanzmann, Schaffner, & Spielberger, 1981),
and psychosomatic complaints by a 24-item scale, (Cronbach alpha
= .88, called B-L’; Zerssen, 1976). The items are rated on 4-
point scales (for further details see Table 1.)
Design and procedures. The experimental condition consisted
of a six-month exercise program taking place 60 minutes once a
week. The content was mixed and designed to train various
aspects of motor performance, namely endurance, strength,
coordination, and flexibility. The typical practice session
started with a warming-up (aerobic exercise) followed by
stretching, exercises for improving strength and coordination,
and by simplified versions of games like basketball or
badminton. In the last five minutes participants regularly
discussed the exercise session with the instructor. The load
intensity of the physical acitivity intervention reached up to
75% (submaximal work load). The control condition consisted of a
waiting group with no exercise or any other supervised activity
during six months. At the beginning (t1) and six months later
(t2) data were collected from the experimental and the control
groups. At Time 2 the control group began the same exercise
Self-Concept and Well-Being 13
program as the experimental group in order to fulfil our promise
and the requirements of ethical considerations .
Results
In order to control for sex differences in dependent
variables the data of the control and the exercise groups at
Time 1 were combined. Three multivariate analyses of variance
were calculated with sex as between subjects factor. One of the
three comparisons (physical self-concept) reached significance,
mF(4,18) = 9.53, p < .001
Further analyses indicated a significant and substantial
difference in one variable, namely perceived physical fitness,
t(32) = 3.82, p < .01, with males scoring higher than females (MM
= 25.0, SD=4.65; MF = 18.0, SD=2.69). No further sex differences
in dependent variables could be found. Thus results show that
male and female participants were similar in dependent variables
at Time 1. The only notable exception is the sex difference in
perceived physical fitness. When comparing the development of
male and female participants over time with 2 (sex) by 2 (time)
analyses of variance no significant interaction effects could be
found. This seems to suggest that males and females did not
develop differently during the six month period.
The impact of exercise on the dependent variables was tested
with a series of factorial multivariate analyses of variance (2
x 2) with treatment (2) as between-subjects factor and time (2)
Self-Concept and Well-Being 14
as repeated measures factor. Physical self-concept (4 scales),
self-esteem (2 scales), and subjective well-being (2 scales)
each were the dependent variables. There were marginally
significant interaction effects for physical self-concept,
Traetment x Time F(1, 33) = 2.09, p=.11) and a significant
three-way interaction for weel-being: Traetment x Time c
Construct, F(1, 33) = 6.25,p < .05, which could be attributed to
psychosomatic complaints F(1, 31) = 8.49, p < .01, eta² = .22,
with nonexercisers at t2 scoring higher but not at t1 (Table 2).No
significant effects were found for trait anxiety F(1, 29) =
0.55. No interaction effects were found for self-esteem. A
significant main effect of treatment was found for well-being
mF(2, 28) = 4.23, p < .05, eta² = .23,
For physical self-concept there was a significant
multivariate main effect of time mF(4, 33) = 4.67, p < .01, eta²
= .36. ). Main effects of time were found for the subscales
positive physical self-worth, F(1, 36) = 6.77, p < 01, eta² = .
16, and perceived physical fitness, F(1, 36) = 10.24, p < .01,
eta² = .22.
The interaction effect „Treatment x Time“ failed to reach
significance, mF(4, 33) = 1.90. Univariate analyses of variance
nevertheless revealed interaction effects on two scales showing
that, compared to the control subjects, participants of the
exercise group decreased in negative physical self-worth, F(1,
Self-Concept and Well-Being 15
36) = 4.72, p < .05, eta² = .12, and in concerns about physical
attractivess, F(1, 36) = 5.23, p < .05, eta² = .13 (descriptive
statistics in Table 2.
A discriminant function analysis with physical self-concept
and psychosomatic complaints at t2 as predictors was able to
classify 80% of the participants correctly into the two
(experimental/control) groups. The remaining dependent variables
could not improve the correct classification.
Discussion
The results of this experiment support the assumption that
mental health may be enhanced by exercise. Contrary to
predictions and to the meta-analyses of Petruzzello et al.
(1991), and of McDonald and Hodgdon (1991) trait anxiety and
self-esteem remained unchanged. Similar to the results of King,
Taylor, Haskell, and DeBusk (1989), changes were seen primarily
in those variables that seem to be most closely associated with
the physical experiences during exercise. The benefits of
exercise were most clearly seen in subjective measures of
physical change, like physical self-concept and less so in
measures of psychological change, namely self-esteem and trait
anxiety. These two constructs may be more stable and enduring
personality characteristics and thus less easily affected by
physical exercise. A more intense exercise program, e.g. twice a
Self-Concept and Well-Being 16
week, might be needed to produce changes in self-esteem and
trait anxiety.
An additional shortcoming of the design that requires some
improvement is the wait-list control group. The control
condition differed from the experimental condition not only in
the amount of exercise but also in some other aspects, mainly
attention and social support by the instructor and the group
members that could have influenced the results.
In addition, larger sample size is preferable because larger
sample size increase the statistical power of the data analyses
(Thomas, Lochbaum, Landers, & He, 1997), and would allow to test
the structural relations between the constructs of the EXSEM.
Therefore, a second experiment was conducted addressing
both, the influence of exercise versus attention on the
dependent measures, and the structure of the EXSEM. Exercise
programs were administered twice a week, and the control groups
were offered alternative intervention programs with the same
amount of attention as the experimental groups. If physical
activity itself was responsible for changes in self-concept and
well-being, and less the attention and the social support
participants got in the experimental condition, then the
experimental groups should improve in the dependent variables
significantly more than the control groups.
Experiment 2
Self-Concept and Well-Being 17
Method
Participants. One hundred eighty-three middle-aged adults
(50 males, 133 females) volunteered to participate in the study
(M = 43.2 yrs, SD = 8.1). Similar to the first experiment the
sample consisted of healthy, middle-aged adults with no exercise
experience during the past 12 months. They had applied for any
kind of intervention program offered by the senior author’s
research lab. Subjects were informed about the programs, the
experimenter controlled group assignment, and about their rights
as research participants before the start of the experiment and
before any measures were administered. Voluntary participation
and confidentiality were assured. Nearly 50% dropped during the
half-year program, leaving 93 participants (24 males and 69
females) in the full study. Attrition rates were similar between
the experimental and the control conditions, and between males
and females. The dropout rate was quite similar to that reported
in the literature, especially when considering the randomization
procedure (cf. Willis & Campbell, 1992, pp. 19-21). Multivariate
and univariate analyses of variances revealed no significant
differences on the dependent measures at Time 1 between dropouts
and non-dropouts (p > .10). Six months after completion of the
intervention programs, follow-up measurement took place.
Seventy-two individuals (21 males, 51 females) were willing to
participate in data assessment (cf. Table 3).
Self-Concept and Well-Being 18
Instruments. The paper-pencil measures were identical to
those in Experiment 1. In addition, trait anger was assessed by
a German version of the Spielberger scale (Schwenkmezger,
Hodapp, & Spielberger, 1992, cf. Table 1). Trait anger was
included because of its health-related significance in previous
research (Jorgensen, Johnson, Kolodziej, & Schreer, 1996; Suls,
Wan, & Costa, 1995).
For reasons of program evaluation, and as a manipulation
check ten motor performance tests were administered. The tests
are supposed to represent a range of basic motor skills and
abilities, namely endurance (fitness-index), flexibility,
balance, speed (shuttle run), strength (jump and reach, throw
distance, sit-ups, push-ups), and coordination (jump and throw).
The fitness-index considers individual walking time on 2000m
relative to the individual’s weight, height and recovering heart
rate.
Design and procedures. All participants took part in an
experimental or in a control group program. All programs lasted
for half a year and took place twice a week for 60 minutes each.
The experimental conditions consisted of two exercise groups.
One group did exclusively aerobic exercise (jogging). This
program was included because aerobic exercise is said to be
better able to change subjective well-being than any other kind
of exercise, as suggested by physiological theories explaining
Self-Concept and Well-Being 19
improvements in well-being (Hatfield, 1991). The activities of
the jogging group were restricted to aerobic exercise, mainly
walking and running, and occasionally cycling and swimming. The
intensity was adapted to participants’ level of fitness and
reached up to 70% of individual maximal workload. At the start
of the program a heart rate of about 130 for at least 10 minutes
per lesson was aimed at, and this increased up to at least 30
minutes per lesson at the program’s completion.
The other experimental condition (fitness group) consisted
of the same exercise program as in Experiment 1.
The control conditions consisted of two programs which were
designed to be different from exercise. But at the same time
they should be able to attract volunteers who were willing to
participate in any of the intervention programs offered. The
programs had to be nearly equally attractive and acceptable, and
they had to be interesting enough to last for half a year.
Therefore, the following two control programs were implemented.
Relaxation training was chosen because meditation/relaxation
is a well-known (psycho)therapeutic method and because it has
been used as a control condition in the exercise literature
(e.g. Bahrke and Morgan, 1978; Greist; 1987). Participants of
the relaxation group learned progressive muscle relaxation and
autogenic training. Each lesson consisted of 15 to 20 minutes of
Self-Concept and Well-Being 20
verbal instruction and discussion. The remainder of the session
was dedicated to relaxation practice.
The second control condition (back group) was chosen because
it is a quite popular activity for health prevention in the
authors’ country, yet with unproven validity. The back group
participated in a training aimed to strengthen the back muscles
and to learn to sit and move in a way that presumably is able to
prevent back pain. Instruction focused as much on practice as on
information about everyday behavior and back pain. Participants
were randomly assigned to one of the four conditions. The only
restriction was a meaningful group size which had to be adapted
to the programs. Therefore, the control groups have slightly
less participants than the experimental groups (cf. Table 3).
Paper-pencil measures were collected three times: at the
beginning of the intervention program, six months later at
program’s completion, and another six months later as a follow-
up. Participants’ motor performance was tested at the beginning
of the program and half a year later at the program’s completion.
Results
Exercise and Self-Esteem Model (EXSEM). The data of the
whole sample at Time 1 were analyzed in order to test the
relationships between the manifest variables of the expanded
version of the EXSEM. The model generally assumes that changes
in perceptions of specific self-concept components generalize to
Self-Concept and Well-Being 21
changes in broader self constructs. In our study we therefore
expected higher correlations between variables of adjacent
levels than between variables of more remote levels. As can be
seen from Table 4, the physical competence variables on the
lowest level („perceived physical fitness“ and „concerns about
physical attractiveness“) correlate only moderately with
positive and negative PSW on the second level, but even lower
with SE (as expected). The latter two correlate highly with SE
(measured by the subscales global and social SE), as
hypothesized. Based on the theoretical assumptions global self-
esteem was regressed on the two PC scales and the two PSW
scales. Only the two PSW scales contributed significantly to SE,
explaining 45% of the variability in global SE, F(2, 161) =
64.67, p < .001 (positive PSW: β = 0.34, T = 4.88, p < .001;
negative PSW: β = -0.42, T = 6.11, p < .001). With regard to
social SE, the PSW scales explain 16.8% of the variability, F(2,
161) = 16.34, p < .001 (positive PSW: β = 0.18, T = 2.09, p < .
05; negative PSW: β = -0.24, T = 3.38, p < .001).
Structural equation models are virtually always multiple
equation models with multiple dependent as well as independent
variables. The simplest extension of the regression model is the
multivariate regression model, in which several dependent
variables are regressed on several predictor variables. In the
Self-Concept and Well-Being 22
EXSEM, global self-esteem is the dependent variable which is
supposed to be predicted by positive and negative PSW, and in
turn PSW by the PC variables of physical fitness and (concern
about) physical attractiveness. Using structural equation
modeling a path analysis was performed (EQS version 5.6;
Bentler, 1997). The results clearly confirm the EXSEM with the
exception of perceived physical fitness which shows a direct
(instead of an indirect) path to SE (Figure 1).
The path model in Figure 1 shows moderate loadings from PSW
and one PC variable (perceived physical fitness) to global SE.
Furthermore, negative PSW is moderated strongly by concerns
about physical attractiveness (λ = .40). There is no direct
influence from this variable to global SE (λ = -.06). The
goodness of fit indices allow to accept the model, χ2 (5, N =
178) = 4.74, p = .44, Bentler-Bonnet normed fit index
(BBNFI)=.922, Bentler-Bonnet nonnormed fit index (BBNNFI) =
1.010, comparative fit index = 1.000).
Motor performance. The motor-performance variables were used
as a manipulation check (control variable). We analyzed the data
of the motor performance tests using 2 (time) x 4 (treatment)
univariate analyses of variance. Due to the number of tests a
probability level of p < .01 was accepted. If the experimental
participants had improved more than the control groups Time x
Self-Concept and Well-Being 23
Treatment interaction effects were to be expected. There was one
significant interaction effect for the endurance test (fitness-
index), F(3,66) = 4.88, p < .01, eta² = .20. This result can be
attributed to the fact that, as expected, the jogging group
improved significantly over time and more than the other three
groups. In five of the ten tests measuring strength (sit-ups,
push-ups, throw distance), speed (shuttle run) and balance
significant main effects of time emerged with eta²-values
between .11 (shuttle run) and .41 (push-ups). Contrary to our
predictions participants’ testscores obviously increased
regardless of the intervention program (Table 4).
Paper-pencil measures of self-concept and well-being.
Similar to Experiment 1, sex differences occurred on physical
self-concept, mF(4, 163) = 29.26, p < .001, and on psychological
well-being, mF(3, 152) = 5.50, p < .001. Again males scored
higher on perceived physical fitness F(1, 166) = 93.39, p < .
001, eta² = .36, M = 24.8 vs. 18.8, and lower on psychosomatic
complaints F(1, 155) = 9.94, p < .01, eta² = .06, M = 16.4 vs.
21.7.
To examine changes in self-concept and well-being after six
months intervention, data of Time 1 and Time 2 were analyzed
with three multivariate 2 (time) x 4 (treatment) analyses of
variance, with physical self-concept, self-esteem, and
Self-Concept and Well-Being 24
psychological well-being as dependent variables. Data of Time 2
and Time 3 were analyzed comparably.
Comparing t1 and t2 the results of the MANOVAS revealed main
effects of time for all three constructs, physical self-concept
mF(4, 74) = 5.13, p < .001, eta² = .22, self-esteem mF(2, 80) =
3.67, p <.05, eta² = .08 and well-being, mF(3, 67) = 3.80, p = .
01, eta² = .15.
In univariate analyses of variance significant main effects
of time were obtained for perceived physical fitness F(1, 83) =
20.75, p < .001, eta² = .20, for negative physical self-worth
F(1, 83) = 4.54, p < .05, eta² = .05, for global self-esteem,
F(1, 82) = 8.47, p < .001, eta² = .09, and for psychosomatic
complaints, F(1, 76) = 11.88, p < .001, eta² = .13. All effects
indicate that participants improved in these variables,
regardless of the kind of intervention program (Table 3). This
seems to suggest that all intervention programs were equally
effective.
Follow-up. When comparing the second and third points of
measurement with MANOVAS no additional changes over time could
be revealed. All multivariate F-values were nonsignificant. This
can be interpreted as indicating neither any further improvement
nor deterioration in participants’ mental health after the end
of the intervention.
Discussion
Self-Concept and Well-Being 25
The purpose of this experiment was to conduct a test of the
hypothesized positive effects of exercise on self-concept and
well-being. Instead of a wait-list control group like in
Experiment 1, in this experiment placebo attention groups were
preferred. All participants took part in some supervised
intervention program with the same amount of invested time and
attention but with quite a different amount of physical
activity. Contrary to our predictions, the control groups
improved as much as the experimental groups on the dependent
variables of self-concept and well-being, and the effects
remained even six months after the program’s completion. Though
we had not expected this result in advance there are some hints
in the literature which cast some doubt on the superiority of
exercise for mental health prevention and which seem to
corroborate our findings that other interventions may be equally
effective. Examples of these successful interventions include
quiet rest (Raglin & Morgan, 1987), meditation (Bahrke & Morgan,
1978), psychotherapy (cf. Craft & Landers, 1998; Greist, 1987),
and relaxation training (Hautzinger & Kleine, 1995). The results
of all these studies point to the fact that the treatment of
controls obviously is a very sensitive issue in exercise
research.
Most studies so far have not included placebo or alternative
treatment groups (Morgan, 1997). But if they do exercise seems
Self-Concept and Well-Being 26
to be as equally effective as other types of intervention. For
example, in their meta-analysis of the effects of exercise on
depression Craft and Landers (1998) report dramatically
different effect sizes (ES) depending on the experimental
design. For wait-list control group studies (like Experiment 1
in this article) they found a large mean ES (0.77), whereas the
mean ES in experiments comparing exercise to alternative
interventions (like in Experiment 2) was close to zero (0.06 and
0.03). This would suggest that the beneficial effects of
exercise may have been overestimated in the past due to the
preponderance of wait-list control groups.
Though Ojanen (1994) has rejected the usefulness of a
placebo treatment and has considerable doubt that exercise and
placebo effects can be separated Morgan (1997) convincingly
argues for the inclusion of placebo and of other intervention
control groups. The design of Experiment 2 thus fulfils the
requirement expressed by Morgan (1997) that „the question of
whether or not physical activity results in psychological
improvement should be expanded to include the question of
‘compared to what’?“ (p. 12). ‘Compared to what’ the results of
this experiment suggest that exercise is able to improve self-
concept and well-being as much as, but also no less than
relaxation and back training. Exercise seems to be as good for
mental health purposes as other preventive strategies, at least
Self-Concept and Well-Being 27
according to the results of this study. We thus should consider
the possibility that exercise is an effective but no superior
strategy to improve self-esteem and subjective well-being. But
it is superior when compared to no activity, as was shown in
Experiment 1 and as it has been shown in a number of meta-
analyses and reviews.
But not only the kind of control group program seems to
affect the impact of exercise on psychological variables. There
are more moderator variables in meta-analyses suggesting that
exercise effects on psychological variables may be lower than
has been assumed in the past. Firstly, as was mentioned in the
introduction, experimental studies produce lower effect sizes
than correlational ones. This too could have contributed to the
no difference result that was found here. Secondly, publication
policy obviously plays an important role. Not surprisingly,
studies with significant and hypothesis confirming differences
are more likely to be published. Craft and Landers (1998) report
that the mean ES from published studies (mES = 0.91) was nearly
twice that from unpublished studies (mES = 0.49). Again this may
help to explain why the psychological effects of exercise could
be overestimated.
An additional limitation of this experiment lies in the
sample size. This surely reduced the power of the statistical
tests, in Experiment 2 in particular, because this design needs
Self-Concept and Well-Being 28
a larger number of participants than the wait-list control group
design of Experiment 1, due to the presumably smaller effect
size (cf. Thomas et al. 1997). Thus with a larger sample we
might have detected significant and meaningful differences
between experimental and control groups and thus larger effects
of exercise.
The small sample size is mainly a consequence of the
attrition rate in our study. In consequence, the
generalizability of the results is restricted to those
participants who are willing to adhere to an intervention for
half a year. Presumably only those participants are willing to
attend the whole program whose expectations are met and who
subjectively benefit from the program. This too could have
contributed to the pattern of results leaving only those
individuals in the experiment who experienced positive changes
in their mental health and accordingly filled in the
questionnaires. Or, as Craft and Landers (1998) put it in
summarizing the results of their meta-analysis on depression,
„what is being observed (...) is that exercise programs produce
changes in depression scores“ (p. 353).
Considering the changes in motor performance it is striking
that only one of ten test results, the fitness test, corresponds
to our assumptions. This is nearly as much as would be expected
by chance. On the other hand, it seems reasonable to conclude
Self-Concept and Well-Being 29
that the endurance measure which includes a 2km walking time may
reflect the real differences between the four groups due to the
different programs. The jogging group, with its exclusive
emphasis on aerobic exercise, improved significantly in
endurance, but not the other groups. This is clearly in
accordance with the program’s content.
Apart from that there were significant main effects of time
in five of the ten motor performance tests, measuring strength,
speed, and balance, an unexpected result. Obviously the
participants of all groups improved in some aspects of motor
performance though the interventions clearly differed in this
respect. The reasons for the similarity in motor performance of
all groups are unclear and desire some speculation about
possible methodological artifacts. As each participant was
tested twice the increase in motor performance might reflect
test effects. In addition, participants were tested in groups,
and this could have contributed to some competitive atmosphere
leading to heightened motivation and performance, similar to the
Hawthorne effect. And finally, demand characteristics, like
being a „good“ subject, might have influenced the participants’
behavior. During the six months intervention period each
participant had been asked to report any self-paced exercise
activities per week. When using this variable as a covariate in
an ANCOVA with motor performance as dependent measures the main
Self-Concept and Well-Being 30
effects of time remained. Though we cannot rule out the
possibility that the participants did not always tell the truth,
the fact remains that all groups improved in motor performance.
At first glance this in turn might help to explain the mental
health effects in this experiment, but the results of other
studies (e.g. King et al, 1989; Sonstroem et al., 1993) show
that changes in physical variables do not necessarily correspond
to changes in psychological variables, and vice versa. Thus,
expectancy effects cannot be ruled out.
The expanded EXSEM (Sonstroem et al., 1994) was partly
supported. The hierarchical structure of (physical) self-concept
is reflected by the correlational pattern and confirmed by the
results of the path analysis (Figure 1). Different from other
studies (Harter, 1996) physical attractiveness is not (highly)
correlated to self-esteem. Instead, it has a significant path to
physical self-worth, thus confirming the results of Sonstroem et
al. (1994, s. also Sonstroem 1997a).
The only contradictory result to the predictions of the
EXSEM is the direct significant path of perceived physical
fitness to SE (and not to PSW). Obviously in our sample physical
competence is still an important contributor to SE even when the
influence of PSW is partialled out. Further research is needed
to clarify if this result is specific for our sample or for this
age group in general.
Self-Concept and Well-Being 31
In summary, it may be said that the results of both
experiments corroborate the assumption that exercise improves
physical self-concept. Participants in the experimental groups
showed significant increases in physical self-concept after six
months of exercise. However, only in Experiment 2 do we find any
increments in self-esteem, and these hold true for all
participants, not only for the experimental groups.
Both studies differ mainly in the ‘treatment’ of the control
groups. In Experiment 1 an exercise group is compared with a
wait-list control group. The results show a positive effect of
exercise on some dependent variables as was hypothesized and as
is suggested by the literature. In Experiment 2, exercise groups
are compared with treatment control groups. Though exercise
again seems to influence self-concept and well-being this time
the control groups show the same response pattern. Thus
alternative interventions may have a similar positive effect on
mental health as exercise.
Self-Concept and Well-Being 32
References
Alfermann, D., & Stoll, O. (1996). Sport im mittleren
Erwachsenenalter: Auswirkungen auf Selbstkonzept, subjektives
Wohlbefinden und Stresstoleranz [Exercise in middle adulthood:
Effects on self-concept, subjective well-being and stress
reactivity] (Research Rep.). Leipzig, Germany: University of
Leipzig, Department of Sport Psychology.
Bahrke, M.S., & Morgan, W.P. (1978). Anxiety reduction
following exercise and meditation. Cognitive Therapy and
Research, 2, 323-333.
Baldwin, M.K., & Courneya, K.S. (1997). Exercise and self-
esteem in breast cancer survivors: An application of the
exercise and self-esteem model. Journal of Sport & Exercise
Psychology, 19, 347-358.
Bentler, P.M. (1997). EQS -Structural Equations Program.
Update Version 5.6. Encino, CA: Multivariate Software
Incorporation.
Caruso, C.M., & Gill, D.L. (1992). Strengthening physical
self-perceptions through exercise. Journal of Sports Medicine
and Physical Fitness, 32, 416-417.
Craft, L.L., & Landers, D.M. (1998). The effect of exercise
on clinical depression and depression resulting from mental
illness: A meta-analysis. Journal of Sport & Exercise
Psychology, 20, 339-357.
Self-Concept and Well-Being 33
Deusinger, I. (1986). Die Frankfurter Selbstkonzeptskalen
(FSKN) [The Frankfurt self-concept scales (FSKN)]. Göttingen:
Hogrefe.
Dishman, R.K. (1994). Introduction: Consensus, problems, and
prospects. In R.K. Dishman (Ed.), Advances in exercise adherence
(pp. 1-28). Champaign, IL: Human Kinetics.
Folkins, C.H., & Sime, W.E. (1981). Physical fitness
training and mental health. American Psychologist, 36, 373-389.
Fox, K.R., & Corbin, C.B. (1989). The physical self-
perception profile: Development and preliminary validation.
Journal of Sport & Exercise Psychology, 14, 1-12.
Greist, J.H. (1987). Exercise intervention with depressed
outpatients. In W.P. Morgan & S.E. Goldston (Eds.), Exercise and
mental health (pp. 117-121). Washington: Hemisphere.
Gruber, J.J. (1986). Physical activity and self-esteem
development in children: A meta-analysis. In G. Stull & H.
Eckert (Eds.), Effects of physical activity on children (pp. 30-
48). Champaign, IL: Human Kinetics and American Academy of
Physical Education.
Harter, S. (1996). Causes, correlates, and the functional
role of global self-worth: A life-span perspective. In R.J.
Sternberg & J. Kolligian, jr. (Eds.), Competence considered (pp.
67-97). New Haven: Yale University Press.
Self-Concept and Well-Being 34
Hatfield, B.D. (1991). Exercise and mental health: The
mechanisms of exercise-induced psychological states. In L.
Diamant (Ed.), Psychology of sports, exercise and fitness (pp.
17-49). New York: Hemisphere.
Hautzinger, M., & Kleine, W. (1995). Sportliche Aktivitaet
und psychisches Wohlbefinden: Zur Wirkung von Sport auf
depressive Symptomatik [Exercise and psychological well-being:
The effects of exercise on depressive symptoms]. Zeitschrift
fuer Gesundheitspsychologie, 3, 255-267.
International Society of Sport Psychology (ISSP) (1992).
Physical activity and psychological benefits: A position
statement. International Journal of Sport Psychology, 23, 86-91.
Jorgensen, R.S., Johnson, B.T., Kolodziej, M.E., & Schreer,
G.E. (1996). Elevated blood pressure and personality: A meta-
analytic review. Psychological Bulletin, 120, 293-320.
Kerr, J.H., & Vlaswinkel, E.H. (1990). Effects of exercise
on anxiety: A review. Anxiety Research, 3, 309-321.
King, A.C., Taylor, C.B., Haskell, W.L., & DeBusk, R.F.
(1989). Influence of regular aerobic exercise on psychological
health: A randomized, controlled trial of healthy middle-aged
adults. Health Psychology, 8, 305-324.
Self-Concept and Well-Being 35
Landers, D.M., & Petruzzello, S.J. (1994). Physical
activity, fitness, and anxiety. In C. Bouchard, R.J. Shephard, &
T. Stephens (Eds.), Physical activity, fitness, and health.
International proceedings and consensus statement (pp. 868-882).
Champaign, IL: Human Kinetics.
Laux, L., Glanzmann, P., Schaffner, P., & Spielberger, C.D.
(1981). STAI. Das State-Trait-Angst-Inventar: Theoretische
Grundlagen und Handweisung [The state-trait-anxiety inventory:
Theoretical foundations and manual]. Weinheim: Beltz
Testgesellschaft.
Leith, L.M. (1994). Foundations of exercise and mental
health. Morgantown, WV: Fitness Information Technology.
Martinsen, E.W., & Stephens, T. (1994). Exercise and mental
health in clinical and free-living populations. In R.K. Dishman
(Ed.), Advances in exercise adherence (pp. 55-72). Champaign,
IL: Human Kinetics.
McAuley, E. (1994). Physical activity and psychosocial
outcomes. In C. Bouchard, R.J. Shephard, & T. Stephens (Eds.),
Physical activity, fitness, and health. International
proceedings and consensus statement (pp. 551-568). Champaign,
IL: Human Kinetics.
McAuley, E., & Rudolph, D. (1995). Physical activity, aging,
and psychological well-being. Journal of Aging and Physical
Activity, 3, 67-96.
Self-Concept and Well-Being 36
McDonald, D.G., & Hodgdon, J.A. (1991). Psychological
effects of aerobic fitness training. Research and theory. New
York: Springer.
Morgan, W.P. (1994). Physical activity, fitness, and
depression. In C. Bouchard, R.J. Shephard, & T. Stephens (Eds.),
Physical activity, fitness, and health. International
proceedings and consensus statement (pp. 851-867). Champaign,
IL: Human Kinetics.
Morgan, W.P. (1997). Methodological considerations. In W.P.
Morgan (Ed.), Physical activity and mental health (pp. 3-32).
Washington, DC: Taylor & Francis.
Ojanen, M. (1994). Can the true effects of exercise on
psychological variables be separated from placebo effects?
International Journal of Sport Psychology, 25, 63-80.
Petruzzello, S.J., Landers, D.M., Hatfield, B.D., Kubitz,
K.A., & Salazar, W. (1991). A meta-analysis on the anxiety-
reducing effects of acute and chronic anxiety. Sports Medicine,
11, 143-182.
Raglin, J.S. (1990). Exercise and mental health. Beneficial
and detrimental effects. Sports Medicine, 9, 323-329.
Raglin, J.S., & Morgan, W.P. (1987). Influence of exercise
and quiet rest on state anxiety and blood pressure. Medicine and
Science in Sports and Exercise, 19, 456-483.
Self-Concept and Well-Being 37
Schwenkmezger, P., Hodapp, V., & Spielberger, C.D. (1992).
Das State-Trait-Ärgerausdrucksinventar (STAXI) [The state-trait-
anger-expression inventory]. Bern: Huber.
Sonstroem, R.J. (1997a). Physical activity and self-esteem.
In W.P. Morgan (Ed.), Physical activity and mental health (pp.
127-143). Washington, DC: Taylor & Francis.
Sonstroem, R.J. (1997b). The physical self-system: A
mediator of exercise and self-esteem. In K.R. Fox (Ed.), The
physical self: From motivation to well-being (pp. 3-26).
Champaign, IL: Human Kinetics.
Sonstroem, R.J. (1998). Physical self-concept: Assessment
and extermal validity. Exercise and Sport Sciences Reviews, 26,
133-144.
Sonstroem, R.J., Harlow, L.L., Gemma, L.M., & Osborne, S.
(1991). Test of structural relations within a proposed exercise
and self-esteem model. Journal of Personality Assessment, 56,
348-364.
Sonstroem, R.J., Harlow, L.L., & Josephs, L. (1994).
Exercise and self-esteem: Validity of model expansion and
exercise associations. Journal of Sport & Exercise Psychology,
16, 29-42.
Sonstroem, R.J., Harlow, L.L., & Salisbury, K.A. (1993).
Path analysis of a self-esteem model across a competitive swim
season. Research Quarterly in Exercise and Sport, 64, 335-342.
Self-Concept and Well-Being 38
Sonstroem, R.J., & Morgan, W.P. (1989). Exercise and self-
esteem: Rationale and model. Medicine and Science in Sports and
Exercise, 21, 329-337.
Sonstroem, R.J., & Potts, S.A. (1996). Life adjustment
correlates of physical self-concepts. Medicine and Science in
Sports and Exercise, 28, 619-625.
Suls, J., Wan, C.K., & Costa, P.T.jr. (1995). Relationship
of trait anger to resting blood pressure: A meta-analysis.
Health Psychology, 14, 444-456.
Thomas, J.R., Lochbaum, M.R., Landers, D.M., & He, C.
(1997). Planning significant and meaningful research in exercise
science: Estimating sample size. Research Quarterly for Exercise
and Sport, 68, 33-43.
Willis, J.D. & Campbell, L.F. (1992). Exercise psychology.
Champaign, IL: Human Kinetics.
Zerssen, D.v. (1976). Die Beschwerdenliste. Manual [List of
complaints. Manual]. Weinheim: Beltz Testgesellschaft.
Self-Concept and Well-Being 39
Author Note
Dorothee Alfermann and Oliver Stoll are with the Faculty of
Sport Sciences at the University of Leipzig, Germany. Experiment
1 was conducted while the authors were with the University of
Giessen, Giessen, Germany.
The research reported herein was financially supported by a
grant from the Ministry of Art and Science Hessen (Experiment
1), and by a grant from the Deutsche Forschungsgemeinschaft (Al
305/3-1; Experiment 2) to the first author.
The authors are indebted to Stevan Hobfoll and two anonymous
reviewers for helpful comments on an earlier version of this
article. Portions of the data were presented at the IXth
European Congress of Sport Psychology 1995 in Brussels, Belgium,
and at the XXVIth International Congress of Psychology 1996 in
Montreal, Canada.
Correspondence concerning this article should be addressed
to Dorothee Alfermann, Faculty of Sport Sciences, University of
Leipzig, Leipzig, Germany 04109. Electronic mail may be sent via
internet to alferman@rz.uni-leipzig.de
Self-Concept and Well-Being 40
Table 1
Cronbach α , Item-Total Correlations and Scale Ranges of the
Subscales Used in Experiments 1 and 2
Instrument Cronbach αrpbis
(range)
Scale
(range)
Body-Self-Concept
Positive physical self-
worth
.86 .46 - .70 12 - 72
Negative physical self-
worth
.69 .40 - .61 7 - 42
Physical fitness .54 .37 - .56 6 - 36
Physical attractiveness .80 .38 - .56 6 - 36
Self-esteem
Global self-esteem .78 .39 - .72 10 - 60
Social self-esteem .57 .38 - .46 6 - 36
Subjective well-being
Trait Anger .88 .39 - .54 10 - 40
Trait anxiety .90 .38 - .68 20 - 80
Psychosomatic complaints .88 .33 - .70 0 - 72
Self-Concept and Well-Being 41
Table 2
Means and Standard Deviations for the Subscales in Experiment 1
as a Function of Time and Condition
Exercise Group (na = 24) Control Group (na = 13)
M SD M SD
Positive physical self-worth
Time 1 50.54 6.74 47.57 10.26
Time 2 53.92 6.57 49.14 9.78
Negative physical self-worth
Time 1 18.33 5.33 18.57 6.35
Time 2 17.71 4.87 20.14 5.82
Perceived physical fitness
Time 1 20.75 4.11 19.29 4.07
Time 2 22.96 4.13 20.29 3.81
Concerns about physical attractiveness
Time 1 19.92 6.36 21.64 8.25
Time 2 18.88 6.80 22.93 8.48
Global self-esteem
Time 1 48.67 5.67 46.92 7.85
Time 2 49.83 5.47 47.31 7.60
(table continues)
Self-Concept and Well-Being 42
Exercise Group (n = 24) Control Group (n = 13)
M SD M SD
Social self-esteem
Time 1 25.83 4.69 26.00 4.32
Time 2 26.91 4.36 25.17 4.45
Psychosomatic complaints
Time 1 18.46 8.74 21.18 10.62
Time 2 18.10 6.49 27.82 9.31
Trait anxiety
Time 1 37.05 9.33 40.50 12.72
Time 2 35.52 7.43 40.17 9.54
Self-Concept and Well-Being 43
Table 3
Means and Standard Deviations for the Subscales in Experiment 2
as a Function of Time and Condition
Experimental ConditionaControl Conditiona
Fitness (F) Jogging (J) Relaxation (R) Back (B)
M SD M SD M SD M SD
Positive physical self-worth
Time 1 51.82 8.30 51.72 6.30 51.12 8.23 54.37 6.07
Time 2 52.73 6.72 52.16 8.24 55.15 7.50 53.79 6.93
Time 3 49.82 11.05 50.27 11.35 48.43 11.75 49.13 13.82
Negative physical self-worth
Time 1 15.52 5.64 17.40 4.71 16.80 5.76 15.37 3.96
Time 2 15.33 5.22 15.64 4.43 15.53 5.62 15.35 4.63
Time 3 16.25 7.49 18.00 6.34 18.86 9.12 16.73 10.28
Perceived physical fitness
Time 1 21.27 4.20 20.31 4.94 20.12 3.59 21.85 4.29
Time 2 22.80 3.32 23.20 3.70 21.27 3.88 21.89 2.64
Time 3 22.74 3.31 22.74 4.22 20.71 2.89 20.88 2.47
Concerns about physical attractiveness
Time 1 20.32 2.69 20.08 3.08 20.31 2.39 19.05 2.86
Time 2 19.27 2.99 19.64 2.81 19.80 2.24 20.70 2.81
Time 3 19.85 3.12 19.61 2.93 20.64 2.50 19.63 2.63
Self-Concept and Well-Being 44
Table 3 (continued)
Experimental ConditionaControl Conditiona
Fitness (F) Jogging (J) Relaxation (R) Back (B)
M SD M SD M SD M SD
Global self-esteem
Time 1 49.55 5.55 47.40 6.28 49.58 10.04 49.43 5.42
Time 2 51.07 5.54 48.00 6.28 50.08 8.95 51.82 4.74
Time 3 51.88 5.43 48.95 6.45 49.55 10.55 48.85 4.45
Social self-esteem
Time 1 26.44 4.50 26.81 4.15 29.08 4.25 26.38 4.31
Time 2 26.42 3.61 26.30 4.76 29.08 2.23 27.23 3.88
Time 3 27.50 4.11 27.55 5.16 29.55 3.47 27.80 4.30
Psychosomatic complaints
Time 1 19.69 6.73 20.00 10.99 20.92 10.82 21.30 9.66
Time 2 18.80 5.83 17.50 9.21 17.67 8.97 17.79 7.97
Time 3 16.85 7.50 17.09 10.30 17.92 11.30 18.94 10.53
Trait anger
Time 1 18.84 4.27 17.88 4.17 17.69 3.55 18.10 3.80
Time 2 18.17 3.82 17.00 4.18 15.40 3.29 19.22 4.08
Time 3 16.82 3.45 16.83 4.33 18.00 3.55 16.69 2.82
(table continues)
Self-Concept and Well-Being 45
Experimental ConditionaControl Conditiona
Fitness (F) Jogging (J) Relaxation (R) Back (B)
M SD M SD M SD M SD
Trait anxiety
Time 1 36.08 8.53 38.00 11.78 32.08 9.28 33.44 8.36
Time 2 33.69 8.60 35.87 10.74 33.25 13.23 35.24 7,60
Time 3 34.67 8.99 34.09 10.64 34.23 14.12 34.06 8.23
Note. aSample sizes at Time 1: n = 57 (F), 50 (J), 36 (R), 40
(B); at Time 2: n = 31 (F), 26 (J), 16 (R), 20 (B); at Time 3: n
= 24 (F), 20 (J), 14 (R), 14 (B); sample sizes may vary in some
scales due to missing values.
Self-Concept and Well-Being 47
Table 4: Means and Standard Deviations for the variables of the motoric performance in
Experiment 2 as a Function of Time and Condition
Experimental ConditionaControl Conditiona
Fitness (F) Jogging (J) Relaxation (R) Back (B) Effect TIME
M SD M SD M SD M SD F p
Fit-Ind. t1 100.8 16.5 95.9 16.2 103.9 12.2 108.3 8.6
2.50 n.s.
Fit-Ind. t2 103.8 16.5 107.0 13.9 97.5 17.5 99.5 16.6
Situps t1 14.6 5.8 14.2 3.3 12.5 2.8 12.9 3.2
38.08 P<.001
Situps t2 16.6 4.7 15.4 4.2 15.5 2.3 16.8 2.7
Pushups t1 11.2 3.2 9.6 2.6 9.6 2.0 9.3 3.5
44.84 P<.001
Pushups t2 13.1 3.1 12.5 2.6 13.4 3.5 11.5 2.1
Throw t1 5.9 1.2 5.8 1.7 5.1 1.0 5.9 1.1
11.59 P<.001
Throw t2 6.1 1.1 5.9 1.4 5.6 1.1 6.0 1.1
Speed t1 11.6 1.3 12.4 0.9 12.9 0.6 11.5 0.6
7.72 P<.01
Speed t2 12.0 1.2 13.3 2.2 13.0 1.3 12.5 1.5
Balance t1 27.9 19.3 27.0 3.5 15.5 11.1 21.8 16.0
9.66 P<.01
Balance t2 45.4 31.9 30.6 8.1 24.5 20.5 33.0 20.1
Self-Concept and Well-Being 48
Table 5
Pearson Correlations among EXSEM Scales at Time 1 (Experiment 1)
Scales 1 2 3 4 5 6
1. Perceived physical fitness - -.15* .12 -.30** .19* .08
2. Concerns about physical
attractiveness
- - -.09 .24** -.13 -.07
3. Positive physical self-worth - - - -.52*** .54*** .34***
4. Negative physical self-worth - - - - -.61*** -.39***
5. Global self-esteem - - - - - .48***
6. Social self-esteem - - - - - -
Notes. *p < .05; **p < .01; ***p < .001
Self-Concept and Well-Being 49
Figure Caption
Figure 1. Path analysis solution.
Method: Maximum Likelihood, χ2=4.74 (df=5), p=.44, Bentler-Bonnet
Normed Fit Index (BBNFI)=.922, Bentler-Bonnet Nonnormed Fit
Index = 1.010, Comparative Fit Index = 1.000; PSW = Physical
self-concept
Global Self-
Esteem
Positive
PSW
Negative PSW
Physical
Fitness
Physical
Attractiveness
.21*
-.18*
.20*
.40*
-.06