JOURNAL OF SPORT & EXERCISE PSYCHOLOGY, 1994,16,43-55

O 1994 Human Kinetics Publishers, Inc.

A Multidimensional Physical Self-concept

and Its Relations to Multiple Components
of Physical Fitness

Herbert W. Marsh Robyn Sutherland Redmayne

University of Western Sydney, University of Sydney
Macarthur

This study examines relations between six components of physical self-

concept (Endurance, Balance, Flexibility, Strength, Appearance, and general
Physical Ability) and five components of physical fitness (Endurance, Bal-
ance, Flexibility, Static Strength, Explosive StrengthfPower) for a sample
(N = 105) of young adolescent girls aged 13 and 14. Hierarchical confirmatory
factor analyses identified the six physical self-concept scales and provided
support for a multidimensional, hierarchical model of physical self-concept.
The pattern of correlations between specific components of physical self-
concept and physical fitness generally supported the construct validity of
the self-concept responses, and the correlation between second-order factors
representing general physical self-concept and general physical fitness (r =
.76) was substantial.

Key words: components of self-concept, Self-Description Questionnaire

The focus of the present investigation is on a multidimensional, hierarchical

physical self-concept and its relation to multiple dimensions of physical fitness.
Historically, self-concept researchers have emphasized a broad, global construct
that did not differentiate among specific domains. In an attempt to remedy
this problem, Shavelson, Hubner, and Stanton (1976) developed a multifaceted,
hierarchical model of self-concept, although they were able to identify no single
instrument that could differentiate among the broad academic, social, and physical
domains that they posited. More recently, researchers have developed self-concept
instruments to measure specific domains that are at least loosely based on an
explicit theoretical model such as the Shavelson et al. model and then used factor
analysis to support these a priori domains (see review by Marsh, 1990). The

Herbert W. Marsh is with the Faculty of Education at the University of Western

Sydney, Macarthur, P.O. Box 555, Campbelltown, NSW 2560, Australia. Robyn Suther-
land Redmayne is with the Faculty of Education at the University of Sydney, Sydney,
NSW 2006, Australia.
44 / Marsh and Redmayne

Self-Description Questionnaires (SDQ; Marsh, 1990) were specifically designed

to test the Shavelson et al. (1976) model and provide the strongest support for
it. Factor analyses have consistently identified the scales that the SDQ instruments
are designed to measure, and construct validity studies have supported a priori
relations between SDQ responses and external criteria (Marsh, 1990). Within
this general self-concept framework, it is also reasonable to posit more detailed
hierarchies that are specific to a particular domain of self-concept such as the
academic domain (Marsh, 1990; Marsh, Byrne, & Shavelson, 1988) and, of
particular relevance to the present investigation, the physical domain (Fox &
Corbin, 1989; Sonstroem, Speliotis, & Fava, 1992).
Support for the Shavelson et al. model is based primarily on multiple
components of academic self-concept, their relation to performances in different
academic areas, and their distinctiveness from other components of self-concept
(Marsh, 1990, 1993a). There is, however, also support for both the model and
the SDQ instruments in the physical domain. Factor analyses (e.g., Marsh, 1990)
have consistently differentiated responses to the SDQ Physical Ability and Physi-
cal Appearance scales from each other and from other SDQ scales. Jackson and
Marsh (1986) demonstrated that athletic participation by high school and young
adult women was substantially related to Physical Ability self-concept, but was
substantially less correlated with nonphysical areas of self-concept. Marsh and
Peart (1988) demonstrated that physical fitness was substantially related to Physi-
cal Ability self-concept, modestly related to Physical Appearance self-concept,
and unrelated to other areas of self-concept. Marsh and Peart also contrasted
competitive and cooperative aerobics interventions. Both interventions led to
substantial increases in physical fitness. However, consistent with a priori predic-
tions, the cooperative intervention led to an increase in Physical Ability self-
concept, the competitive intervention led to a decline in Physical Ability self-
concept, and neither intervention had any substantial effect on nonphysical SDQ
scales. Marsh, Richards, and Barnes (1986a, 1986b) demonstrated that participa-
tion in Outward Bound had significant effects on those SDQ factors most relevant
to the program-particularly Physical Ability self-concept-and that the size and
pattern of these effects were stable over an 18-month follow-up. Collectively,
these studies support the construct validity of responses to the SDQ physical
scales.
Using data from the Australian Health and Fitness survey, Marsh (1993b)
related a single-item physical fitness self-concept scale to a diverse set of physical
fitness indicators. For a nationally representative sample of Australian boys and
girls aged 9 to 15, physical self-concept was significantly correlated to a variety of
components of physical fitness. Although the size of relations increased somewhat
with age, the pattern of relations was consistent for responses by boys and by girls.
Commenting on limitations of the study, Marsh specifically noted that multiple
dimensions of physical self-concept should be considered and that a multifaceted,
hierarchical structure of physical self-concept is consistent with the Marsh/Shavelson
model of self-concept, particularly their more differentiated, multidimensional model
of academic self-concept (Marsh, Byrne, & Shavelson, 1988).
Marsh (1993b) proposed that a particularly fruitful direction for such research
would be to relate a multidimensional prof11e of physical fitness indicators to a
parallel set of multidimensional physical self-conceptscales. For this purpose, Marsh
proposed the potential usefulness of the widely used Physical Self-Perception Profile
 Multidimensional Physical Self-concept / 45

(PSPP; Fox, 1990; Fox & Corbii, 1989) and the unpublished Sutherland and Marsh
(1982) Physical Ability and Self Description Scale, which is the basis of the present
investigation.
The PSPP (Fox & Corbin, 1989; Sonstroem, Speliotis, & Fava, 1992)
measures four physical subdomains (Bodily Attractiveness, Sports Competence,
Physical Strength, and Physical ConditioninglExercise)as well as a global Physi-
cal Self-Worth scale. Factor analyses identified the four subdomains (although
the global physical scale was not included in these factor analyses) and the self-
concept responses predicted degree and type of physical activity involvement.
Also, the pattern of relations among the self-concept scales supported their
hierarchical model of physical self-concept and provided further support for the
Shavelson et al. (1976) model. Fox and Corbin (1989) noted, however, that the
Physical Attractiveness scale was not substantially related to physical activity,
although it was more strongly correlated with both their Physical Self-Worth
scale and a general Self-Esteem scale (see also Marsh, 1990). Fox and Corbin
focused primarily on the internal structure of physical self-concept, and tests of
the divergent validity were not emphasized.
Potentially useful directions for future research with the PSPP and other
multidimensional physical self-concept scales include stronger tests of (a) the hierar-
chical model based on hierarchical confiiatory factor analyses (e.g., Marsh &
Hocevar, 1985), (b) the divergent validity of the physical selfconcept responses
based on relations to specific criteria logically related to each scale, and (c) relations
between their physical self-concept hierarchy and a hierarchy of physical fitness
measures. The purpose of the present investigation is to pursue these issues by
examining relations between responses to the Sutherland and Marsh physical self-
concept instrument and multiple dimensions of physical fitness.

Methods
Subjects
Subjects (Ss) were 105 girls attending one of two private girls' schools in
metropolitan Sydney. All participants were in the eighth grade and were either
13 or 14 years of age. The girls came from primarily middle- and upper-middle-
class backgrounds. After obtaining consent from the school and parents, the
physical fitness tests and the physical self-concept instrument were administered
during physical education classes. The average performances on the physical
fitness tests by these girls were well below the normative average performances
of 14-year-old girls published by Fleishman (1964).

Measures
The physical fitness tests included four tests from Fleishman's Basic Fitness
Tests (1964) and the 12-Minute Run Test recommended by Cooper (1968). The
tests, administered according to instructions presented by Fleishman and by
Cooper, were Static Strength using a hand grip dynamometer, Balance (gross
body equilibrium) based on ability to remain balanced on a 1.91 cm rail, using
their preferred foot, with hands on hips and their eyes closed; shuttle run test
(Explosive Strength/Power) in which Ss ran back and forth five times between
46 / Marsh and Redmayne

two lines 18.29 metres apart (i.e., a total of 91.44 metres); dynamic Flexibility
in which Ss bent forward without bending their knees to touch an X marked on
the floor, straightened up, and without moving their feet, twisted to the left and
touched an X located about shoulder height directly behind them for as many
cycles as possible in 20 seconds; and 12-minute run (Endurance), which was
substituted for the 600 yard runlwalk test originally recommended by Fleishman
(1964) because it was deemed a more valid test of endurance (Cooper, 1968).
The physical self-concept items that comprise the Sutherland and Marsh
(1982) instrument and its a priori scales are presented in Figure 1. Instructions,
format, and the 6-point response scale were based on the SDQII (Marsh, 1990),
as were the general Physical Ability and Physical Appearance scales. The Balance,
Flexibility, and Endurance self-concept scales were designed to match and be
most highly correlated with the corresponding fitness tests of Balance, Flexibility,
and Endurance, whereas the Strength self-concept factor was designed to corre-
spond to and be most highly correlated with the Static Strength and shuttle run
(Explosive Strength) tests. The general Physical Ability scale from the SDQII
was expected to be substantially correlated with all the physical fitness compo-
nents, but to be most highly correlated with Endurance-the most general compo-
nent of fitness. The Physical Appearance self-concept scale based on the SDQII
was not predicted to be substantially correlated to any of the fitness tests. Coeffi-
cient alpha estimates of reliability were consistently high for all six physical self-
concept scales (.84 to .92). An exploratory factor analysis-a principal axis factor
extraction with a Kaiser normalization followed by an oblique rotation--clearly
identified each of the six a priori scales in that each measured variable loaded
more substantially on its intended scale than on any other scale.

Statistical Analyses
As recommended for analyses of SDQ responses (e.g., Marsh, 1990, 1992,
1993a), all factor analyses were conducted on item-pair responses in which the
first two items in each scale (as shown in Figure 1) were averaged to form the
first item pair, the second two items were averaged to form the second item pair,
and so forth. The use of item pairs is recommended because item-pair scores are
more reliable and contain less idiosyncratic variance and because the ratio of the
number of measured variables to the number of Ss is doubled. This is particularly
important in the present investigation in which the number of Ss (N = 105) is
small in relation to the number of items (50), so that even the ratio of Ss to item-
pairs (105125) is marginal.
Confirmatory factor analyses (CFA), performed with LISREL 7 (Jore-
skog & Sorbom, 1988), were used to test the a priori factor structure underlying
the self-concept responses and physical fitness tests. In CFA the researcher posits
an a priori structure and tests the ability of a solution based on this structure to
fit the data by demonstrating that (a) the solution is well defined, (b) parameter
estimates are consistent with theory and a priori predictions, and (c) the x2
and subjective indices of fit are reasonable (Marsh, Balla, & McDonald, 1988;
McDonald & Marsh, 1990). For present purposes the Relative Noncentrality
Index (RNI) and the Tucker-Lewis Index (TLI) recommended by McDonald and
Marsh (1990) are considered. Both indices vary along a 0-to-1 continuum in
which values greater than .9 are typically taken to reflect an acceptable fit. The
 Multidimensional Physical Self-concept / 47

Physical Appearancea
I am attractive for my age.
I believe I have a nicely shaped body.
*I am ugly.
I am good looking.
I have a nice looking face.
*Nobody thinks that I'm good looking.
My body looks nice.
I'm better looking than most of my friends.
Physical Abilitya
*I am awkward at things like sport, gym, and dance.
*I hate things like sport, gym, and dance.
I can run a long way without stopping.
I enjoy things like sport, gym, and dance.
I am good at things like sport, gym, and dance.
*I try to get out of sports and physical education classes whenever I can.
*I am lazy when it comes to sports and hard physical exercise.
I am better than most of my friends at things like sport, gym, and dance.
Strength
I am stronger than most girls my age.b
*I lack the strength needed to run fast.
I am good at lifting heavy objects.
*I am weak in physical tasks.
I would do well in a test of strength.
*I am weak and have no muscles.
*I have always been weak in physical activities.
I have a lot of power in my body.
Balance
I would be successful on a test of balance.
*I fall over often when trying to balance.
I think I could hold a balanced position for a long time.
I rarely lose my balance.
*I am not a very good balancer.
When necessary I think I could keep my body in a stable position.
*I have little balancing ability.
I have a good sense of balance.
Flexibility
*I have never been able to bend, twist, or tum my body easily.
I think I am flexible enough for most sports.
*It is difficult for me to bend and move quickly in different directions.

Figure 1 -The Sutherland and Marsh (1982) Instrument: 50 items and 6 a priori
scales. Note. Asterisks denote negatively worded items that should be reverse scored.
"Thesetwo scales were from the SDQII instrument. bFor more general use, the word
girl should be replaced with people. (Continued on next page)
48 1 Marsh and Redmayne

Flexibility (Continued)
I am quite good at bending, twisting, and turning my body.
*My body is stiff and inflexible.
My body parts bend and move in most directions well.
I think I would perform well on a test measuring flexibility.
I can bend and turn my body easily in games and sports.
Endurance
*I never last long in physical activities.
I am fit and can last a long time during physical activities.
*I have poor endurance in physical tasks.
I can last a long time in most physical activities.
*I am exhausted quickly by physical tasks.
I think I could run a long way without getting tired.
I can exert myself for a long period of time if I have to.
*I have very little stamina.
I do not tire easily in physical activities.
I get tired quickly if I start to run.

Figure 1 -(Continued)

RNI contains no penalty for a lack of parsimony so that the addition of new
parameters automatically leads to an improved fit that may reflect capitalization
on chance, whereas the TLI contains a penalty for a lack of parsimony.

Results
The first-order factor structure underlying the six components of physical
self-concept and the five components of physical fitness (Table 1) was well-
defined, the goodness of fit was reasonable (RNI = .911), and all the factor
loadings relating the measured variables to their latent constructs were statistically
significant and substantial (see Table 1).
Correlations among the six components of physical self-concept were sub-
stantial, varying from .448 to 399 (Table 1). Not surprisingly,the general Physical
Ability scale from the SDQII was substantiallycorrelated with the specific compo-
nents of physical self-concept. Among the specific components, correlations
involving Endurance tended to be the highest, whereas those involving Physical
Appearance were lowest. Correlations among the five components of physical
fitness were substantially smaller, varying from .024 to .394. The largest correla-
tions involved the Endurance component of physical fimess, whereas the smallest
involved the Balance component.
Correlations between the physical self-concept and physical fitness factors
varied from -.052 to 543 (Table 1). Consistent with a priori predictions, Endur-
ance self-concept was most highly correlated with Endurance fitness (r = .643),
and Strength self-concept was most highly correlated with the Static Strength
(r = 4 2 ) and Shuttle Run (r = .441) tests. Although Flexibility self-concept
was significantly correlated with the Flexibility test (r = .21 I), it was more highly
 Multidimensional Physical Self-concept / 49

Table 1 Factor Analysis of Self-concept (SC) and Physical Fitness (Fit)

Measures: First-Order Factors

Variable Factor loadings Unique Covaf

SC Appear1
SC Appear2
SC Appear3
SC Appear4
SC Physicall
SC Physical2
SC Physical3
SC Physical4
SC Strong1
SC Strong2
SC Strong3
SC Strong4
SC Balance1
SC Balance2
SC Balance3
SC Balance4
SC Flexible1
SC Flexible2
SC Flexible3
SC Flexible4
SC Endure1
SC Endure2
SC Endure3
SC Endure4
SC Endure5
SC Fit Strongb
SC Fit Balanceb
SC Fit Shuttleb
SC Fit Flexibleb
SC Fit Endureb

Factor correlations
SC Appear -
SC Physical 455 -
SC Strong 538 825 -
SC Balance 529 449 481 -
SC Flexible 549 786 775 723 -

(Continued)
50 / Marsh and Redrnayne

Table 1 (Continued)

SC Endure 448 899 844 530 809 -

Fit Strong 183 253 442 153 300 196 -
Fit Balance -052 -014 040 097 001 014 096 -
Fit Shuttle 044 454 441 107 363 492 152 024 -
Fit Flexible -063 235 249 216 211 336 376 049 294 -
Fit Endure 134 594 401 315 489 643 221 102 437 394 -

Note. All parameter estimates are presented in standardized form without decimal
points. All estimated factor loadings, uniquenesses, and residual covariances are
statistically significant ( p < .05), as are factor correlations greater than .20. All
nontarget factor loadings were fixed at zero and were not estimated. The ability of this
model to fit the data was good, ~'(353)= 545.8, RNI = .911, TLI = 290. All
indicators are scored so that higher scores reflect higher levels of fitness and self-
concept.
"Residual covariances were fit relating the SC Appear 1 and 4, and relating SC
Strength 2 and 3. "The five physical fitness tests were treated as single-item factors
and were assigned an arbitrary, but plausible, reliability estimate of .90 by fixing the
uniqueness values to be .10 (see Joreskog & Sorbom, 1988).

correlated with the Endurance, Shuttle Run, and Static Strength tests. Balance
self-concept was not significantly related to the Balance test (r = .097), although
the Balance test was not significantly related to any other measure of fitness or
self-concept. The general Physical Ability scale was significantly correlated with
all the fitness tests except Balance, but the largest correlation was with Endurance
(r = .594). Physical Appearance self-concept was not significantly correlated
with any of the fitness tests.
The first hierarchicalmodel (Model A in Figure 2) posited a single hierarchi-
cal component of physical self-concept and a single hierarchical component of
physical fitness. The very high correlation between the general fitness and general
physical self-concept factors (r = .76) offered strong support for the convergent
validity of the physical self concept responses in relation to physical fitness.
Model A did not fit the data as well as the first-order factor (RNIs of .911 vs.
.883), but it did surprisingly well given its parsimony.
Inspection of the first-order factor suggested that there was a clear relation
between the two strength components, in addition to a general agreement based
particularly on the endurance components. In Model B (Figure 2), higher order
strength factors were posited for both the self-concept and fitness sides of the model.
Although this model was able to fit the data better than Model A (RNIs of .892 vs.
.883), inspection of LISREL's modification indices (Joreskog & Sorbom, 1988)
suggested the need for a correlation between the residual variances associated with
the Balance and Flexibility self-concept factors, as shown in Figure 2. This model
provided an acceptable fit to the data (RNI = .902), and the TLI, which took into
account model parsimony, was actually better for this model (391) than for the
first-order factor (.890). Although the correlation between the general physical self-
concept and physical fitness factors was still substantial (r = .77), the correlation
between the two strength factors (r = .45) was also large.
 Multidimensional Physical Self-concept 1 51

Model A: x2(396) = 649, RNI = .883, TLI = .872

Model B: x2(392) = 605, RNI = .902, TLI = .891

Figure 2 -Three models of relations between multiple dimensions of physical self-
concept and multiple components of physical fitness. Except for higher order factor
loadings of Balance, all parameter estimates are significant ( p c .05). RNI = Relative
Noncentrality Index. TLI = Tucker-Lewis Index. (Continued on next page)
52 / Marsh and Redmayne

Model C: ~'(392)= 601, RNI = .904, TLl = .893.

Figure 2 - (Continued)

In an alternative conceptualization of relations between physical self-con-

cept and fitness (Model C), correlations between residual first-order factors on
each side of the model were evaluated. Thus, for example, this model tested
whether there was residual variance in the physical fitness Strength component
(i.e., variance in addition to that explained by the general physical fitness factor)
that was related to the residual variance in the Strength self-concept factor. In
an initial model, six correlated uniquenesses were posited, but in Model C shown
in Figure 2, only those that were statistically significant were retained. After
partialling out variance attributable to the general factors, Endurance fitness was
significantly related to Endurance and Physical Ability self-concepts, whereas
Strength fitness was significantly related to Strength self-concept. Models B
and C were very similar in their ability to fit the data and resulted in similar
interpretations.

Discussion and Implications

The present investigation supports a multidimensional, hierarchical repre-
sentation of the physical self-concept domain. Exploratory and confirmatory
factor analyses clearly identified the six a priori physical self-concept scales.
Consistent with a hierarchical representation, the general Physical Ability scale
from the SDQII was substantially correlated with all the specific components of
physical self-concept. There were also substantial correlations between the self-
concept responses and five components of physical fitness indicators. General
(second-order) factors reflecting the two constructs were substantially correlated
(r = .76). Also, self-concepts of Endurance, Strength, and Flexibility were all
 Multidimensional Physical Self-concept / 53

significantly correlated with matching components of physical fitness (Table 1).

Physical Appearance was not posited to be correlated with any of the physical
fitness indicators, so this lack of correlation may be consistent with support for
the divergent validity of the self-concept responses (also see Fox & Corbin,
1989). The lack of correlation between the Balance self-concept and fitness may,
however, call into question the validity of responses to this component of physical
self-concept.
Interpretations of these results should be evaluated in relation to limitations
of this study, the most serious being the small, unrepresentative sample of young
adolescent girls who apparently were physically unfit. Although the implications of
this limitation are unknown, support for the construct validity of the physical self-
concept responses would probably be better for a more diverse sample of older
children. Relations between self-concept and external criteria typically become
stronger with age (Marsh, 1990, 1993a, 1993b), and physical self-concepts may
be particularly volatile for 13-year-old girls who are experiencing so many other
physiological and psychological changes associated with the onset of puberty.
Also, because the girls in this sample were young and appeared to be
physically unfit, they may have had a limited basis for forming self-perceptions
in relation to specific components such as balance and flexibility. This apparent
lack of physical fitness may also explain why the Endurance components domi-
nated the general factors of physical fitness and physical self-concept. A second
general limitation in the present investigation is the small number of physical
fitness tests that were considered. The inclusion of multiple indicators of each
component of physical fitness would have provided a stronger application of the
CFA approach, an evaluation of the construct validity of the components of
fitness that were considered for this particular sample, and tests of some alternative
explanations of the results proposed here.
It is also relevant to compare the present findings with those from the Fox
and Corbin (1989) study. Both studies provided support for a multidimensional,
hierarchical physical self-concept construct, and there is an apparent similarity
in the scales that were considered. At least superficially, 4 of 6 scales considered
here (Endurance, Strength, Physical Appearance, and Physical Ability) match
those on the Fox (1990) PSPP instrument (Physical Condition, Physical Strength,
Body Attractiveness, and Physical Self-Worth). The PSPP Physical Self-Worth
scale, however, may reflect mood-related affect as well as general self-evaluations
of physical competence. Consistent with this speculation is the fact that this is
apparently the least reliable and least stable of the five PSPP scales (Fox, 1990,
p. 9) even though the Shavelson et al. (1976) model suggests that more global
constructs should be more stable. Also, the PSPP Physical Condition scale appar-
ently reflects a combination of physical endurance like the Endurance scale
considered here and an adherenceto physical exercise. The Flexibility and Balance
scales considered here have no direct counterparts in the PSPP.
Interestingly, neither study considered a physical health scale that may be
an important aspect of physical self-concept. Although Fox and Corbin (1990)
reported moderate relations between self-reported physical activity and physical
self-concept responses, they did not pursue construct validity studies in which
their scales were related to external criteria that were specific to particular scales
as in the present investigation. Given the apparent similarity between the instru-
ments, it would be useful to conduct a multitrait-multimethod study based on
54 / Marsh and Redmayne

both instruments and, perhaps, a range of external criteria that are logically
related to the specific scales from each instrument (e.g., the Marsh, Byme, &
Shavelson, 1988, study of the academic self-concept domain).

References
Bentler, P.M. (1990). Comparative fit indices in structural models. Psychological Bulletin,
107, 238-246.
Cooper, K. (1968). Aerobics. New York: Evans.
Fleishman, F.A. (1964). The structure and measurement physical fitness. Englewood
Cliffs, NJ: Prentice Hall.
Fox, K.R. (1990). The Physical SeZf-F'erceptio Profile manual. DeKalb, IL: Northern
Illinois University, Office for Health Promotion.
Fox, K.R., & Corbin, C.B. (1989). The Physical Self-Perception Profile: Development
and preliminary validation. Journal of Sport & Exercise Psychology, 11,408-430.
Joreskog, K.G., & Sorbom, D. (1988). LISREL 7: A guide to the program and applications.
Chicago: SPSS.
Jackson, S., & Marsh, H.W. (1986). Athletic or antisocial: The female sport experience.
Journal of Sport Psychology, 8, 198-211.
Marsh, H.W. (1990). A multidimensional,hierarchical self-concept: Theoretical and empir-
ical justification. Educational Psychology Review, 2, 77-171.
Marsh, H.W. (1992). SelfDescription Questionnaire II: A Theoretical and empirical basis
for the measurement of multiple dimensions of adolescent self-concept. A test manual
and a research monograph. Sydney, Australia: University of Western Sydney,
Faculty of Education.
Marsh, H.W. (1993a). Academic self-concept: Theory measurement and research. In J.
Suls (Ed.), Psychological perspectives on the self (Vol. 4, pp. 59-98). Hillsdale,
NJ: Erlbaum.
Marsh, H.W. (1993b). Physical fitness self-concept: Relations to field and technical indica-
tors of physical fitness for boys and girls aged 9-15. Journal of Sports & Exercise
Psychology, 15, 184-206.
Marsh, H.W., Balla, J.R., & McDonald, R.P. (1988). Goodness-of-fit indices in confirma-
tory factor analysis: The effect of sample size. Psychological Bulletin, 102, 391-
410.
Marsh, H.W., Byrne, B.M., & Shavelson, R. (1988). A multifaceted academic self-concept:
Its hierarchical structure and its relation to academic achievement. Journal of
Educational Psychology, 80, 366-380.
Marsh, H.W., Hocevar, D. (1985). The application of confirmatory factor analysis to the
study of self-concept: First and higher order factor structures and their invariance
across age groups. Psychological Bulletin, 97, 562-582.
Marsh, H.W., & Peart, N. (1988). Competitive and cooperative physical fitness training
programs for girls: Effects on physical fitness and on multidimensional self-con-
cepts. Journal of Sport & Exercise Psychology, 10, 390-407.
Marsh, H.W., Richards, G., & Barnes, J. (1986a). Multidimensional self-concepts: A
long term follow-up of the effect of participation in an Outward Bound program.
Personality and Social Psychology Bulletin, 12, 475-492.
Marsh, H.W., Richards, B., & Barnes, J. (1986b). Multidimensional self-concepts: The
effect of participation in an Outward Bound program. Journal of Personality and
Social Psychology, 45, 173-187.
.LaupAs JO Ll!sla~!unaqL Aq pueIlaqlnS uAqoa 01 papnme SVM ley] uo!lmnpg 1e3!sAqd
u! ( s ~ o u o ~uo!le3npa
) jo lolaq3eg jo aasap aql ~ o smaura1!nbal
j aql jo luawIIglnj
1e!md se pa~l!urqns s!saql paqs!lqndun ue urorj a m 3 Lpnls s!yl u! pasn elea