You are on page 1of 20

59

The
British
Psychological
British Journal of Educational Psychology (2011), 81, 5977

C 2010 The British Psychological Society
Society

www.wileyonlinelibrary.com

Academic self-concept and academic


achievement: Relations and causal ordering

Herbert W. Marsh1 and Andrew J. Martin2


1
Oxford University, UK
2
University of Sydney, Australia

Background. A positive self-concept is valued as a desirable outcome in many


disciplines of psychology as well as an important mediator to other outcomes.
Aims. The present review examines support for the reciprocal effects model (REM)
that posits academic self-concept (ASC) and achievement are mutually reinforcing, each
leading to gains in the other and its extension to other achievement domains.
Method. We review theoretical, methodological, and empirical support for the
REM. Critical features in this research are a theoretical emphasis on multidimensional
perspectives that focus on specific components of self-concept and a methodological
focus on a construct validity approach to evaluating the REM.
Results. Consistent with these distinctions, REM research and a comprehensive meta-
analysis show that prior ASC has direct and indirect effects on subsequent achievement,
whilst the effects of self-esteem and other non-academic components of self-concept
are negligible. We then provide an overview of subsequent support for the generality
of the REM for: young children, cross-cultural, health (physical activity), and non-elite
(gymnastics) and elite (international swimming championships) sport.
Conclusion. This research is important in demonstrating that increases in ASC
lead to increases in subsequent academic achievement and other desirable educational
outcomes. Findings confirm that not only is self-concept an important outcome variable
in itself, it also plays a central role in affecting other desirable educational outcomes.
Implications for educational practice are discussed.

There is a revolution sweeping psychology, one that emphasizes a positive psychology


and focuses on how healthy, normal, and exceptional individuals can get the most
from life (e.g., Marsh & Craven, 2006; Seligman & Csikszentmihalyi, 2000; Vallerand
et al., 2003). Consistent with this emphasis, a positive self-concept is valued as a
desirable outcome in many disciplines of psychology such as educational, developmental,
sport/exercise, health, social, and personality psychology, as well as in a broad array
of other social science disciplines. Self-concept is regarded as a highly important and

Correspondence should be addressed to Professor Herbert W. Marsh, Department of Education, University of Oxford,
University Offices, 15 Norham Gardens Rd., Oxford OX2 6PY, UK (e-mail: herb.marsh@education.ox.ac.uk).

DOI:10.1348/000709910X503501
60 Herbert W. Marsh and Andrew J. Martin

influential factor in that it is closely associated with peoples behaviours and various
emotional and cognitive outcomes such as anxiety, academic achievement, happiness,
suicide, deficient self-esteem, etc (Branden, 1994). Self-concept enhancement is seen
as a central goal of education and an important vehicle for addressing social inequities
experienced by disadvantaged groups (see Marsh & Craven, 2006). In their model of
effective schools, Brookover and Lezotte (1979) emphasized that maximizing academic
self-concept (ASC), self-reliance, and academic achievement should be the major
outcome goals of schooling. Recognizing this role of self-concept, the Organisation for
Economic Co-operation and Development (OECD) noted that self-concepts are closely
tied to students economic success and long-term health and wellbeing (OECD, 2003,
p. 9) and play a critical part in students interest in and satisfaction at school, underpin
their academic achievement, and constitute a very influential platform for pathways
beyond school (Ackerman, 2003; Marsh, 2007; Marsh, Hau, Artelt, Baumert, & Peschar,
2006), leading Chamorro-Premuzic and Furnham (2006) to postulate that ASCs both
mediate and moderate the effects of aptitudes on learning and academic performance.
The present review addresses the role of self-concept in academic achievement and
its extension to other achievement domains. We examine theoretical, methodological,
and empirical support for the reciprocal effects model (REM) that posits ASC and
achievement are mutually reinforcing, each leading to gains in the other. We then
attend to the generality of the REM by assessing the hypothesized process in relation
to self-concept and achievement/performance in cross-cultural settings, health (physical
activity), and non-elite (gymnastics) and elite (international swimming championships)
sport. We then conclude by summarizing implications of the REM for educational
practice.

Construct definition of self-concept: A multidimensional,


hierarchical construct
Historically, self-concept measurement, theory, research, and application have been
plagued by the poor quality of both theoretical models and self-concept measurement
instruments (e.g., Shavelson, Hubner, & Stanton, 1976; Wells & Marwell, 1976; Wylie,
1979). In an attempt to remedy this situation, Shavelson et al. (1976) reviewed existing
self-concept research and instruments, proposed a new theoretical model of self-concept,
and provided a blueprint for the development of a whole new generation of multidimen-
sional self-concept instruments (see review by Marsh & Hattie, 1996). Self-concept,
broadly defined by Shavelson et al. (1976), is a persons self-perceptions formed through
experience with and interpretations of ones environment. Self-concept according to
Shavelson et al. is multifaceted and hierarchically organized with perceptions of personal
behaviour in specific situations at the base of the hierarchy, inferences about self in
broader domains (e.g., social, physical, and academic) at the middle of the hierarchy,
and a global self-concept (also known as self-esteem) at the apex.

Theoretical background: A multidimensional, hierarchical model


of self-concept
Definition of self-concept and self-esteem
Based upon their extensive review of self-concept theory and research, Shavelson
et al. (1976) noted the plethora of theoretical definitions of self-concept and the
Academic self-concept 61

potential confusion between self-concept and self-esteem. Based on their integration of


different theoretical models, they defined self-concept to be a persons self-perceptions
that are formed through experience with and interpretations of ones environment.
They emphasized the importance of social influences and self-attributions, and asserted
that although self-concept is a hypothetical construct, it can nonetheless be useful in
explaining and predicting behaviour. Extending upon this, they suggested that behaviour
and self-perceptions have reciprocal relations one basis for the REM that is the emphasis
here.
Particularly relevant to the present investigation is the distinction between self-
concept and self-esteem. This distinction has caused much confusion and controversy
as informal interpretations in the lay population take the two terms to be synonymous.
Particularly since the development of Shavelson et al. (1976) model, researchers (e.g.,
Blascovich, & Tomaka, 1991; Hattie, 1992; Marsh, 2007) have viewed general self-esteem
as a global and relatively stable construct, reflecting the broad view that an individual
has about him or herself. Marsh (2007) argued that self-esteem items such as those on
the widely used Rosenbergs Self-Esteem instrument (1965) are constructed so that they
do not refer to any specific domain. Historically, some theoretical models distinguished
between self-esteem as the evaluative component of self-concept whilst self-concept
was posited to be descriptive. Shavelson et al. (1976) addressed this issue, arguing
that self-concept has both a descriptive and an evaluative aspect such that individuals
may describe themselves (I am happy) and evaluate themselves (I do well in sports).
Evaluations can be made against an absolute ideal (e.g., the 5 minute-mile), the relative
performance of others, a personal, internal standard (a personal best), or other standards
of comparison (e.g., expectations of others). Following Shavelson et al., it is generally
accepted that self-concept is both descriptive and evaluative (e.g., Byrne, 1996a, b; Marsh,
2007) so that this is not a useful distinction between self-concept and self-esteem. In
the context of the Shavelson et al. multidimensional, hierarchical model of self-concept,
self-esteem is the global construct at the apex of the hierarchy, whilst self-concept refers
to specific components within this model (e.g., ASC, physical self-concept, social self-
concept). In this sense, we treat the terms global self-concept, self-esteem, and global
self-esteem as synonymous. This usage is somewhat analogous to the use of IQ as a
term for general or global intelligence that appears at the hierarchy of multidimensional,
hierarchical models of intelligence (e.g., Vernon, 1950). As the Shavelson et al. (1976)
multidimensional, hierarchical model has been so important in subsequent theoretical
and methodological advances in self-concept research, we now consider it in more
detail.

A multidimensional, hierarchical model of self-concept


The distinction between self-concept and self-esteem is also fundamental to under-
standing the distinction between multidimensional and unidimensional perspectives
to self-concept. Unidimensional perspectives emphasize a single, global domain of self-
concept, typically referred to as self-esteem. Multidimensional perspectives emphasize
multiple, relatively distinct components of self-concept. Historically, a unidimensional
perspective dominated self-concept research. Importantly, these two perspectives are
both consistent with a multidimensional, hierarchical model. The relatively distinct
domains of self-concept support its multidimensionality whilst the hierarchical aspect is
consistent with a focus on self-esteem. However, subsequent factor analytic research
(e.g., Marsh, Byrne, & Shavelson, 1988; Marsh & Hattie, 1996) showed that the
62 Herbert W. Marsh and Andrew J. Martin

hierarchical aspect of the multidimensional, hierarchical model proposed by Shavelson


et al. (1976) was much weaker than originally hypothesized. In particular, specific
components of self-concept were more differentiated and less highly correlated with
each other than anticipated, so that much of the variance in domain-specific factors
of self-concept could not be explained in terms of higher-order self-concept factors
or self-esteem. Thus, for example, the hierarchy of self-concept domains (with self-
esteem at the apex) is much weaker than the hierarchy of abilities (with IQ at
the apex).
Marsh and Craven (2006) reported that the acceptance of a multidimensional rather
than a unidimensional perspective of self-concept varies substantially across social
science disciplines and within subdisciplines in psychology. However, its broadest
acceptance and strongest support comes from educational psychology with its focus on
ASC and its relation to academic achievement, school grades, student learning, and other
academic outcomes. Thus, Marsh and Craven (2006; also see Byrne, 1996a, b; Marsh,
1993) reviewed a large body of research showing that diverse academic outcomes were
systematically related to ASC but nearly unrelated (or even negatively related) to global
self-esteem and other non-academic components of self-concept.
This extreme multidimensionality and domain specificity of self-concept was convinc-
ingly demonstrated by factor analysis of adolescent responses to a recent adaptation of
the multidimensional Self-Description Questionnaire (SDQ II; Marsh, Trautwein, L udtke,
Koller, & Baumert, 2006). The 17 self-concept factors that the instrument was designed to
measure were clearly identified and the average correlation among the 17 self-concept
factors even after controlling for unreliability was only .14. They found a well-
defined multivariate pattern of relations between the 17 self-concept factors, personality
constructs (e.g., Big Five personality factors, positive and negative affect, life satisfaction),
and academic criteria (e.g., school grades, coursework selection in different school
subjects). Consistent with theory and previous research, math and verbal self-concepts
were somewhat negatively related to each other and this extreme domain specificity
was reflected in the systematic and substantial relations with academic criteria measures,
whilst non-academic components were nearly unrelated to the achievement measures.
For example, math self-concept was substantially related to math school grades (r =
.71), math standardized achievement test scores (r = .59), and taking advanced math
courses (r = .51). In contrast, the academic outcomes were nearly unrelated to global
self-esteem (rs ranging from .03 to .05) as well as nine other non-academic domains
of self-concept. Although specific components of self-concept explained substantial
amounts of variance in the personality factors, very little variance was uniquely due to
self-esteem. This highly differentiated multivariate pattern of relations argues against the
unidimensional perspective of self-concept that is still prevalent in some disciplines (for
further discussion, see Marsh, 2007; Marsh & Craven, 2006).

Methodological background: A construct validity approach


Following from the Shavelson review, self-concept researchers (e.g., Byrne, 1996a, b;
Marsh & Hattie, 1996; Wylie, 1989) have routinely evaluated responses to self-concept
instruments through the application of: (a) confirmatory factor analysis (CFA) to evaluate
the structure of self-concept; (b) structural equation models (SEMs) to relate self-
concept to other constructs; and (c) multitraitmultimethod analyses to establish the
convergent and discriminant validity of self-concept responses. Early research based
Academic self-concept 63

on the SDQ instruments provided strong support for the multidimensionality of self-
concept responses (e.g., Marsh, Smith, Barnes, & Butler, 1983; see reviews by Byrne,
1996a, b). In support of a multidimensional perspective, this research also showed that
the proposed hierarchy was weak and that the specific components of self-concept
(e.g., social, academic, physical, emotional) were highly differentiated (Marsh & Craven,
1997).
Important advances in self-concept research have come through the systematic
application of a construct validity approach. As a hypothetical construct, self-concept
is best understood through investigations of construct validity. The within-construct
aspects of construct validity examine the relations between self-concept domains, while
between-construct studies attempt to establish the relationship between the multiple
dimensions of self-concept and a host of other constructs (Shavelson et al., 1976). Marsh
(2007) noted that The essence of the construct validity approach is to look for areas of
convergence and non-convergence in measures of the same construct across multiple
methods: multiple indicators, multiple outcomes, multiple independent variables,
multiple methodologies, multiple analytical approaches, and multiple settings (p. 81).
In this regards, a particularly strong basis for testing the convergent and discriminant
validity of self-concept interpretations is to show that relevant indicators of achievement
are substantially related to ASC but almost unrelated to self-esteem, as shown by Marsh,
Trautwein, et al. (2006) and discussed earlier. In our review, we extend this test of
convergent and discriminant validity to evaluation of longitudinal relations between
self-concept and achievement.

Self-concept in the context of other psycho-educational factors


The importance of ASC in educational research was also highlighted by results of
OECD-Programme for International Student Assessment (PISA) Students Approaches
to Learning instrument (SAL; Marsh, Hau, et al., 2006). Through a rigorous process of
selecting educational psychologys most useful affective constructs, it was constructed to
measure 14 factors that assess self-regulated learning strategies, self-beliefs, motivation,
and learning preferences. Marsh, Hau, et al. evaluated SAL responses from nationally
representative samples of approximately 4,000 15-year-olds from each of 25 countries
(N = 107,899). Across this set of 14 factors, mathematics and verbal achievement were
consistently more strongly correlated with the three (math, verbal, and academic) self-
concept measures than any of the other SAL constructs. Furthermore, formal tests
of factorial invariance showed that the pattern, direction and relative sizes of these
correlations were relatively invariant across the 26 countries. Whilst these results support
the importance of ASC in educational settings, it is important to reiterate that correlations
based on a single wave of data as in PISA study do not provide any basis for inferring
causality. To address this issue, we turn to longitudinal studies specifically designed to
evaluate the causal ordering of ASC and performance.

Causal ordering of self-concept and academic performance


Calsyn and Kenny (1977) contrasted self-enhancement and skill development models
of the self-concept/achievement relation. According to the self-enhancement model,
self-concept is a primary determinant of academic achievement, thus supporting
the self-concept enhancement interventions explicit or implicit in many educational
64 Herbert W. Marsh and Andrew J. Martin

programmes (e.g., Hattie, Marsh, Neill, & Richards, 1997; Marsh & Peart, 1988; Marsh
& Richards, 1988; Marsh, Richards, & Barnes, 1986). In contrast, the skill development
model implies that ASC emerges principally as a consequence of academic achievement
so that ASC is enhanced by developing stronger academic skills. Byrne (1984) proposed
three criteria that studies addressing such issues must satisfy: (a) a statistical relationship
must be established, (b) a clearly established time precedence must be evident, and (c) a
causal model must be tested using appropriate statistical techniques such as use of SEMs.

Reciprocal effects model


A question commonly posed is, Which comes first ASC or academic achievement?
Not surprisingly, eitheror answers to this question are too simplistic and a growing
body of research supports a REM in which ASC both affects and is affected by academic
achievement (Marsh, 1990, 2007; Marsh, Byrne, & Yeung, 1999; Marsh & Craven, 2006;
Marsh & Scalas, in press). Theoretically, as clearly articulated by Shavelson et al. (1976)
and many others, prior academic accomplishments are important in the formation of
subsequent ASC. Hence, it makes no theoretical sense to argue that this linkage does not
exist. Rather, the critical issue is whether the linkage from self-concept to subsequent
achievement also exists.
Most causal ordering studies rely on longitudinal panel data in which both self-
concept and achievement are measured on at least two occasions (i.e., a two-wave
two-variable design) and preferably three or more. More recent developments in the
application of SEM have evolved for the analysis of such longitudinal panel designs.
Figure 1 presents a prototypical REM designed to test the causal ordering of ASC and
achievement. The critical issue is whether there are statistically significant paths leading
from prior self-concept to subsequent achievement (in support of self-enhancement
predictions) and from prior achievement to subsequent self-concept (in support of skill
development predictions). Support for the REM requires that both sets of paths are
statistically significant, but from the perspective of self-concept theory and practice, the
linkages from prior self-concept to subsequent achievement are particularly important.
In the introduction of the REM, Marsh (1990) tested the causal ordering of ASC and
academic achievement with data from the large, nationally representative US Youth
in Transition (YIT) study (Figure 2). Data were considered from Times 1 (early 10th
Grade), 2 (late 11th Grade), 3 (late 12th Grade), and 4 (1 year after normal high school
graduation). Three latent constructs were considered: academic ability inferred on the
basis of four standardized test scores, ASC inferred from self-report responses, and school
grades. Of particular importance are the effects of latent constructs in one wave on latent
constructs in subsequent waves (Figure 2). At T2, ASC is influenced by academic ability
and T1 ASC, but not T1 grades. At T2, school grades are influenced both by T1 ASC and
by T1 school grades. Similarly, school grades at T3 are influenced significantly both by
T2 ASC and by T2 grades. ASC at T4 was influenced significantly by ASC at T2 (there was
no T3 ASC measure), but not by T3 school grades. The findings provide strong support
for the effect of prior self-concept on subsequent school grades as they spanned two
intervals.
Based in part on this study, Marsh (2007; Marsh & Craven, 1997, 2006) provided
an overview of important design features for the ideal REM studies: (a) measure ASC
and academic achievement (school performance, standardized test scores, or preferably
both) at least twice (i.e., a two-wave study) and preferably more frequently; (b) infer all
latent constructs on the basis of multiple indicators; (c) consider a sufficiently large and
Academic self-concept 65

T1-ASC T2-ASC T3-ASC

T1-ACH T2-ACH T3-ACH

Figure 1. Prototype causal-ordering model for testing self-enhancement, skill development, and
reciprocal-effects models. Notes. In this full-forward, multiwave, multivariable model, multiple indicators
of academic self-concept (ASC) and achievement (ACH) are collected in three successive waves (T1,
T2, and T3). Each latent construct (represented by ovals) has paths leading to all latent constructs
in subsequent waves. Within each wave, ASC and ACH are assumed to be correlated; in the first
wave, this correlation is a covariance between two latent constructs, and in subsequent waves, it is a
covariance between residual factors. Curved lines at the top and bottom of the figure reflect correlated
uniquenesses between responses to the same measured variable (represented by boxes) collected on
different occasions. Paths connecting the same variable on multiple occasions reflect stability (the solid
black paths), but these coefficients typically differ from the corresponding test-retest correlations (which
do not include the effects of other variables). Light gray arrows reflect effects of prior achievement
on subsequent academic self-concept, whereas dark gray arrows reflect the effects of prior academic
self-concept on subsequent achievement. Adapted with permission from Marsh (2007).

diverse sample to justify the use of CFA and the generality of the findings; and (d) fit the
data to a variety of CFA models that incorporate measurement error and test for likely
residual covariation among measured variables. If both test scores and school grades
are collected in the same study, then they should be considered as separate constructs
unless there is empirical support for combining them to form a single construct. If any of
the latent constructs are measured with a single measured variable, an a priori estimate
of reliability should be used and the sensitivity analysis should be conducted on the full
model to determine the generality of the conclusions in relation to reliability estimates
and potential correlated uniquenesses.
The reciprocal pattern of relations between self-concept and performance posited
in the REM is also represented in many other theoretical accounts of related self-belief
constructs (e.g., Bandura, 1997; Byrne, 1996a, b; Eccles & Wigfield, 2002; Harter, 1998;
Hattie, 1992; Skaalvik, 1997; Valentine & DuBois, 2005; Wigfield & Eccles, 2002). Thus,
for example, expectancy-value theorists (Eccles & Wigfield, 2002) hypothesize academic
self-beliefs to be a function of prior academic successes and to affect subsequent
academic success directly or indirectly through their influence on other mediating
constructs. More generally, in their theoretical review and meta-analysis of empirical
research, Valentine and DuBois concluded that reciprocal effects relating academic self-
beliefs and achievement are consistent with theories of learning and human development
that view the self as a causal agent (e.g., Bandura, 1997; Carver & Scheier, 1981; Deci &
Ryan, 1985).
66 Herbert W. Marsh and Andrew J. Martin

.10
.10
9 .10
.13
5
13
.41 .28 .50 .39 .95
.95 .95
1 2 3 4

.57
Grades-T2 .46
.77 .85 .71 .78
Grades-T1
.22 Grades-T3
.57
Ability-T1 .42 .22 .17
.20
.38
.17 ASC-T4
.66
ASC-T1 .67 ASC-T2
.76 .54
.71 .78 .52 .74 .81 .58
14 15
6 7 8 10 11 12 .23 .68
.32 .25 .82 .31 .24 .74 .04 .39
.45
.04 .18
.37
.04

Figure 2. Structural equation model of results from a longitudinal-panel-design study relating ASC on
multiple occasions (T1, T2, T3, and T4). Notes. The model shows the standardized effects of ASC on
subsequent school grades and ASC. The 13 boxes represent multiple indicators of the latent constructs
(the ovals). Straight lines connecting the latent constructs, represent path coefficients. Non-significant
paths are excluded for purposes of clarity. The curved lines represent correlated residuals. Of particular
relevance are paths (highlighted) leading from prior ASC to future grades and those leading from prior
grades to future ASC. Adapted with permission from Marsh (1990, p. 650).

Meta-analysis of studies of the REM


The strongest support for the generalizability and robustness of the REM comes
from the extensive meta-analysis conducted by Valentine and colleagues (Valentine &
DuBois, 2005; Valentine, DuBois, & Cooper, 2004). Whereas REM studies have mostly
focused specifically on measures of ASC, Valentine and colleagues considered self-
belief constructs more generally. In their meta-analysis, they began with a thorough
search of all published and unpublished research that allowed them to determine
the relation between T1 self-beliefs and T2 achievement whilst controlling for T1
achievement. This resulted in a total of 55 publications, including of 60 independent
samples, and 282 separate effect sizes. They found that the effect of prior self-beliefs
on subsequent achievement after controlling for the effects of prior achievement was
highly significant overall and positive in 90% of the studies in their meta-analysis. These
results led Valentine and Dubois to conclude that their meta-analysis provided clear
support for the REM and that any claims that prior self-beliefs are unrelated or detri-
mental to subsequent student achievement is inconsistent with the results of empirical
research.
A particular strength of meta-analysis is its ability to evaluate the generalizability of
the results across different study characteristics, something that is typically not possible
in a single study. Valentine and colleagues (Valentine & DuBois, 2005; Valentine et al.,
Academic self-concept 67

2004) considered a wide variety of potential moderators of the REM effects: year the
study was published/reported; base year of data collection; sample size of the study;
stability of the achievement measure (i.e., T1T2 stability coefficient); reliability of the
self measure; the number of variables used as controls in the analysis; whether the effect
size was from an analysis of manifest or latent variables; use of a convenience sample
versus random selection from a known population; age of students; type of achievement;
time interval between the collection of T1 and T2 measures; and country from which the
sample of students came. However, none of these potential moderators had a significant
effect in the size of the REM. These meta-analysis results provide compelling support for
the robustness and generalizability of the REM in a way that could not be achieved on
the basis of any single primary study.
One design characteristic, the globality of the self-belief measure, did have a
substantial moderating effect on the REM effect sizes. In support of a multidimensional
perspective that is a theoretical underpinning for the REM, Valentine and colleagues
(Valentine & DuBois, 2005; Valentine et al., 2004) found that the effects of prior self-
beliefs were significantly stronger when the self-belief measure was based on academic
self-beliefs rather than on global measures such as self-esteem, and when the self-belief
and achievement measures were matched in terms of subject area (e.g., mathematics
achievement and math self-concept). In particular, they reported little evidence of any
effects of global or generalized self-beliefs (e.g., self-esteem) on academic achievement.
Valentine and colleagues (Valentine & DuBois, 2005; Valentine et al., 2004) also
found that the strength of the REM was weaker in studies where students experienced
a normative school transition (e.g., from elementary school to middle school). In
particular, the effects of prior self-beliefs (collected prior to the transition) on subsequent
achievement (collected after the transition) were smaller than in studies where there
was no transition.
In summary, this meta-analysis of self-belief research provides clear support for
the REM, the robustness and generalizability of the effect, the theoretical focus on
a multidimensional perspective, and the methodological focus on a construct validity
approach that has been central to REM studies and self-concept research more generally.

Extension of the REM


In their review of REM research, Marsh et al. (1999; also see Valentine et al., 2004)
provided clear support for reciprocal effects of ASC and achievement. With the hindsight
of 15 years experience, Marsh et al. offered commentary on potential methodological
issues and directions for further research. Here, we summarize some subsequent research
in response to needs identified by Marsh et al. (also see Marsh & Craven, 2006).

Cross-cultural generalizability
Partly in response to Marsh et al. (1999), subsequent research demonstrated that support
for the REM of ASC and achievement generalized to different cultural/national settings
in a large nationally representative sample of Hong Kong students (Marsh, Hau, & Kong,
2002) and large samples of East and West German students at the time of the fall of the
Berlin Wall (Marsh & K oller, 2003; Marsh, K
oller, & Baumert, 2001). Support for the
generalizability also comes from research based on French-speaking Canadian primary
students (Guay, Marsh, & Boivin, 2003), German high school students (Marsh, Trautwein,
L
udtke, K oller, & Baumert, 2005), and 487 high school students (Grades 7, 8, and 9)
68 Herbert W. Marsh and Andrew J. Martin

from mainland China (Yeung & Lee, 1999). More generally, in their meta-analysis of
REM studies, Valentine and colleagues (Valentine & DuBois, 2005; Valentine et al., 2004)
considered the country from which the sample was drawn as a moderator variable.
However, they found that support for the REM did not differ as a function of country.
Whilst this research provides cross-national and cross-cultural support for the REM, we
note that the majority of the research comes from Western and industrialized countries
so that it is premature to claim that support for the REM is universal.

Developmental perspectives on the REM


Based on developmental theory, some researchers have suggested that the reciprocal
pattern of relations in support of the REM found with adolescents is unlikely to generalize
to preadolescents (see Wigfield & Karpathian, 1991). However, both the review of REM
studies by Marsh and colleagues (Marsh et al., 1999) and the meta-analysis by Valentine
and colleagues (Valentine & DuBois, 2005; Valentine et al., 2004) concluded that there
was not sufficient good quality research with young children to support this conclusion.
Guay et al. (2003) addressed this issue about developmental trends in REM research.
They used a multicohortmultioccasion design, a methodological approach that is
especially well-suited to address this issue (as depicted in Figure 3 of the present
investigation). In particular, they considered responses by students who at T1 were
in Grades 24 (i.e., three age cohorts 810 years of age). Responses for all three
cohorts were then collected annually for the next 3 years (i.e., the three measurement
occasions). They found strong support for the REM for all three age cohorts, and
these results were reasonably invariant when rigorously tested with multigroup tests
of invariance across the three age cohort groups. This multicohortmultioccasion
design is particularly appropriate for evaluating the development of the REM, but
there is need for further research that considers different age groups and longer time
intervals.

Mediating variables: The role of intrinsic motivation


Implicit in the rationale of the REM is the largely untested assumption that the effect
of prior self-concept on subsequent achievement is mediated by student characteristics
such as increased conscientious effort, persistence in the face of difficulties, enhanced
intrinsic motivation, academic choice, and coursework selection (see Marsh et al., 1999).
Thus, for example, Marsh and Yeung (1997a, b) found that coursework selection partially
mediated the effects of prior ASC in a specific school subject on subsequent achievement
in the same subject (e.g., high math self-concept led to taking more advanced math
courses, which led to higher levels of math achievement). Indeed, Marsh and Yeung
found that whereas ASC, academic achievement, and coursework selection were all
highly correlated, prior ASC was a much better predictor of subsequent coursework
choice than was prior academic achievement.
Pursuing this line of thinking, Marsh et al. (1999) suggested that intrinsic motivation
might serve this mediating role. Marsh et al. (2005) took on the methodological challenge
of testing this suggestion with SEMs of longitudinal data based on two large, nationally
representative samples of German high school students. They expanded the typical
causal ordering REM model to include academic interest and two different measures
of achievement (grades and achievement test scores) as well as ASC. In both studies,
they found clear support for the REM based on ASC and achievement, demonstrating
Academic self-concept 69

Second grade Third grade Fourth grade Fifth grade Sixth grade
.25*

.44* .57*
T1-ASC T2-ASC T3-ASC
.26* .05 .05
First cohort
.18* .13* .03
T1-ACH T2-ACH T3-ACH
.67*a .46*
.31*
.25*

.44* .57*
T1-ASC T2-ASC T3-ASC
.26* .05 .05
Second cohort
.18* .13* .03
T1-ACH T2-ACH T3-ACH
.67*a .46*
.31* .25*

.44* .58*
T1-ASC T2-ASC T3-ASC
.05
.26* .05
Third cohort
.22* .14* .03
T1-ACH T2-ACH T3-ACH
.52*b .41*
.33*

Figure 3. Test of reciprocal effects model across multiple cohorts of young children in Grades 2,
3, and 4, tested in each of three successive years (T1, T2, T3). Note. ASC, academic self-concept;
ACH, academic achievement. Adapted with permission from Marsh (2007). Path coefficients with are
statistically significant at p < .05.

that the effect of prior math self-concept was substantial for subsequent math school
performance as well as for math test scores. Extending previous results, prior self-concept
also significantly influenced subsequent measures of academic interest beyond the effects
of earlier measures of school performance, achievement test scores, and academic
interest. However, prior academic interest had only a small effect on subsequent ASC
and little or no effect on either school performance or test scores beyond what could be
explained by ASC. Thus, the reciprocal effects of ASC and achievement were mediated
by academic interest only to a small degree, but any effects of academic interest on
achievement were substantially mediated by ASC. More strongly than previous SEM
research, the results demonstrated the positive effects of ASC on academic interest as well
as achievement based both on standardized test scores and school-based performance
measures.
70 Herbert W. Marsh and Andrew J. Martin

Generalizability to other self-concept domains


Although there is a growing body of research based on ASC and academic achievement,
Marsh et al. (1999) noted that there were few tests of the REM in non-academic
domains. Existing research shows that there are little or no reciprocal effects of academic
accomplishments and non-academic domains of self-concept. This, of course, does not
preclude the possibility that non-academic domains of self-concept do have reciprocal
effects with competence and accomplishments in the matching domain. Indeed, this is a
natural extension of the REM and also results from the meta-analysis research described
earlier.
Sport is well suited to test the generalizability of the REM to non-academic settings
because feedback about ones sport performance comes largely from social comparison
with the performances of peers (e.g., competitors), direct feedback from peers, and a
variety of sources that are not directly related to school. There have been several tests of
the REM in the physical domain for general populations and elite athletes. Marsh, Chanal,
Sarrazin, and Bois (2006) demonstrated REM support for gymnastics self-concept and
performance measures collected before and after a 10-week gymnastics programme. As
predicted by the REM, the results in this short longitudinal study showed that gymnastics
self-concept and gymnastics performance were both determinants and consequences of
each other.
Recognizing the critical importance of health-related physical activity in children and
adolescents, Marsh, Papaioannou, and Theodorakis (2006) adapted the REM in a study
of the causal ordering of physical self-concept and exercise behaviour. The study was
based on a large sample of primary and secondary Greek physical education students
(2,786 students, 200 classes, 67 teachers) and data collected early (T1) and late (T2) in
the school year. There was clear support for the REM as there were significant effects
of T1 physical self-concept on T2 exercise behaviour and T1 exercise behaviour on T1
physical self-concept. Physical self-concept was both an effect and a cause of exercise
behaviour.
Adding a developmental perspective to this research, Marsh, Gerlach, Trautwein,
Ludtke, and Brettschneider (2007) tested the generalizability of the REM with pre-
adolescent children in the physical domain. They used longitudinal data for young
boys and girls (N = 1,135; M age = 9.67 years) to show that physical self-concept
is both a cause and a consequence of physical accomplishments. After controlling for
prior physical performance (physical performance-based tests and teacher assessments in
Grade 3), physical self-concept had a positive effect on subsequent physical performance
in both Grade 4 and subsequently in secondary school. Coupled with previous REM
research based largely on studies of adolescents in the academic domain, this study
supported the REMs generalizability over gender, self-concept domain, pre-adolescent
ages, and the transition from primary to secondary school.
Following from this, we might ask how well does the REM generalize to elite
athletes? Marsh and Perry (2005) tested the effects of sport self-concept on subsequent
performance for 270 elite swimmers from 30 countries participating in the Pan Pacific
Swimming Championships and the World Short Course Championships. Whereas subse-
quent championship performance was highly related to prior personal best performances
(r = .90), SEMs demonstrated that elite athlete self-concept contributed significantly to
the prediction of subsequent championship performance, explaining approximately 10%
of the residual variance after controlling for personal best performances. Because each
swimmer typically competed in at least two different events, the authors were also able
to show that support for the REM was nearly identical for both events. In summary, good
Academic self-concept 71

athletic self-concepts contributed to winning gold medals in the top echelons of elite
sport.

The Baumeister et al. challenge to REM research and the value


of self-beliefs
Although the REM findings are now widely accepted in educational psychological
theory, research and practice, the findings have been contested in other disciplines.
In particular, in a set of highly influential reviews published in the Scientific American
and Psychological Science in the Public Interest, Baumeister, Campbell, Krueger, and
Vohs (2003, 2005) challenged the value of positive self-beliefs and, perhaps, the positive
psychology movement more generally. Baumeister et al. (2003) concluded that self-
esteem per se is not the social panacea that many people hoped it was (p. 38). Moreover,
in apparent contradiction to the REM findings, Baumeister et al. (2005) concluded that
efforts to boost peoples self-esteem are of little value in fostering academic achievement
or preventing undesirable behaviour (p. 84).
In response to this challenge, Marsh and Craven (2006) argued that Baumeister
et al. had taken an overly narrow focus: adapting a unidimensional perspective that
included only studies based on self-esteem (excluding all ASC studies considered here)
and only considering studies conducted prior to 1990 that did not incorporate current
statistical methodology and conceptual advances in self-concept theory. Furthermore,
Marsh and Craven argued that from a multidimensional perspective, it is logical that there
are essentially no reciprocal links between academic achievement and self-esteem (as
reported by Baumeister et al.), whereas consistent reciprocal relations existed between
ASC and achievement (as reported by Marsh and Craven). Consistent with this point of
view, the meta-analysis conducted by Valentine and DuBois (2005) indicated that the
effect on subsequent school performance was stronger for academic self-beliefs than for
global self-beliefs (such as global self-esteem). In summary, the apparent controversy is
easily resolved by placing it within an appropriate theoretical and statistical perspective.
Ironically, there was almost no overlap in the studies considered by Marsh and Craven
(2006; as well as those in the Valentine et al., 2004) meta-analysis, and those considered
by Baumeister et al. (2003, 2005). However, both Marsh et al. and Baumeister et al. cited
classic studies based on the YIT database as providing particularly strong support of
their respective claims. The YIT research emphasized by Baumeister et al. was an early
study by Bachman and OMalley (1977) that examined self-esteem but not ASC, whereas
Marsh and Craven (2006) emphasized the Marsh (1990) study summarized earlier (see
Figure 2) that included ASC but not global self-esteem. In a definitive test of these counter
claims, Marsh and OMara (2008) reanalysed this YIT data including both self-esteem
(emphasized by Baumeister et al.), ASC (emphasized by Marsh & Craven), and post-
secondary educational attainment (emphasized by Bachman & OMalley) based on all
five waves of data, using stronger statistical methods than used in any of the previous
studies. Consistent with REM results, they found positive reciprocal effects between ASC
and GPA, but also found positive reciprocal links relating ASC and educational attainment
not previously reported in this research literature. Consistent with the Baumeister et al.
review (and also meta-analyses by Valentine and colleagues), they found only weak and
inconsistent linkages between self-esteem and either GPA or attainment. These new
results provided clear support for Marsh and Cravens proposed rapprochement in their
debate with Baumeister et al., integrating apparently contradictory results into a single
72 Herbert W. Marsh and Andrew J. Martin

theoretical framework based on a multidimensional perspective of self-concept and


supporting the REM.

Implications and directions for further research


The results of causal modelling studies provide a clear affirmative answer to the question
Do changes in ASC lead to changes in subsequent academic achievement? This research
is important in that it has established that increases in ASC lead to increases in subsequent
academic achievement and other desirable educational outcomes. Hence, not only is
ASC an important outcome, but it also plays a central role in mediating the effects of
other desirable educational outcomes. It is important to emphasize that the direction of
causality between ASC and achievement also has very important practical implications
for educators. If the direction of causality were from ASC to achievement (the self-
enhancement model), then teachers might be justified in placing more effort into
enhancing students self-concepts rather than fostering achievement. On the other
hand, if the direction of causality were from achievement to self-concept (the skill
development model), then teachers should focus primarily on improving academic skills
as the best way to improve self-concept. In contrast to both these apparently overly
simplistic (eitheror) models, the REM implies that ASC and academic achievement are
reciprocally related and mutually reinforcing. Improved ASCs leads to better achievement
and improved achievement leads to better ASCs. For example, if teachers enhance
students ASCs without improving achievement, then the gains in self-concept are
likely to be short-lived. However, if teachers improve students academic achievement
without also fostering students self-beliefs in their academic capabilities, then the
achievement gains are also unlikely to be long lasting. If teachers focus on either one
of these constructs to the exclusion of the other, then both are likely to suffer. Hence,
according to the REM, teachers should strive to improve simultaneously both ASC and
achievement.
Research reviewed here suggests a number of fruitful directions for further research.
The meta-analysis by Valentine and colleagues (Valentine & DuBois, 2005; Valentine
et al., 2004) suggested that support for the REM was similar for standardized test scores
and school grades. However, Marsh and Craven (2006) suggested that the effects of
ASC should be stronger on school grades than on test scores. Whilst a number of
studies have evaluated the strength of the REM effects with general ASC and domain
specific measures, there is not clear consensus about which gives the strongest results.
Also, there is research on the internal/external frame of reference model showing that
the effect of prior math achievement is positive on math self-concept (consistent with
REM predictions) but negative on verbal self-concept, whilst the effect of prior verbal
achievement is positive for verbal self-concept but negative on math self-concept. Putting
together these two models might suggest that the effects of prior self-concept should
be positive for achievement in the matching domain but negative for achievement
in a contrasting domain (see Marsh & K oller, 2003). However, there has been little
research testing these counter-intuitive predictions based on the integration of these two
models. The REM implies a causal relation between priori self-concept and subsequent
achievement. Whilst the use of longitudinal data clearly provides a stronger basis for
causal inferences than cross-sectional data, trying to prove causality is always a very
tricky undertaking. As new and better methodological approaches to evaluate causal
inferences are developed, there will be stronger tests of REM predictions. In relation
Academic self-concept 73

to this issue, we also note that the Haney and Durlak (1998) meta-analysis of self-
concept interventions concluded consistent with REM inferences that interventions
specifically designed to enhance self-concept not only had significant effects on self-
concept, but also had positive effects on academic achievement. Finally, although there
is evidence for the cross-cultural validity of the REM, we note that most studies are based
on research conducted in Western and industrialized countries. There is not sufficient
research to argue for the universality of the REM.
More sophisticated methodological extensions might consider the interface of individ-
ual and context in self-concept effects. For example, cross-level multi-level models (e.g.,
Goldstein, 2003; Marsh et al., 2009; Raudenbush & Bryk, 2002) might examine the causal
effects of individual self-concept on school or class-level such as aggregate achievement.
Similarly, transactional models (Sameroff & Chandler, 1975; Sameroff, 2009) would argue
for the interplay of, for example, student ASC on teacher pedagogical self-concept (and
vice versa). Also using multi-level approaches it is possible that individual students
self-concept trajectories may differentially predict important outcomes and so multi-
level growth modelling (e.g., Holt, 2008) of self-concept over time may reveal distinct
intra- and inter-person patterns of self-concept development that affect outcomes in
distinct ways. Yet another application of longitudinal methods might consider intensive
longitudinal modelling (Walls & Schafer, 2006). Whereas much self-concept research
examines longitudinal effects separated by relatively lengthy intervals (e.g., one academic
year), intensive longitudinal modelling (e.g., ratings over the course of the day and
across days) may help uncover potential real-time casual variance attributable to
self-concept.
REM research provides a particularly appropriate methodology for evaluating causal
hypotheses that a particular psychosocial variable has a significant effect on subsequent
measures of achievement. Based on this review, it seems as if there is strong evidence for
the REM that generalizes across academic specific self-beliefs such as ASC and academic
self-efficacy (but also see Marsh, Walker, & Debus, 1991) particularly when there is a
clear match between domain specificity of the measures (e.g., math self-concept with
math achievement). However, there was no support for REM effects when the self-
beliefs were global or general measures such as general self-esteem. The juxtaposition of
these two sets of findings support both the convergent and discriminant validity of REM
predictions. Marsh et al. (2005) extended this logic to studies of academic interest and
intrinsic motivation more generally. Noting that there was clear evidence that intrinsic
motivation is correlated with academic achievement, there was almost no research that
applied the REM methodology to measures of intrinsic motivation. At least in their
study, they found that there was only limited support for REM effects between intrinsic
motivation and academic achievement, and even the small effects that they found were
largely mediated by ASC. Hence, the methodological approaches used in REM studies of
ASC can advantageously be applied to the entire spectrum of psychosocial variables that
are prevalent in educational psychology research and practice.

Conclusion
The present review has examined the role of ASC in academic achievement and
its extension to other achievement and performance domains. There is theoretical,
methodological, and empirical support for the REM positing that ASC and achievement
are mutually reinforcing, each leading to gains in the other. There is also support
74 Herbert W. Marsh and Andrew J. Martin

for the generality of the REM in developmental research, cross-cultural settings, and
health and sporting domains. Findings are relevant to researchers seeking to assess
longitudinal patterns of ASC and achievement and for practitioners seeking to enhance
the educational outcomes of children and young people outcomes that rely on domain-
specific ASC.

Acknowledgements
This research was supported in part by a grant to the first author from the UK Economic and
Social Research Council.

References
Ackerman, P. L. (2003). Cognitive ability and non-ability trait determinants of expertise. Educa-
tional Researcher, 32, 1520. doi:10.3102/0013189X032008015
Bachman, J. G., & OMalley, P. M. (1977). Self-esteem in young men: A longitudinal analysis of
the impact of educational and occupational attainment. Journal of Personality and Social
Psychology, 35, 365380.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.
Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. D. (2003). Does high self-esteem cause
better performance, interpersonal success, happiness, or healthier lifestyles? Psychological
Science in the Public Interest, 4, 144. doi:10.1111/1529-1006.01431
Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. D. (2005). Exploding the self-esteem
myth. Scientific American, 292, 8492. doi:10.1038/scientificamerican0105-84
Blascovich, J., & Tomaka, J. (1991). The self-esteem scale. In J. P. Robinson, P. R., Shaver, &
L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (Vol. I,
pp. 115160). San Diego, CA: Academic Press.
Branden, N. (1994). Six pillars of self-esteem. New York: Bantam.
Brookover, W. B., & Lezotte, L. W. (1979). Changes in schools characteristics coincident with
changes in student achievement. East Lansing, MI: Michigan State University. (ERIC Document
Reproduction Service No. ED 181 005).
Byrne, B. M. (1984). The general/academic self-concept nomological network: A review of
construct validation research. Review of Educational Research, 54, 427456.
Byrne, B. M. (1996a). Academic self-concept: Its structure, measurement, and relation to academic
achievement. In B. A. Bracken (Ed.), Handbook of self-concept: Developmental, social, and
clinical considerations (pp. 287316). New York: Wiley.
Byrne, B. M. (1996b). Measuring self-concept across the life span: Issues and instrumentation.
Washington, DC: American Psychological Association.
Calsyn, R. J., & Kenny, D. A. (1977). Self-concept of ability and perceived evaluation of others:
Cause or effect of academic achievement? Journal of Educational Psychology, 69, 136145.
doi:10.1037/0022-0663.69.2.136
Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A control theory approach
to human behavior. New York: Springer-Verlag.
Chamorro-Premuzic, T., & Furnham, A. (2006). Intellectual competence and the intelligent
personality: A third way in differential psychology. Review of General Psychology, 10, 251
267. doi:10.1037/1089-2680.10.3.251
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human
behavior. New York: Plenum Press.
Eccles, J. S., & Wigfield, A. (2002). Motivational beliefs, values, and goals. Annual Review of
Psychology, 53, 109132. doi:10.1146/annurev.psych.53.100901.135153
Goldstein, H. (2003). Multilevel statistical models (3rd ed.). London: Hodder Arnold.
Academic self-concept 75

Guay, F., Marsh, H. W., & Boivin, M. (2003). Academic self-concept and academic achievement:
Developmental perspectives on their causal ordering. Journal of Educational Psychology, 95,
124136. doi:10.1037/0022-0663.95.1.124
Haney, P., & Durlak, J. A. (1998). Changing self-esteem in children and adolescents: A meta-analytic
review. Journal of Clinical Child Psychology, 27, 423433. doi:10.1207/s15374424jccp2704
6
Harter, S. (1998). Developmental perspectives on the self-system. In N. Eisenberg (Vol. Ed.),
Handbook of child psychology (5th ed., Vol. 3, pp. 553618). New York: Wiley.
Hattie, J. (1992). Self-concept. Hillsdale, NJ: Erlbaum.
Hattie, J., Marsh, H. W., Neill, J. T., & Richards, G. E. (1997). Outward Bound and adventure
education: Out-of-class experiences that make a lasting difference. Review of Educational
Research, 67, 4387.
Holt, J. K. (2008). Modeling growth using multilevel and alternative approaches. In OConnell, A.
A., & D. B. McCoach (Eds.), Multilevel modeling of educational data (pp. 111159). Charlotte,
NC: Information Age Publishing.
Marsh, H. W. (1990). The causal ordering of academic self-concept and academic achievement:
A multiwave, longitudinal panel analysis. Journal of Educational Psychology, 82, 646656.
doi:10.1037/0022-0663.82.4.646
Marsh, H. W. (1993). Academic self-concept: Theory measurement and research. In J. Suls (Ed.),
Psychological perspectives on the self (Vol. 4, pp. 5998). Hillsdale, NJ: Erlbaum.
Marsh, H. W. (2007). Self-concept theory, measurement and research into practice: The role of
self-concept in educational psychology. Leicester: British Psychological Society.
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, 366380. doi:10.1037/0022-0663.80.3.366
Marsh, H. W., Byrne, B. M., & Yeung, A. S. (1999). Causal ordering of academic self-concept and
achievement: Reanalysis of a pioneering study and revised recommendations. Educational
Psychologist, 34, 155167. doi:10.1207/s15326985ep3403 2
Marsh, H. W., Chanal, J. P., Sarrazin, P. G., & Bois, J. E. (2006). Self-belief does make a difference:
A reciprocal effects model of the causal ordering of physical self-concept and gymnastics
performance. Journal of Sport Sciences, 24, 101111. doi:10.1080/02640410500130920
Marsh, H. W., & Craven, R. (1997). Academic self-concept: Beyond the dustbowl. In G. Phye (Ed.),
Handbook of classroom assessment: Learning, achievement, and adjustment (pp. 131198).
Orlando, FL: Academic Press.
Marsh, H. W., & Craven, R. G. (2006). Reciprocal effects of self-concept and performance from
a multidimensional perspective: Beyond seductive pleasure and unidimensional perspectives.
Perspectives on Psychological Science, 1, 133163. doi:10.1111/j.1745-6916.2006.00010.x
Marsh, H. W., Gerlach, E., Trautwein, U., L udtke, U., & Brettschneider, W.-D. (2007). Longitudinal
study of preadolescent sport self-concept and performance: Reciprocal effects and causal
ordering. Child Development, 78, 16401656. doi:10.1111/j.1467-8624.2007.01094.x
Marsh, H. W., & Hattie, J. (1996). Theoretical perspectives on the structure of self-concept. In B.
A. Bracken (Ed.), Handbook of self-concept (pp. 3890). New York: Wiley.
Marsh, H. W., Hau, K.-T., Artelt, C., Baumert, J., & Peschar, J. L. (2006). OECDs brief self-
report measure of educational psychologys most useful affective constructs: Cross-cultural,
psychometric comparisons across 25 countries [Special issue]. International Journal of
Testing, 6, 311360. doi:10.1207/s15327574ijt0604 1
Marsh, H. W., Hau, K. T., & Kong, C. K. (2002). Multilevel causal ordering of academic self-concept
and achievement: Influence of language of instruction (English compared with Chinese) for
Hong Kong students. American Educational Research Journal, 39, 727763. doi:10.3102/
00028312039003727
Marsh, H. W., & K oller, O. (2003). Bringing together two theoretical models of relations
between academic self-concept and achievement. In H. W. Marsh, R. G. Craven, & D. M.
76 Herbert W. Marsh and Andrew J. Martin

McInerney (Eds.), International advances in self research (Vol. 1, pp. 1748). Greenwich,
CT: Information Age Publishing.
Marsh, H. W., K oller, O., & Baumert, J. (2001). Reunification of East and West German
school systems: Longitudinal multilevel modeling study of the big-fishlittle-pond effect on
academic self-concept. American Educational Research Journal, 38, 321350. doi:10.3102/
00028312038002321
Marsh, H. W., Ludtke, O., Robitzsch, A., Trautwein, U., Asparouhov, T., Muthen, B., . . . Nagengast,
B. (2009). Doubly-latent models of school contextual effects: Integrating multilevel and
structural equation approaches to control measurement and sampling error. Multivariate
Behavioral Research, 44, 764802. doi:10.1080/00273170903333665
Marsh, H. W., & OMara, A. (2008). Reciprocal effects between academic self-concept, self-
esteem, achievement, and attainment over seven adolescent years: Unidimensional and
multidimensional perspectives of self-concept. Personality and Social Psychology Bulletin,
34, 542552. doi:10.1177/0146167207312313
Marsh, H. W., Papaioannou, A., & Theodorakis, Y. (2006). Causal ordering of physical self-concept
and exercise behavior: Reciprocal effects model and the influence of physical education
teachers. Health Psychology, 25, 316328. doi:10.1037/0278-6133.25.3.316
Marsh, H. W., & Peart, N. (1988). Competitive and cooperative physical fitness training programs
for girls: Effects on physical fitness and on multidimensional self-concepts. Journal of Sport
and Exercise Psychology, 10, 390407.
Marsh, H. W., & Perry, C. (2005). Does a positive self-concept contribute to winning gold
medals in elite swimming? The causal ordering of elite athlete self-concept and championship
performances. Journal of Sport and Exercise Psychology, 27, 7191.
Marsh, H. W., & Richards, G. (1988). The Outward Bound Bridging Course for low achieving high-
school males: Effect on academic achievement and multidimensional self-concepts. Australian
Journal of Psychology, 40, 281298. doi:10.1080/00049538808260049
Marsh, H. W., Richards, G., & Barnes, J. (1986). Multidimensional self-concepts: The effect of
participation in an Outward Bound program. Journal of Personality and Social Psychology,
45, 173187. doi:10.1037/0022-3514.45.1.173
Marsh, H. W., & Scalas L. F. (in press). Self-concept and learning: Reciprocal effects model between
academic self-concept and academic achievement. In E. Baker, B. McGaw, & P. P. Peterson
(Eds.). International encyclopedia of education (3rd ed.), Elsevier.
Marsh, H. W., Smith, I. D., Barnes, J., & Butler, S. (1983). Self-concept: Reliability, stability,
dimensionality, validity and the measurement of change. Journal of Educational Psychology,
75, 772790. doi:10.1037/0022-0663.75.5.772
Marsh, H. W., Trautwein, U., L udtke, O., K oller, O., & Baumert, J. (2005). Academic self-concept,
interest, grades and standardized test scores: Reciprocal effects models of causal ordering.
Child Development, 76, 297416. doi:10.1111/j.1467-8624.2005.00853.x
Marsh, H. W., Trautwein, U., L udtke, O., K oller, O., & Baumert, J. (2006). Integration of
multidimensional self-concept and core personality constructs: Construct validation and
relations to well-being and achievement. Journal of Personality, 74, 403456. doi:10.1111/j.
1467-6494.2005.00380.x
Marsh, H. W., Walker, R., & Debus, R. (1991). Subject-specific components of academic self-
concept and self-efficacy. Contemporary Educational Psychology, 16, 331345. doi:10.1016/
0361-476X(91)90013-B
Marsh, H. W., & Yeung, A. S. (1997a). Causal effects of academic self-concept on academic achieve-
ment: Structural equation models of longitudinal data. Journal of Educational Psychology, 89,
4154. doi:10.1037/0022-0663.89.1.41
Marsh, H. W., & Yeung, A. S. (1997b). Coursework selection: The effects of academic self-concept
and achievement. American Educational Research Journal, 34, 691720.
OECD. (2003). PISA literacy skills for the world of tomorrow further results from PISA 2000.
Paris: OECD and UNESCO.
Academic self-concept 77

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data
analysis methods (2nd ed.). Thousand Oaks, CA: Sage Publications.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University
Press.
Sameroff, A. J. (Ed.), (2009). The transactional model of development: How children and contexts
shape each other. Washington, DC: American Psychological Association.
Sameroff, A. J., & Chandler, M. J. (1975). Reproductive risk and the continuum of caretaking
causality. In D. Cicchetti (Ed.), The emergence of a discipline: Rochester Symposium on
Developmental Psychopathology (Vol. 1, pp. 4168). Hillsdale, NJ: Erlbaum.
Seligman, M., & Csikszentmihalyi, M. (2000). Positive psychology: An introduction. American
Psychologist, 55, 514. doi:10.1037/0003-066X.55.1.5
Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976). Self-concept: Validation of construct
interpretations. Review of Educational Research, 46, 407441.
Skaalvik, E. M. (1997). Issues in research on self-concept. In M. L. Maehr & P. R. Pintrich (Eds.),
Advances in motivation and achievement (Vol. 10, pp. 5198). Greenwich, CT: JAI Press.
Valentine, J. C., & DuBois, D. L. (2005). Effects of self-beliefs on academic achievement and
vice-versa: Separating the chicken from the egg. In H. W. Marsh, R. G. Craven, & D. M.
McInerney (Eds.), International advances in self research (Vol. 2, pp. 5378). Greenwich,
CT: Information Age Publishing.
Valentine, J. C., DuBois, D. L., & Cooper, H. (2004). The relations between self-beliefs
and academic achievement: A systematic review. Educational Psychologist, 39, 111133.
doi:10.1207/s15326985ep3902 3
Vallerand, R. J., Blanchard, C., Mageau, G. A., Koestner, R., Ratelle, C., Leonard, M., . . . Marsolais,
J. (2003). Les passions de lame: On obsessive and harmonious passion. Journal of Personality
and Social Psychology, 85, 756767. doi:10.1037/0022-3514.85.4.756
Vernon, P. E. (1950). The structure of human abilities. London: Muethon.
Walls, T. A., & Schafer, J. L. (Eds.), (2006). (Eds.), Models for intensive longitudinal data. Oxford:
Oxford University Press.
Wells, L. E., & Marwell, G. (1976). Self-esteem: Its conceptualization and measurement. Beverly
Hills, CA: Sage Publications.
Wigfield, A., & Eccles, J. S. (2002). The development of competence beliefs, expectancies for
success, and achievement values from childhood through adolescence. In A. Wigfield & J.
S. Eccles (Eds.), Development of achievement motivation (pp. 173195). San Diego, CA:
Academic Press.
Wigfield, A., & Karpathian, M. (1991). Who am I what can I do: Childrens self-concepts and
motivation in achievement situations. Educational Psychologist, 25, 233261. doi:10.1207/
s15326985ep2603&4 3
Wylie, R. C. (1979). The self-concept (Vol. 2). Lincoln, NE: University of Nebraska Press.
Wylie, R. C. (1989). Measures of self-concept. Lincoln, NE: University of Nebraska Press.
Yeung, A. S., & Lee, F. L. (1999). Self-concept of high school students in China: Confirmatory
factor analysis of longitudinal data. Educational and Psychological Measurement, 59, 431
450. doi:10.1177/00131649921969965

Received 23 November 2009; revised version received 12 April 2010


Copyright of British Journal of Educational Psychology is the property of Wiley-Blackwell and its content may
not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

You might also like