40
Educational Psychology Review, Vol. 15, No. 1, March 2003 ( C 2003) Academic Self-Concept and Self-Efficacy: How Different Are They Really? Mimi Bong 1,3 and Einar M. Skaalvik 2 Academic motivation researchers sometimes struggle to decipher the distinc- tive characteristics of what appear to be highly analogous constructs. In this article, we discuss important similarities between self-concept and self-efficacy as well as some notable differences. Both constructs share many similarities such as centrality of perceived competence in construct definition; use of mas- tery experience, social comparison, and reflected appraisals as major informa- tion sources; and a domain-specific and multidimensional nature. Both predict motivation, emotion, and performance to varying degrees. However, there are also important differences. These differences include integration vs. separation of cognition and affect, heavily normative vs. goal-referenced evaluation of competence, aggregated vs. context-specific judgment, hierarchical vs. loosely hierarchical structure, past vs. future orientation, and relative temporal stabil- ity vs. malleability. We argue that self-efficacy acts as an active precursor of self-concept development and suggest that self-concept research separate out its multiple components and subprocesses and invest more effort toward mak- ing students less preoccupied with normative ability comparisons in school. KEY WORDS: self-concept; self-efficacy; self-esteem; motivation. Researchers in personality and social psychology have long been interested in the role of self-related perceptions. Individuals who are otherwise similar feel differently about themselves and choose different courses of action, depending on how they construe themselves—what attributes they think 1 Department of Educational Psychology, University of South Carolina, South Carolina, Columbia. 2 Department of Education, Norwegian University of Science and Technology, Trondheim, Norway. 3 To whom correspondence should be addressed at Department of Educational Psychol- ogy, 135 Wardlaw Hall, University of South Carolina, Columbia, South Carolina. E-mail: [email protected]. 1 1040-726X/03/0300-0001/0 C 2003 Plenum Publishing Corporation

Bong Skaalvvik Se-sc

Embed Size (px)

DESCRIPTION

xx

Citation preview

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Educational Psychology Review, Vol. 15, No. 1, March 2003 ( C 2003)

    Academic Self-Concept and Self-Efficacy: HowDifferent Are They Really?

    Mimi Bong1, 3 and Einar M. Skaalvik2

    Academic motivation researchers sometimes struggle to decipher the distinc-tive characteristics of what appear to be highly analogous constructs. In thisarticle, we discuss important similarities between self-concept and self-efficacyas well as some notable differences. Both constructs share many similaritiessuch as centrality of perceived competence in construct definition; use of mas-tery experience, social comparison, and reflected appraisals as major informa-tion sources; and a domain-specific and multidimensional nature. Both predictmotivation, emotion, and performance to varying degrees. However, there arealso important differences. These differences include integration vs. separationof cognition and affect, heavily normative vs. goal-referenced evaluation ofcompetence, aggregated vs. context-specific judgment, hierarchical vs. looselyhierarchical structure, past vs. future orientation, and relative temporal stabil-ity vs. malleability. We argue that self-efficacy acts as an active precursor ofself-concept development and suggest that self-concept research separate outits multiple components and subprocesses and invest more effort toward mak-ing students less preoccupied with normative ability comparisons in school.

    KEY WORDS: self-concept; self-efficacy; self-esteem; motivation.

    Researchers in personality and social psychology have long been interestedin the role of self-related perceptions. Individuals who are otherwise similarfeel differently about themselves and choose different courses of action,depending on how they construe themselveswhat attributes they think1Department of Educational Psychology, University of South Carolina, South Carolina,Columbia.

    2Department of Education, Norwegian University of Science and Technology, Trondheim,Norway.

    3To whom correspondence should be addressed at Department of Educational Psychol-ogy, 135 Wardlaw Hall, University of South Carolina, Columbia, South Carolina. E-mail:[email protected].

    1

    1040-726X/03/0300-0001/0 C 2003 Plenum Publishing Corporation

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    2 Bong and Skaalvik

    they possess, what roles they presume they are expected to play, what theybelieve they are capable of, how they view they fare in comparison withothers, and how they judge they are viewed by others. Without doubt, theseare beliefs and perceptions about self that are heavily rooted in ones pastachievement and reinforcement history. Yet it is these subjective convictionsabout oneself, once established, which play a determining role in individualsfurther growth and development (Bandura, 1997; Markus and Nurius, 1986).

    It is only reasonable that these self-perceptions have received a greatdeal of attention in educational research (Byrne, 1984). Children with dif-ferent self-beliefs demonstrate different levels of cognitive, social, and emo-tional engagement in school. Because school-related experience makes upa major portion of childrens lives and shapes the early paths to importantlife outcomes, educational researchers try to grasp the meaning of self instudents minds. Various models and theories of self-related cognition havebeen proposed and tested within the context of school learning. Self-conceptand self-efficacy are the two self-constructs that have received a lot of at-tention. During the past couple of decades, numerous studies in educationalresearch have resorted to either self-concept or self-efficacy to explain thefunction of self in school contexts. These studies produced abundant evi-dence on the potency of each self-belief. The field now struggles to decipherthe distinguishing characteristics and comparative usefulness of the two be-lief systems.

    Making a clear and irrefutable distinction between beliefs of self-concept and self-efficacy is not an easy task. However, it is nonetheless possi-ble to illuminate some of the similarities and differences between these twoconceptions. This is the goal of this article. While more recent reviews on thistopic highlighted differences between the two (e.g., Bong and Clark, 1999),we try to deduce also important similarities underlying the formulation ofthe two self-beliefs. In doing so, our hope is that the theory and researchin this area become more integrated to give educational researchers andpractitioners better understandings of students perception of self and whatit does to their cognitive and psychological well-being in school.

    DEFINITIONS OF CONSTRUCTS

    Self-Concept and Self-Efficacy

    Self-concept is colloquially defined as a composite view of oneself.Rosenberg (1979) defined self-concept as . . . the totality of the indi-viduals thoughts and feelings having reference to himself as an object(p. 7). Shavelson, Hubner, and Stanton (1976) provided a similar

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 3

    definition of self-concept that formed the theoretical foundation of con-temporary self-concept research:

    In very broad terms, self-concept is a persons perception of himself. . . . We donot claim an entity within a person called self-concept. Rather, we claim thatthe construct is potentially important and useful in explaining and predicting howone acts. Ones perceptions of himself are thought to influence the ways in whichhe acts, and his acts in turn influence the ways in which he perceives himself. . . .Seven features can be identified as critical to the construct definition. Self-conceptmay be described as: organized, multifaceted, hierarchical, stable, developmental,evaluative, and differentiable. (p. 411)

    Self-concept is formed through experiences with the environment and isinfluenced especially by environmental reinforcements and significant others(Shavelson et al., 1976). Skaalvik (1997a) identified some key antecedents toself-concept in his recent review (see also Rayner and Devi, 2001; Skaalvikand Skaalvik, in press):

    (1) Frames of reference. Self-concept is heavily influenced by framesof reference or standards against which to judge ones own traitsand accomplishments. Social comparison often serves as the mostpotent source of information for self-concept. Frames of referenceplay a particularly important role in the development of academicself-concept (Marsh, 1986, 1987).

    (2) Causal attributions. The factors to which people attribute their suc-cesses and failures are hypothesized to influence descriptive and af-fective aspects of their self-concept. Self-concept and attributions arerelated in a reciprocal manner such that the types of causal attribu-tions made for previous successes and failures influence subsequentself-concept and the self-concept thus formed affects later attribu-tions (Skaalvik, 1997a; Stipek, 1993; Tennen and Herzberger, 1987).

    (3) Reflected appraisals from significant others. Several self-concept re-searchers suggested that people come to view themselves as theybelieve how others view them. Sullivan (1947) stated, The self maybe said to be made up of reflected appraisals (p. 10). Rosenberg(1979) also claimed that . . . there is probably no more critical andsignificant source of information about ourselves than other peoplesviews of us, referring to Meads conception that in communicationwe take the role of the other. (Mead, 1934)

    (4) Mastery experiences. Self-schemas are created from individuals pastexperiences in a particular domain. Relevant information and expe-riences are subsequently processed by these self-schemas (Markusand Nurius, 1986). Although self-concept researchers do not explic-itly emphasize the role of mastery experiences in self-concept for-mation, Skaalvik (1997a) suggested that prior mastery experiences

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    4 Bong and Skaalvik

    might be of comparable importance to the formation of self-conceptas they are to the formation of self-efficacy (Bandura, 1986).

    (5) Psychological centrality. Rosenberg (1979), in his analysis of self-esteem, claimed that self-esteem is based on self-assessments of qual-ities that are perceived as important or psychologically central byindividuals. Skaalviks (1997a) review found mixed evidence to sup-port this notion. However, Harter and Mayberry (1984) providedevidence that supports the effects of psychological centrality on self-concept. These investigators asked fifth to seventh graders to rateboth the importance of five different areas (i.e., school, sports, socialrelations, physical appearance, and behavior) and their own compe-tency within these areas. Self-esteem was the highest among studentswho rated their best areas as also the most important.

    Historically, self-concept research has emphasized a global constructsuch as general self-concept. Typically, a composite score was computed bysumming self-concept responses from standardized instruments toward var-ious aspects of life and was then treated as an indicator of ones self-concept(e.g., Piers and Harris, 1964; see Marsh, 1990a, for an overview). These globalassessments of self-regard that were detached from any specific context con-tributed to earlier views of self-concept research as an ill-disciplined fieldand difficult to conceptualize and operationalize (Hansford and Hattie,1982). Harter (1982) also observed that Typically, constructs such as self-concept and self-esteem are vaguely defined at the conceptual level andtherefore do not point to any clear operational definition (p. 87). Owingmostly to this ambiguity, the average relationship between the self (variouslytermed as self, self-concept, and self-esteem) and academic achievement in-dexed in 128 studies located by Hansford and Hattie was only 0.212. Thiswas a disappointing result in light of intuitive assumptions and theoreticalarguments that positive self-beliefs should result in improved performance.

    The global nature of self-concept has since been criticized as diminish-ing its power to explain behavior (Bandura, 1981) and overlooking impor-tant distinctions children make across activity domains (Harter, 1982). Afterdecades of research with a myriad of global and undifferentiated views ofself, the field has come to realize that any sound understanding of childrensself-concept and its impact on their functioning in school must take into ac-count the effects of domain on these judgments (Marsh, 1993). As can beseen in the definition by Shavelson et al. (1976), self-concept is now viewedas perceptions of oneself that are multidimensional. Recent self-conceptstudies that focus on the domain-specific self-concepts have documentedthat globality is not necessarily inherent in the construct definition (Byrne,1996). At the same time, perceived competence emerged as a key component

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 5

    that has particularly significant bearing on students motivation and learn-ing among the array of information available in their complex school-relatedself-conceptions (Harter, 1990). The work of Marsh and his associates, forexample, reflects both of these recent developments fairly well. Conductedprimarily in the framework of the Shavelson hierarchy, this line of workhas produced more consistent and encouraging results regarding the self-concept effect (Marsh, 1990d, 1993). A recent meta-analysis on math self-concept also showed that studies published after 1986 reported particularlystronger relations between self-concept and achievement (Ma and Kishor,1997).

    Compared with the self-concept research, research in self-efficacy ischaracterized by its relatively short history. Bandura (1977) offered a formaltheoretical definition of self-efficacy:

    Perceived self-efficacy refers to beliefs in ones capabilities to organize and executethe courses of action required to produce given attainments. . . . Such beliefs influencethe course of action people choose to pursue, how much effort they put forth ingiven endeavors, how long they will persevere in the face of obstacles and failures,their resilience to adversity, whether their thought patterns are self-hindering orself-aiding, how much stress and depression they experience in coping with taxingenvironmental demands, and the level of accomplishments they realize. (p. 3)

    Like self-concept, self-efficacy is presumed to explain and predict onesthought, emotion, and action. However, efficacy judgment is less concernedwith what skills and abilities individuals possess. It considers more importantwhat individuals believe they can do with whatever skills and abilities theymay possess. This provides a point of comparison with a self-concept judg-ment, which routinely calls for an evaluation of the skills and abilities. Whileself-concept represents ones general perceptions of the self in given do-mains of functioning, self-efficacy represents individuals expectations andconvictions of what they can accomplish in given situations. For example, theexpectation that one can high-jump 6 ft is an efficacy judgment (Bandura,1986). It is not a judgment of whether one is competent in high-jumpingin general but a judgment of how strongly a person believes that he or shecan successfully jump that particular height under the given circumstances.Self-efficacy researchers thus emphasize the role played by specific contextsin efficacy appraisals.

    Information for shaping self-efficacy beliefs comes from the followingfour major sources (Bandura, 1986, 1997):

    (1) Enactive mastery experience. Ones prior experiences with the tasksin question provide the most reliable source of information for ef-ficacy beliefs. Successes strengthen self-efficacy, whereas repeatedfailures undermine it. A firm sense of efficacy built on the basis ofpast successes is believed to withstand temporary failures.

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    6 Bong and Skaalvik

    (2) Vicarious experience. People also establish their self-efficacy beliefson the basis of similar others performance on the tasks. Modelingthus serves as another effective source of efficacy information. Vi-carious experience exerts greater influence on self-efficacy formationwhen there are no absolute measures of adequacy and when peopleperceive similarity between the model and themselves (Schunk andHanson, 1985; Schunk, Hanson, and Cox, 1987).

    (3) Verbal persuasion. Persuasive communication and evaluative feed-back from significant others also influence ones judgment of self-efficacy. Verbal persuasion is most effective when people who con-vey the efficacy information are viewed knowledgeable and credibleand when the information is viewed realistic. However, disconfirm-ing mastery experience easily outweighs self-efficacy beliefs createdsolely on the basis of verbal persuasion.

    (4) Physiological reactions. Heightened physiological arousals such assweating, heartbeats, fatigue, aches, pain, and mood changes alsosend a signal to people that affects their efficacy appraisal. Recog-nition of these somatic symptoms leads to self-efficacy adjustmentsthrough their effects on cognitive processing.

    As can be seen, self-concept and self-efficacy share many of the pre-sumed antecedents such as past experience, social comparison, and rein-forcements from significant others. They share many of the presumed out-comes related to cognitive, affective, and behavioral functioning as well.However, there are also differences in how they are conceptualized and op-erationalized in research. We discuss some of the noticeable trends in moredetail as they pertain to the domain of academic functioning.

    Academic Self-Concept and Academic Self-Efficacy

    Academic self-concept and academic self-efficacy refer to individualsself-concept and self-efficacy beliefs that are formed specifically toward aca-demic (as distinct from nonacademic, general, social, emotional, or phys-ical) domains. More specifically, academic self-concept refers to individu-als knowledge and perceptions about themselves in achievement situations(Byrne, 1984; Shavelson and Bolus, 1982; Wigfield and Karpathian, 1991).Academic self-efficacy refers to individuals convictions that they can suc-cessfully perform given academic tasks at designated levels (Schunk, 1991).

    Both constructs received much attention from educational researchersbecause of their purported influence on students academic functioning. Nu-merous studies reported how positive self-concept or self-efficacy facilitated

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 7

    students academic engagement, goal-setting, task choice, persistence andeffort, intrinsic motivation, strategy use, performance and achievement, andeven career selection (see discussion under the Predictive Outcomes be-low). Despite the vast volume of evidence attesting to the powerful natureof these constructs, it is not always easy to locate specific factors or work-able strategies to enhance these beliefs to realize such desirable outcomes.This difficulty is in part due to the hazy distinction between self-concept andself-efficacy, which thwarts any synthesis or integration efforts of the sort.

    The rather subtle conceptual distinction between self-concept and self-efficacy applies equally to these academic self-perceptions. Because nowthey are both dealing with the same academic domain, it is conceivablymore difficult to identify the critical distinction between these two constructs.Theoretical definitions alone are often not enough to point out specific di-mensions on which they are believed to be similar or different. It becomesmuch easier to distinguish academic self-concept from academic self-efficacyand vice versa when provided with operational definitions of each. On onehand, the clearer divergence of operational definitions may indicate that thedifferences between the two constructs have been exaggerated because ofthe different assessment and analytic strategies that are associated with eachtheory (Bong and Clark, 1999; Skaalvik and Rankin, 1996a). On the otherhand, operational definitions are manifestation of implicit and explicit the-oretical tenets and, as such, may reflect genuine differences between thesetwo constructs. Rather than comparing conceptual definitions that are some-times obscure on how best to capture the construct in question, we start withanalyzing the current and representative operationalizations of academicself-concept and academic self-efficacy.

    By far, the most commonly used method of measuring both constructsis self-reports. Items that are typically used to assess academic self-conceptinclude Schoolwork is easy for me, I have always done well in (a subject),and Compared with others my age, Im good at (a subject). Students indi-cate how much they agree with each of these statements on 15, 16, or 17response scales. It is worth noting that there exist different views among re-searchers regarding whether academic self-concept also includes emotionalreactions to the tasks such as interest, enjoyment, and satisfaction. Someregard these as part of self-concept, whereas others consider them a distinctconstruct. Researchers who endorse the former view add items such as I aminterested in (a subject) and I look forward to (a subject) to academic self-concept assessment (Marsh, 1999a, 1999b). Other researchers (e.g., Ecclesand Wigfield, 1995; Eccles, Wigfield, and Schiefele, 1998; Wigfield, Eccles,Mac Iver, Reuman, and Midgley, 1991) make clear conceptual distinctionsbetween ability- or expectancy-related perceptions and task-value compo-nents (e.g., interest, importance, usefulness). The issue of whether or not

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    8 Bong and Skaalvik

    the competence and affective components of self-concept are empiricallydistinguishable has not been resolved.

    Another procedure that is frequently used to measure academic self-concept of young children involves presenting two contrasting descriptionsof hypothetical children. For example, a statement, Some kids do very wellat their classwork, is written on the left column of a page. An oppositestatement, Other kids dont do very well at their classwork, is on the rightcolumn. Children first select which of the two statements describes thembetter. They then judge whether the selected statement is really true forthem or just sort of true for them (Harter, 1982; Harter and Pike, 1984). Thisassessment procedure yields self-concept response scores that range from 1to 4 on each item.

    The standard method of measuring academic self-efficacy is to presentproblems that are similar to the actual problems students must solve. Stu-dents estimate their confidence that they can solve each problem correctly(e.g., Bandura and Schunk, 1981). Alternatively, academic self-efficacy itemsmay include written descriptions of problems or tasks in place of actualproblemsfor example, How sure are you that you can correctly spell allwords in a one-page story or composition? (Pajares, Miller, and Johnson,1999), How confident are you that you can successfully solve equationscontaining square roots? (Bong, 2002) or How confident are you that youwill get a grade better than a B in mathematics at the end of this term?(Zimmerman and Bandura, 1994). Students rate the degree of their confi-dence for successfully accomplishing each task on a 0100 or 10100 scalein 10-unit intervals. A score of 0 represents absolute lack of confidence anda 100 represents complete confidence. Other self-efficacy items include Iexpect to do very well in (a subject) class and I am sure that I can doan excellent job on the problems and tasks assigned for (a subject) class(Pintrich and De Groot, 1990). As was the case with self-concept items, re-spondents mark how much they agree with each of these efficacy statementson a Likert-type response scale.

    A quick glance at these items makes evident several features that canbe easily overlooked when given only the theoretical definition of each con-struct. First of all, despite several apparent differences in assessment proce-dures, both constructs seem to call for a subjective judgment of perceivedcompetence in reference to some target domain or activities. In addition tothe cognitive appraisal of ones competence, academic self-concept assess-ment also often inquires about students affective reactions to the recognizedself and its attributes. Items such as I enjoy doing work in (a subject), Ihate (a subject), and I never want to take another (subject) course (Marsh,1990b) exemplify this. In judging self-efficacy, respondents make largelycognitive evaluations of their perceived capability without deliberately

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 9

    reflecting on their feelings generated by those evaluations (Zimmerman,1996).

    The nature of self-concept and self-efficacy evaluations differ from oneanother. Assessing ones capability in academic self-concept relies heavilyon social comparative information and reflected appraisals from significantothers. Items such as Compared with others my age, Im good at (a sub-ject) or In (a subject), I am one of the best students in my class are com-monly found in self-concept scales (e.g., Marsh, 1999a). Some self-conceptresearchers suggest that students further compare their academic capabil-ity in one domain to their capability in other domains. Such ipsative compa-rison makes performance improvement in one domain cause decrease in self-concepts in other areas (Marsh, 1986). In contrast, self-efficacy items solicitgoal-referenced evaluation and do not directly ask students to compare theirability to those of others. Although normative information wields tangible in-fluence on self-efficacy estimation at times, efficacy is gauged mainly againstconcrete performance standards (Bong and Clark, 1999; Zimmerman,1996).

    Academic self-concept items typically refer to specific school subjects,whereas self-efficacy items most often refer to specific tasks. Both constructsare thus closely tied to academic content areas. However, it is noticeablethat the expected performance or features of the outcome against which toevaluate ones competence are not explicitly stated in self-concept items.The lack of context-specific information leads students to make some aggre-gated judgments of their competence in the given area. Self-efficacy itemsprovide respondents with a specific description of required performance asa referent against which to appraise their competence. Judgments formedas a result of such appraisal are not only specific to certain academic con-tent but also specific to given performance contexts. Self-concept is orga-nized in multidimensional and hierarchical fashion such that self-conceptsin more specific domains are subsumed under self-concepts in more generaldomains (Shavelson et al., 1976). Self-efficacy beliefs are also multidimen-sional in the sense that students form differentiated perceptions of capa-bility across diverse tasks and domains. Relationships among these beliefsare only loosely hierarchical because self-efficacy in more general areas maynot sufficiently incorporate particularities of diverse contexts that influenceself-efficacy judgments toward more specific tasks.

    The theoretical and operational definitions of the constructs, when com-pared, also create the impression that self-concept embodies fairly stableperceptions of the self that are past-oriented, whereas self-efficacy repre-sents relatively malleable and future-oriented conceptions of the self and itspotential. Despite these differences, self-concept and self-efficacy are usedto predict a fairly similar set of outcomes including motivation, emotion, and

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    10 Bong and Skaalvik

    Table I. Comparison Between Academic Self-Concept and Academic Self-Efficacy

    Comparison Academic Academicdimensions self-concept self-efficacy

    1. Working definition Knowledge and perceptionsabout oneself inachievement situations

    Convictions forsuccessfully performinggiven academic tasks atdesignated levels

    2. Central element Perceived competence Perceived confidence3. Composition Cognitive and affective

    appraisal of selfCognitive appraisal of self

    4. Nature of competenceevaluation

    Normative and ipsative Goal-referenced andnormative

    5. Judgment specificity Domain-specific Domain-specific andcontext-specific

    6. Dimensionality Multidimensional Multidimensional7. Structure Hierarchical Loosely hierarchical8. Time orientation Past-oriented Future-oriented9. Temporal stability Stable Malleable

    10. Predictive outcomes Motivation, emotion, andperformance

    Motivation, emotion,cognitive andself-regulatory processes,and performance

    performance. In addition, self-efficacy predicts cognitive and self-regulatoryprocesses. Table I lists the working definition of academic self-concept andself-efficacy used in this article and summarizes the key dimensions of com-parison including central element, composition, nature of competence evalu-ation, judgment specificity, dimensionality, structure, temporal stability, andpredictive outcomes. Below we elaborate and present evidence for the pur-ported similarities and differences on each of these dimensions.

    CENTRAL ELEMENT

    One of the most glaring similarities between the conceptualization ofself-concept and self-efficacy is the central role played by ones competenceperceptions. Perceived competence in defined domains or activities comprisesthe single most critical element in both self-beliefs (Eccles et al., 1998). Con-temporary academic self-concept researchers assert that students percep-tions of competence in given areas provide key ingredients to their self-concepts (e.g., Harter, 1982; Marsh, 1990a, 1992; Shavelson and Bolus, 1982;Wigfield et al., 1997; Wigfield and Karpathian, 1991). Perceived capabilityin reference to specific academic tasks and domains is also the principalconstituent of academic self-efficacy judgments (Pajares, 1996).

    Many researchers recognize that academic self-concept includes a self-efficacy component and that this component may be the most important

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 11

    building block in ones self-concept (Bong and Clark, 1999; Schunk, 1991).Pajares (1996) suspected that at the domain level of specificity, academicself-concept and self-efficacy beliefs might not be separable. Thus far, re-searchers have reported that students responses to the Self DescriptionQuestionnaire, one of the popular self-concept scales, formed two separatefactors: cognitive and motivational (Skaalvik and Rankin, 1996a; Tanzer,1996). The cognitive academic self-concept factor was empirically indistin-guishable from the academic self-efficacy factor (Pietsch, 1999; Skaalvikand Rankin, 1996a). Because few studies have addressed the equivalenceof self-concept and self-efficacy responses systematically, it is still prematureto draw any firm conclusion regarding the nature of relationship betweenthese two constructs. At minimum, many empirical investigations need tobe conducted.

    At present, there is still some room for debate whether the perceivedcompetence component in self-concept is indeed identical to percepts ofself-efficacy. For example, different information sources have been knownto affect the two self-systems to different degrees, as is shown later in thisarticle. The two self-beliefs, in turn, have sometimes yielded different psy-chological and behavioral outcomes. Nevertheless, it seems reasonable toassume, on the basis of limited available evidence, that there is at least con-siderable overlap in the makeup of academic self-concept and academicself-efficacy and that perception of academic capability is the major com-mon denominator between the two.

    COMPOSITION

    Although perceived capability constitutes the core in contemporaryviews of academic self-concept, self-concept has long been recognized to re-flect more than ones competence perceptions. Scheirer and Kraut (1979), forexample, argued that self-concept consists of at least four distinguishable as-pects. These include descriptive categorization of self in terms of social rolesand personality traits, evaluation of the self-attributes according to social de-sirability, comparison of qualities through which individuals determine theirranking relative to other people on a specific dimension, and emotional at-titudes toward the selfcalled self-esteem. More recently, Skaalvik (1997a)distinguished between descriptive, evaluative, and affective/motivationalaspects of self-concept. However, consistent with the observation ofShavelson et al. (1976), he claimed that a clear empirical distinction be-tween self-description and self-evaluation often could not be made. He wrote

    . . . self-conceptions like I am tall and I learn mathematics easily include both de-scriptive and evaluative aspects. The descriptive or cognitive component represents

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    12 Bong and Skaalvik

    knowledge and beliefs that a person has about herself or himself in different areas(see Markus, 1977); for example, the belief that he or she learns mathematics easily.However, a persons belief that he or she learns mathematics easily is also the resultof an evaluation. When a person describes himself as clever in mathematics thisdescription can therefore not be distinguished from the persons evaluation of heror his mathematics abilities. I shall therefore refer to this aspect of self-concept asdescriptive/evaluative. (p. 53)

    With specific regard to self-concept in the academic domain, Skaalvik(1997a) argued that its descriptive/evaluative aspect (e.g., I learn mathe-matics easily) could be distinguished from its affective/motivational aspect(e.g., I am proud of my mathematics ability or I like mathematics). Ina similar vein, Bong and Clark (1999) suggested that academic self-conceptconsists of cognitive and affective dimensions and that a cognitive dimen-sion is further differentiated into descriptions and evaluations of self and itsattributes. We conjecture that the cognitive dimension of self-concept givesrise to the affective/motivational reactions. For instance, conceiving one-self as smart almost unanimously engenders positive emotional responses(Covington, 1984b, 1992). Because ability to achieve competently is highlyvalued, students who regard themselves as smart and competent usually feelgood about that aspect of their self-description. These affective evaluationstoward the self are believed to have important implications for further mo-tivation. As Wigfield and Karpathian (1991) noted, children avoid academictasks and situations that are likely to make them feel bad about themselvesin an attempt to maintain positive self-regard (see also Covington, 1984a).

    The tendency to avoid negative information about the self is in line withpredictions of the social comparison theory. Academic self-concept is largelydetermined by the result of social comparison and such comparison with sim-ilar others is believed to result in strong emotional consequences (Festinger,1954). These presumed emotional reactions might explain why some of thetheoretical as well as operational definitions of academic self-concept in-clude a mixture of components that deal with students cognition, affect,and motivation in domains under consideration. Instruments intended tomeasure academic self-concept vary with respect to which of these differentaspects they stress. For example, although Harter (1998) fully acknowledgedthe importance of affect in many theoretical conceptualizations of self andits integration with cognitive and social processes, she nonetheless made adistinction between perceived competence, anxiety, and motivational ori-entation (Silon and Harter, 1985). Her self-concept instrument, the Per-ceived Competence Scale, concentrates on childrens perceptions of compe-tence, the dimension she believed most central to childrens self-evaluation(Harter, 1982). Expectancy-value theorists (e.g., Eccles and Wigfield, 1995;Eccles et al., 1998; Wigfield et al., 1991) also distinguish between perceivedcompetence from task-value perceptions.

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 13

    In comparison, the Self Description Questionnaire developed by Marshand his associates (e.g., Marsh, 1999b) combines the cognitive (e.g., I dobadly in tests of mathematics) and affective/motivational aspects of self-concept (e.g., I hate mathematics). Recent research suggests that thesetwo dimensions of self-concept do form separate factors (Pietsch, 1999;Skaalvik and Rankin, 1996b; Tanzer, 1996). Chapman and Tunmer (1995,Experiments 3 and 4) also reported that a three-factor model best explainedstudents responses to the Reading Self-Concept Scale. The three factorswere perceptions of difficulty with reading, perceptions of competence inreading, and attitudes toward reading. Intercorrelations among these sub-components and their relations with measures of reading skills differed indifferent age groups. Again, these questions on the internal compositionof academic self-conceptwhether cognitive and affective components areboth part of self-concept or whether they need to be treated as separateconstructsare a relatively recent issue in self-concept research. Althoughinvestigators cited above reported evidence of separability between thesecomponents, most existing academic self-concept scales have not yet incor-porated such a distinction explicitly. As such, it seems reasonable to say thatacademic self-concept measures tend to reflect multiple aspects of the self,including some forms of cognitive evaluations and affective reactions.

    Measures of academic self-efficacy are designed to tap exclusively thecognitive aspect of students self-perceptions. Self-efficacy measures ask stu-dents to judge how well they can execute particular courses of actions. Obvi-ously, this judgment involves an evaluation of what one is and is not capableof performing. Hence, the cognitive aspects of both self-concept and self-efficacy measures include a strong evaluative component, although theremay be a distinct difference in the nature of these evaluations and the emo-tions that are generated thereafter. Even though measures of academic self-efficacy never refer to affective or motivational responses directly, the the-ory presumes that self-efficacy beliefs determine subsequent motivation andemotion through a self-regulatory mechanism (Bandura, 1986; Pajares andMiller, 1994; Schunk and Zimmerman, 1997; Zimmerman, 2000). Bandura(1986) claimed that . . . those who regard themselves as inefficacious . . .suffer much anxiety and stress (p. 395). Several studies showed that self-efficacy is indeed a strong predictor of anxiety and depression (Bandura,Pastorelli, Barbaranelli, and Caprara, 1999; Pajares et al., 1999; Pajares andKranzler, 1995; Pajares and Miller, 1994).

    In sum, both academic self-concept and academic self-efficacy percep-tions are related to how students feel about themselves. Self-concept re-searchers traditionally tend to view these emotional and motivational orien-tations as an important aspect of self-image that needs to be integrated in thedefinition of the construct. Self-efficacy researchers similarly acknowledge

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    14 Bong and Skaalvik

    the close link between cognitive self-perceptions and affective self-reactions.However, they consider these motivational and emotional responses asmainly a correlate or consequence of perceived self-efficacy and not a nec-essary constituent for defining self-efficacy beliefs. Self-processes such asintrinsic interest/value or self-satisfaction/affect are linked to self-efficacywithin the cyclical self-regulatory processes (Schunk and Zimmerman, 1997;Zimmerman, 2000).

    NATURE OF COMPETENCE EVALUATION

    The social comparison theory of Festinger (1954) suggests that, whenobjective standards of comparison are not provided, people appraise them-selves using significant others in their immediate environment as the basesof comparison. Many self-concept investigations demonstrated the effectsof social comparison on academic self-concept. For example, Rogers, Smith,and Coleman (1978) rank-ordered and assigned students to high-, medium-,and low-achieving groups either on the basis of their within-classroom rank-ings or on the basis of their achievement scores irrespective of their within-classroom standing. Across reading and math, significant group differenceson various academic and nonacademic self-concepts were observed onlywhen the trichotomy was conducted in the context of students classrooms.The investigators thus concluded, the most meaningful way to understandthe relation between academic achievement and self-concept is within thecontext of the social comparison group or classroom (p. 56). Because per-formance standards are only implicitly alluded to in self-concept assessment,students often engage in social comparison processes as an alternative wayof evaluating how good they are or how well they do academically.

    Social comparison effects on self-concept were documented withspecial populations. Renick and Harter (1989) found that a majority oflearning disabled students spontaneously compared themselves to regularclassroom students when reporting their self-concepts. When they did, theirself-concepts suffered. Coleman and Fults (1982) reported analogous find-ings with gifted students. Students identified as gifted and who subsequentlyparticipated in the special gifted program soon formed less favorable viewsof themselves, presumably because of their new, comparably performingpeers. Those who stayed in the regular classes maintained their high self-concept. Because social comparison is one of the most powerful sources ofevaluative information for judging self-concept (Marsh, 1990d, 1993), stu-dents in the high-ability schools often experience loss in their academic self-concept. Marsh (1987) observed that, after the difference in individual abilitywas controlled for, school-average ability demonstrated negative effects on

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 15

    students academic self-concept. In other words, students judge themselvesless capable in the environment with highly able students and more capablein the environment with less able peers. Marsh termed this social comparisoneffect on self-concept the big-fishlittle-pond effect.

    Another comparative frame of reference thought to influence aca-demic self-concept is internal comparison. Marsh (1986) argued that stu-dents base their academic self-concepts in a particular subject not only onhow their ability compares with those of other students (i.e., social or ex-ternal comparison) but also on how their ability in that subject compareswith their abilities in other subjects (i.e., internal comparison). Internalcomparison is presumed to create a negative relationship between achieve-ment in one domain and self-concept in other domains. For example, asstudents achievement levels in math improve, their math self-concepts alsoimprove. At the same time, their recent success in math is more likely tomake them believe that their math ability is better than their verbal abil-ity. This recognition subsequently lowers their verbal self-concepts. Thejoint effects of internal and external comparisons (hence the I/E model)are assumed to balance each other out, resulting in a near-zero correla-tion between math and verbal self-concepts. The external or social com-parison tends to result in a positive correlation between students verbaland math self-concepts because verbal and math achievements are oftenhighly correlated. The internal comparison presumably yields a negativecorrelation between students verbal and math self-concepts. Dependingon the weight assigned to each comparison, relationships between verbaland math self-concepts can be positive or negative but almost always sub-stantially reduced in magnitude from the corresponding relationships be-tween achievements. Predictions of the I/E model are supported in a numberof studies (Marsh, 1990b; Skaalvik and Rankin, 1992, 1995; but see Bong,1998).

    Peoples inferences about themselves are also affected by how othersperceive them. Reflected appraisals from significant others provide usefulinformation for molding ones self-concept. In his classic volume on self-concept, Rosenberg (1979) refers to a large body of research indicating thatindividuals actually tend to view themselves as they are seen by others.Students are believed to shape their academic self-concept in part on thebasis of their impressions of how their parents, teachers, and peers appraisetheir academic ability (Harter, 1990). In the absence of absolute standardsagainst which to estimate ones capabilities, students determine how goodthey are in the given subject by comparing their ability to those of theirpeers and, at the same time, monitoring other peoples appraisals of theirability. Reflected appraisals function as an important source of evaluativeinformation in academic self-concept formation.

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    16 Bong and Skaalvik

    In contrast, a sense of academic self-efficacy is most heavily affectedby ones previous encounters with the same or similar tasks (Bong, 1997;Bong and Clark, 1999; Zimmerman, 1995). As discussed earlier, individualsown prior mastery experiences carry heavier weight in self-efficacy appraisalthan vicarious information, verbal persuasion, or physiological reactions(Bandura, 1977). Because students gauge their confidence for success againstgoals and standards, there is less reason for them to engage in vigorous socialcomparison. Instead, they calibrate their chances of successfully carrying outthe described performance at designated levels (Schunk, 1991). Self-efficacyjudgments are hence goal-referenced evaluations that are less affected byrelativistic impressions. Although self-efficacy is less influenced by socialcomparison than is self-concept, social comparative information providescritical information for judging efficacy under certain circumstances. Stu-dents often gather efficacy-related cues by observing teachers and peers,especially when the task is novel or when there exists no clear and immedi-ate standard for evaluating performance (e.g., Schunk, 1981; Schunk et al.,1987; Schunk and Hanson, 1985). In general, if the observers judge theirability to be comparable to the models capability, successes and failures ofthe model have stronger effect on the observers self-efficacy (Schunk et al.,1987).

    The internal comparison processes described by the I/E model (Marsh,1986) do not seem relevant in self-efficacy estimation. Predictions from theI/E model are not supported by academic self-efficacy measures (Bong, 1998;Marsh, Walker, and Debus, 1991; Skaalvik and Rankin, 1990). More specifi-cally, verbal and math self-efficacy perceptions usually demonstrate a strongpositive correlation that is commensurate with the corresponding correla-tion between verbal and math achievements. Moreover, high achievementin the verbal area does not necessarily lower efficacy judgments in math orvice versa.

    On the other hand, reflected appraisals are implicit in self-efficacy judg-ments. Verbal persuasion by credible others is known to influence percep-tions of self-efficacy. Verbal persuasion, in effect, is a concrete manifestationof how a person is perceived or evaluated by significant others. It was pointedout above that when the task is novel or when the criteria for success are notclear, students estimate their efficacy perceptions primarily on the basis of so-cial comparative information (Bandura, 1977). Under such circumstances,their efficacy beliefs are also more heavily swayed by verbal persuasionof significant and knowledgeable others. However, percepts of efficacy in-stilled purely by verbal persuasion can only be maintained when followed bysuccessful mastery experiences. Self-efficacy increase to an unrealistic levelwanes quickly by disappointing failures. Therefore, the difference betweenself-concept and self-efficacy regarding social comparison and reflected

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 17

    appraisals is one of degree. These sources of information are regarded lesspowerful for self-efficacy adjustment than for self-concept development. In-ternal comparison only influences self-concept.

    JUDGMENT SPECIFICITY

    Students express academic self-concept and academic self-efficacy thatare both domain-specific. Domain-specificity of self-perceptions is ascer-tained when these perceptions are differentiated clearly across different con-tent areas and when they relate only to relevant outcomes in the same contentarea and not to those in different areas. Because academic self-perceptionsare commonly assessed at the school subject level (e.g., math self-concept),domain-specificity is often viewed synonymous to subject-specificity. Al-though subject-specificity certainly attests to the domain-specificity of a con-struct, the term domain-specificity should not be equated to a particularmeasurement level. Rather, a domain can represent from relatively limitedskill areas such as reading comprehension in English to broader contentareas such as social science.

    Although both constructs are associated with certain a degree ofdomain-specificity, traditional measures of self-concept and self-efficacy dif-fer with respect to the level of measurement specificity (Pajares, 1996). Aca-demic self-concept, even when assessed in reference to particular domains,has been measured at more general levels. Students typically report theiroverall feelings of doing well or poorly in given subject areas. Compared withthe self-concept assessment, beliefs of self-efficacy have been examined atmore specific levels, usually in the context of performing specific tasks withina particular domain. Self-efficacy has also been measured at a more generallevel beyond particularized tasks or academic subjects. The primary reasonfor assessing self-efficacy at different levels of specificity, both specific andgeneral, has been to ensure correspondence between self-efficacy percep-tions and performance criterion. For example, when the researchers goalis to predict performance of broader scope such as course grades and over-all grade point averages, perceived self-efficacy at correspondingly broaderlevels are assessed (e.g., Pajares and Miller, 1995; Randhawa, Beamer, andLundberg, 1993; Zimmerman, Bandura, and Martinez-Pons, 1992).

    Researchers express little disagreement as regards the purported dif-ferences between task-specific academic self-efficacy and subject-specificacademic self-concept (e.g., Marsh et al., 1991; Pajares, 1996). However,when the two constructs are put side by side at the same level of measure-ment specificity, the opposing arguments collide. Academic self-efficacy re-searchers express pessimistic views that self-concept can ever be assessed at

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    18 Bong and Skaalvik

    task-specific or problem-specific levels (Bong and Clark, 1999; Pajares, 1996).Academic self-concept researchers, on the other hand, question the practicalutility of self-efficacy judgments beyond what they view as microlevel anal-yses of performance. The problem worsens because both self-concept andself-efficacy theories contend that their construct can be assessed at varyinglevels of measurement specificity (Bandura, 1997; Shavelson et al., 1976).

    Self-concept items rarely specify what constitutes successful academicperformance. This omission forces respondents to come up with aggregatedevaluations of themselves in the particular domain. Rosenberg (1968) oncestated that . . . a mans global self-esteem is not based solely on his as-sessment of his constituent qualities; it is based on his self-assessments ofqualities that count (p. 339). This seems true for global self-esteem and ap-plicable to domain-specific self-concepts. Students are asked to make evalua-tions of their competence in academic domains without being provided withexplicit information about criteria. Specific performance criteria that indi-viduals should take into account in appraising their competence are largelyleft to the individuals to decide. As a result, competence information thatis most salient and readily accessible in ones self-schema in the domain ofinterest tends to dominate the perceptions of self.

    An important requirement in the self-efficacy measurement is that itshould be tailored so as to directly correspond to the specific target per-formance (Bandura, 1997; Pajares, 1996; Zimmerman, 1995). The problemof aggregating different dimensions and impressions regarding the self thusbecomes fairly irrelevant in academic self-efficacy estimation. Ordinarily, im-portant features of tasks that could wield tangible influence on performanceoutcomes are clearly spelled out in self-efficacy items (Bandura, 1997). Thiscontext-specificity helps respondents to focus on the suggested dimensionsand to reach more accurate assessment of their capabilities regarding theparticular tasks (Mischel, 1977). Stronger relations of academic self-efficacybeliefs with diverse performance measures have been reported when thecontent and specificity of self-efficacy measures corresponded closely withcriterial performance (Joo, Bong, and Choi, 2000; Multon, Brown, and Lent,1991; Pajares and Miller, 1995).

    One recent study provides a good example that highlights the differ-ences between the measurements of two constructs. Lau, Yeung, Jin, andLow (1999) assessed four skill-specific self-concepts in English: listening,speaking, reading, and writing self-concepts. If we strictly concern ourselveswith the measurement specificity of the scale, these skill-specific academicself-concepts appear to be at the same level of specificity as task-specificacademic self-efficacy (e.g., writing self-efficacy; Pajares et al., 1999). Whenwe examine the questions, however, we soon realize that the differenceis more profound than the measurement level per se. Lau et al. assessed

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 19

    students self-concepts in different skill areas by substituting the school sub-ject portion of the Self Description Questionnaire items with each skill area.For example, English speaking self-concept was assessed with items such asCompared with other students, Im good at speaking (in English), Imhopeless when it comes to speaking (in English), and I have always donewell in speaking (in English). Respondents were students in Hong Kongwho were enrolled in English-as-a-second-language program.

    In providing definitions of these skills, the authors wrote, to these stu-dents, listening typically refers to understanding English speeches in formaland social situations and in academic and nonacademic contexts; speakingrefers to activities such as the delivery of a talk or having a conversation withanother person in class and out of class; reading refers to the comprehensionof written prose, understanding of vocabulary, and study for academic andnonacademic purposes; and writing refers to written work leading to essays,reports, and all other work in the written form as required academically intheir respective disciplines at the university (p. 749). However, these areassumptions made by the researchers on how each skill would be interpretedby respondents. When students respond to an item, Compared with otherstudents, Im good at speaking in English, some of them may try to evaluatetheir competence on the basis of their capability for carrying out casual En-glish conversations, whereas others may concentrate on their ineptness formaking public speeches and class presentations in English. The assessmentlevels are now skill-specific, but the aggregated judgments of competence ineach skill area are still being solicited.

    It is perhaps a useful exercise to try to come up with self-efficacy itemsfor this particular skill area. An assumption is made that the target of pre-diction is students proficiency in speaking in English in their everyday life.Three items are readily conceivable on the basis of the authors definition ofspeaking in English (Lau et al., 1999): How confident are you that you cansuccessfully deliver a talk in English in front of your class? How confidentare you that you can carry out English conversations in class? and Howconfident are you that you can successfully carry out conversations in Englishoutside your class? More detailed examples of academic self-concept andself-efficacy items at task-specific and subject-specific levels of measurementspecificity are provided in Tables II and III, respectively. Again, regardless ofwhether the measurement level is specific or general, self-concept items seekout students overall reactions toward the given area, whereas self-efficacyitems specify different aspects or levels in the expected target performance.

    Earlier, we suggested that domain-specificity is further demonstratedwhen self-perceptions in one area relate only to relevant outcomes in thesame content area and not to those in different areas. Academic self-conceptresearchers have repeatedly demonstrated that students self-concept in a

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    20 Bong and Skaalvik

    Table II. Sample Academic Self-Concept and Self-Efficacy Items at Task-Specific MeasurementLevels (e.g., Writing)

    Writing self-concept Writing self-efficacy

    I have always done well inwriting.

    How confident are you that you can . . .

    Work in writing is easy for me. correctly spell all words in a one-page story orcomposition?

    Compared with others my age Iam good at writing.

    correctly punctuate a one-page story or composition?

    I get good marks in writing. correctly use parts of speech such as nouns, verbs,adjectives, or adverbs?

    I learn things quickly in writing. write a simple sentence with good grammar?Im hopeless when it comes to

    writing.acorrectly use singulars and plurals, verb tenses,

    prefixes, and suffixes?It is important to me to do well

    in writing.write a strong paragraph that has a good topic sentence

    or main idea?I am satisfied with how well I

    do in writing.write a paragraph with details that support the topic

    sentence or main idea?organize sentences into a paragraph that clearly

    expresses an idea?write a well-organized and well-sequenced paper that

    has a good introduction, body, and conclusion?

    Note. Self-concept items were adapted from the Academic Self Description Questionnaire I(Marsh, 1999a); Self-efficacy items were reprinted from Pajares, Miller, and Johnson (1999)with permission from the first author.aNegatively worded items.

    particular school subject relates most strongly with achievement indexesin the same subject area. Its relations to achievement measures in otherschool subjects are considerably weaker (e.g., Byrne and Shavelson, 1986;Marsh, 1992; Marsh, Byrne, and Shavelson, 1988; Skaalvik and Rankin, 1995;Skaalvik and Vals, 1999). Self-efficacy investigators typically include mea-sures that belong to a single academic domain and, as such, have not fre-quently tested whether the within-domain relations between self-efficacyand performance are stronger than their cross-domain relations. However,several recent studies reported evidence of strong content-specificity of aca-demic self-efficacy beliefs that is comparable to that obtained in academicself-concept research. Joo et al. (2000), for instance, measured students biol-ogy self-efficacy, Internet self-efficacy, written biology test performance, andInternet biology test performance during Web-based instruction in biology.Students biology self-efficacy predicted their written biology test scores,whereas students self-efficacy for using the Internet predicted their biologytest performance based on the Internet search. Bong (2002) also reportedthat when multiple self-efficacy and achievement indexes in English andmath entered the same predictive equation, English self-efficacy emerged asthe sole predictor of English performances, with math self-efficacy as the sole

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 21

    Table III. Sample Academic Self-Concept and Self-Efficacy Items at Subject-SpecificMeasurement Levels (e.g., Math)

    Math self-concept Math self-efficacy

    Mathematics is one of my bestsubjects

    How confident are you that you can . . .

    I often need help inmathematics.a

    pass mathematics at the end of this term?

    I look forward to mathematicsclasses.

    pass mathematics at the end of this term with a gradebetter than a D?

    I have trouble understandinganything with mathematics init.a

    get a grade better than a D+ in mathematics?

    I enjoy studying formathematics.a

    get a grade better than a C in mathematics?

    I do badly in tests ofmathematics.a

    get a grade better than a C in mathematics?

    I get good marks inmathematics.

    get a grade better than a C+ in mathematics?

    I never want to take anothermathematics course.a

    get a grade better than a B in mathematics?

    I have always done well inmathematics.

    get a grade better than a B in mathematics?

    I hate mathematics.a get a grade better than a B+ in mathematics?get a grade better than an A in mathematics?get an A in mathematics?

    Note. Self-concept items were reprinted from the Self Description Questionnaire II (Marsh,1999b); Self-efficacy items were adapted from Zimmerman and Bandura (1994) withpermission from the first author.aNegatively worded items.

    predictor of math performances. Relations of self-efficacy in one academicdomain to performance scores in the other domain were not significant.Therefore, although the context-specificity issue is still a ground for muchdebate, evidence is fairly consistent that both academic self-beliefs reflectdomain-specific judgments.

    DIMENSIONALITY AND STRUCTURE

    Self-concept is a multidimensional construct that is differentiated acrossdomains of functioning. These domain-specific perceptions are hierarchicallystructured with the most general perceptions at the apex of the hierarchy(Shavelson et al., 1976). With regard to the academic arena, Shavelson etal. hypothesized that a general academic self-concept would subsume morearea-specific self-concepts. Although their basic tenets of multidimensional-ity and hierarchy of self-concept still hold, researchers later discovered thatthe nature of academic self-concept hierarchy was slightly different fromwhat Shavelson et al. originally envisioned. Specifically, students academic

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    22 Bong and Skaalvik

    self-concepts in the verbal and math areas are so weakly correlated that theycannot be represented by a single general academic self-concept (Byrne andShavelson, 1986; Marsh, 1990c; Marsh et al., 1988). Consequently, the aca-demic portion of the Shavelson hierarchy was revised to represent twoverbal and mathhigher order academic self-concept factors (Marsh andShavelson, 1985).

    Although the multidimensionality of self-concept is rarely disputed, re-searchers do not always agree on the hierarchical nature of self-conceptstructure (Harter, 1990). Many different orientations exist as regards theinternal structure of self-concept (Byrne, 1984, 1996). Harter (1998) ques-tioned the validity of self-concept hierarchy stating that . . . one has toask whether the statistical structure extracted does, in fact, mirror the psy-chological structure as it is phenomenologically experienced by individuals(p. 579). Evidence is not conclusive (Marsh and Yeung, 1998) but tends tosupport potential self-concept hierarchy (Byrne and Shavelson, 1986; Byrneand Worth Gavin, 1996; Vispoel, 1995). Several recent studies demonstratedthat skill-specific self-concepts within a domain (i.e., speaking, reading, andwriting English self-concepts) formed a higher order English self-conceptfactor. Moreover, this second-order English self-concept factor was foundto be equivalent to an independently assessed global English self-conceptfactor. English self-concepts that were empirically extracted from more spe-cific self-concept factors demonstrated correlation coefficients close to 1.0with directly assessed English self-concepts (Lau et al., 1999; Yeung et al.,2000, Studies 3 and 5). These investigations provide much stronger supportfor the hierarchical nature of academic self-concept.

    Evidence suggests that academic self-efficacy perceptions may also forma multidimensional and what can be described as a loosely hierarchicalstructure. Students make reliable differentiation between their self-efficacyjudgments across different academic domains (Bong, 1997; Bong and Hoce-var, in press). The degree of such differentiation varies somewhat dependingon gender, grade, and levels of prior knowledge (Bong, 1999, 2001a). Stu-dents also make a distinction, within a given subject area, between theirefficacy beliefs at different levels of measurement specificity (Bong, 2001b;Lent, Brown, and Gore, 1997) or toward different aspects of required skills(Shell, Colvin, and Bruning, 1995; Shell, Murphy, and Bruning, 1989). Aswas the case with academic self-concept, two higher order factors, verbaland quantitative academic self-efficacy, normally embrace more area-specificacademic self-efficacy beliefs. This finding is consistently observed regardlessof whether self-efficacy beliefs were assessed with specific problems (Bong,1997) or with subject-level self-efficacy statements (e.g., Im certain that Ican do an excellent job on the problems and tasks assigned for [a specificsubject] class; Bong, 2001a).

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 23

    However, questions still remain as to whether the internal structure ofself-efficacy belief resembles the hierarchical organization of self-concept.Although Bongs series of studies (Bong, 1997, 1999, 2001a, 2001b) pro-vided enough evidence to confirm the generality and multidimensionalityof academic self-efficacy beliefs, they have not yet provided direct evidenceto confirm the hierarchical organization of self-efficacy beliefs. It needs tobe demonstrated, as self-concept researchers have (Lau et al., 1999; Yeunget al., 2000), that the common factor underlying more specific self-efficacybeliefs is equivalent in content to the self-efficacy beliefs that are directlyassessed at the more general level. Even when such an attempt could besuccessful empirically, Bandura (1986) warns against the danger in comingup with such a simplified measure and questions its usefulness in predict-ing behavior. He wrote, The most informative efficacy analysis requiresdetailed assessment of the level, strength, and generality of perceived self-efficacy commensurate with the particularity and perceptions with whichperformance is measured. . . . particularized measures of self-perceptsof efficacy surpass global measures in explanatory and predictive power(p. 397).

    There has been at least one consistent discrepancy between what appearto be otherwise similar internal structures. While academic self-concepts inverbal and math areas are nearly uncorrelated (Byrne and Shavelson, 1986;Marsh et al., 1988; Skaalvik and Rankin, 1995), verbal and math academicself-efficacy are almost always highly correlated (Bong, 1997, 2001a; Marshet al., 1991; Skaalvik and Rankin, 1995). Whether this difference reflects atrue construct-related difference or some artifact of methodological proce-dures is not yet known. As discussed previously, the internal and externalframes of reference model of academic self-concept (Marsh, 1986; Marshet al., 1991) explains the near-zero correlation between verbal and mathself-concepts as a result of simultaneous operation of internal and externalcomparison processes. Students do not undergo internal comparison pro-cesses when judging their self-efficacy (Bong, 1998) and, therefore, expressefficacy beliefs in different domains that are more highly correlated.

    Recently, Bong and Hocevar (in press) compared three academic self-efficacy scales that differed in terms of measurement specificity, using a mul-titrait, multimethod framework. Academic self-efficacy factors at differentlevels of specificity were positively correlated within each domain. Further,the types of problems/tasks included in the measure and the subject ar-eas from which these problems/tasks were drawn concomitantly determinedstudents responses to problem-specific and task-specific self-efficacy items.In contrast, students responses to subject-specific self-efficacy items weremore or less uniform within each academic domain and did not differ muchby the individual items. More interestingly, students percepts of efficacy in

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    24 Bong and Skaalvik

    different school subjects were most highly correlated when assessed with spe-cific problems and least highly correlated when assessed with subject-levelstatements. Compared with self-efficacy studies, an overwhelming majorityof contemporary academic self-concept studies use more general-level mea-sures (e.g., Self Description Questionnaire, Perceived Competence Scale).This finding, therefore, suggests the possibility that the difference in thestrengths of relations between verbal and math self-concepts might havebeen created, at least in part, by the different assessment specificity.

    TIME ORIENTATION

    It is worth noting that most academic self-concept items begin withphrases that read I am good . . ., I am hopeless . . ., or I have donewell . . . (see Byrne, 1996). Self-efficacy items usually start with Howconfident are you that you can . . .? How well can you . . .? or I amconfident that I will be able to . . . (see also Pajares, 1996). The wordingof self-concept items tends to direct the attention of respondents towardtheir past accomplishments, whereas that of self-efficacy items focuses theattention of students on their future expectancies (see Wigfield and Eccles,2000, for related discussion).

    Although self-concept and self-efficacy items make salient the past orthe future time frames, respectively, both types of judgments are primarilya product of past experiences. Even when self-concept items refer to thecurrent self, for example, Mathematics is easy for me, such judgments canonly be formed on the basis of ones mathematics achievements in the past.As pointed out by Markus and Nurius (1986), self-concepts are past-orientedbecause relevant information and experiences need to be processed by self-schemas and these schemas are created from individuals past experiences ina particular domain. Self-efficacy perceptions are inherently future-orientedbecause they represent individuals confidence for successfully accomplish-ing the imminent tasks. Yet these expectations, too, are in large part resultsof self-schemas that are created from their earlier experiences.

    The same previous experiences in the domain provide vital informationfor carving both ones self-concept and self-efficacy beliefs. However, indi-viduals do not necessarily reach the same conclusion. Because self-efficacyitems make an explicit reference to outcomes in the upcoming future, thereis ample room for the same individual or for different individuals with sim-ilar achievement records to arrive at drastically different expectations forsuccess. Consider two students who believe that they have always done wellin mathematics and that they are good at mathematics compared with oth-ers their age. These two students may or may not express similar strength

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 25

    of certitude for accomplishing such tasks as correctly solving given equa-tion problems or getting a grade better than a B+ in mathematics at theend of the term. Math self-concept reflects students evaluations of theirgeneral competence in math, whereas math self-efficacy represents theirjudgments of what they could do with their competence for accomplish-ing the specified math tasks. Depending on how students analyze and com-pare the given achievement situation with previous ones involving simi-lar tasks, their confidence for successfully performing each math task canbe strengthened or weakened. On the same token, two students who feelequally efficacious that they can successfully perform the particular mathtasks may or may not regard themselves as equally competent in math(Pajares, 1996).

    The relative emphasis on the past and the future is inevitably intertwinedwith how much specific aspects of the prospective situation should be takeninto account in coming up with a final judgment. When the bases of judgmentsthat are being called for are mostly experiences in the past, there is nocompelling reason either for researchers to provide a detailed description ofthe current situation or for respondents to pay attention to those particulars.Schemas, by definition, are a constellation of commonalities extracted frommany isolated experiences. Individuals overall views of themselves in thearea based on the past self-schema will not change much by the specificsof any single event (Markus, 1977). On the other hand, if students are toreport their likelihood of success on some impending tasks that are yet tobe performed, they need to consider all the available information regardingthese tasks. Otherwise, their judgments cannot be accurate because theirperformance on these tasks could well be determined by the situationalaffordances and constraints. This difference of the past vs. future orientationbetween academic self-concept and self-efficacy logically extends itself totheir difference in temporal stability.

    TEMPORAL STABILITY

    Self-schemata is cognitive generalizations about the self which,when well articulated, should demonstrate cross-situational consistency(Markus, 1977). Consistent with this claim, one of the features thatShavelson et al. (1976) identified as critical to the definition of self-conceptis its stability. Shavelson and Bolus (1982) subsequently reported stabilitycoefficients between 0.56 and 0.81 with general and subject-matter academicself-concepts assessed over a 4-month time lag. More important, these self-concepts appeared more stable than the corresponding achievements. Marshand Yeung (1998, Study 2) also reported that subject-specific as well as global

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    26 Bong and Skaalvik

    academic self-concepts that were assessed in two consecutive years demon-strated high stability coefficients, mostly above 0.70. Presumably because ofits resistance to change, self-concept does not lend itself easily to short-termexperimental manipulations. For example, Craven, Marsh, and Debus (1991)found that, although students domain-specific academic self-concepts wereimproved somewhat by adaptive attributional feedback, these changes wereonly modest at best. All these attest to the relatively unchanging na-ture of academic self-concept. There may be developmental differences inthe stability of these perceptions such that younger students self-conceptsare more flexible, whereas older students self-concepts are more firmly es-tablished (Skaalvik and Hagtvet, 1990; Wigfield et al., 1997). As childrengrow older, their self-concepts also become more highly correlated withothers evaluation of their competence (Wigfield et al., 1997).

    It is interesting to note that stability of self-efficacy beliefs has rarelybeen investigated. Bandura (1997) stated that, once established, perceptionof self-efficacy is resilient to temporary failures. Nevertheless, he emphasizedthat it is fundamentally a context-specific construct that should not be viewedas one of the personality traits. Supporting this claim, Pajares and Graham(1999) reported that math self-concept demonstrated slightly higher stabilitythan math self-efficacy when assessed with a 6-month interval. While mathself-concept scores did not change significantly during this period, math self-efficacy scores did. The authors noted that this might have been because ofthe more demanding nature of the second self-efficacy assessment items andthat more research with measures of similar difficulty are needed. However,this exemplifies well the context-specific and malleable nature of academicself-efficacy judgments. Self-efficacy is a predictive construct that should beassessed before the target performance (Zimmerman, 1996) because thesebeliefs could change greatly upon receiving contextual information.

    In discussing the stability of self-concepts, Shavelson et al. (1976) notedthat as one descends the self-concept hierarchy and assesses self-concepts inmore specific situations, the self-concepts become less stable. Self-concepts atthe apex of the hierarchy are more resistant to change, whereas self-conceptsat lower levels are expected to vary considerably with situations. Self-efficacyis frequently measured at levels that correspond to the situation-specific lev-els of the Shavelson hierarchy. As such, self-efficacy, as typically assessed,represents relatively malleable perceptions. Schunk and his colleagues re-ported repeated successes in experimentally augmenting students efficacyperceptions in a relatively short period of time and in areas where theywere experiencing great difficulty (Schunk, 1982, 1983, 1984; Schunk et al.,1987; Schunk and Cox, 1986; Schunk and Hanson, 1985, 1989; Schunk andSwartz, 1993). These experiments are strong evidence of the dynamic natureof self-efficacy beliefs.

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 27

    PREDICTIVE OUTCOMES

    Both academic self-concept and self-efficacy research underscore thatthe construct is important as a desirable outcome in itself as well as a poten-tial mediator of academic motivation and performance. While recognizingthe conceptual and methodological differences between the two constructs,Marsh et al. (1991) wrote, both self-efficacy and self-concept responses areposited to reflect more than just an objective assessment of existing achieve-ment levels. . . . In this sense, self-efficacy and self-concept measuresevenafter partialling out the effects of prior achievementare likely to con-tribute to the prediction of subsequent behaviors that are dependent onactive choice, motivation, and sustained effort (p. 336).

    In accordance with this self-enhancement view, numerous studies havedocumented strong relations between measures of academic self-concept oracademic self-efficacy and a variety of motivational and performance indi-cators. Academic self-concept has been shown to relate systematically toteachers ratings of level of engagement and persistence in classroom activ-ities (Skaalvik and Rankin, 1996b; Skinner, Wellborn, and Connell, 1990),students effort ratings (Skaalvik and Rankin, 1995), help-seeking behavior(Ames, 1983), course-selection (Marsh and Yeung, 1997b), intrinsic moti-vation (Gottfried, 1990; Harter, 1982; Mac Iver, Stipek, and Daniels, 1991;Meece, Blumenfeld, and Hoyle, 1988; Skaalvik, 1997b, 1998; Skaalvik andRankin, 1996b), and achievement (Marsh, 1992; Marsh et al., 1988; Marshand Yeung, 1997a; Shavelson and Bolus, 1982; Skaalvik and Hagtvet, 1990;Skaalvik and Vals, 1999).

    Academic self-efficacy beliefs have been found to strongly relate to taskchoice (Bandura and Schunk, 1981; Pajares and Miller, 1995), career selec-tion (Betz and Hackett, 1981, 1983), persistence and performance (Bandura,Barbaranelli, Caprara, and Pastorelli, 1996; Lent, Brown, and Larkin, 1986;Multon et al., 1991; Pajares and Miller, 1994; Pajares et al., 1999; Pajaresand Johnson, 1996; Schunk, 1981, 1982, 1983, 1984; Schunk and Cox, 1986;Schunk and Hanson, 1985, 1989; Schunk and Swartz, 1993), grade goalsand academic aspirations (Bandura et al., 1996; Zimmerman et al., 1992;Zimmerman and Bandura, 1994), cognitive strategy use and self-regulation(Pintrich and De Groot, 1990; Wolters and Pintrich, 1998), perceived value(Bong, 2001b; Meece, Wigfield, and Eccles, 1990), mastery goal orientation(Bong, 2001a; Meece and Holt, 1993; Roeser, Midgley, and Urdan, 1996;Skaalvik, 1997b), and intrinsic interest and self-satisfactions (Zimmermanand Kitsantas, 1997, 1999).

    Although both theories emphasize the predictive and explanatory roleof these self-judgments, academic self-concept and self-efficacy have tra-ditionally been paired with slightly different sets of outcomes (Bong and

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    28 Bong and Skaalvik

    Clark, 1999). Favorite outcomes of academic self-concept research includecourse grades, standardized achievement test scores, intrinsic motivation,and anxiety. Self-efficacy investigations normally include measures of goalsetting, persistence, effort expenditure, and specific task performance. Morerecent studies in both areas are less bound by these traditional outcomes.For example, Skaalvik (1997b, 1998) demonstrated that self-concept relatedpositively to student goal setting. Skaalvik and Skaalvik (2000) showedthat both self-concept and self-efficacy correlated positively with masteryand self-enhancing ego (performance-approach) goal orientation and nega-tively with self-defeating ego (performance-avoid) goal orientation. Pajaresand his colleagues (Pajares et al., 1999; Pajares and Kranzler, 1995; Pajaresand Miller, 1994) demonstrated that self-efficacy is a significant predictor ofanxiety.

    In general, self-concept better predicts affective reactions such as anxi-ety, satisfaction, and self-esteem, whereas self-efficacy better predicts cogni-tive processes and actual performance. Such relative superiority notwith-standing, both constructs have been found useful for predicting similaroutcomes. Because self-efficacy researchers have used both correlationaland experimental designs, self-efficacy effects are more clearly established(see Pajares, 1997, for an overview). Self-concept researchers, primarily us-ing survey designs and correlational analyses, are still debating the causalrelations between self-concept and achievement (see Skaalvik, 1997a, for anoverview). Nevertheless, academic self-concept and academic self-efficacyresearch shares the basic premise that the construct plays a significantrole in enhancing students intrinsic motivation, positive emotion, andperformance.

    WHAT NOW? SOME DIRECTIONS FOR FUTURE RESEARCH

    The purpose of this article was to illuminate conceptually the similar-ities and differences between academic self-concept and self-efficacy, theself-constructs known to wield critical influence on students academic at-tainment and psychological well-being in school. The two conceptualizationsshare important similarities such as their treatment of perceived competenceas the most integral element in construct definition and assessment. Bothself-beliefs use prior mastery experience, social comparison, and reflectedappraisals as major information sources. Beliefs of academic self-conceptand self-efficacy are also domain-specific and multidimensional such thatstudents hold perceptions that are unique to each academic domain andreasonably differentiated across diverse areas. These two constructs predictsubsequent motivation, emotion, and performance to varying degrees.

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    Self-Concept vs. Self-Efficacy 29

    There are also important differences. Some definitions of academic self-concept include cognitive evaluations of capability along with affective re-actions toward results of such evaluations. Self-efficacy theorists make cleardistinction between these components and consider affective reactions as aseparate construct that is mainly a consequence of self-efficacy perceptions.While competence evaluation in self-concept relies heavily on social com-parison and hence tends to be normative, self-efficacy evaluation is primarilygoal-referenced and most strongly affected by ones enactive experiences.Academic self-concept reflects an aggregated judgment or overall impres-sion of ones competence in given academic domains. As such, self-conceptbeliefs tend to be past-oriented, stable over time, and resistant to change.Academic self-efficacy reflects a highly context-specific judgment of onescompetence, although repeated successes or failures make these beliefs moredurable. The dynamic and malleable nature of self-efficacy perceptions ren-ders them more amenable to experimental procedures aiming at efficacyenhancement.

    Need to Separate Multiple Components of Academic Self-Concept

    Presently, the dominant view of academic self-concept is that it is acollection of a host of related perceptions: competence, self-worth, interest,enjoyment, and intentions, to name a few. It is conceivable that self-conceptmeasures, which reflect this complexity, better predict outcomes that arejointly influenced and determined by these factors. Such outcomes tend toinvolve choice and performance measures at more general levels of speci-ficity. Although this composite view toward academic self-concept may in-deed mirror students actual thought patterns in certain situations, severalresearchers demonstrated the need as well as usefulness of separating outsome of these components. Wigfield et al. (1997), for example, reported thateven elementary school children were able to differentiate their perceivedability and interest within the same activity domains. Studies that distinguishbetween competence and task-value perceptions provide evidence that eachpredicts certain outcomes better than the other. Generally, perceived com-petence predicts academic performance better, whereas task-value predictschoice behaviors better (e.g., Meece et al., 1990).

    Self-efficacy researchers have argued that self-efficacy is the most usefulself-construct because it predicts subsequent motivation and performancebetter than the other constructs, including self-concept (e.g., Bandura, 1986;Pajares and Miller, 1994). One reason for its stronger predictive power owesto the fact that it purposefully avoids intermixing different components un-der the rubric of self-efficacy. Instead, it concentrates on students subjective

  • Educational Psychology Review [jepr] pp686-edpr-455576 November 21, 2002 18:35 Style file version June 4th, 2002

    30 Bong and Skaalvik

    judgments of capability to perform given academic tasks successfully at des-ignated levels (Schunk, 1991). Rather than assessing omnibus views of selfthat include perceived competence and affect, efficacy researchers study var-ious self-processes (e.g., self-evaluation, self-satisfaction, affect) separatelyfrom self-efficacy within the cyclical phases of self-regulation (Zimmerman,2000). For example, self-efficacy affects goal setting, which influences self-evaluation and self-satisfaction/affect during the subsequent self-reflectionphase, the results of which, in turn, influence intrinsic interest/value, outcomeexpectations, and subsequent self-efficacy.

    We believe that academic self-concept research would also benefit fromseparating perceived competence components from other elements and ex-amining the specific contributions of each major constituent. This approachshould generate specific guidelines for how these components are linkedwithin the broader self-system and for when each of them is most useful forpredictive and explanatory purposes.

    Self-Efficacy as an Active Precursor of Self-Concept

    The previous discussion on the separability of multiple self-conceptcomponents and the centrality of perceived competence among those com-ponents bring to light yet another closely related set of issues. These issuesinclude whether the perceived competence components of academic self-concept are equivalent to self-efficacy judgments and, if so, how self-efficacybeliefs influence the development of self-concept.

    Lent et al. (1997) investigated whether self-concept subsumed self-efficacy components by subjecting various measures of academic self-concept and self-efficacy to confirmatory factor analysis. In that study,self-concept and self-efficacy formed correlated but separate factors. Theresearchers thus concluded that self-concept did not appear to subsumeself-efficacy. However, they did not examine th