Recent research on self-concept has emphasized multidimensionality such that self-concept and its relation to other constructs are distinct in each specific domain. The present study tests the hypothesis that self-concepts in creative arts are not only multidimensional, but are also highly related to family backgrounds in specific domains. Confirmatory factor analysis of survey responses from 312 Australian university students in teacher education showed that, in support for domain specificity, self-concepts in music, visual art, dance and drama were distinct constructs. Each of these distinct self-concepts was more positively correlated with family background in its corresponding domain than with other domains. Family background was also found to correlated more negatively with anxiety in corresponding domains than with other domains. The results show that, in creative arts education, family background may be an important factor for the formation of self-concept and anxiety. There is also further support for the multidimensionality and domain specific relations of self-concept with other constructs.

Since Shavelson, Hubner, and Stanton (1976) proposed a multidimensional, hierarchical model of self-concept, researchers have attempted to investigate the structure of students' self-concept. Recent research has provided strong support for the multidimensionality of self-concepts in various areas (e.g., Byrne & Gavin, 1996; Marsh, Hey, Johnson, & Perry, 1997; Marsh & Yeung, 1997a, 1998; Vispoel, 1993, 1995). In particular, the relations of self-concept with other educational outcomes and psychological constructs have been found to be very domain specific (e.g., Marsh & Yeung, 1996, 1997a, 1997b; Yeung, Chui, & Lau, in press; Yeung & Lee, 1999). Even within a curriculum domain such as creative arts, recent research has suggested that the multidimensionality is so strong that students are able to distinguish between their self-concepts in different artistic areas such as music, visual art, dance and drama (Vispoel, 1993, 1995; Yeung, McInerney, & Russell-Bowie, 1999). Given general speculations that artists in a specific field are often a product derived from a combination of inborn capabilities and family influence such that the musician's offspring is more likely to become a musician whereas the actor's offspring is more likely to become an actor, the question of whether there is a close relation between family background and self-concept is important to pursue. Furthermore, if self-concept is associated with family background, then whether such a relation is also domain specific is important both for the understanding of artistic self-concept and for practical purposes of self-concept enhancement. In the present study, we examined the relations among self-concept, anxiety, and family background in four creative arts areas: music, visual art, dance and drama. Specifically, we examined whether (a) family background and self-concept in each artistic areas are positively correlated, (b) both family background and self-concept are negatively correlated with anxiety, and whether (c) these relations are domain specific.

Self-concept has been recognized as both an important educational outcome and an important factor that contributes to other desired educational outcomes (Marsh, 1993). Recent research has shown close relations between academic self-concept and academic achievement (e.g., Chapman & Tunmer, 1997; Hay, 1997; Helmke & Aken, 1995; Marsh & Yeung, 1997a; Muijs, 1997; Yeung & Lee, 1999). Extending studies on expectancy-value models of achievement motivation (e.g., Eccles & Wigfield, 1995; Wigfield & Eccles, 1992), Marsh and Yeung (1997b; also see Yeung, Chui, & Lau, in press) found significant effects of academic self-concept on high school students' choice of coursework. More importantly, the relations between academic self-concept and academic affects and behaviour have been found to be domain specific (e.g., Marsh & Yeung, 1996, 1997a). Thus, for example, students clearly distinguished between self-concepts in English and in maths, and these self-concepts influenced academic performance and behaviour only in their respective curriculum domains. On the basis of these findings, recent research on self-concept and related issues has emphasized domain specificity and the multidimensional nature of self-concept (Byrne & Gavin, 1996; Cross & Markus, 1994; Harter, 1996; Hattie, 1992; Marsh, 1993; Marsh, Byrne, & Shavelson, 1988; Marsh & Yeung, 1997a; Yeung & Lee, 1999; but also see Lau, Yeung, Jin, & Low, in press). This emphasis has also led to the development of various instruments for measuring domain-specific self-concepts (e.g., Marsh, 1992, 1993; Marsh, Hey, Johnson, & Perry, 1997; Vispoel, 1993, 1995).

Researchers have generally assumed an important relationship between creative arts and self-concept development (e.g., Liebler, 1997; Lowenfield, 1987; Vander Ark, 1989). For example, in early childhood education, creative opportunities for infants and toddlers are generally perceived as a powerful source for enhancing self-concept (Hendrick, 1990). Researchers have indicated significant relations between specific creative arts areas and self-concept or self-esteem. For example, Kalliopuska (1989; 1991) found that involvement in classical ballet by girls in the Finnish Junior Ballet Company promotes the development of sound self-esteem and increased self-confidence. Berryman-Miller (1988) found that dance and movement programs tended to raise the self-esteem of retired adults. Drama activities have been found to improve the self-concepts of 4^th grade emotionally disturbed students (Buege, 1993), children with severe behavior disorders (Warger & Kleman, 1986), and hearing impaired students (Davies, 1984). Huntsman (1982) also found that drama increased college students' self-confidence. From the perspective of a multidimensional self-concept, the question is whether such a relationship is so domain specific that a musical family background, for example, would be highly related with musical self-concept rather than self-concept in other creative arts areas. For example, whereas Poll (1979) claimed that dance has the potential to enhance children's self-concepts, there was no indication of whether we should expect children's self-concept in general or self-concept specifically in the dance domain to be enhanced. In their study comparing 20 high school students participating in drama with 20 control students, Beales and Zemel (1990) did not find significant difference between their self-esteem. Whereas self-esteem is often taken as a global representation of one's self-concepts (Marsh & Yeung, 1999), perhaps the effects of creative arts activities on the development of self-concept is more domain specific than in general terms.

Until recently, few studies have seriously examined students' self-concept in artistic curriculum domains with the application of strong methodology. Two of the few studies with a strong research design were Vispoel's (1993, 1995) studies that showed that students not only distinguished their artistic self-concept from their self-concepts in other curriculum areas, but also distinguished among the four different areas of creative arts considered. This strong domain specificity may be consistent with what most people have assumed about the association between family background and the performance and perceptions of artists in respective artistic areas. For example, one may assume that a child from a musical family background would probably feel more musical and are more likely to develop into a musician than a child from a non-musical family background. At least, history has shown that most of the famous musicians such as Beethoven, Mozart, Bach, Stravinsky and Johann Strauss all came from very musical families and were steeped in music-making from a very young age (Microsoft Corporation, 1997). It would be surprising to find any instance where a child from an extremely musical family became a visual artist of world renown instead of becoming a musician.

There has been at least some support in empirical research for the relation between parental support in creative arts and children's performance and perceptions. For example, Russell-Bowie (1993) has shown that family background factors such as income, language background, and urban living were significant factors related to the amount of musical experiences children have at home, the priority their local primary schools give to music, and the priority their parents give to music. In a study with 168 kindergarten children, Liikanen (1975) found that the more interest the parents had in art and creative activities, the more creative their children tended to be. Again, however, it was unclear whether such relations would be domain specific such that a child from a strong family background of visual art would have more positive perceptions about themselves in visual art than in other creative arts areas.

Although recent research has shown noteworthy relations of academic self-concept with academic achievement (e.g., Chapman & Tunmer, 1997; Hay, 1997; Helmke & Aken, 1995; Marsh & Yeung, 1997a; Muijs, 1997; Yeung & Lee, 1999) and with other psychological constructs (e.g., Marsh & Yeung, 1997b, Yeung, Chui, & Lau, in press), few studies have examined the relation between self-concept and anxiety, particularly from a multidimensional perspective. Previous studies have suggested associations between creative arts and anxiety but it was unclear whether these associations should be domain specific. For example, Davis (1985) found that creative drama tended to lower the anxiety level of older adults. Liebler (1997) claimed that moving, singing and having fun could help children release stress and tension. In their study on 107 drug users in Kuwait, Alnajjar and Clarke (1996) found a significantly negative relationship between levels of self-esteem and anxiety. Examining the relations between self-concept and depression, Tarlow and Haaga (1996) found that low self-esteem was associated with anxiety and depression. However, none of these studies has examined the potentially domain specificity of such relations.

Marsh and Yeung (1996) has shown that academic affects such as whether the student looks forward to maths classes, feels that maths will be useful, and is afraid to ask questions in maths are domain specific such that the student can distinguish these academic affects between curriculum domains. It is therefore reasonable to expect that anxiety in creative arts should be domain specific, similar to self-concept and perhaps family background. Partly in support of this speculation, Bandalos, Yates, and Thorndikechrist (1995) found that maths self-concept was negatively related to test anxiety both in general and in a specific domain such as statistics, which is logically associated with maths. Also, based on Hay, Ashman, and Vankraayenoord (1998) who found in their study with 515 preadolescents from 18 schools that students with high self-concept were lower in anxiety and had more supportive families, we hypothesized that (a) self-concepts in creative arts are domain specific in four different areas, viz., music, visual art, dance and drama, (b) the relations between family background, self-concept, and anxiety are domain specific, and (c) in each creative arts area, self-concept and family background are positively correlated but both are negatively correlated with anxiety.

The participants were 329 university students enrolled in a teacher education program in a university in Sydney, Australia (90% females), where Creative Arts is one of the key learning areas in the schools of the state and comprise a crucial component in the teacher education program. They were training to be generalist primary school teachers, not specialists in any of the particular creative arts areas. Consent to participate in the study was obtained from the participants before they completed the survey. After listwise deletion of missing data, 231 completed surveys were analyzed.

The 60 items pertaining to three a priori scales for each of four Creative Arts domains are listed in Appendix. For each domain, there were four items for family background, six for anxiety, and five for self-concept. To each of the response items the participants responded on a 5-point scale (1 = strongly disagree; 5 = strongly agree). Thus, higher scores mean stronger family background, higher anxiety, and higher self-concept, respectively for the domain-specific measures in respective areas.

Preliminary analysis included alpha estimates of internal consistency of each of three measures in each domain and confirmatory factor analysis (CFA) measurement models for each scale in each domain separately. The conduct of CFA has been described elsewhere (e.g., Bollen, 1989; Byrne, 1998; Joreskog & Sorborm, 1993; Marsh, 1992; Marsh & Hocevar, 1985; Pedhazur & Schmelkin, 1991) and is not further detailed here. All analyses throughout this paper were conducted with the SPSS version of PRELIS and LISREL (Joreskog & Sorbom, 1988). The goodness of fit of models is evaluated based on suggestions of Marsh, Balla, and McDonald (1988) and Marsh, Balla, and Hau (1996) with an emphasis on the Tucker-Lewis index (TLI), but we present also the chi-square test statistic and the relative noncentrality index (RNI). For acceptable model fit (TLI > .9), for parallel items (with similar wording) correlated uniqueness were included in the models (Marsh, 1993; Marsh & Hau, 1996).

A series of CFA models, based on a 60 x 60 covariance matrix, tested whether the Family factor is domain specific or can be taken as one single factor (Model 1), whether the Anxiety factor is domain specific (Model 2), whether Self-concept is domain specific (Model 3). Then based on Models 1, 2 and 3, we tested in Model 4 the distinctiveness of all 12 factors (3 factors x 4 domains) and examined the correlations of the three constructs in matching and nonmatching domains. Finally, we tested the hypothesis that the paths from Family to Anxiety and Self-concept should be statistically significant only for matching domains. Thus for example, family background in music was expected to lead to significantly lowered anxiety in music and significantly higher self-concept in music but should have negligible impacts on anxiety and self-concept in nonmatching domains.

Preliminary Analysis

The alpha reliability estimate for each of the 12 a priori scales was good (see Appendix). In one-factor measurement models tested with each scale separately, modification indices provided by LISREL suggested the inclusion of one correlated uniqueness in each of the Anxiety scales for improving model fit, and this suggestion was consistent across the four domains. Similarly, for each of the Self-concept scales, one correlated uniqueness was suggested. Although each of the one-factor models provided a reasonable fit to the data even without the correlated uniqueness, we included these correlated uniquenesses in subsequent models.

Models 1, 2 and 3 followed the same logic in testing the multidimensionality of the Family, Anxiety, and Self-concept scales. In family background in music, for example, we first present a four-factor model without correlated uniquenesses (Model 1a) for comparison with the a priori four-factor model with the inclusion of correlated uniquenesses (Model 1b). Because strictly parallel items were used across all four domains, there were a total of 6 x 4 = 24 correlated uniquenesses. Model 1c tested the possibility that all 16 Family items formed a single Family factor. Support for the multidimensionality of the Family scales requires Model 1b positing domain-specific factors to fit the data better than Model 1c positing a single factor. Table 1 shows the goodness of fit of the models tested. All models converged to proper solutions with no unreasonable parameter estimates. Model 1b provided a much better fit (TLI = .940) than Model 1a, providing evidence for the necessity of correlated uniquenesses for model fit; and a much better fit than Model 1c, providing support for the domain specificity of the Family constructs. The factor coefficients for Model 1b ranged from .55 to .80, and the factor correlations ranged from .35 to .54.

Similar analyses were conducted for the Anxiety and Self-concept scales in Models 2 and 3. Again because strictly parallel items were used across domains, in addition to the one correlated uniqueness found in preliminary one-factor models for each Anxiety factor, there were a total of 6 x 6 + 4 = 40 correlated uniquenesses for Model 2. Similarly, there were a total of 6 x 5 + 4 = 34 correlated uniquenesses for Self-concept in Model 3. The pattern of results for Models 2 and 3 was similar to that for Model 1. That is, Model 2b (TLI = .946) provided a better fit than Models 2a and 2c (factor coefficients for Model 2b ranged from .49 to .82, factor correlations ranged from .48 to .72), and Model 3b (TLI = .968) provided a better fit than Models 3a and 3c (factor coefficients for Model 3b ranged from .39 to .79, factor correlations ranged from .39 to .58). These results supported the domain specific nature of the Family, Anxiety, and Self-concept constructs when considered separately.

Model 4 further tested the distinctiveness and domain specificity of all the 12 scales considered here. Model 4 (Table 1) had a total of 98 correlated uniquenesses included. It converged to a proper solution with an acceptable fit (TLI = .900) and reasonable parameter estimates (factor coefficients ranging from .37 to .81, factor correlations ranging from -.60 to .93). These mostly low to moderate correlations among constructs support the distinctiveness of the 12 scales, although the correlation between Family and Self-concept in each domain tends to be high (rs > .8) and three of them > .9. An inspection of the correlations among the constructs (same estimates as in Model 5 presented in Table 2) found that the correlations between Family and Anxiety were all negative but were more negative in matching domains (-.58, -.56, -.48, and -.54, respectively in music, visual art, dance and drama) than in nonmatching domains (e.g., Family in the music domain and Anxiety in the Dance domain). The correlations between Family and Self-concept were all positive but were more positive in matching (.93, .91, .93, and .80 in music, visual art, dance and dram, respectively) than in nonmatching domains. These results suggest strong domain specificity of the relations of family background with anxiety and self-concept in Creative Arts.

Model 5 provided further evidence for the domain specificity of the relations of family background with anxiety and self-concept. Model 5 is a full path model including all possible paths from the Family constructs in four domains to the other eight constructs (2 constructs x 4 domains). Strong domain specificity of the relations of the constructs requires the path from Family to Anxiety to be more negative and from Family to Self-concept to be more positive in matching domains that in nonmatching domains. The results show that, as expected, the paths from Family to Anxiety were significantly negative only in matching domains (-.65, -.56, -.41, -.60, respectively in the music, visual art, dance, and drama domains), and the paths from Family to Self-concept were significantly positive only in matching domains (.92, .97, .93, and .79, respectively in music, visual art, dance, and drama). All the other paths were statistically nonsignificant. These result provided further support for the hypothesis that family background had significant relations with anxiety and self-concept and these relations were domain specific.

In general, the results supported our hypothesis in that (a) family background and self-concept in each artistic areas were positively correlated, (b) both family background and self-concept were negatively correlated with anxiety, and (c) these relations were very domain specific. These results provided extended support for the multidimensionality of self-concept found in previous research (Byrne & Gavin, 1996; Marsh, Hey, Johnson, & Perry, 1997; Marsh & Yeung, 1997a, 1998; Vispoel, 1993, 1995). Particularly, the domain specificity of family background, self-concept and anxiety in the four creative arts areas supported Vispoel's (1993, 1995) findings that not only was artistic self-concept distinct from self-concepts in other curriculum areas, but the self-concepts in subdomain areas such as music, visual art, dance and drama were distinct psychological constructs. Thus the domain specificity of these constructs should not be ignored in designing instruction in respective areas. The domain specificity of the family background, self-concept, and anxiety factors in four different creative arts domains calls for the attention of curriculum designers and practitioners in the creative arts field to consider the specific characteristics of these different areas when designing and implementing programs.

The strong relation of family background with self-concept and anxiety suggests the important need for parental involvement in the development of children in artistic domains. Very young children exhibit a natural interest in movement, music and drawing and such interest can either be nurtured or ignored by the parents or other caregivers and can have far reaching effects on the growing child's self-concept in relation to each of the arts domains. Parents are important in shaping the child's attitude to creative arts and therefore should be strongly encouraged to be involved in arts activities with their children.

Family background is highly related to self-concept because it is one of the most salient sources of reinforcement for children's behaviour. Not only are these positive attitudes to music, dance, drama or visual arts modelled in the home environment, but generally if parents are interested in one particular art form they will ensure their children have exposure to a wide variety of experiences in this area, both as a spectator and as a participant. Parents with positive attitudes to the arts are more likely to have their children involved in extra curricular art activities such as music, dance, drama and art lessons and to take them to art galleries, the theatre and concerts. Also they are more likely to have their child attend a school where the arts are a priority, so that the home influence is followed through within the school environment.

The present study shows that student teachers who have had this nurturing environment towards one of the arts from their families and schools tended to have a more positive self-concept and tended to be less anxious about teaching this art form in the classroom. Our schools are reaping the results of many, many years of inadequate arts education in most primary schools as children who have come through this system are now back in schools as teachers or are training to be teachers. Because of the lack of supportive family background and adequate training in each of the arts areas teachers are anxious about, and lack confidence about teaching each of these subjects and therefore often end up omitting the subjects from their program (Russell-Bowie, 1993). The seriousness of the situation has been reflected repeatedly in numerous reports into Arts Education over the past 35 years. Repeating findings in previous reports, the report of the Senate Environment, Recreation, Communications and the Arts References Committee (1995) reiterated that "Generalist primary classroom teachers, because of their own poor arts experience at school, and because of inadequate teacher training, lack confidence to teach the arts. As a result... there is a strong impulse to marginalise the arts in their teaching". This situation has certainly called, in vain, for the serious attention of policy makers, curriculum designers, and teacher educators.

The results of the present investigation with potential teachers also have important implications for teacher education. Student teachers enter their training institution with attitudes to music, dance, drama and visual arts developed over their lifetime. Most of them arrive having had little formal arts education in school and often have negative attitudes and low self-esteem in relation to their ability to teach these subjects (Russell-Bowie, 1997). This will surely affect their confidence in teaching creative arts in the classroom. It is crucial that teacher educators realise this nexus and attempt to break down the negative attitudes and low self-esteem by providing students with positive and successful learning experiences in each of the arts domains. Successful experiences may help break down the barriers and develop confidence and expertise in teaching the various creative arts subjects in the classroom.

Such intervention, even for adults, can be very productive in changing anxiety levels and self-concept in relation to the arts. Positive learning experiences in each of the arts areas have been found to be a significant factor in breaking down barriers, changing students' attitudes, lessening their anxiety, giving them confidence and new teaching ideas, and developing their self-esteem in relation to teaching each of the subjects in their classroom (Russell-Bowie, 1997). Obviously it is better that children receive encouragement and positive modelling from an early age both in the family and the school situation, but this often does not occur, especially when children come from homes of a non-English-speaking background or of low socioeconomic status, particularly those from rural areas (Russell-Bowie, 1993).

There is support for the general speculation that family background in a particular domain would probably influence self-concept which would probably lead to subsequent effective teaching programs in this particular domain. Thus a positive nurturing family background in the area of visual arts would increase a student teacher's self-concept and decrease their anxiety levels both in relation to visual arts and to teaching visual arts. Similarly a positive nurturing family background in the area of music would increase a student teacher's self-concept and decrease their anxiety levels both in relation to music and to teaching music.

Although the multidimensionality and domain specificity of the constructs are supported and the patterns of relations among the constructs are clear and consistent across the four domains, there may be two major limitations in the interpretations of the results. First, the correlation between family background and self-concept in each corresponding creative arts domain was very high (.93, .91, .93, and .80 for music, visual art, dance, and drama, respectively). Thus although the constructs of family background and self-concept were distinguishable, there was doubt about how much overlap there may be between them. Second, although the structural model showed a consistent pattern of a significantly positive impact of family background on self-concept and a significantly negative impact on anxiety in each of the four domains, extreme caution has to be taken in interpretations of these results in terms of causal relationships. To test the causal relations among these constructs, longitudinal data would be required.

Alnajjar, M. & Clarke, D. D. (1996). Self-esteem and trait anxiety in relation to drug misuse in Kuwait. Substance Use & Misuse, 31, 937-943.

Bandalos, D. L., Yates, K., Thorndikechrist, T. (1995). Effects of math self-concept, perceived self-efficacy, and attributions for failure and success on test anxiety. Journal of Educational Psychology, 87, 611-623.

Beales, J. N., & Zemel, B. (1990). The effects of high school drama on social maturity. School Counselor, 38(1), 46-51.

Berryman-Miller, S. (1988). Dance movement: Effects on elderly self-concept. Journal of Physical Education, Recreation & Dance, 59(5), 42-46.

Bollen, K.A. (1989). Structural equations with latent variables. New York: Wiley.

Buege, C. (1993). The effect of mainstreaming on attitude and self-concept using creative drama and social skills training. Youth Theatre Journal, 7(3), 19-22.

Byrne, B. M. (1998). Structural equation modeling with LISREL, PRELIS, and SIMPLIS: Basic concepts, applications, and programming. Mahwah, NJ: Erlbaum.

Byrne, B. M., & Gavin, D. A. W. (1996). The Shavelson model revisited: testing for the structure of academic self-concept across pre-, early, and late adolescents. Journal of Educational Psychology, 88, 215-228.

Chapman, J. W., & Tunmer, W. E. (1997). A longitudinal study of beginning reading achievement and reading self-concept. British Journal of Educational Psychology, 67, 279-291.

Cross, S. E., & Markus, H. R. (1994). Self-schema, possible selves, and competent performance. Journal of Personality and Social Psychology, 86, 423-438.

Davies, D. G. (1984). Utilization of creative drama with hearing-impaired youth. Volta Review, 86(2), 106-113.

Eccles, J. S., & Wigfield, A. (1995). In the mind of the actor: The structure of adolescents' achievement task values and expectancy-related beliefs. Personality and Social Psychology Bulletin, 21, 215-225.

Harter, S. (1996). Historical roots of contemporary issues involving self-concept. In B. A. Bracken (Ed.), Handbook of self-concept (pp. 1-37). New York: Wiley.

Hattie, J. (1992). Self-concept, Hillsdale, NJ: Erlbaum.

Hay, I. (1977). Investigating the influence of achievement on self-concept using an intra-class design and a comparison of the PASS and SDQ-1 self-concept tests. British Journal of Educational Psychology, 67, 311-321.

Hay, I., Ashman, A. F., & Vankraayenoord, C. E. (1998). Educational characteristics of students with high or low self-concept. Psychology in the School, 35, 391-400.

Helmke, A., & van Aken, M. A. G. (1995). The causal ordering of academic achievement and self-concept of ability during elementary school: A longitudinal study. Journal of Educational Psychology, 87, 624-637.

Hendrick, J. (1990). Total learning. Developmental curriculum for the young child (3^rd ed.). Columbus, OH: Merrill.

Huntsman, K. H. (1982). Improvisational dramatic activities: Key to self-actualization? Children's Theatre Review, 31(2), 3-9.

Joreskog, K. G., & Sorbom, D. (1988). LISREL 7: A guide to the program and applications. Chicago: SPSS.

Joreskog, K. G., & Sorbom, D. (1993). LISREL 8: Structural equation modeling with SIMPLIS command language. Chicago: Scientific Software International.

Kalliopuska, M. (1989). Empathy, self-esteem and other personality factors among junior ballet dancers. Perceptual and Motor Skills, 69, 1227-1234.

Kalliopuska, M. (1991). Empathy, self-esteem and other personality factors among junior ballet dancers. British Journal of Projective Psychology, 36(2), 47-61.

Lau, I. C., Yeung, A. S., Jin, P., & Low, R. (in press). Toward a hierarchical, multidimensional English self-concept. Journal of Educational Psychology.

Liebler, S. (1997). The joy of moving, singing, and being silly. Children & Families, 16(3), 20-23.

Liikanen, P. (1975). Increasing creativity through art education among pre-school children. Jyvaskyla Studies in Education, Psychology & Social Research, 29. (ERIC Document Reproduction Service ED114198).

Lowenfield, V. (1987). Therapeutic aspects of art education. American Journal of Art Therapy, 25(4), 111-146.

Marsh, H. W. (1992). Content specificity of relations between academic achievement and academic self-concept. Journal of Educational Psychology, 84, 35-42.

Marsh, H. W. (1993). Academic self-concept. Theory measurement and research. In J. Suls (Ed.), Psychological perspectives on the self (Vol. 4, pp. 59-98), Hillsdale, NJ: Erlbaum.

Marsh, H. W., Balla, J. R., & Hau, K. T. (1996). An evaluation of incremental fit indices: A clarification of mathematical and empirical process: In G. A. Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling techniques (pp. 315-353). Hillsdale, NJ: Erlbaum.

Marsh, H. W., Balla, J. R., & McDonald, R. P. (1988). Goodness-of-fit indices in comfirmatory factor analyses: the effect of sample size. Psychological Bulletin, 103, 391-410.

Marsh, H. W., Byrne, B., & Shavelson, R. J. (1988). A multifaceted academic self-concept: its hierarchical structure and its relations to academic achievement. Journal of Educational Psychology, 80, 366-380.

Marsh, H. W., & Hau, K.-T. (1996). Assessing goodness of fit: Is parsimony always desirable? The Journal of Experimental Education, 64, 364-380.

Marsh, H. W., Hey, J., Johnson, S., & Perry, C. (1997). Elite athlete Self Description Questionnaire: Hierarchical confirmatory factor analysis of responses by two distinct groups of elite athletes. International Journal of Sport Psychology, 28, 237-258.

Marsh, H. W., & Hocevar, D. (1985). The application of confirmatory factor analysis to the study of self-concept. First and higher order factor structures and their invariance across age groups. Psychological Bulletin, 97, 562-582.

Marsh, H. W., & Yeung, A. S. (1996). The distinctiveness of affects in specific school subjects: An application of confirmatory factor analysis with the National Educational Longitudinal Study of 1988. American Educational Research Journal, 33, 665-689.

Marsh, H. W., & Yeung, A. S. (1997a). Causal effects of academic self-concept on academic achievement - structural equation models of longitudinal data. Journal of Education Psychology, 89, 41-54.

Marsh, H. W., & Yeung, A. S. (1997b). Coursework selection - relations to academic self-concept and achievement. American Educational Research Journal, 34, 691-720.

Marsh, H. W., & Yeung, A. S. (1998). Longitudinal structural equation models of academic self-concept and achievement: Gender differences in the development of math and English constructs. American Educational Research Journal, 35, 705-738.

Marsh, H. W., & Yeung, A. S. (1999). The lability of psychological ratings: The chameleon effect in global self-esteem. Personality and Social Psychology Bulletin, 25, 49-64.

Microsoft Corporation (1997). Encarta 97 Encyclopedia. U.S.: Microsoft Corporation.

Muijs R. D. (1997). Predictors of academic achievement and academic self-concept - a longitudinal perspective. British Journal of Education Psychology, 67, 263-277.

Pedhauzur, E. J., & Schmelkin, L. P. (1991). Measurement, design, and analysis: An integrated approach. Hillsdale, NJ: Erlbaum.

Poll, T. (1979). Dance, self esteem, and motor acquisition. Journal of Physical Education & Recreation, 50(1), 64-65.

Russell-Bowie, D. (1993) Policy and Practice in Music Education in NSW State Primary Schools. Unpublished PhD Thesis, University of Wollongong.

Russell-Bowie , D. (1997, July). New sounds in primary school teacher education. Proceedings from the ASME (Australian Society of Music Education) National Conference, Sydney.

Senate Environment, Recreation, Communications and the Arts References Committee (1995). Arts education. Canberra : Commonwealth of Australia.

Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976). Self-concept: validation of construct interpretations. Review of Educational Research, 46, 407-441.

Tarlow, E. M., & Haaga, D. A. F. (1996). Negative self-concept-Specificity to depressive symptoms and relation to positive and negative affectivity. Journal of Research in Personality, 30(1), 120-127.

Vander Ark, S. (1989). Self-esteem, creativity, and music: Implications and directions for research. Suncoast Music Education Forum on Creativity, 105-121 (ERIC Document Reproduction Service ED380342).

Vispoel, W. P. (1993). The development and validation of the Arts Self-Perception Inventory for Adolescents. Educational and Psychological Measurement, 53, 1023-1033.

Vispoel, W. P. (1995). Self-concept in the arts. An extension of the Shavelson model. Journal of Educational Psychology, 87, 134-145.

Warger, C. L., & Kleman, D. (1986). Developing positive self-concepts in institutionalized children with severe behavior disorders. Child Welfare, VLXV(2), 165-176.

Wigfield, A., & Eccles, J. S. (1992). The development of achievement task values: A theoretical analysis. Developmental Review, 12, 1-41.

Yeung, A. S., Chui, H. S., & Lau I. C-Y. (in press). Hierarchical and multidimensional academic self-concept of commercial students. Contemporary Educational Psychology.

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.

Yeung, A. S., McInerney, D. M., & Russel-Bowie, D. (1999, February). Potential teachers' hierarchical, multidimensional artistic self-concept. Paper presented at the International Conference on Teacher Education at the Hong Kong Institute of Education, Hong Kong.

Table 1. Goodness-of-fit Summary for Models

Model c 2 df TLI RNI Null c 2 df

1a. 4 Family factors, no CU 773.58 98 .578 .655 2079.37 120

b. 4 Family factors, 24 CU 147.09 74 .940 .963 2079.37 120

c. 1 Family factor, 24 CU 788.47 80 .458 .638 2079.37 120

2a. 4 Anxiety factors, no CU 2388.47 246 .470 .527 4807.83 276

b. 4 Anxiety factors, 40 CU 388.27 206 .946 .960 4807.83 276

c. 1 Anxiety factor, 40 CU 1629.39 216 .601 .688 4807.83 276

3a. 4 Self-concept factors, no CU 1304.71 164 .453 .528 2606.76 190

b. 4 Self-concept factors, 34 CU 183.50 130 .968 .978 2606.76 190

3. 1 Self-concept factor, 34 CU 931.50 140 .556 .672 2606.76 190

4. 12 factors, 98 CU 2457.46 1542 .900 .913 12320.39 1770

Note: N = 231. RNI= Relative noncentrality index. TLI= Tucker-Lewis index. CU = correlated uniquenesses.

Table 2. CFA Solution for Model 5: Paths from Domain-specific Family to Anxiety and Self-concept

MUFAM MUANX MUSEL ARFAM ARANX ARSEL DAFAM DAANX DASEL DRFAM DRANX DRSEL

Variable Factor Coefficients

Item1 .67* .78* .56* .80* .71* .56* .72* .81* .54* .75* .76* .50*

Item2 .52* .74* .60* .57* .66* .70* .58* .77* .66* .69* .69* .70*

Item3 .73* .81* .46* .67* .73* .49* .72* .76* .38* .70* .76* .37*

Item4 .79* .67* .74* .73* .59* .70* .73* .72* .71* .66* .69* .76*

Item5 -- .56* .66* -- .54* .75* -- .58* .73* -- .60* .66*

Item6 -- .55* -- -- .61* -- -- .60* -- -- .53* --

Item1 .55* .40* .69* .36* .49* .68* .48* .34* .70* .45* .42* .75*

Item2 .73* .46* .64* .67* .57* .51* .66* .40* .56* .52* .52* .50*

Item3 .46* .35* .79* .55* .47* .76* .48* .43* .85* .52* .42* .87*

Item4 .37* .56* .45* .46* .65* .51* .47* .48* .50* .56* .53* .42*

Item5 -- .69* .56* -- .70* .44* -- .66* .46* -- .64* .57*

Item6 -- .70* -- -- .63* -- -- .64* -- -- .72* --

Path Coefficients (From row to column variables)

MUFAM -- -.65* .92* -- .01 -.07 -- -.09 -.01 -- .05 -.09

ARFAM -- .14 .01 -- -.56* .97* -- -.01 -.06 -- .01 .02

DAFAM -- -.10 .04 -- -.04 .01 -- -.41* .93* -- .04 .08

DRFAM -- .06 -.02 -- .03 -.08 -- -.06 .05 -- -.60* .79*

MUFAM --

MUANX -.58* --

MUSEL .93* -.57* --

ARFAM .46* -.15 .43* --

ARANX -.25* .48* -.22* -.56* --

ARSEL .33* -.15 .42* .91* -.60* --

DAFAM .39* -.27* .39* .36* -.23* .29* --

DAANX -.29* .68* -.23* -.23* .60* -.20* -.48* --

DASEL .35* -.26* .45* .30* -.28* .38* .93* -.49* --

DRFAM .54* -.26* .50* .52* -.28* .39* .49* -.32* .47* --

DRANX -.25* .53* -.18* -.26* .59* -.29* -.23* .72* -.29* -.54* --

DRSEL .38* -.19* .51* .42* -.37* .55* .44* -.28* .60* .80* -.58* --

Note: N = 231. The Creative Arts domains are Music (MU), Visual Art (AR), Dance (DA), and Drama (DR). In each domain there are three factors, namely, Family (FAM), Anxiety (ANX), and Self-concept (SEL). This full path model posits paths from Family factors in all 4 domains to Anxiety and Self-concepts in all 4 domains. Parameters estimates are completely standardized. *p < .05

Parallel items for four Creative Arts domains in the survey:

1. In my leisure time I am often involved in (musical) activities.

2. My family is often involved socially in (musical) activities.

3. Outside of school, I have had a lot of (music) lessons.

4. I have had a good background in (music).

5. I would feel anxious if I had to teach a (music) lesson.

6. I would feel anxious if I had to assess children in relation to (music) activities.

7. I would feel anxious about being assessed in relation to my ability in (music).

8. I would feel anxious if I had to demonstrate the following to a Primary/Elementary class: (singing a song).

9. I would feel anxious if I had to demonstrate the following to my peers in a lecture: (singing a song).

10. I would feel anxious if I had to play a musical instrument for an audience.

11. I feel inadequate when I attend (music) lectures.

12. I would like to be (a musician).

13. I am not particularly interested in (music).

14. I consider myself to be (a musician).

15. If I worked hard, I could be good at (music).

-.65* Music Anxiety

Music Family

.92*

Music Self

Visual Art Anxiety

Visual Art Family .97*

Visual Art Self

Dance Anxiety

Dance Family .93*

Dance Self

Drama Anxiety

-.54*

Drama Family .80*

Drama Self