Rhonda G. Craven, Raymond L. Debus and Herbert W. Marsh

Aims

All self-concept, self-attribution and achievement scores were standardised as z-scores. To examine the effects of the experimental interventions on self-concept, self-attributions and achievement a series of multiple regression analyses were undertaken. To ensure the strongest possible statistical test was employed, a number of covariates were incorporated in the multiple regression analysis. All significant interaction effects were graphed to aid interpretation. Regions of significance were calculated using the Neyman-Johnson (1936) technique as described by Pedhazur and Schmelkin (1991) to interpret interaction effects based on comparisons of the experimental and control groups. The effects of the intervention were examined at time 2 (at the conclusion of the intervention) and time 3 (after a 4 month lapse in time after the conclusion of the intervention).

A series of multiple regression analyses were conducted to test for significant differences between the within-class control group and the external diffusion class control group at time 2 and time 3, with time 1 scores used as the covariate.

Comparison of self-concept scores of the within-class control group with the external diffusion control group at time 2 revealed main effects for group were present for school (t(769)=2.18, p < .05), general (t(769)=2.66, p < .01) and combined academic (t(769)=3.37, p < .01) self-concept. Participants in within-class control groups had higher self-concepts in school, general and combined academic self-concept at time 2.

Main effects for group were not present for reading and mathematics self-concept at time 2, however significant aptitude treatment interaction effects were present for mathematics (t(769)=-2.30, p < .05) and reading self-concept (t(769)=-2.73, p < .01) with prior levels of self-concept at time 2. These significant aptitude treatment interaction effects at time 2 suggest that diffusion effects in mathematics and reading self-concept for students in the within-class control group are greater for students with prior low self-concepts in these facets.

This diffusion effect implies that teachers assigned to experimental classes are perhaps not focusing the enhancement treatment on experimental subjects and the target facets of self-concept. Perhaps teachers are integrating the intervention techniques into their natural reinforcement repertoire.

Analyses based on teachers and researchers ratings of how well the intervention was focussed on target participants also showed that participants whose teacher had a low focus rating had generally higher self-concept and self-attribution scores at time 2 and time 3 compared to teachers who were rated with a high focus (see Craven, 1996).

The identification of a diffusion effect in relation to the within-class class control group is predicted to weaken tests of significance in that within-class control group seems to have been contaminated, resulting in significant gains in academic facets of self-concept in comparison to the external-class control group. Subsequent analyses were focused on experimental groups compared to a combination control group that consisted of both within-class controls and external-class controls and a non-corrupted control group consisting solely of the external-class control group.

Academic Self-Concept

All researcher-mediated interventions significantly enhanced self-concept facets targeted by each intervention. These significant effects were present at time 2 for all interventions except the researcher-mediated intervention in reading, where a significant effect on reading self-concept was present at time 3. Some support was found for the maintenance of gains on target self-concept facets over time which implies that the treatment may have longer term benefits.

Consistent with previous research, researcher-mediated interventions were more effective than teacher-mediated interventions and produced a more consistent pattern of results. Using a construct validity approach to the study of intervention effects, the specificity of the intervention was supported for the researcher-mediated intervention in mathematics and to a lesser extent the specificity of the combined researcher intervention was supported. The specificity of the researcher-mediated intervention in reading was not supported for comparisons based on the combined control group for students as a significant main effect was present for mathematics self-concept at time 3. However, the results of comparisons with the uncontaminated external diffusion control group offer support for the specificity of the intervention in only enhancing facets of self-concept related to the goals of the intervention (see Craven, 1996). These results offer further support for the effectiveness of these interventions.

Some support was also found for the presence of transfer effects to school or general self-concept for students with prior low self-concepts in these facets. Since none of the researcher-mediated treatments enhanced both school and general self-concept, and path coefficients for the direct effect of the intervention on general self-concept were positive but small, support is limited to influencing only one factor of the theorised hierarchy at a time and it cannot be demonstrated definitively that significant effects on general self-concept were mediated by school self-concept.

It was postulated that students in the combined researcher-mediated intervention would have higher combined academic self-concepts compared to students who had experienced one aspect of the intervention solely. It was found that participants in the combined researcher-mediated intervention with prior medium and high academic self-concepts relative to the range of self-concepts in the sample, had significantly higher academic self-concepts than participants in the single domain researcher-administered interventions. This result might be expected since the combined academic score included both mathematics and reading self-concept scores.

Teacher-administered intervention seemed far less effective than the researcher-administered intervention. In most self-concept facets where main effects were predicted only interaction effects were present.

The teacher-mediated intervention in mathematics by time 3 had significantly enhanced mathematics self-concept. Effects were also present for general self-concept scores at time 3 for some categories of students and hence support for transfer effects for at least some categories of students for one higher order hypothesised factor is present. The specificity of the intervention was also supported at time 3. This pattern of results demonstrates that teachers can enhance mathematics self-concept utilising the strategies employed in this investigation (see Craven, 1996).

Significant interaction effects were also present for the teacher-mediated treatment in reading. The teacher-mediated intervention in reading produced a significant effect on reading self-concept at time 2 for students with prior low or medium reading self-concepts. Some support was found for the maintenance of this effect across both occasions although effects were not significant at time 3. Transfer effects were present for general and school self-concept at time 2 for students with prior low and medium self-concepts in these areas. Whilst these effects seemed to have been maintained at time 3 for school self-concept, they were not maintained over time for general self-concept. The specificity of this intervention was not supported for some categories of students in that at time 2 significant interactions were present for the nontargeted academic facet of mathematics self-concept and the nontargeted variable of combined nonacademic self-concept. These results suggest that the effects of the intervention were not confined to targeted academic facets of self-concept for students with prior low and medium nonacademic and mathematics self-concepts. The results also suggest that teachers may have extended the application of the intervention to nontarget subject areas.

In contrast to the single domain teacher-mediated interventions the combined teacher-mediated intervention was not successful in enhancing targeted facets of self-concept. It is interesting to note that in contrast to the combined teacher-mediated intervention, the combined researcher-mediated intervention was successful in enhancing both mathematics and reading self-concept. Perhaps the complexity of the intervention caused difficulty in delivering a complex intervention in two subject domains.

The pattern of results for the teacher-mediated intervention in reading also suggests that students with prior low and medium self-concepts may benefit more from this treatment. Given that self-concept changes are greatest for this category of students and this category of students is of the most practical significance, these results offer support for the salience of the treatment in educational settings. The results also offer support for the multidimensionality of self-concept, the theory on which this model is based and some support is evident for the hierarchical model of self-concept proposed by Shavelson.

The researcher-mediated interventions were all successful in enhancing some aspects of self-attribution relevant to the goals of the intervention (see Craven, 1996). A number of interaction effects were present that suggested that students with prior low and medium self-attribution scores benefited most from the intervention. The interpretation of these interactions was consistent with the goals of the intervention.

Support was also found for some changes in self-attributions being maintained over time which indicates some durability of the effects of the treatment. Some support was found for the specificity of the single domain interventions in that the majority of significant results were present in the appropriate target area. The presence of significant results in nontarget areas, however, qualifies this evidence.

Significant results were more common for success attributions. Given that there was less opportunity to utilise enhancement strategies based on failure feedback due to the graded nature of tasks, perhaps it is reasonable to expect failure attributions to be less affected by the researcher-administered intervention. A number of expected effects for self-attribution were not present suggesting that the effects of the interventions on hypothesised self-attributions are weaker than initially expected.

Findings in relation to strategy attributions in success situations seem in general to be more frequently present in all three interventions than for other facets and more consistent for reading attributions than for other facets of self-attribution. These results suggest that strategy attributions may be more easily changed than other self-attribution domains.

The single domain researcher-mediated treatments were successful in enhancing success attributions to ability. The specificity of effects on this variable were supported for both groups. At time 3 a significant effect for reading attributions to ability in success situations was present for the combined researcher-mediated intervention though no significant effects were present for mathematics attributions to ability. These results suggest that all three researcher-administered interventions were successful at enhancing this category of attributions, though the effects for the combined intervention were weaker compared to the single domain treatments.

The pattern of results for success attributions to effort is similar for the pattern of results to ability though no significant effects were present for the combined researcher-mediated intervention. For reading attributions a significant effect was not present until time 3. The researcher-mediated intervention in mathematics increased mathematics attributions to effort in success situations at time 2 and these increases were maintained over time, thus supporting the strength of this intervention to enhance attributions associated with the goals of the intervention.

The pattern of results for the effects of the combined researcher-mediated intervention on self-attributions seemed weaker than for the single domain treatments. Perhaps attributional feedback is less credible when applied to more than one academic subject area in withdrawn assistance groups.

A similar pattern of results is present for self-concept and self-attribution findings for the researcher-mediated interventions that supports the goals of the intervention. The results therefore offer some support for effectiveness of the intervention to instigate changes in both constructs.

The self-attribution results for the teacher-mediated interventions contrast with the reasonably consistent pattern of results for the researcher-mediated interventions though some patterns of results suggest positive changes in self-attributions have taken place. Some aspects of both reading and mathematics self-attributions were changed by the combined teacher-mediated intervention.

The teacher-mediated intervention in reading was successful in changing four attributions relevant to the goals of the intervention. Since a number of expected significant effects are not present, strong support for the effectiveness of this treatment to change self-attributions is not evident. Based on comparisons with the combined control group, neither of the single domain teacher-mediated treatments was very successful in changing self-attributions relevant to the goals of the intervention. However, supplemental analyses undertaken with the external diffusion control group demonstrated some significant effects were present though the number of effects was less than for the parallel researcher-mediated interventions.

The results for the combined teacher-mediated intervention and the teacher-mediated intervention in reading also extend previous research in school settings in that the success of these treatments in changing at least some facets of self-attribution relevant to the goals of the intervention suggests that teachers can enhance some aspects of self-attributions over a relatively short period.

Achievement

The researcher-mediated intervention in mathematics enhanced mathematics achievement at time 2 and some support was found for these gains being maintained over time though gains at time 3 were not significant (see Craven, 1996). These results were associated with significant increases in effort/persistence on the mathematics test at both time 2 and time 3. No effects were present for reading achievement scores supporting the specificity of the intervention using a construct validity approach. Since significant effects were predicted not to be present until time 3, these significant results at time 2 demonstrate that the treatment principles may be stronger than previously thought for at least the researcher-mediated treatment in mathematics. The results also demonstrate that changes in mathematics self-concept, self-attribution, achievement and effort/persistence on a mathematics test are associated.

The researcher-mediated intervention in reading produced a significant change in reading achievement and effort on the ACAP test for comparisons of time 3 scores with time 2 scores. These results suggest that gains in reading achievement and effort on the ACAP test were made at time 3 in comparison to time 2 scores, however, these gains were not significant at time 3. A significant change was also present for some categories of students for TORCH achievement scores at time 2. Students with prior low and medium TORCH achievement scores had higher scores on this variable at time 2 compared to the diffusion control group. Support was also present to suggest that these gains may have been maintained over time, though gains were not significant at time 3. The nontarget outcomes of mathematics achievement and effort/persistence were unaffected by the intervention, supporting the specificity of the intervention effects. These results provide some support for the usefulness of the intervention to produce some gains in reading achievement and effort/persistence scores. Further, the results for ACAP reading vocabulary scores suggest that achievement in a less discrete subject area may be more difficult to change and it may be expected that significant effects for such less discrete areas may emerge after the lapse of the intervention.

The combined researcher-mediated intervention was not successful in enhancing mathematics achievement and most aspects of reading achievement. The combined researcher-mediated intervention was successful in enhancing time 3 TORCH effort/persistence scores in comparison to time 2 scores for students with high prior scores on this variable. The pattern of results for the combined researcher-administered intervention was considerably weaker than for the single domain treatments. Interestingly, these results are associated with weaker effects on self-concept and self-attribution which may suggest that academic achievement may not be significantly improved without associated changes in self-concept and self-attribution.

The teacher-mediated intervention in mathematics positively affected mathematics effort/persistence scores and support was found for these gains being maintained over time for at least students with prior medium to high scores on this variable. Whilst effort/persistence increased, there was no evidence of this increase being associated with significant gains in mathematics achievement. Hence this intervention was not successful in enhancing mathematics achievement. The specificity of the intervention was not supported as some significant effects were present for TORCH reading achievement and effort/persistence scores.

The effectiveness of the teacher-mediated intervention in reading was supported in enhancing ACAP effort/persistence at time 2 for students with low scores on this variable, and TORCH reading achievement and effort/persistence for students with prior low to medium scores on these variables. A significant aptitude treatment interaction effect was also present at time 3 for ACAP achievement scores for students with medium and high prior scores on this variable. Again the specificity of the intervention was not supported in that interaction effects were present for mathematics achievement at time 2 and the stability score for this variable which suggests that teachers may not have been focusing the intervention solely on the target subject area of reading.

The pattern of results for achievement for the combined teacher-mediated intervention compared to the diffusion control group extended the previous findings based on the pattern of results for the comparison to the combined control group. Significant interaction effects were present for TORCH effort/persistence scores at time 2 and TORCH achievement scores at time 3. These interactions suggested that students with prior low and medium scores had higher scores on these variables compared with the diffusion control group. A significant main effect was also present for ACAP effort/persistence scores at time 3 based on comparisons with the combined control group. No significant effects were present for mathematics achievement or effort/persistence scores, indicating that the combined teacher-mediated intervention was not successful in enhancing scores for these variables. The results however, do show that the combined teacher-mediated treatment was effective in enhancing some aspects of reading achievement and effort/persistence for at least some categories of students.

The pattern of results for the teacher-mediated intervention is less consistent than the pattern of results for the researcher-mediated interventions in terms of the specificity of intervention effects. If teachers focus on academic skills training in the classroom in addition to self-concept enhancement, it could be expected that gains in academic achievement would be greater for the teacher-administered interventions compared to the control group. Since positive effects are present but are not specific to the target areas, perhaps the teachers in this study were unable to limit the focus of the intervention to target subject areas.

Some support was found for teachers' ability to enhance academic achievement associated with the intervention. The single domain researcher-mediated interventions also demonstrated that changes in academic achievement were associated with changes in self-attribution and self-concept. These results imply that enhancing specific facets of self-concept and self-attribution can result in significant gains in achievement. Hence, self-concept enhancement should continue to be a major goal of schooling both as a desirable goal in itself and for the effects that such programs can have on other desirable educational outcomes.

The identification of a student diffusion effect in which teachers applied the intervention techniques beyond the intended target students, implies that teachers may readily be able to incorporate the self-concept enhancement techniques employed in this study into their natural reinforcement repertoire. Teachers may also adapt and apply the techniques to other nontarget facets of self-concept such that a diffusion effect across subject domains may be present. For example significant interaction effects were present for mathematics and nonacademic self-concept for the teacher-mediated intervention in reading, although these were nontarget facets.

The utilisation of the techniques employed in this study seems to produce a positive effect that diffuses the treatment benefits to nontarget students. For example, if a teacher chose to focus the intervention techniques on a specific group of students, the benefits of the intervention may be likely to some extent to diffuse to other students in the class. The results of this study suggest that teacher-mediated interventions aimed at selected students are more likely to benefit both target and nontarget students. Given that this study has demonstrated that changes in self-concept are associated with some desirable changes in self-attributions and academic achievement, it seems that enhancing self-concept should be considered a critical outcome of schooling systems that aim to maximise individual potential.

Students whose teachers were not trained in intervention techniques had lower scores on a number of facets of self-concept and self-attribution compared to the within-class control group. Although teachers of the diffusion control group were advised of their students' self-concept levels, the results of this study suggest that teachers who are not trained in intervention techniques are not as able as teachers who have experienced training to enhance children's self-concepts or do not recognise the importance of doing so as a critical goal of schooling. Currently there are still too many teachers, in too many schools who over rely on general praise strategies that target general self-concept. Given the history of self-concept research, it is not surprising that common reinforcement strategies are based on outdated unidimensional models of self-concept. What is surprising is that education publishers still publish material that encourages teachers to utilise general 'feel good' reinforcement strategies that solely focus on general self-concept although the research literature has demonstrated that this approach is flawed. Consideration needs to be given to designing inservice and preservice teacher education courses to educate teachers about the structure, measurement and enhancement of self-concept. Such courses should include an appreciation of the multidimensionality of self-concept, the relationship of self-concept to other constructs, the need to focus enhancement programs on specific facets of self-concept and enhancing teachers' skills in administering self-concept enhancement strategies.

The present investigation demonstrated that interventions that employ attributional feedback and internally focused feedback can change some aspects of self-concept, self-attributions and academic achievement related to the goals of the interventions. These results support the further use of the treatments employed in this study in programs designed to enhance self-concept. Designing interventions to be administered in naturalistic settings is a desirable goal since this is the target setting where interventions have most direct practical significance. Given that Hattie (1992) has found that teachers are least likely to enhance self-concept compared to other instructors, thorough training methods may need to be instigated to ensure teachers comprehend how to implement a self-concept intervention and recognise the value in doing so.

The study demonstrates that self-concept and its relation to other constructs cannot be adequately understood without attention to the multidimensional structure of the construct. The results of some aspects of this study also support a multidimensional theoretical position that suggests self-concept, self-attribution and achievement are interrelated. They attest to the need for schools to not only train students to think but ensure they have adaptive motivations to do this. Such motivations include appropriate attributions which foster persistence on challenging tasks and a sense of personal control and maximal self-concept in specific facets.

Australian Council for Educational Research (1991). TORCH Manual: Tests of reading comprehension. Victoria: Australian Council for Educational Research.

Australian Council for Educational Research. (1984). Teachers handbook maths; Progressive achievement tests in mathematics. Victoria: Australian Council for Educational Research.

Byrne, B. (1996). Measuring self-concept across the life span: Issues and instrumentation. American Psychological Association. Washington, DC.

Commonwealth of Australia. (1979). Australian Co-operative Assessment Project (ACAP), Test of General Reading Vocabulary. Canberra: Commonwealth of Australia.

Craven, R.G. (1989). An examination of self-concept: The interrelationship of teachers', parents and children's perceptions of self-concept, and their influence in enhancing self-concept. Unpublished B.A. Honours Thesis, University of Sydney, Australia.

Craven, R. G. (1996). Enhancing Academic Self-Concept: A Large-Scale Longitudinal Study in an Educational Setting. Doctoral thesis submitted to the University of Sydney. USA:UMI.

Craven, R.G., Marsh, H. W., and Debus, R. L. (1991). Effects of internally focused feedback and attributional feedback on the enhancement of academic self-concept. Journal of Educational Psychology, 83, 17-27.

Hattie, Janet. (1986). Enhancing Self-Concept. Unpublished Master of Education Thesis, University of New England, Australia.

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

Marsh, H. W. (1984). Relationships among dimensions of self-attribution, dimensions of self-concept, and academic achievements. Journal of Educational Psychology, 76, 1291-1380.

Marsh, H. W. (1990a). Causal ordering of academic self-concept and academic achievement: A multiwave, longitudinal panel analysis. Journal of Educational Psychology, 82, 646-656.

Marsh, H. W. (1990b). The Self Description Questionnaire-I: SDQ-I manual. Sydney: University of Western Sydney.

Marsh, H. W. (1990c). The structure of academic self-concept: The Marsh/Shavelson Model. Journal of Educational Psychology, 82, 623-636.

Marsh, H. W. (1992a). The Self Description Questionnaire-II: SDQ-II manual. Sydney: University of Western Sydney.

Marsh, H. W. (1992b). The Self Description Questionnaire-III: SDQ-III manual. Sydney: University of Western Sydney.

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., Byrne, B. M. , and Shavelson, R. (1988). A multifaceted academic self-concept: Its hierarchical structure and its relation to academic achievement. Journal of Educational Psychology, 80, 366-380.

Marsh, H. W. and Craven, R.G. (1997). Academic self-concept: Beyond the dustbowl. In G. Phye (Ed.) Handbook of classroom assessment: Learning, achievement and adjustment. US: Academic Press.

Marsh, H. W., and Richards, G. (1988). The Outward Bound Bridging Course for low achieving high-school males: Effect on academic achievement and multidimensional self-concepts. Australian Journal of Psychology, 40, 281-298.

Marsh, H. W., and Shavelson, R. J. (1985). Self-concept: Its multifaceted, hierarchical structure. Educational Psychologist, 20, 107-125.

Marsh, H. W., Cairns, L., Relich, J., Barnes, J., and Debus, R. L. (1984). The relationship between dimensions of self-attribution and dimensions of self-concept. Journal of Educational Psychology, 76, 3-32.

Mossenson, L., Hill, P., and Masters, G. (1987). TORCH: Tests of reading comprehension manual. Victoria: Australian Council for Educational Research.

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

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

Shavelson, R. L., and Marsh, H. W. (1986). On the structure of self-concept. In R. Schwarzer (Ed.), Anxiety and cognitions. NJ.: Earlbaum.